HK1229003B - Pancreatic cancer biomarkers and uses thereof - Google Patents

Description

Pancreatic cancer biomarkers and their uses

This application is a divisional application of Chinese patent application 201180049408.2, filed on August 12, 2011, entitled “Pancreatic Cancer Biomarkers and Uses Thereof”.

Related applications

This application claims the benefit of U.S. Provisional Application Serial No. 61/373,687, filed August 13, 2010, U.S. Provisional Application Serial No. 61/418,689, filed December 1, 2010, U.S. Provisional Application Serial No. 61/482,347, filed May 4, 2011, and U.S. Provisional Application Serial No. 61/482,480, filed May 4, 2011, each of which is incorporated herein by reference in its entirety.

Field of the Invention

The present application relates generally to the detection of biomarkers and the diagnosis of cancer in an individual, and more specifically to one or more biomarkers, methods, devices, reagents, systems, and kits for diagnosing cancer, more particularly pancreatic cancer, in an individual.

Background Art

The following description provides a summary of information relevant to the present application and is not an admission that any of the information provided or publications cited herein are prior art to the present application.

Pancreatic cancer is the fourth leading cause of cancer-related death in the United States. Although the 5-year survival rate is only 5%, it has been shown to increase with early surgical intervention: in the 20% of individuals who are suitable for "radical" resection, the survival rate increases to 15-20%. At the time of diagnosis, more than half of patients have distant disease, while another 25% have regional spread. This is because the disease is notoriously difficult to diagnose in its early stages. Approximately 20% of patients with "operable" disease (stage IIb or less) undergo "radical" resection, and the 5-year survival rate increases from less than 5% to 15-20%.

Pancreatic cancer can arise from both the exocrine and endocrine parts of the pancreas. Of pancreatic tumors, 95% develop from the exocrine part of the pancreas, which includes ductal epithelium, acinar cells, connective tissue, and lymphoid tissue. Approximately 75% of all pancreatic cancers occur in the head or neck of the pancreas, 15-20% occur in the body, and 5-10% occur in the tail.

Recurrence can be local (same place or near its beginning) or distal (diffusion to organs such as liver, lung or bone).When exocrine pancreatic cancer recurs, it is mainly treated in the same manner as metastatic cancer, and if the patient can tolerate, may include chemotherapy.Usually, pancreatic cancer first metastasizes to regional lymph nodes, then to liver, and less commonly, to lung. It can also directly invade surrounding visceral organs, such as duodenum, stomach and colon, or be transferred to any surface in the abdominal cavity by peritoneal diffusion. Ascites can be caused, and this has a poor prognosis. Pancreatic cancer can spread to the skin as a painful nodule metastasis. Pancreatic cancer rarely metastasizes to bone.

Two clinical applications of blood-based pancreatic cancer testing are for preclinical diagnosis in asymptomatic high-risk populations and for differential diagnosis in symptomatic populations. The clinical uses for these two indications are listed below.

Screening in asymptomatic high-risk groups: In 2010, there were an estimated 43,140 new cases of pancreatic cancer and 36,800 deaths in the United States. Genetics, family history, chronic pancreatitis, smoking, and heavy alcohol consumption increase the risk of pancreatic cancer, as does cystic fibrosis. Increased risk is reported for:

Smoking: <25/day is 2x risk, >25/day is 3x risk

Alcohol: More than 3 drinks/day resulted in a 1.6-fold increased risk

Family history: First degree association with the disease yields a 5x increase

Adults with cystic fibrosis: 31x risk

BRCA2 gene mutation 10x risk

In the absence of an effective screening paradigm, cancer is simply detected in asymptomatic but at-risk individuals when symptoms appear. This can be too late. The availability of early detection tests would increase the proportion of patients suitable for radical surgery. Currently, the cure rate for 20% of individuals with early detection is only 4% of the overall population. If the eligibility for radical surgery were increased from the current 20% in the asymptomatic population through early detection, the total number of patients who can be cured would increase, and the number of lives saved each year would also increase. Because pancreatic cancer is a low-prevalence disease, high specificity is an important attribute of a screening test, even in this high-risk population. A low false-positive rate is crucial for reducing the costs of unnecessary follow-up procedures and alleviating patient anxiety.

Differential diagnosis in symptomatic patients. Pancreatic cancer can be difficult to distinguish from benign conditions such as pancreatitis or gastrointestinal disorders. The differential diagnosis of primary exocrine pancreatic cancer includes chronic pancreatitis, pancreatic endocrine tumors, autoimmune pancreatitis, lymphoma, and various other rare conditions. Common but nonspecific symptoms associated with pancreatic cancer include:

Abdominal pain – especially if it radiates to the back

Obstructive jaundice

Sudden unexplained diabetes

weight loss

Anorexia, fatigue

Nausea and vomiting

Acute or chronic pancreatitis

The table below shows the number of patients who presented to emergency rooms and hospitals with at least two of these related symptoms; the first symptom is any one of the symptoms listed, and the second symptom is a symptom listed in the table. The emergency room data is from: (http://hcupnet.ahrq.gov/), and the outpatient data is from the CDC 2008 National Ambulatory Medical Care Survey 2006 (number 8).

Sensitive detection of resectable disease is very important for the clinical application of this indication. Rapid detection of pancreatic cancer increases the chances of diagnosing a curable disease. Diagnosis of pancreatic cancer is usually performed by radiographic imaging of a mass found in the pancreas, which often obstructs the pancreatic duct or bile duct. However, imaging can be invasive and expensive. Determining which patients require follow-up blood tests, including diagnostic imaging, would benefit the patients and simplify diagnosis.

Biomarker selection for a specific disease state involves first identifying markers that have measurable and statistically significant differences in the disease population compared to the control population for a specific medical application. Biomarkers can include secreted or shed molecules that parallel disease development or progression and readily diffuse into the bloodstream from pancreatic cancer tissue or from surrounding tissues and circulating cells in response to the tumor. The identified biomarker or set of biomarkers is typically clinically validated or confirmed to be a reliable indicator for the original intended use for which it was selected. Biomarkers can include small molecules, peptides, proteins, and nucleic acids. Some key issues affecting biomarker identification include over-fitting of the available data and data bias.

Various methods have been used to attempt to identify biomarkers and diagnose diseases. For protein-based markers, these methods include two-dimensional electrophoresis, mass spectrometry, and immunoassays. For nucleic acid markers, these methods include mRNA expression profiling, microRNA profiling, FISH, serial analysis of gene expression (SAGE), and large-scale gene expression arrays.

The application of two-dimensional electrophoresis is limited by the following problems: low detection sensitivity; problems related to protein solubility, charge and hydrophobicity; gel reproducibility; and the possibility that a single spot represents multiple proteins. For mass spectrometry, depending on the format used, limitations arise around sample processing and separation, sensitivity to low-abundance proteins, signal-to-noise considerations, and the inability to immediately identify the detected protein. The limitation of immunoassay methods for biomarker discovery centers on the inability of antibody-based multiplexed assays to measure a large number of analytes. Arrays of high-quality antibodies can be easily printed and analytes bound to these antibodies can be measured without the need for sandwiches. (This would be the equivalent of using whole genome nucleic acid sequences to measure all DNA or RNA sequences in an organism or cell by hybridization. Hybridization experiments are possible because hybridization can be a stringent test of identity. Even very good antibodies are not stringent enough in the selection of their binding partners to work in a blood or even cell extract environment because the protein ensembles in those matrices have extremely different abundances.) Therefore, a different immunoassay-based approach must be used to discover biomarkers - multiplexed ELISA assays (i.e., sandwiches) are needed to obtain sufficient stringency to measure many analytes simultaneously and thus determine which analytes are indeed biomarkers. Sandwich immunoassays do not scale up to high levels, so biomarkers cannot be discovered with stringent sandwich immunoassays using standard array formats. Finally, antibody reagents are subject to considerable batch variability and reagent instability. The protein biomarker discovery platform of the present invention overcomes this problem.

Many of these methods rely on or require the classification of some type of sample before analysis. Therefore, the sample preparation required for conducting sufficiently effective studies designed to identify and discover statistically relevant biomarkers in a series of well-defined sample populations is extremely difficult, expensive, and time-consuming. During the classification period, a wide range of variability can be introduced into the various samples. For example, a potential marker may be unstable for the method, the concentration of the marker may vary, inappropriate aggregation or disaggregation may occur, and inadvertent sample contamination may occur, thereby masking subtle changes in expected early-stage disease.

It is widely accepted that biomarker discovery and detection methods using these technologies have serious limitations for identifying diagnostic biomarkers. These limitations include the inability to detect low-abundance biomarkers, the inability to consistently cover the full dynamic range of the proteome, irreproducibility in sample processing and fractionation, and overall irreproducibility and lack of robustness of the methods. In addition, these studies have introduced biases in the data and have not adequately addressed the complexity of the sample populations, including appropriate controls, in terms of the distribution and randomization required to identify and validate biomarkers within target disease populations.

Although efforts to discover new and effective biomarkers have been underway for decades, these efforts have largely been unsuccessful. Biomarkers for various diseases are typically identified in academic laboratories, often discovered by chance while conducting basic research on some disease process. Based on these discoveries and a small amount of clinical data, published papers suggest the identification of new biomarkers. However, most of these proposed biomarkers have not been proven to be true or useful biomarkers, primarily because the small amount of clinical samples tested provides only weak statistical evidence for the discovery of effective biomarkers. In other words, the initial identification was not rigorous with respect to the basic elements of statistics. In each year from 1994 to 2003, a search of the scientific literature revealed that thousands of references to biomarkers were published. However, during the same period, the FDA approved a maximum of three new protein biomarkers for diagnostic use per year, and no new protein biomarkers were approved in several years.

Based on the history of failed biomarker discovery efforts, mathematical theories have been proposed to further the general understanding that biomarkers for disease are few and difficult to find. Biomarker studies based on 2D gels or mass spectrometry support these views. Very few useful biomarkers have been identified by these methods. However, 2D gels and mass spectrometry are generally overlooked for measuring proteins present in blood at concentrations of approximately 1 nM or higher, the population of which is likely the least likely to vary with disease. With the exception of the biomarker discovery platform of the present invention, no proteomic biomarker discovery platform exists that can accurately measure protein expression levels at much lower concentrations.

Many biochemical pathways are known about the complexities of human biology. Many biochemical pathways culminate or begin with secreted proteins that act locally within the pathology, such as secreting growth factors to stimulate replication of other cells in the pathology, and secreting other factors to evade the immune system. While many of these secreted proteins act in a paracrine manner, some operate at distal locations in the body. Those skilled in the art with a basic understanding of biochemical pathways will understand that many pathology-specific proteins should be present in the blood at concentrations below (or even far below) the detection limits of 2D gels and mass spectrometry. Prior to the identification of this relatively abundant number of disease biomarkers, a proteomic platform must be available that can analyze proteins at concentrations below those detectable by 2D gels or mass spectrometry.

Therefore, there is a need for biomarkers, methods, devices, reagents, systems, and kits that allow for (a) differentiation of pancreatic cancer from benign disease states; (b) screening of asymptomatic individuals at high risk for pancreatic cancer; (c) detection of pancreatic cancer biomarkers; and (d) diagnosis of pancreatic cancer.

SUMMARY OF THE INVENTION

This application includes biomarkers, methods, reagents, devices, systems, and kits for detecting and diagnosing cancer, more particularly pancreatic cancer. The biomarkers of this application were identified using the multiplex aptamer-based assay detailed in Example 1. By using the aptamer-based biomarker identification methods described herein, this application describes a surprisingly large number of pancreatic cancer biomarkers that can be used to detect and diagnose pancreatic cancer, as well as a large number of cancer biomarkers that can be used to detect and diagnose more general cancers. In identifying these biomarkers, over 800 proteins were measured from hundreds of individual samples, some of which had concentrations in the low femtomolar range. This is approximately four orders of magnitude lower than biomarker discovery experiments performed using 2D gels and/or mass spectrometry.

While some of the pancreatic cancer biomarkers described herein can be used individually to detect and diagnose pancreatic cancer, the methods described herein are used to group multiple subsets of pancreatic cancer biomarkers for use as a panel of biomarkers. Once an individual biomarker or subset of biomarkers has been identified, detection or diagnosis of pancreatic cancer in an individual can be accomplished using any assay platform or format capable of measuring differences in the levels of the selected biomarker or biomarkers in a biological sample.

However, the identification of the pancreatic cancer biomarkers disclosed herein was only possible by using the aptamer-based biomarker identification method described herein, in which over 800 individual potential biomarker values were individually screened from a large number of individuals who had previously been diagnosed with or without pancreatic cancer. This discovery approach is in stark contrast to biomarker discovery from conditioned media or lysed cells because it interrogates a more patient-relevant system that does not require translation to human pathology.

Thus, in one aspect, the present application provides one or more biomarkers for use alone or in various combinations to diagnose pancreatic cancer, or to allow for the differential diagnosis of pancreatic cancer from benign gastrointestinal (GI) conditions such as acute or chronic pancreatitis (or both), pancreatic obstruction, GERD, gallstones, or abnormal imaging that is later found to be benign. Exemplary embodiments include the biomarkers provided in column 2 of Table 1, which, as described above, were identified using the multiplex aptamer-based assays generally described in Example 1 and more specifically described in Example 2. The markers provided in Table 1 can be used to diagnose pancreatic cancer in a high-risk asymptomatic population and to differentiate acute or chronic pancreatitis (or both), pancreatic obstruction, GERD, gallstones, or abnormal imaging that is later found to be benign from pancreatic cancer.

While some of the pancreatic cancer biomarkers described herein can be used individually to detect and diagnose pancreatic cancer, the methods described herein can also be used to group multiple subsets of pancreatic cancer biomarkers, each of which can be used as a panel of two or more biomarkers. Thus, various embodiments of the present application provide a combination comprising N biomarkers, where N is at least 2 biomarkers. In other embodiments, N is selected from any number between 2 and 65 biomarkers.

In still other embodiments, N is selected from any number between 2 and 7, 2 and 10, 2 and 15, 2 and 20, 2 and 25, 2 and 30, 2 and 35, 2 and 40, 2 and 45, 2 and 50, 2 and 55, or 2 and 65. In other embodiments, N is selected from any number between 3 and 7, 3 and 10, 3 and 15, 3 and 20, 3 and 25, 3 and 30, 3 and 35, 3 and 40, 3 and 45, 3 and 50, 3 and 55, or 3 and 65. In other embodiments, N is selected from any number between 4 and 7, 4 and 10, 4 and 15, 4 and 20, 4 and 25, 2 and 30, 2 and 35, 2 and 40, 4 and 45, 4 and 50, 4 and 55, or 4 and 65. In other embodiments, N is selected from any number between 5 and 7, 5 and 10, 5 and 15, 5 and 20, 5 and 25, 5 and 30, 5 and 35, 5 and 40, 5 and 45, 5 and 50, 5 and 55, or 5 and 65. In other embodiments, N is selected from any number between 6 and 10, 6 and 15, 6 and 20, 6 and 25, 6 and 30, 6 and 35, 6 and 40, 6 and 45, 6 and 50, 6 and 55, or 6 and 65. In other embodiments, N is selected from any number between 7 and 10, 7 and 15, 7 and 20, 7 and 25, 7 and 30, 7 and 35, 7 and 40, 7 and 45, 7 and 50, 7 and 55, or 7 and 65. In other embodiments, N is selected from any number between 8 and 10, 8 and 15, 8 and 20, 8 and 25, 8 and 30, 8 and 35, 8 and 40, 8 and 45, 8 and 50, 8 and 55, or 8 and 65. In other embodiments, N is selected from any number between 9 and 15, 9 and 20, 9 and 25, 9 and 30, 9 and 35, 9 and 40, 9 and 45, 9 and 50, 9 and 55, or 9 and 65. In other embodiments, N is selected from any number between 10 and 15, 10 and 20, 10 and 25, 10 and 30, 10 and 35, 10 and 40, 10 and 45, 10 and 50, 10 and 55, or 10 and 65. It should be understood that N can be selected from ranges containing similar but higher order ranges.

In another aspect, the present invention provides a method of diagnosing pancreatic cancer in an individual, the method comprising detecting in a biological sample from the individual at least one biomarker value, the at least one biomarker value corresponding to at least one biomarker selected from the group of biomarkers provided in Table 1, Col. 2, wherein the individual is classified as having pancreatic cancer based on the at least one biomarker value.

In another aspect, the present invention provides a method of diagnosing pancreatic cancer in an individual, the method comprising detecting, in a biological sample from the individual, biomarker values, each of the biomarker values corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 1, Col. 2, wherein a likelihood that the individual has pancreatic cancer is determined based on the biomarker values.

In another aspect, the present invention provides a method of diagnosing pancreatic cancer in an individual, the method comprising detecting, in a biological sample from the individual, biomarker values, each of the biomarker values corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 1, Col. 2, wherein the individual is classified as having pancreatic cancer based on the biomarker values, and wherein N = 2-10.

In another aspect, the present invention provides a method of diagnosing pancreatic cancer in an individual, the method comprising detecting, in a biological sample from the individual, biomarker values, each of the biomarker values corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 1, Col. 2, wherein a likelihood that the individual has pancreatic cancer is determined based on the biomarker values, and wherein N = 2-10.

In another aspect, the present invention provides a method of diagnosing a person who does not have pancreatic cancer, the method comprising detecting in a biological sample from the person at least one biomarker value, the at least one biomarker value corresponding to at least one biomarker selected from the group of biomarkers listed in Table 1, Col. 2, wherein the person is classified as not having pancreatic cancer based on the at least one biomarker value.

In another aspect, the present invention provides a method of diagnosing that an individual does not have pancreatic cancer, the method comprising detecting, in a biological sample from the individual, biomarker values, each of the biomarker values corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 1, Col. 2, wherein the individual is classified as not having pancreatic cancer based on the biomarker values, and wherein N = 2-10.

In another aspect, the present invention provides a method of diagnosing pancreatic cancer, the method comprising detecting, in a biological sample from an individual, biomarker values, each of the biomarker values corresponding to a biomarker from a set of N biomarkers, wherein the biomarkers are selected from the group of biomarkers listed in Table 1, Col. 2, wherein a classification of the biomarker values indicates that the individual has pancreatic cancer, and wherein N = 3-10.

In another aspect, the present invention provides a method of diagnosing pancreatic cancer, the method comprising detecting, in a biological sample from an individual, biomarker values, each of the biomarker values corresponding to a biomarker from a set of N biomarkers, wherein the biomarkers are selected from the group of biomarkers listed in Table 1, Col. 2, wherein a classification of the biomarker values indicates that the individual has pancreatic cancer, and wherein N = 3-10.

In another aspect, the present invention provides a method of diagnosing pancreatic cancer, the method comprising detecting, in a biological sample from an individual, biomarker values, each of the biomarker values corresponding to a biomarker in a panel of biomarkers selected from the group of biomarkers listed in Tables 2-11, wherein a classification of the biomarker values indicates that the individual has pancreatic cancer.

In another aspect, the present invention provides a method of diagnosing the absence of pancreatic cancer, the method comprising detecting, in a biological sample from an individual, biomarker values, each of the biomarker values corresponding to a biomarker from a set of N biomarkers, wherein the biomarkers are selected from the group of biomarkers listed in Table 1, Col. 2, wherein a classification of the biomarker value indicates the absence of pancreatic cancer in the individual, and wherein N = 3-10.

In another aspect, the present invention provides a method of diagnosing the absence of pancreatic cancer, the method comprising detecting, in a biological sample from an individual, biomarker values, each of the biomarker values corresponding to a biomarker from a set of N biomarkers, wherein the biomarkers are selected from the group of biomarkers listed in Table 1, Col. 2, wherein a classification of the biomarker value indicates the absence of pancreatic cancer in the individual, and wherein N = 3-10.

In another aspect, the present invention provides a method of diagnosing the absence of pancreatic cancer, the method comprising detecting, in a biological sample from an individual, biomarker values, each of the biomarker values corresponding to a biomarker in a panel of biomarkers selected from the group of biomarkers provided in Tables 2-11, wherein a classification of the biomarker values indicates the absence of pancreatic cancer in the individual.

In another aspect, the present invention provides a method of diagnosing pancreatic cancer in an individual, the method comprising detecting, in a biological sample from the individual, a biomarker value corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 1, Col. 2, wherein the individual is classified as having pancreatic cancer based on a classification score that deviates from a predetermined threshold, and wherein N = 2-10.

In another aspect, the present invention provides a method of diagnosing the absence of pancreatic cancer in an individual, the method comprising detecting, in a biological sample from the individual, a biomarker value corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 1, Col. 2, wherein the individual is classified as not having pancreatic cancer based on a classification score that deviates from a predetermined threshold, and wherein N = 2-10.

In another aspect, the present invention provides a computer-implemented method for indicating a likelihood of pancreatic cancer. The method comprises: retrieving, on a computer, biomarker information for an individual, wherein the biomarker information comprises biomarker values, each of the biomarker values corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 1, Col. 2, where N is as defined above; performing, on a computer, a classification on each of the biomarker values; and indicating a likelihood that the individual has pancreatic cancer based on a plurality of classifications.

In another aspect, the present invention provides a computer-implemented method for classifying an individual as having or not having pancreatic cancer. The method comprises: retrieving, on a computer, biomarker information for the individual, wherein the biomarker information comprises biomarker values, each of the biomarker values corresponding to one of at least N biomarkers selected from the group of biomarkers provided in Table 1, Col. 2; performing, on a computer, a classification on each of the biomarker values; and indicating whether the individual has pancreatic cancer based on a plurality of classifications.

In another aspect, the present invention provides a computer program product that indicates the likelihood of pancreatic cancer. The computer program product includes a computer-readable medium containing program code, the program code executable by a processor of a computing device or system, the program code comprising: code for retrieving data attributed to a biological sample from an individual, wherein the data includes biomarker values, each of the biomarker values corresponding to one of at least N biomarkers in the biological sample selected from the group of biomarkers listed in Table 1, Col. 2, where N is as defined above; and code for executing a classification method that indicates the likelihood that the individual has pancreatic cancer as a function of the biomarker values.

In another aspect, the present invention provides a computer program product for indicating pancreatic cancer status in an individual. The computer program product includes a computer-readable medium containing program code, the program code executable by a processor of a computing device or system, the program code comprising: code for retrieving data attributed to a biological sample from an individual, wherein the data includes biomarker values, each of the biomarker values corresponding to one of at least N biomarkers in the biological sample selected from the group of biomarkers provided in Table 1, Col. 2; and code for executing a classification method that indicates pancreatic cancer status in the individual as a function of the biomarker values.

In another aspect, the present invention provides a computer-implemented method for indicating a likelihood of pancreatic cancer. The method comprises: retrieving, on a computer, biomarker information for an individual, wherein the biomarker information comprises a biomarker value corresponding to a biomarker selected from the group of biomarkers listed in Table 1, Col. 2; classifying, on a computer, the biomarker value; and indicating a likelihood that the individual has pancreatic cancer based on the classification.

In another aspect, the present invention provides a computer-implemented method for classifying an individual as having or not having pancreatic cancer. The method comprises: retrieving biomarker information for the individual from a computer, wherein the biomarker information comprises a biomarker value corresponding to a biomarker selected from the group of biomarkers provided in Table 1, Col. 2; classifying the biomarker value using the computer; and indicating whether the individual has pancreatic cancer based on the classification.

In another aspect, the present invention provides a computer program product that indicates the likelihood of pancreatic cancer. The computer program product includes a computer-readable medium containing program code, the program code executable by a processor of a computing device or system, the program code comprising: code for retrieving data attributed to a biological sample from an individual, wherein the data includes biomarker values corresponding to biomarkers in the biological sample selected from the group of biomarkers listed in Table 1, Col. 2; and code for executing a classification method that indicates the likelihood that the individual has pancreatic cancer as a function of the biomarker values.

In another aspect, the present invention provides a computer program product for indicating pancreatic cancer status in an individual. The computer program product includes a computer-readable medium containing program code, the program code executable by a processor of a computing device or system, the program code comprising: code for retrieving data attributed to a biological sample from an individual, wherein the data includes biomarker values corresponding to biomarkers in the biological sample selected from the group of biomarkers provided in Table 1, Col. 2; and code for executing a classification method that indicates pancreatic cancer status in the individual as a function of the biomarker values.

While some of the described cancer biomarkers can be used individually to detect and diagnose cancer, the methods described herein are used to group multiple subsets of cancer biomarkers for use as a panel of biomarkers. Once an individual biomarker or subset of biomarkers has been identified, detection or diagnosis of cancer in an individual can be accomplished using any assay platform or format capable of measuring differences in the levels of the selected biomarker or biomarkers in a biological sample.

However, the identification of the cancer biomarkers disclosed herein was only possible by using the aptamer-based biomarker identification method described herein, in which over 800 individual potential biomarker values were individually screened from a large number of individuals who had been previously diagnosed with or without cancer. This discovery approach is in stark contrast to biomarker discovery from conditioned media or lysed cells because it interrogates a more patient-relevant system that does not require translation to human pathology.

Thus, in one aspect of the present invention, one or more biomarkers are provided for use alone or in various combinations to diagnose cancer. Exemplary embodiments include the biomarkers provided in Table 19, which were identified using the multiplex aptamer-based assays generally described in Example 1 and more specifically described in Example 7. The markers provided in Table 19 can be used to distinguish individuals with cancer from individuals without cancer.

While some of the described cancer biomarkers can be used individually to detect and diagnose cancer, the methods described herein are also useful for grouping multiple subsets of cancer biomarkers, each of which can be used as a panel of three or more biomarkers. Thus, various embodiments of the present application provide a combination comprising N biomarkers, where N is at least three biomarkers. In other embodiments, N is selected from any number between 3 and 65 biomarkers.

In other embodiments, N is selected from any number between 3 and 7, 3 and 10, 3 and 15, 3 and 20, 3 and 25, 3 and 30, 3 and 35, 3 and 40, 3 and 45, 3 and 50, 3 and 55, 3 and 60, or 3 and 65. In other embodiments, N is selected from any number between 4 and 7, 4 and 10, 4 and 15, 4 and 20, 4 and 25, 4 and 30, 4 and 35, 4 and 40, 4 and 45, 4 and 50, 4 and 55, 4 and 60, or 4 and 65. In other embodiments, N is selected from any number between 5 and 7, 5 and 10, 5 and 15, 5 and 20, 5 and 25, 5 and 30, 5 and 35, 5 and 40, 5 and 45, 5 and 50, 5 and 55, 5 and 60, or 5 and 65. In other embodiments, N is selected from any number between 6 and 10, 6 and 15, 6 and 20, 6 and 25, 6 and 30, 6 and 35, 6 and 40, 6 and 45, 6 and 50, 6 and 55, 6 and 60, or 6 and 65. In other embodiments, N is selected from any number between 7 and 10, 7 and 15, 7 and 20, 7 and 25, 7 and 30, 7 and 35, 7 and 40, 7 and 45, 7 and 50, 7 and 55, 7 and 60, or 7 and 65. In other embodiments, N is selected from any number between 8 and 10, 8 and 15, 8 and 20, 8 and 25, 8 and 30, 8 and 35, 8 and 40, 8 and 45, 8 and 50, 8 and 55, 8 and 60, or 8 and 65. In other embodiments, N is selected from any number between 9 and 15, 9 and 20, 9 and 25, 9 and 30, 9 and 35, 9 and 40, 9 and 45, 9 and 50, 9 and 55, 9 and 60, or 9 and 65. In other embodiments, N is selected from any number between 10 and 15, 10 and 20, 10 and 25, 10 and 30, 10 and 35, 10 and 40, 10 and 45, 10 and 50, 10 and 55, 10 and 60, or 10 and 65. It should be understood that N can be selected from a range that includes similar but higher order ranges.

In another aspect, the present invention provides a method of diagnosing cancer in an individual, the method comprising detecting in a biological sample from the individual at least one biomarker value, the at least one biomarker value corresponding to at least one biomarker selected from the group of biomarkers provided in Table 19, wherein the individual is classified as having cancer based on the at least one biomarker value.

In another aspect, the present invention provides a method of diagnosing cancer in an individual, the method comprising detecting, in a biological sample from the individual, biomarker values, each of the biomarker values corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 19, wherein a likelihood that the individual has cancer is determined based on the biomarker values.

In another aspect, the present invention provides a method of diagnosing cancer in an individual, the method comprising detecting, in a biological sample from the individual, biomarker values, each of the biomarker values corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 19, wherein the individual is classified as having cancer based on the biomarker values, and wherein N=3-10.

In another aspect, the present invention provides a method of diagnosing cancer in an individual, the method comprising detecting, in a biological sample from the individual, biomarker values, each of the biomarker values corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 19, wherein a likelihood that the individual has cancer is determined based on the biomarker values, and wherein N=3-10.

In another aspect, the present invention provides a method of diagnosing that an individual does not have cancer, the method comprising detecting in a biological sample from the individual at least one biomarker value, the at least one biomarker value corresponding to at least one biomarker selected from the group of biomarkers listed in Table 19, wherein the individual is classified as not having cancer based on the at least one biomarker value.

In another aspect, the present invention provides a method of diagnosing that an individual does not have cancer, the method comprising detecting, in a biological sample from the individual, biomarker values, each of the biomarker values corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 19, wherein the individual is classified as not having cancer based on the biomarker values, and wherein N=3-10.

In another aspect, the present invention provides a method of diagnosing cancer, the method comprising detecting, in a biological sample from an individual, biomarker values, each of the biomarker values corresponding to a biomarker from a set of N biomarkers, wherein the biomarkers are selected from the group of biomarkers listed in Table 19, wherein a classification of the biomarker value indicates that the individual has cancer, and wherein N=3-10.

In another aspect, the present invention provides a method of diagnosing cancer, the method comprising detecting, in a biological sample from an individual, biomarker values, each of the biomarker values corresponding to a biomarker from a set of N biomarkers, wherein the biomarkers are selected from the group of biomarkers listed in Table 19, wherein a classification of the biomarker value indicates that the individual has cancer, and wherein N=3-10.

In another aspect, the present invention provides a method of diagnosing cancer, the method comprising detecting, in a biological sample from an individual, biomarker values, each of the biomarker values corresponding to a biomarker in a panel of biomarkers selected from the group of biomarkers listed in Tables 20-29, wherein a classification of the biomarker values indicates that the individual has cancer.

In another aspect, the present invention provides a method of diagnosing the absence of cancer, the method comprising detecting in a biological sample from an individual biomarker values, the biomarker values each corresponding to a biomarker from a set of N biomarkers, wherein the biomarkers are selected from the group of biomarkers listed in Table 19, wherein a classification of the biomarker value indicates the absence of cancer in the individual, and wherein N=3-10.

In another aspect, the present invention provides a method of diagnosing the absence of cancer, the method comprising detecting in a biological sample from an individual biomarker values, the biomarker values each corresponding to a biomarker from a set of N biomarkers, wherein the biomarkers are selected from the group of biomarkers listed in Table 19, wherein a classification of the biomarker value indicates the absence of cancer in the individual, and wherein N=3-10.

In another aspect, the present invention provides a method of diagnosing the absence of cancer, the method comprising detecting, in a biological sample from an individual, biomarker values, each of the biomarker values corresponding to a biomarker in a panel of biomarkers selected from the group of biomarkers provided in Tables 20-29, wherein a classification of the biomarker values indicates the absence of cancer in the individual.

In another aspect, the present invention provides a method of diagnosing cancer in an individual, the method comprising detecting, in a biological sample from the individual, a biomarker value corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 19, wherein the individual is classified as having cancer based on a classification score that deviates from a predetermined threshold, and wherein N=3-10.

In another aspect, the present invention provides a method of diagnosing the absence of cancer in an individual, the method comprising detecting, in a biological sample from the individual, a biomarker value corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 19, wherein the individual is classified as not having cancer based on a classification score that deviates from a predetermined threshold, and wherein N=3-10.

In another aspect, the present invention provides a computer-implemented method for indicating a likelihood of having cancer. The method comprises: retrieving, on a computer, biomarker information for an individual, wherein the biomarker information comprises biomarker values, each of the biomarker values corresponding to one of at least N biomarkers selected from the group of biomarkers listed in Table 19, where N is as defined above; performing, on a computer, a classification on each of the biomarker values; and indicating a likelihood that the individual has cancer based on a plurality of classifications.

In another aspect, the present invention provides a computer-implemented method for classifying an individual as having or not having cancer. The method comprises: retrieving, on a computer, biomarker information for the individual, wherein the biomarker information comprises biomarker values, each of the biomarker values corresponding to one of at least N biomarkers selected from the group of biomarkers provided in Table 19; performing, on a computer, a classification on each of the biomarker values; and indicating whether the individual has cancer based on a plurality of classifications.

In another aspect, the present invention provides a computer program product that indicates a likelihood of having cancer. The computer program product includes a computer-readable medium containing program code executable by a processor of a computing device or system, the program code comprising: code for retrieving data attributed to a biological sample from an individual, wherein the data includes biomarker values, each of the biomarker values corresponding to one of at least N biomarkers in the biological sample selected from the group of biomarkers listed in Table 19, where N is as defined above; and code for executing a classification method that indicates a likelihood that the individual has cancer as a function of the biomarker values.

In another aspect, the present invention provides a computer program product for indicating a cancer status in an individual. The computer program product includes a computer-readable medium containing program code executable by a processor of a computing device or system, the program code comprising: code for retrieving data attributed to a biological sample from an individual, wherein the data includes biomarker values, each of the biomarker values corresponding to one of at least N biomarkers in the biological sample selected from the group of biomarkers provided in Table 19; and code for executing a classification method that indicates the cancer status of the individual as a function of the biomarker values.

In another aspect, the present invention provides a computer-implemented method for indicating a likelihood of having cancer. The method comprises: retrieving, on a computer, biomarker information for an individual, wherein the biomarker information comprises biomarker values corresponding to a biomarker selected from the group of biomarkers listed in Table 19; classifying, on a computer, the biomarker values; and indicating a likelihood that the individual has cancer based on the classification.

In another aspect, the present invention provides a computer-implemented method for classifying an individual as having or not having cancer. The method comprises: retrieving biomarker information for the individual from a computer, wherein the biomarker information comprises biomarker values corresponding to biomarkers selected from the group of biomarkers provided in Table 19; classifying the biomarker values using a computer; and indicating whether the individual has cancer based on the classification.

In another aspect, the present invention provides a computer program product that indicates a likelihood of having cancer. The computer program product includes a computer-readable medium containing program code executable by a processor of a computing device or system, the program code comprising: code for retrieving data attributed to a biological sample from an individual, wherein the data includes biomarker values corresponding to biomarkers in the biological sample selected from the group of biomarkers listed in Table 19; and code for executing a classification method that indicates a likelihood that the individual has cancer as a function of the biomarker values.

In another aspect, the present invention provides a computer program product for indicating a cancer status in an individual. The computer program product includes a computer-readable medium containing program code executable by a processor of a computing device or system, the program code comprising: code for retrieving data attributed to a biological sample from an individual, wherein the data includes biomarker values corresponding to biomarkers in the biological sample selected from the group of biomarkers provided in Table 19; and code for executing a classification method that indicates the cancer status of the individual as a function of the biomarker values.

In another aspect, the present invention provides a method of diagnosing pancreatic cancer, the method comprising detecting the tumor marker CA 19-9 in a biological sample from an individual in addition to biomarker values, the biomarker values each corresponding to a biomarker in a panel of biomarkers selected from the group of biomarkers listed in Table 1, wherein a classification of the combined CA 19-9 and biomarker values indicates that the individual has pancreatic cancer.

BRIEF DESCRIPTION OF THE DRAWINGS

1A is a flow chart of an example method for detecting pancreatic cancer in a biological sample.

FIG. 1B is a flow chart of an example method for detecting pancreatic cancer in a biological sample using a naive Bayesian classification method.

FIG2 shows the ROC curve of the single biomarker CTSB using a naive Bayes classifier for the test to detect pancreatic cancer.

FIG3 shows ROC curves for biomarker panels of 2 to 10 biomarkers using a naive Bayes classifier for a test to detect pancreatic cancer.

FIG4 illustrates the increase in classification score (AUC) when the number of biomarkers increases from 1 to 10 using Naive Bayes classification for a pancreatic cancer panel.

Figure 5 shows the measured biomarker distribution of CTSB as cumulative distribution function (cdf) in log-transformed RFU format for the combined GI and normal controls (solid line) and pancreatic cancer disease group (dashed line), and their curve fit to the normal cdf (dashed line) for training the naive Bayes classifier.

FIG6 illustrates an example computer system for use with the various computer-implemented methods described herein.

7 is a flow chart of a method of indicating the likelihood that an individual has pancreatic cancer, according to an embodiment.

8 is a flow chart of a method of indicating the likelihood that an individual has pancreatic cancer, according to an embodiment.

9 illustrates an example aptamer assay that can be used to detect one or more pancreatic cancer biomarkers in a biological sample.

FIG10 shows a histogram of frequencies of which biomarkers were used from an aggregated set of potential biomarkers to build a classifier to distinguish pancreatic cancer from GI and normal controls.

FIG11A shows a pair of bar graphs summarizing all possible single-protein naive Bayes classifier scores (AUC) using the biomarkers listed in Table 1 (solid lines) and a collection of random markers (dashed lines).

FIG11B shows a pair of bar graphs summarizing all possible two-protein naive Bayes classifier scores (AUC) using the biomarkers listed in Table 1 (solid lines) and a collection of random markers (dashed lines).

FIG11C shows a pair of bar graphs summarizing all possible three-protein naive Bayes classifier scores (AUC) using the biomarkers listed in Table 1 (solid lines) and a collection of random markers (dashed lines).

Figure 12 shows the AUC of the Naive Bayes classifier using 2-10 markers selected from the full set (diamonds), as well as the scores obtained by discarding the best 5, 10, and 15 markers during classifier generation.

FIG13 shows the performance of three different classifiers: CA19-9 alone, the SOMAmer panel, and the combination of SOMAmers and CA19-9.

Figure 14 shows the performance of CA19-9 plus one (HAMP) or two (HAMP and CTSB) SOMAmer biomarkers.

FIG15 shows the performance of a 10-label random forest classifier.

FIG16A shows a set of ROC curves modeled from the data in Table 14 for panels of 1 to 5 markers.

FIG. 16B shows a set of ROC curves calculated from the training data for the 2 to 5 labeled groups of FIG. 12 .

Figures 17A and 17B show a comparison of the performance of 10 biomarkers selected by the greedy selection method (Table 19) and a 1,000-sample random set of 10 "non-marker" biomarkers. The average AUC for the 10 biomarkers in Table 19 is shown as a dashed vertical line. In Figure 17A, the set of 10 biomarkers was randomly selected from all 10 analytes present in all three cancer studies that were not selected by the greedy method. In Figure 17B, the same method as in Figure 17A was used; however, sampling was limited to the remaining 55 biomarkers from Table 1 that were not selected by the greedy method.

Figure 18 shows the receiver operating characteristic (ROC) curves for the three Naive Bayes classifiers listed in Table 19. For each study, the area under the curve (AUC) is also shown next to the figure legend.

Detailed Description of the Invention

Representative embodiments of the present invention will now be described in detail. Although the present invention is described in conjunction with the embodiments listed, it should be understood that the present invention is not limited to these embodiments. On the contrary, the present invention is intended to encompass all substitutions, modifications, and equivalents that may be included within the scope of the invention as defined in the claims.

One skilled in the art will recognize many methods and materials similar or equivalent to those described herein, which could be used and are within the scope of the practice of the present invention. The present invention is not limited in any way to the methods and materials described.

Unless otherwise defined, technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the invention belongs. Although any methods, devices, and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, the preferred methods, devices, and materials are now described.

All publications, published patent documents, and patent applications cited in this application are indicative of the state of the art in the field to which this application pertains. All publications, published patent documents, and patent applications cited herein are incorporated by reference to the same extent as if each individual publication, published patent document, or patent application was specifically and individually indicated to be incorporated by reference.

As used in this application, including the appended claims, unless otherwise indicated, the singular forms "a," "an," and "the" include plural forms and are used interchangeably with "at least one" and "one or more." Thus, reference to "an aptamer" includes a mixture of aptamers, reference to "a probe" includes a mixture of probes, and so on.

As used herein, the term "about" represents an insignificant modification or variation in a value such that the basic function of the item to which the value relates is not changed.

As used herein, the terms “comprises,” “comprising,” “includes,” “including,” “contains,” “containing” and any variations thereof are intended to cover a non-exclusive inclusion whereby a process, method, method-defined product or composition of matter that comprises, includes or contains an element or series of elements includes not only those elements but may include other elements not expressly listed or inherent to such process, method, method-defined product or composition of matter.

The present application includes biomarkers, methods, devices, reagents, systems and kits for detecting and diagnosing pancreatic cancer, and cancer more generally.

In one aspect, the present invention provides one or more biomarkers that are useful alone or in various combinations for diagnosing pancreatic cancer, allowing for differential diagnosis of pancreatic cancer from non-malignant GI conditions, including acute or chronic pancreatitis (or both), pancreatic obstruction, GERD, gallstones, or abnormal imaging that is later found to be benign, monitoring for pancreatic cancer recurrence, or addressing other clinical indications. As described in detail below, exemplary embodiments include the biomarkers provided in Table 1, column 2, which were identified using a multiplex aptamer-based assay, which is generally described in Example 1 and more specifically described in Example 2.

Table 1, column 2 lists the findings obtained from the following analysis: hundreds of individual blood samples from pancreatic cancer cases, and hundreds of equivalent individual blood samples from GI and normal controls. The GI and normal control groups were designed to match the populations that a pancreatic cancer diagnostic test would have the greatest benefit from, including asymptomatic individuals and symptomatic individuals. The normal control group represents asymptomatic individuals at high risk for pancreatic cancer. High risks for pancreatic cancer include family history of pancreatic cancer, obesity, smoking, diabetes, cystic fibrosis, chronic or hereditary pancreatitis, BRCA mutation carriers, p16 mutations, and Pohl-Jeghers syndrome (Brand E et al. Gut 2007:56:1460). The GI control group included nonspecific abdominal symptoms such as acute or chronic pancreatitis (or both), pancreatic obstruction, GERD, gallstones, or abnormal imaging that was later found to be benign. Samples from the normal controls were combined with the GI controls to discover biomarkers that can be used to screen for differential diagnosis of high-risk asymptomatic individuals and symptomatic individuals. Potential biomarkers were measured in separate samples rather than in mixed disease and control blood; this allows for a better understanding of individual and group variations in phenotypes associated with the presence and absence of disease (in this case, pancreatic cancer). Since 823 protein measurements were made for each sample, and several hundred samples from each disease and control population were measured individually, Table 1, column 2, results from analysis of a very large data set. The measurements were analyzed using the methods described in the "Classification of Biomarkers and Calculation of Disease Scores" section herein. Table 1, column 2, lists 65 biomarkers found to be useful in distinguishing samples from individuals with pancreatic cancer from samples from GI and normal controls. GI controls included individuals with acute or chronic pancreatitis (or both), pancreatic obstruction, GERD, gallstones, or abnormal imaging that was later found to be benign.

While some of the pancreatic cancer biomarkers described herein can be used individually to detect and diagnose pancreatic cancer, methods are also described herein for grouping multiple subsets of pancreatic cancer biomarkers, wherein each grouping or subset selection can be used as a set of three or more biomarkers, which are interchangeably referred to herein as a "biomarker panel" and a set. Thus, various embodiments of the present application provide a combination comprising N biomarkers, where N is at least 2 biomarkers. In other embodiments, N is selected from 2 to 65 biomarkers.

In other embodiments, N is selected from any number between 2 and 7, 2 and 10, 2 and 15, 2 and 20, 2 and 25, 2 and 30, 2 and 35, 2 and 40, 2 and 45, 2 and 50, 2 and 55, or 2 and 65. In other embodiments, N is selected from any number between 3 and 7, 3 and 10, 3 and 15, 3 and 20, 3 and 25, 3 and 30, 3 and 35, 3 and 40, 3 and 45, 3 and 50, 3 and 55, or 3 and 65. In other embodiments, N is selected from any number between 4 and 7, 4 and 10, 4 and 15, 4 and 20, 4 and 25, 2 and 30, 2 and 35, 2 and 40, 4 and 45, 4 and 50, 4 and 55, or 4 and 65. In other embodiments, N is selected from any number between 5 and 7, 5 and 10, 5 and 15, 5 and 20, 5 and 25, 5 and 30, 5 and 35, 5 and 40, 5 and 45, 5 and 50, 5 and 55, or 5 and 65. In other embodiments, N is selected from any number between 6 and 10, 6 and 15, 6 and 20, 6 and 25, 6 and 30, 6 and 35, 6 and 40, 6 and 45, 6 and 50, 6 and 55, or 6 and 65. In other embodiments, N is selected from any number between 7 and 10, 7 and 15, 7 and 20, 7 and 25, 7 and 30, 7 and 35, 7 and 40, 7 and 45, 7 and 50, 7 and 55, or 7 and 65. In other embodiments, N is selected from any number between 8 and 10, 8 and 15, 8 and 20, 8 and 25, 8 and 30, 8 and 35, 8 and 40, 8 and 45, 8 and 50, 8 and 55, or 8 and 65. In other embodiments, N is selected from any number between 9 and 15, 9 and 20, 9 and 25, 9 and 30, 9 and 35, 9 and 40, 9 and 45, 9 and 50, 9 and 55, or 9 and 65. In other embodiments, N is selected from any number between 10 and 15, 10 and 20, 10 and 25, 10 and 30, 10 and 35, 10 and 40, 10 and 45, 10 and 50, 10 and 55, or 10 and 65. It should be understood that N can be selected from a range that includes similar but higher order ranges.

In one embodiment, the number of biomarkers that can be used in a biomarker subset or panel is based on the sensitivity and specificity values for a particular combination of biomarker values. As used herein, the terms "sensitivity" and "specificity" refer to the ability to correctly classify an individual as having or not having pancreatic cancer based on the values of one or more biomarkers detected in a biological sample from that individual. "Sensitivity" refers to the performance of a biomarker or multiple biomarkers in correctly classifying individuals as having pancreatic cancer. "Specificity" refers to the performance of a biomarker or multiple biomarkers in correctly classifying individuals as not having pancreatic cancer. For example, a panel of markers with a specificity of 85% and a sensitivity of 90% for testing a panel of control samples and pancreatic cancer samples means that 85% of the control samples were correctly classified as control samples by the panel, and 90% of the pancreatic cancer samples were correctly classified as pancreatic cancer samples by the panel. Desired or preferred minimum values can be determined as described in Example 3. Representative panels are shown in Tables 4-11, which illustrate a range of 100 different panels of 3-10 biomarkers, each with the indicated specificity and sensitivity levels. The total number of occurrences of each marker in each of these groups is shown at the bottom of each table.

In one aspect, pancreatic cancer is detected or diagnosed in an individual by assaying a biological sample from the individual and detecting biomarker values, each of which corresponds to at least one of the biomarkers CTSB, C5a, or C5 and at least N additional biomarkers selected from the list of biomarkers in Table 1, Col. 2, wherein N is equal to 2, 3, 4, 5, 6, 7, 8, or 9. In another aspect, pancreatic cancer is detected or diagnosed in an individual by assaying a biological sample from the individual and detecting biomarker values, each of which corresponds to one of the biomarkers CTSB, C5a, or C5 and at least N additional biomarkers selected from the list of biomarkers in Table 1, Col. 2, wherein N is equal to 1, 2, 3, 4, 5, 6, or 7. In another aspect, pancreatic cancer is detected or diagnosed in an individual by assaying a biological sample from the individual and detecting biomarker values, each of which corresponds to the biomarker CTSB and one of at least N additional biomarkers selected from the list of biomarkers in Table 1, Col. 2, wherein N is equal to 2, 3, 4, 5, 6, 7, 8, or 9. In another aspect, pancreatic cancer is detected or diagnosed in an individual by assaying a biological sample from the individual and detecting biomarker values, each of which corresponds to the biomarker C5a and one of at least N additional biomarkers selected from the list of biomarkers in Table 1, Col. 2, wherein N is equal to 2, 3, 4, 5, 6, 7, 8, or 9. In another aspect, pancreatic cancer is detected or diagnosed in an individual by assaying a biological sample from the individual and detecting biomarker values, each of which corresponds to the biomarker C5 and one of at least N additional biomarkers selected from the list of biomarkers in Table 1, Col. 2, wherein N is equal to 2, 3, 4, 5, 6, 7, 8, or 9.

The pancreatic cancer biomarkers identified herein represent a selection of a larger number of subsets or panels of biomarkers that can be used to effectively detect or diagnose pancreatic cancer. The desired number of such biomarkers will depend on the specific combination of biomarkers selected. It is important to remember that a panel of biomarkers for detecting or diagnosing pancreatic cancer may also include biomarkers not found in Table 1, Column 2, and that the inclusion of additional biomarkers not found in Table 1, Column 2 may reduce the number of biomarkers selected from a particular subset or panel of biomarkers from Table 1, Column 2. The number of biomarkers from Table 1, Column 2 used in a subset or panel may also be reduced if additional biomedical information is used in conjunction with the biomarker values to establish acceptable sensitivity and specificity values for a given assay.

Another factor that can influence the number of biomarkers used in a subset or panel of biomarkers is the method used to obtain biological samples from individuals undergoing pancreatic cancer diagnosis. In carefully controlled sample acquisition environments, the number of biomarkers required to meet desired sensitivity and specificity values may be lower than in situations where there may be more variation in sample collection, processing, and storage. In studying the list of biomarkers listed in column 2 of Table 1, multiple sample collection sites were used to collect data for classifier training. This provides more robust biomarkers that are less sensitive to variations in sample collection, processing, and storage, but may also require a larger number of biomarkers in a subset or panel if the training data were all obtained under very similar conditions.

One aspect of the present application can be generally described with reference to Figures 1A and 1B. A biological sample is obtained from one or more individuals of interest. The biological sample is then assayed to detect the presence of one or more (N) biomarkers of interest, and a biomarker value (referred to as the marker RFU in Figure 1B) is determined for each of the N biomarkers. Once the biomarkers are detected and assigned a biomarker value, each marker is scored or classified as described in detail herein. The marker scores are then combined to provide an overall diagnostic score that represents the likelihood that the individual from whom the sample was obtained has pancreatic cancer.

"Biological sample," "sample," and "test sample" are used interchangeably herein to refer to any material, biological fluid, tissue, or cell obtained from or otherwise derived from an individual. This includes blood (including whole blood, leukocytes, peripheral blood mononuclear cells, buffy coat, plasma, and serum), sputum, tears, mucus, nasal washes, nasal aspirates, breath, urine, semen, saliva, peritoneal lavage fluid, ascites, cystic fluid, meningeal fluid, amniotic fluid, glandular fluid, pancreatic juice, lymphatic fluid, pleural fluid, nipple aspirates, bronchial aspirates, bronchial brushings, synovial fluid, joint aspirates, organ secretions, cells, cell extracts, and cerebrospinal fluid. It also includes experimentally separated fractions of all the above materials. For example, a blood sample can be fractionated into serum, plasma, or fractions containing a specific type of blood cell, such as red blood cells or white blood cells (leukocytes). If necessary, the sample can be a combination of samples from an individual, such as a combination of tissue and liquid samples. The term "biological sample" also includes materials containing homogeneous solid materials, such as materials from fecal samples, tissue samples or tissue biopsy samples. The term "biological sample" also includes materials derived from tissue culture or cell culture. Any suitable method for obtaining a biological sample can be used; exemplary methods include phlebotomy, swabs (such as oral swabs) and fine needle aspiration biopsy methods. Exemplary tissues susceptible to fine needle aspiration include lymph nodes, lungs, lung washes, BAL (bronchoalveolar lavage fluid), thyroid, breast, pancreas and liver. Samples can also be collected by microdissection (such as laser capture microdissection (LCM) or laser microdissection (LMD)), bladder washing, smears (such as PAP smears) or ductal lavage. "Biological samples" obtained from or derived from an individual include any such samples that have been processed by any suitable means after being obtained from the individual.

Furthermore, it should be appreciated that a biological sample can be obtained by taking biological samples from a number of individuals and pooling them or pooling aliquots of each individual's biological sample. The pooled sample can be treated as a sample from a single individual, and if the presence of cancer is determined in the pooled sample, the biological sample from each individual can then be tested to determine which individual(s) have pancreatic cancer.

For the purposes of this specification, the phrase "data attributed to a biological sample from an individual" means that the data is derived in some form from or generated using the individual's biological sample. The data may be reformatted, modified, or mathematically altered to some extent after generation, for example, by conversion from units in one measurement system to units in another measurement system; however, it is understood that the data is derived from or generated using the biological sample.

"Target," "target molecule," and "analyte" are used interchangeably herein to refer to any molecule of interest that may be present in a biological sample. A "molecule of interest" includes any minor change in a particular molecule, such as, in the case of a protein, a minor change in the amino acid sequence, disulfide bond formation, glycosylation, lipidation, acetylation, phosphorylation, or any other manipulation or modification, such as coupling with a labeling component that does not substantially change the identity of the molecule. A "target molecule," "target," or "analyte" is a set of copies of a type or kind of molecule or multimolecular structure. "Target molecule," "target," and "analyte" refer to one or more such molecules. Exemplary target molecules include proteins, polypeptides, nucleic acids, carbohydrates, lipids, polysaccharides, glycoproteins, hormones, receptors, antigens, antibodies, affybodies, antibody mimics, viruses, pathogens, toxic substances, substrates, metabolites, transition state analogs, cofactors, inhibitors, drugs, dyes, nutrients, growth factors, cells, tissues, and any fragment or portion of any of the foregoing.

As used herein, "polypeptide," "peptide," and "protein" are used interchangeably herein to refer to amino acid polymers of any length. The polymer may be linear or branched, it may contain modified amino acids, and it may be interrupted by non-amino acids. The term also encompasses amino acid polymers that have been modified naturally or by intervention; for example, disulfide bond formation, glycosylation, lipidation, acetylation, phosphorylation, or any other manipulation or modification, such as conjugation with a labeling component. The definition also includes, for example, polypeptides containing one or more analogs of amino acids (including, for example, non-natural amino acids, etc.), as well as other modifications known in the art. The polypeptide may be a single chain or an associated chain. The definition also includes preproteins and intact mature proteins; peptides or polypeptides derived from mature proteins; fragments of proteins; splice variants; recombinant forms of proteins; protein variants with amino acid modifications, deletions, or substitutions; digestion; and post-translational modifications, such as glycosylation, acetylation, phosphorylation, and the like.

As used herein, "marker" and "biomarker" are used interchangeably to refer to a target molecule that indicates or is a sign of a normal or abnormal process in an individual, or a disease or other disease condition in an individual. More specifically, a "marker" or "biomarker" is an anatomical, physiological, biochemical, or molecular parameter that is associated with the presence of a specific physiological state or process, whether normal or abnormal, and, if abnormal, whether chronic or acute. Biomarkers can be detected and measured by various methods, including laboratory assays and medical imaging. When the biomarker is a protein, the expression of the corresponding gene can also be used as a surrogate measure of the amount or presence or absence of the corresponding protein biomarker in a biological sample, or the methylation state of the gene encoding the biomarker or the protein that controls the expression of the biomarker.

As used herein, "biomarker value," "value," "biomarker level," and "level" are used interchangeably to refer to a measurement made using any analytical method to detect a biomarker in a biological sample that indicates the presence, absence, absolute amount or concentration, relative amount or concentration, titer, level, expression level, ratio of measured levels, etc. of, for, or corresponding to a biomarker in the biological sample. The precise nature of the "value" or "level" depends on the specific design and components of the particular analytical method used to detect the biomarker.

When a biomarker is indicative of an abnormal process or disease or other disease condition in an individual, or is a sign of an abnormal process or disease or other disease condition in an individual, the biomarker is generally described as being overexpressed or underexpressed when compared to the expression level or value of a biomarker that is indicative of a normal process or absence of a disease or other disease condition in the individual, or is a sign of a normal process or absence of a disease or other disease condition in the individual. "Upregulated," "upregulated," "overexpressed," "overexpressed," and any variations thereof are used interchangeably to refer to a value or level of a biomarker in a biological sample that is higher than the value or level (or range of values or levels) of the biomarker typically detected in a similar biological sample from a healthy or normal individual. The term may also refer to a value or level of a biomarker in a biological sample that is higher than the value or level (or range of values or levels) of the biomarker detected at different stages of a particular disease.

"Downregulated," "downregulated," "underexpression," or "underexpressed," and any variations thereof, are used interchangeably to refer to a value or level of a biomarker in a biological sample that is lower than the value or level (or range of values or levels) of the biomarker typically detected in a similar biological sample from a healthy or normal individual. The term may also refer to a value or level of a biomarker in a biological sample that is lower than the value or level (or range of values or levels) of the biomarker detected at different stages of a particular disease.

Additionally, an overexpressed or underexpressed biomarker can also refer to being "differentially expressed" or having a "different level" or "different value" compared to a "normal" expression level or value of the biomarker, which is indicative of or indicative of a normal process or absence of a disease or other condition in an individual. Thus, "differential expression" of a biomarker can also refer to a difference from a "normal" expression level of the biomarker.

The terms "differential gene expression" and "differential expression" are used interchangeably and refer to the expression of a gene (or its corresponding protein expression product) being activated to a higher or lower level relative to its expression in a normal or control subject in a subject suffering from a specified disease. The term also includes the expression of a gene (or its corresponding protein expression product) being activated to a higher or lower level at different stages of the same disease. It should also be understood that differentially expressed genes can be activated or inhibited at the nucleic acid level or protein level, or can undergo alternative splicing to obtain different polypeptide products. Such differences can be confirmed by many changes, including mRNA levels, surface expression, secretion or other partitioning of the polypeptide. Differential gene expression can include comparing the expression between two or more genes or their gene products; or comparing the ratio of expression between two or more genes or their gene products; or even comparing two differently processed products of the same gene, which are different between normal subjects and diseased subjects or between different stages of the same disease. Differential expression includes quantitative and qualitative differences in the temporal or cellular expression pattern of genes or their expression products, for example, in normal and diseased cells or cells experiencing different disease events or disease stages.

As used herein, "subject" refers to a test subject or patient. The subject can be a mammal or a non-mammal. In many embodiments, the subject is a mammal. The mammalian subject can be human or non-human. In many embodiments, the subject is human. A healthy or normal subject is one in which the disease or condition of interest (including, for example, pancreatic disease, pancreatic-related disease, or other pancreatic disease condition) is not detectable by conventional diagnostic methods.

"Diagnose," "diagnosing," "diagnosis," and variations thereof refer to the detection, determination, or identification of a health or disease state of an individual based on one or more signs, symptoms, data, or other information associated with the individual. The health state of an individual can be diagnosed as healthy/normal (i.e., diagnosed as the absence of a disease or disease state) or diagnosed as diseased/abnormal (i.e., diagnosed as the presence of a disease or disease state or an evaluation of the characteristics of a disease or disease state). For a particular disease or disease state, the terms "diagnose," "diagnosing," "diagnosis," and the like encompass the initial detection of the disease; the characterization or classification of the disease; the detection of progression, remission, or recurrence of the disease; and the detection of disease response after treatment or therapy is administered to an individual. Diagnosis of pancreatic cancer includes distinguishing individuals with cancer from those without cancer. It also includes distinguishing GI and normal controls from pancreatic cancer.

"Prognose," "prognosing," "prognosis," and variations thereof refer to predicting the future course of a disease or condition in an individual having the disease or condition (e.g., predicting patient survival), and such terms encompass evaluating the response of a disease after administration of treatment or therapy to an individual.

"Evaluate," "evaluating," "evaluation," and variations thereof encompass "diagnosis" and "prognosis," and also encompass determining or predicting the future course of a disease or condition in an individual who is not ill and determining or predicting the likelihood of recurrence of the disease or condition in an individual who has apparently been cured of the disease. The term "evaluate" also includes evaluating an individual's response to a therapy, e.g., predicting whether an individual is likely to respond favorably to a therapeutic agent, or is unlikely to respond to a therapeutic agent (or will, for example, experience toxicity or other undesirable side effects); selecting a therapeutic agent to administer to an individual; or monitoring or determining an individual's response to a therapy that has already been administered to the individual. Thus, "evaluating" pancreatic cancer can include, for example, any of the following: prognosing the future course of pancreatic cancer in an individual; predicting the recurrence of pancreatic cancer in an individual who has apparently been cured of pancreatic cancer; or determining or predicting an individual's response to a pancreatic cancer therapy; or selecting a pancreatic cancer therapy to administer to an individual based on determining a biomarker value derived from a biological sample from the individual.

Any of the following instances may be referred to as "diagnosing" or "evaluating" pancreatic cancer: initially detecting the presence or absence of pancreatic cancer; determining the specific stage, type, or subtype or other classification or characteristics of pancreatic cancer; determining whether a suspicious mass is a benign lesion or a malignant pancreatic tumor; or detecting/monitoring pancreatic cancer progression (e.g., monitoring tumor growth or metastatic spread), remission, or recurrence.

As used herein, "additional biomedical information" refers to one or more assessments of an individual's risk for cancer, or more specifically, pancreatic cancer, made in addition to the use of any of the biomarkers described herein. "Additional biomedical information" includes any of the following: a physical description of the individual, including a pancreatic mass as observed by any contrast-enhanced multislice (multi-detector) spiral computed tomography (CT) with three-dimensional reconstruction, percutaneous or endoscopic ultrasound (US or EUS), endoscopic retrograde cholangiopancreatography (ERCP), magnetic resonance imaging (MRI), MR cholangiopancreatography (MRCP), or abdominal ultrasound; the individual's height and/or weight; weight change; the individual's race; occupational history; family history of pancreatic cancer (or other cancers); the presence of genetic markers in the individual or family members that are associated with an increased risk for pancreatic cancer (or other cancers); the presence or absence of a pancreatic or other abdominal mass; the size of the mass; the location of the mass; the morphology of the mass and associated abdominal area (e.g., by imaging observations); clinical symptoms such as abdominal pain, weight loss, anorexia, early satiety, diarrhea, or steatorrhea, jaundice, recent atypical diabetes mellitus, recent but unexplained history of thrombophlebitis, or previous attacks of pancreatitis; gene expression values; physical description of the individual, including physical descriptions observed by radiographic imaging; height and/or weight of the individual; sex of the individual; race of the individual; smoking history; alcohol use history; occupational history; exposure to known carcinogens (such as exposure to any asbestos, radon, chemicals, smoke from fires, and air pollution, which can include emissions from stationary or mobile sources, such as industrial/factory or automotive/shipping/aircraft emissions); exposure to secondhand smoke; and family history of pancreatic or other cancers. Testing of biomarker levels in conjunction with evaluation of any additional biomedical information, including other laboratory tests (e.g., CA19-9, serum bilirubin concentration, alkaline phosphatase activity, presence of anemia) can, for example, improve the sensitivity, specificity, and/or AUC for detecting pancreatic cancer (or other pancreatic cancer-related uses) compared to testing of the biomarker alone or evaluating any specific item of additional biomedical information alone (e.g., ultrasound imaging alone). The additional biomedical information can be obtained from the individual using conventional techniques known in the art, such as from the individual themselves through the use of routine patient questionnaires or health history questionnaires, or from a medical practitioner, etc. Testing of biomarker levels in conjunction with evaluation of any additional biomedical information can, for example, improve the sensitivity, specificity, and/or AUC for detecting pancreatic cancer (or other pancreatic cancer-related uses) compared to testing of the biomarker alone or evaluating any specific item of additional biomedical information alone (e.g., CT imaging alone).

Cancer-associated antigen 19-9 (CA 19-9) is a known blood marker for pancreatic cancer. The reported sensitivity and specificity of CA 19-9 for pancreatic cancer are 80-90%, respectively. However, these values are closely related to tumor size. The accuracy of CA 19-9 in identifying patients with small, surgically resectable cancers is limited. CA 19-9 requires the presence of Lewis blood group antigens (glycosyltransferases) for expression. In individuals with a Lewis-negative phenotype (an estimated 5-10% of the population), CA 19-9 levels are not a useful tumor marker. The specificity of CA 19-9 is also limited. CA 19-9 is often elevated in patients with various benign pancreatic and biliary conditions. The degree of CA 19-9 elevation (at initial presentation and in the postoperative setting) is associated with long-term prognosis. In addition, in patients who appear to have potentially resectable disease, the magnitude of CA 19-9 levels can also help predict the presence of radiographically occult metastatic disease. Serial monitoring of CA 19-9 levels can be used to follow up patients after potentially curative surgery and for patients receiving chemotherapy for advanced disease. Elevated CA 19-9 levels often precede the appearance of radiographic recurrent disease, but confirmation of disease progression should be performed by imaging studies and/or biopsy. Detection of biomarker levels in combination with CA 19-9 can, for example, improve sensitivity, specificity, and/or AUC for detecting pancreatic cancer (or other pancreatic cancer-related uses) compared to CA 19-9 alone.

The term "area under the curve" or "AUC" refers to the area under the receiver operating characteristic (ROC) curve, both of which are well known in the art. The AUC measurement can be used to compare the accuracy of a classifier within the complete data range. A classifiers with larger AUCs have greater ability to correctly classify unknown cases between two groups of interest (such as pancreatic cancer samples and normal or control samples). ROC curves can be used to plot the performance of a specific feature (such as any biomarker described herein and/or any additional biomedical information project) to distinguish between two populations (such as cases with pancreatic cancer and controls without pancreatic cancer). Typically, the feature data for an entire population (such as cases and controls) is incrementally classified based on the value of a single feature. Then, for each value of that feature, the true positive rate and false positive rate of the data are calculated. The true positive rate is determined by counting the number of cases above the feature value and then dividing it by the total number of cases. The false positive rate is determined by counting the number of controls above the feature value and then dividing it by the total number of controls. In some embodiments, the present invention relates to a method for measuring the sensitivity of a patient to a specific disease or condition. The method of measuring the sensitivity of a patient to a specific disease or condition is used to determine the specific disease or condition. The method of measuring the sensitivity of a patient to a specific disease or condition is used to determine the specific disease or condition. The method of measuring the sensitivity of a patient to a specific disease or condition is used to determine the specific disease or condition. The method of measuring the sensitivity of a patient to a specific disease or condition is used to determine the specific disease or condition.

As used herein, "detecting" or "determining" a biomarker value includes the use of equipment necessary to observe and record a signal corresponding to the biomarker value and materials necessary to generate the signal. In various embodiments, biomarker values are detected using any suitable method, including fluorescence, chemiluminescence, surface plasmon resonance, surface acoustic waves, mass spectrometry, infrared spectroscopy, Raman spectroscopy, atomic force microscopy, scanning tunneling microscopy, electrochemical detection methods, nuclear magnetic resonance, quantum dots, and the like.

"Solid support" as used herein refers to any support having a surface to which molecules can be attached, directly or indirectly, by covalent or non-covalent bonds. "Solid support" can have various physical forms and can include, for example, membranes; chips (such as protein chips); slides (such as slides or coverslips); columns; hollow, solid, semi-solid, porous or cavity particles, such as beads; gels; fibers, including optical fiber materials; matrices; and sample containers. Exemplary sample containers include sample wells, tubes, capillaries, vials, and any other container, groove, or depression capable of holding a sample. Sample containers can be contained on multi-sample platforms, such as microtiter plates, slides, microfluidic devices, and the like. The support can be composed of natural or synthetic materials, organic or inorganic materials. The composition of the solid support to which the capture agent is attached generally depends on the attachment method (such as covalent attachment). Other exemplary containers include droplets and microfluidic controlled or bulk oil/aqueous emulsions in which assays and related operations can be performed. Solid supports can be used for example to prepare solid supports.Suitable solid supports include, for example, plastics, resins, polysaccharides, silica or silica-based materials, functionalized glass, modified silicon, carbon, metal, inorganic glass, film, nylon, natural fiber (such as silk, wool and cotton), polymer etc. The material comprising solid supports can include reactive groups, such as carboxyl, amino or hydroxyl for the attachment of trapping agents.Polymeric solid supports can include, for example, polystyrene, polyethylene terephthalate, polyvinyl acetate, polyvinyl chloride, polyvinyl pyrrolidone, polyacrylonitrile, polymethyl methacrylate, polytetrafluoroethylene, butyl rubber, styrene butadiene rubber, natural rubber, polyethylene, polypropylene, (poly) tetrafluoroethylene, (poly) vinylidene fluoride, polycarbonate and polymethylpentene.Operable suitable solid support particles include, for example, coded particles, such as Luminex-type coded particles, magnetic particles and glass particles.

Exemplary Uses of Biomarkers

In many exemplary embodiments, the present invention provides methods for diagnosing pancreatic cancer in an individual by detecting one or more biomarker values corresponding to one or more biomarkers present in the individual's circulation, such as serum or plasma, and by any number of analytical methods, including any analytical methods described herein. These biomarkers are, for example, differentially expressed in individuals with pancreatic cancer compared to individuals without pancreatic cancer. Detection of differential expression of biomarkers in an individual can be used, for example, to allow early diagnosis of pancreatic cancer, to distinguish between benign and malignant masses (e.g., masses observed on computed tomography (CT), MRI, or ultrasound), to monitor for pancreatic cancer recurrence, or to differentially diagnose other clinical conditions such as acute or chronic pancreatitis (or both), pancreatic obstruction, GERD, gallstones, or abnormal imaging that is later found to be benign.

Any of the biomarkers described herein can be used for various clinical indications of pancreatic cancer, including any of the following: detecting pancreatic cancer (e.g., in high-risk individuals or populations); characterizing pancreatic cancer (e.g., determining the type, subtype, or stage of pancreatic cancer), such as by distinguishing pancreatic cancer (cancer) from acute or chronic pancreatitis (or both), pancreatic obstruction, GERD, gallstones, or abnormal imaging that is later found to be benign and/or distinguishing adenocarcinoma from other malignant cell types (or otherwise facilitating histopathology); determining whether a pancreatic mass is a benign or malignant pancreatic tumor; determining pancreatic cancer prognosis; monitoring pancreatic cancer progression or remission; monitoring pancreatic cancer recurrence; monitoring metastasis; treatment selection; monitoring response to therapeutic agents or other treatments; stratification of individuals for endoscopic ultrasound (EUS) screening (e.g., identifying those individuals at higher risk for pancreatic cancer who would be most likely to benefit from radiographic screening, thereby increasing the positive predictive value of EUS); combining biomarker testing with additional biomedical information such as smoking or alcohol history, or CA 19-9 level, the presence of a genetic marker indicating a high risk of pancreatic cancer, etc., or with mass size, morphology, the presence of ascites, etc. (e.g., to provide an assay with increased diagnostic performance compared to a CA 19-9 test or other biomarker test or with mass size, morphology, etc.); facilitate the diagnosis of an abdominal mass as malignant or benign; facilitate clinical decisions once an abdominal mass is observed on CT, MRI, PET, or EUS (e.g., if the abdominal mass is considered low risk, e.g., if the biomarker-based test is negative, with or without classification of mass size, then repeat radiographic scanning, or if the mass is considered intermediate-to-high risk, e.g., if the biomarker-based test is positive, with or without classification of mass size or degree of tissue invasion, then consider biopsy); or facilitate decisions regarding clinical follow-up (e.g., whether to perform a repeat radiographic scan, fine needle biopsy, or surgery after an abdominal mass is observed on imaging). Biomarker testing alone can improve the positive predictive value (PPV) of EUS screening in high-risk individuals. In addition to combined EUS screening, the biomarkers described herein can also be used in conjunction with any other imaging modality used to detect pancreatic cancer, such as CT, MRI, or PET scans. Furthermore, the biomarkers can be used to allow certain applications before signs of pancreatic cancer are detected by imaging or other clinical correlates, or before symptoms develop. This also includes differentiating acute or chronic pancreatitis (or both), pancreatic obstruction, GERD, gallstones, or abnormal imaging that is later found to be benign from pancreatic cancer.

An exemplary manner in which any of the biomarkers described herein can be used to diagnose pancreatic cancer is that differential expression of one or more of the biomarkers in an individual not known to have pancreatic cancer can indicate that the individual has pancreatic cancer, thereby enabling detection of pancreatic cancer at an earlier stage of the disease when treatment is most effective, perhaps before pancreatic cancer is detected by other means or before symptoms appear. Overexpression of one or more biomarkers during pancreatic cancer can indicate progression of the cancer, such as pancreatic tumor growth and/or metastasis (and thus, a poor prognosis); whereas a decrease in the degree of differential expression of one or more biomarkers (i.e., expression levels in the individual trending toward or approaching "normal" expression levels on subsequent biomarker testing) can indicate remission of the cancer, such as pancreatic tumor shrinkage (and thus, a good or better prognosis). Similarly, an increase in the degree of differential expression of one or more biomarkers during pancreatic cancer treatment (i.e., the expression levels in the individual move further away from "normal" expression levels in subsequent biomarker tests) can indicate progression of the pancreatic cancer and, therefore, that the treatment is ineffective; whereas a decrease in the differential expression of one or more biomarkers during pancreatic cancer treatment can indicate remission of the pancreatic cancer and, therefore, that the treatment is successful. Furthermore, an increase or decrease in the differential expression of one or more biomarkers after an individual appears to have been cured of pancreatic cancer can indicate recurrence of pancreatic cancer. In such cases, for example, treatment can be restarted in the individual at an early stage (or, if the individual is maintained on treatment, the treatment regimen can be modified to increase the dose and/or frequency), where recurrence of pancreatic cancer would otherwise not be detected until a later stage. Furthermore, the differential expression levels of one or more biomarkers in an individual can be predictive of the individual's response to a particular therapeutic agent. In monitoring pancreatic cancer recurrence or progression, changes in biomarker expression levels can indicate the need for repeat imaging (e.g., repeat EUS), for example, to determine pancreatic cancer activity or to determine the need for a change in treatment regimen.

Detection of any of the biomarkers described herein can be particularly useful after or in conjunction with pancreatic cancer treatment, such as to assess the success of treatment or to monitor remission, recurrence, and/or progression (including metastasis) of pancreatic cancer after treatment. Treatment of pancreatic cancer can include, for example, administering a therapeutic agent to a subject, performing surgery (e.g., surgically removing at least a portion of a pancreatic tumor or removing the pancreas and surrounding tissue), administering radiation therapy, or any other type of pancreatic cancer treatment method used in the art, as well as any combination of these treatments. For example, any biomarker can be detected at least once after treatment, or can be detected multiple times after treatment (e.g., periodically), or can be detected before and after treatment. Differential expression levels of any biomarker in a subject over time can indicate progression, remission, or recurrence of pancreatic cancer, examples of which include: an increase or decrease in the expression level of the biomarker after treatment compared to before treatment; an increase or decrease in the expression level of the biomarker at a later time point after treatment compared to an earlier time point after treatment; and a difference in the expression level of the biomarker at a time point after treatment compared to a normal level of the biomarker.

As a specific example, biomarker levels of any of the biomarkers described herein can be determined in serum or plasma samples before and after surgery (e.g., 2-8 weeks after surgery). An increase in biomarker expression levels in the post-operative sample compared to the pre-operative sample can indicate progression of pancreatic cancer (e.g., an unsuccessful surgery); whereas a decrease in biomarker expression levels in the post-operative sample compared to the pre-operative sample can indicate regression of pancreatic cancer (e.g., a successful surgery to remove a pancreatic tumor). Similar analyses of biomarker levels can be performed before and after other forms of treatment, such as radiation therapy or before and after administration of a therapeutic agent or cancer vaccine.

In addition to testing biomarker levels as a stand-alone diagnostic test, biomarker levels can also be combined with the determination of SNPs or other genetic lesions or variations that indicate increased risk of disease susceptibility (see, e.g., Amos et al., Nature Genetics 40, 616-622 (2009)).

In addition to testing biomarker levels as a stand-alone diagnostic test, biomarker levels can also be combined with radiological screening. In addition to testing biomarker levels as a stand-alone diagnostic test, biomarker levels can also be combined with relevant symptoms or genetic testing. The detection of any biomarker described herein can be used to assist in the diagnosis of pancreatic cancer and guide appropriate individual clinical care after a pancreatic mass has been observed by imaging, including care by appropriate surgical experts or through palliative care in unresectable patients. In addition to testing biomarker levels in combination with relevant symptoms or risk factors, information about biomarkers can also be evaluated in combination with other types of data, particularly data indicating individual pancreatic cancer risk (e.g., patient clinical history, symptoms, family history of pancreatic cancer, smoking or drinking history, sudden onset of diabetes, jaundice, risk factors such as the presence of genetic markers, and/or the status of other biomarkers, etc.). These different data can be evaluated by automated methods, such as computer programs/software, which can be implemented in computers or other devices/apparatuses.

In addition to combining biomarker levels with radiological screening tests in high-risk individuals (e.g., evaluating biomarker levels in combination with the size or other characteristics of a pancreatic mass observed on an imaging scan), information about biomarkers can also be evaluated in combination with other types of data, particularly data indicative of an individual's risk for pancreatic cancer (e.g., patient clinical history, symptoms, family history of cancer, risk factors such as whether the individual is a smoker, alcoholic, and/or other biomarker status, etc.). These various data can be evaluated by automated methods, such as computer programs/software, which can be implemented on a computer or other device/apparatus.

Any of the biomarkers can also be used in imaging tests. For example, an imaging agent can be coupled to any of the biomarkers, which can be used to assist in pancreatic cancer diagnosis, monitor disease progression/remission or metastasis, monitor disease recurrence, or monitor response to treatment.

Detection and determination of biomarkers and biomarker values

The biomarker values of the biomarkers described herein can be detected using any known analytical method. In one embodiment, the biomarker values are detected using a capture reagent. As used herein, a "capture agent" or "capture reagent" refers to a molecule that is capable of specifically binding to a biomarker. In many embodiments, the capture reagent can be exposed to the biomarker in solution, or can be exposed to the biomarker while the capture reagent is immobilized on a solid support. In other embodiments, the capture reagent contains a feature that reacts with a second feature on the solid support. In these embodiments, the capture reagent can be exposed to the biomarker in solution, and then the feature on the capture reagent can combine with the second feature on the solid support to immobilize the biomarker on the solid support. The capture reagent is selected based on the type of analysis being performed. Capture agents include, but are not limited to, aptamers, antibodies, adnectins, ankyrins, other antibody mimetics and other protein scaffolds, autoantibodies, chimeras, small molecules, F(ab') ₂ fragments, single-chain antibody fragments, Fv fragments, single-chain Fv fragments, nucleic acids, lectins, ligand-binding receptors, affybodies, nanobodies, imprinted polymers, avimers, peptidomimetics, hormone receptors, cytokine receptors, and synthetic receptors, as well as modifications and fragments of these substances.

In some embodiments, biomarker values are detected using a biomarker/capture reagent complex.

In other embodiments, the biomarker value is derived from a biomarker/capture reagent complex and is detected indirectly, for example, as a result of a reaction subsequent to the biomarker/capture reagent interaction, but dependent upon the formation of the biomarker/capture reagent complex.

In some embodiments, biomarker values are detected directly from the biomarker in the biological sample.

In one embodiment, biomarkers are detected using a multiplex format, which allows for the simultaneous detection of two or more biomarkers in a biological sample. In one embodiment of the multiplex format, capture reagents are directly or indirectly, covalently or non-covalently immobilized at discrete locations on a solid support. In another embodiment, the multiplex format utilizes separate solid supports, each having a unique capture reagent, such as quantum dots, associated therewith. In another embodiment, a separate device is used to detect each of the multiple biomarkers to be detected in a biological sample. The separate devices can be configured to allow for simultaneous processing of each biomarker in a biological sample. For example, a microtiter plate can be used, whereby each well in the plate is used to uniquely analyze one of the multiple biomarkers to be detected in a biological sample.

In one or more of the foregoing embodiments, a fluorescent tag can be used to label components of the biomarker/capture complex to allow detection of the biomarker value. In many embodiments, a fluorescent label can be coupled to a capture reagent specific for any of the biomarkers described herein using known techniques, and the fluorescent label can then be used to detect the corresponding biomarker value. Suitable fluorescent labels include rare earth element chelates, fluorescein and its derivatives, rhodamine and its derivatives, dansyl, allophycocyanin, PBXL-3, Qdot 605, Lissamine, phycoerythrin, Texas Red, and other such compounds.

In one embodiment, fluorescent labeling is a fluorescent dye molecule. In some embodiments, fluorescent dye molecule includes at least one substituted indole ring (indolium ring) system, wherein the substituent on the 3-carbon of indole ring contains chemically reactive groups or coupled substances. In some embodiments, dye molecule includes AlexFluor molecule, such as AlexaFluor 488, AlexaFluor 532, AlexaFluor 647, AlexaFluor 680 or AlexaFluor 700. In other embodiments, dye molecule includes the dye molecule of the first type and the second type, such as two different AlexaFluor molecules. In other embodiments, dye molecule includes the dye molecule of the first type and the second type, and two kinds of dye molecules have different emission spectra.

Fluorescence can be measured using a variety of instruments compatible with a wide range of assay formats. For example, spectrofluorometers have been designed to analyze microtiter plates, microscope slides, printed arrays, cuvettes, and the like. See Principles of Fluorescence Spectroscopy, by J.R. Lakowicz, Springer Science+Business Media, Inc., 2004. See Bioluminescence & Chemiluminescence: Progress & Current Applications; Philip E. Stanley and Larry J. Kricka, editors, World Scientific Publishing Company, January 2002.

In one or more of the foregoing embodiments, a chemiluminescent tag can optionally be used to label components of the biomarker/capture complex to allow detection of the biomarker value. Suitable chemiluminescent materials include any of oxalyl chloride, Rodamin 6G, Ru(bipy) ₃ ²⁺ , TMAE (tetrakis(dimethylamino)ethylene), 1,2,3-trihydroxibenzene, lucigenin, peroxyoxalate, aryl oxalate, acridinium ester, dioxetane, and the like.

In other embodiments, the detection method comprises an enzyme/substrate combination that produces a detectable signal corresponding to the biomarker value. Typically, the enzyme catalyzes a chemical change in a chromogenic substrate that can be measured using a variety of techniques, including spectrophotometry, fluorescence, and chemiluminescence. Suitable enzymes include, for example, luciferase, luciferin, malate dehydrogenase, urease, horseradish peroxidase (HRPO), alkaline phosphatase, β-galactosidase, glucoamylase, lysozyme, glucose oxidase, galactose oxidase, and glucose-6-phosphate dehydrogenase, urate oxidase, xanthine oxidase, lactoperoxidase, microperoxidase, and the like.

In other embodiments, the detection method can be a combination of fluorescence, chemiluminescence, radionuclide, or enzyme/substrate combinations that produce a measurable signal.Multiple modalities of signaling can have unique and advantageous characteristics in biomarker assay formats.

More specifically, the biomarker values of the biomarkers described herein can be detected using known analytical methods, including single-plex aptamer assays, multiplex aptamer assays, single-plex or multiplex immunoassays, mRNA expression profiling, miRNA expression profiling, mass spectrometry analysis, histological/cytological methods, etc., which are described in detail below.

Determining biomarker values using aptamer-based assays

Assays for detecting and quantifying physiologically significant molecules in biological and other samples are important tools in scientific research and healthcare. One class of such assays involves the use of microarrays comprising one or more aptamers immobilized on a solid support. Each of the aptamers is capable of binding to a target molecule with high specificity and very high affinity. See, for example, U.S. Patent No. 5,475,096, entitled "Nucleic Acid Ligands"; see also, for example, U.S. Patent No. 6,242,246, U.S. Patent No. 6,458,543, and U.S. Patent No. 6,503,715, all entitled "Nucleic Acid Ligand Diagnostic Biochips." Once the microarray is contacted with the sample, the aptamers bind to their respective target molecules present in the sample, thereby allowing the determination of a biomarker value corresponding to the biomarker.

As used herein, "aptamer" refers to a nucleic acid with specific binding affinity for a target molecule. It should be understood that the key to affinity interactions is degree; however, in this article, the "specific binding affinity" of an aptamer to its target refers to the fact that the aptamer generally binds to its target with a higher degree of affinity than its affinity for other components in the test sample. An "aptamer" is a series of copies of a nucleic acid molecule of a type or species that has a specific nucleotide sequence. An aptamer can contain any suitable number of nucleotides, including any number of chemically modified nucleotides. An "aptamer" refers to more than one such series of molecules. Different aptamers can have the same or different numbers of nucleotides. An aptamer can be DNA or RNA or a chemically modified nucleic acid and can be single-stranded, double-stranded, or contain a double-stranded region, and can contain a higher-order structure. An aptamer can also be a photoaptamer, wherein the aptamer contains a photoreactive or chemically reactive functional group to allow it to be covalently linked to its corresponding target. Any of the aptamer methods disclosed herein can include the use of two or more aptamers that specifically bind to the same target molecule. As further described below, the aptamers can include a tag. If an aptamer includes a tag, all copies of the aptamer do not need to have the same tag. In addition, if different aptamers each include a tag, the different aptamers can have the same tag or different tags.

Aptamers can be identified using any known method, including the SELEX method. Once identified, aptamers can be prepared or synthesized according to any known method, including chemical synthesis methods and enzymatic synthesis methods.

As used herein, "SOMAmer" or low off-rate modified aptamer refers to an aptamer with improved off-rate characteristics. SOMAmers can be produced using the improved SELEX method described in U.S. Publication No. 2009/0004667, entitled "Methods for Producing Aptamers with Improved Off-Rates."

The terms "SELEX" and "SELEX method" are used interchangeably herein and generally refer to a combination of (1) selecting aptamers that interact with a target molecule in a desired manner, such as binding to a protein with high affinity, and (2) amplifying those selected nucleic acids. The SELEX method can be used to identify aptamers with high affinity for a specific target or biomarker.

SELEX generally comprises preparing a candidate mixture of nucleic acids; binding the candidate mixture to a desired target molecule to form an affinity complex; separating the affinity complex from unbound candidate nucleic acids; separating nucleic acids from the affinity complex and isolating the nucleic acids; purifying the nucleic acids; and identifying specific aptamer sequences. The method may include multiple cycles to further refine the affinity of the selected aptamer. The method may include an amplification step at one or more points in the method. See, for example, U.S. Patent No. 5,475,096, entitled “Nucleic Acid Ligands.” The SELEX method can be used to generate aptamers that covalently bind to the target of the aptamer, as well as aptamers that non-covalently bind to the target of the aptamer. See, for example, U.S. Patent No. 5,705,337, entitled “Systematic Evolution of Nucleic Acid Ligands by Exponential Enrichment: Chemi-SELEX.”

The SELEX method can be used to identify the high-affinity aptamers containing modified nucleotides, and the modified nucleotides give the aptamer improved features, such as improved in vivo stability or improved delivery characteristics. The example of such modifications includes the chemical replacement of ribose and/or phosphate and/or base positions. The aptamers containing modified nucleotides identified by the SELEX method are described in the U.S. Patent No. 5,660,985 entitled "High-affinity nucleic acid ligands containing modified nucleotides", which describes the oligonucleotides containing chemically modified nucleotide derivatives at the 5'-and 2'-positions of pyrimidines. See above, U.S. Patent No. 5,580,737 describes high-specificity aptamers, which contain one or more nucleotides modified with 2'-amino (2'-NH2), 2'-fluoro (2'-F) and/or 2'-O-methyl (2'-OMe). See also US Patent Application Publication 20090098549 entitled "SELEX and PHOTOSELEX," which describes nucleic acid libraries with extended physical and chemical properties and their use in SELEX and photoSELEX.

SELEX can also be used to identify aptamers with desired off-rate characteristics. See U.S. Patent Application Publication 20090004667, entitled “Methods for Producing Aptamers with Improved Off-Rates,” which describes an improved SELEX method for producing aptamers that can bind to target molecules. Methods for producing aptamers and photoaptamers that have slower off-rates with their respective target molecules are described. The method includes contacting a candidate mixture with a target molecule; allowing a nucleic acid-target complex to form; and performing a slow off-rate enrichment process, wherein nucleic acid-target complexes with fast off-rates dissociate and no longer form, while complexes with slow off-rates remain intact. In addition, the method includes using modified nucleotides in generating a candidate nucleic acid mixture to produce aptamers with improved off-rate properties.

A variation of this assay uses aptamers that contain photoreactive functional groups, which allow the aptamers to covalently bind or "photocrosslink" to their target molecules. See, for example, U.S. Patent No. 6,544,776, entitled "Nucleic Acid Ligand Diagnostic Biochip." These photoreactive aptamers are also referred to as photoaptamers. See, for example, U.S. Patent No. 5,763,177, U.S. Patent No. 6,001,577, and U.S. Patent No. 6,291,184, all entitled "Systematic Evolution of Nucleic Acid Ligands by Exponential Enrichment: Photoselection of Nucleic Acid Ligands and Solution SELEX"; see also, for example, U.S. Patent No. 6,458,539, entitled "Photoselection of Nucleic Acid Ligands." After the microarray is contacted with the sample and the photoaptamers have an opportunity to bind to their target molecules, the photoaptamers are photoactivated and the solid support is washed to remove any non-specifically bound molecules. Stringent wash conditions can be used because target molecules bound to the photoaptamer are typically not removed due to the covalent bond created by the photoactivated functional group on the photoaptamer.In this manner, the assay allows for the detection of biomarker values corresponding to the biomarkers in the test sample.

In both assay formats, the aptamer is immobilized on a solid support prior to contact with the sample. However, in some cases, immobilizing the aptamer prior to contact with the sample may not provide the optimal assay. For example, pre-immobilizing the aptamer may result in ineffective mixing of the aptamer and the target molecule on the surface of the solid support, which may result in lengthy reaction times and, therefore, extended incubation times to allow the aptamer to effectively bind to its target molecule. In addition, when photoaptamers are used for the assay and depending on the material used as the solid support, the solid support may tend to disperse or absorb the light used to achieve the covalent bond formation between the photoaptamer and its target molecule. Furthermore, depending on the method used, detection of the target molecule bound to the aptamer may be inaccurate because the surface of the solid support may also be exposed to and affected by any labeling agent used. Finally, immobilization of the aptamer on the solid support typically includes an aptamer preparation step (i.e., immobilization) before the aptamer is exposed to the sample, and this preparation step may affect the activity or functionality of the aptamer.

Aptamer assays have also been described that allow aptamers to capture their targets in solution, followed by a separation step designed to remove specific components of the aptamer-target mixture prior to detection (see U.S. Patent Application Publication 20090042206, entitled "Multiplexed Analysis of Test Samples"). The aptamer assay method allows for the detection and quantification of non-nucleic acid targets (such as protein targets) in a test sample by detecting and quantifying nucleic acids (i.e., aptamers). The method generates nucleic acid surrogate (i.e., aptamers) to detect and quantify non-nucleic acid targets, thereby allowing many nucleic acid techniques, including amplification, to be applied to a wider range of desired targets, including protein targets.

Aptamers can be constructed to facilitate separation of assay components from aptamer-biomarker complexes (or photoaptamer-biomarker covalent complexes), as well as to allow separation of aptamers for detection and/or quantification. In one embodiment, these constructs can include cleavable or releasable elements within the aptamer sequence. In other embodiments, additional functionality can be introduced into the aptamer, such as a labeled or detectable component, a spacer component, or a specific binding tag or immobilization element. For example, the aptamer can include a tag, a label, or a spacer component separating a label from a cleavable moiety that is attached to the aptamer via a cleavable moiety. In one embodiment, the cleavable element is a photocleavable linker. The photocleavable linker can be attached to a biotin moiety and a spacer segment, can include an NHS group for amine derivatization, and can be used to introduce a biotin group into the aptamer, thereby allowing the aptamer to be released later in the assay method.

Homogeneous assays, using all assay components in solution, do not require separation of sample and reagents before signal detection. These methods are rapid and easy to use. These methods generate signals based on molecular capture or binding reagents that react with their specific targets. For pancreatic cancer, the molecular capture reagents are aptamers or antibodies, and the specific targets are pancreatic cancer biomarkers listed in Table 1, column 2.

In one embodiment, a signal generation method utilizes the change in anisotropy signal caused by the interaction of a fluorophore-labeled capture agent with its specific biomarker target. When the labeled capture agent reacts with its target, the increased molecular weight causes the rotational motion of the fluorophore attached to the complex to slow down, thereby changing the anisotropy value. By monitoring the change in anisotropy, the binding event can be used to quantitatively measure the biomarker in solution. Other methods include fluorescence polarization assays, molecular beacon methods, time-resolved fluorescence quenching, chemiluminescence, fluorescence resonance energy transfer, and others.

An exemplary solution-based aptamer assay that can be used to detect a biomarker value corresponding to a biomarker in a biological sample comprises the following steps: (a) preparing a mixture by contacting the biological sample with an aptamer comprising a first tag and having a specific affinity for the biomarker, wherein an aptamer affinity complex is formed when the biomarker is present in the sample; (b) exposing the mixture to a first solid support comprising a first capture element and allowing the first tag to bind to the first capture element; and (c) removing any components of the mixture that are not bound to the first solid support. ; (d) attaching a second tag to the biomarker component of the aptamer affinity complex; (e) releasing the aptamer affinity complex from the first solid support; (f) exposing the released aptamer affinity complex to a second solid support comprising a second capture element and allowing the second tag to bind to the second capture element; (g) removing any uncomplexed aptamers from the mixture by separating them from the aptamer affinity complex; (h) eluting aptamers from the solid support; and (i) detecting the biomarker by detecting the aptamer component of the aptamer affinity complex.

Any method known in the art can be used to detect biomarker values by detecting the aptamer component of the aptamer affinity complex. Many different detection methods can be used to detect the aptamer component of the affinity complex, for example, hybridization assays, mass spectrometry, or qPCR. In some embodiments, nucleic acid sequencing methods can be used to detect the aptamer component of the aptamer affinity complex, thereby detecting the biomarker value. Simply put, a test sample can be subjected to any type of nucleic acid sequencing method to identify and quantify the sequence or sequences of one or more aptamers present in the test sample. In some embodiments, the sequence includes the entire aptamer molecule or any portion of the molecule that can be used to uniquely identify the molecule. In other embodiments, the identification sequence is a specific sequence added to the aptamer; such sequences are often referred to as "tags," "barcodes," or "zip codes." In some embodiments, the sequencing method includes an enzymatic step to amplify the aptamer sequence, or to convert any type of nucleic acid (including RNA and DNA containing chemical modifications at any position) into any other type of nucleic acid suitable for sequencing.

In some embodiments, the sequencing method comprises one or more cloning steps. In other embodiments, the sequencing method comprises a direct sequencing method without cloning.

In some embodiments, the sequencing method comprises a direct approach with specific primers targeting one or more aptamers in the test sample. In other embodiments, the sequencing method comprises a shotgun approach targeting all aptamers in the test sample.

In some embodiments, the sequencing method includes an enzymatic step to amplify the molecules targeted for sequencing. In other embodiments, the sequencing method directly sequences a single molecule. An exemplary nucleic acid sequencing-based method that can be used to detect biomarker values corresponding to biomarkers in a biological sample includes the following steps: (a) converting chemically modified nucleotides into unmodified nucleic acids by an enzymatic step; (b) sequencing the unmodified nucleic acids obtained by shotgun sequencing using a massively parallel sequencing platform, such as the 454 Sequencing System (454 Life Sciences/Roche), the Illumina Sequencing System (Illumina), the ABI SOLiD Sequencing System (Applied Biosystems), the HeliScope Single Molecule Sequencer (Helicos Biosciences), or the Pacific Biosciences Real-Time Single Molecule Sequencing System (Pacific BioSciences) or the Polonator G Sequencing System (Dover Systems); and (c) identifying and quantifying the aptamers present in the mixture by specific sequencing and sequencing counts.

Determination of biomarker values using immunoassays

Immunoassays are based on the reaction of antibodies with their corresponding targets or analytes, and depending on the specific assay format, can detect analytes in samples. In order to improve the specificity and sensitivity of immunoreactivity-based assays, monoclonal antibodies are typically used due to their specific epitope recognition. Polyclonal antibodies have been successfully used in various immunoassays due to their increased target affinity compared to monoclonal antibodies. Immunoassays have been designed for use with a wide range of biological sample matrices. Immunoassay formats have been designed to provide qualitative, semi-quantitative, and quantitative results.

Quantitative results are generated by using a standard curve generated using known concentrations of the specific analyte to be detected. The response or signal from an unknown sample is plotted against the standard curve and the amount or value corresponding to the target in the unknown sample is determined.

Many immunoassay formats have been designed. ELISA or EIA can quantitatively detect analytes. This method relies on the attachment of a label to the analyte or antibody, and the label component directly or indirectly includes an enzyme. The ELISA test can be designed to detect analytes directly, indirectly, competitively, or sandwich. Other methods rely on labels, such as radioisotopes (I125) or fluorescence. Other techniques include, for example, agglutination reaction, turbidimetry, turbidimetry, Western blotting, immunoprecipitation, immunocytochemistry, immunohistochemistry, flow cytometry, Luminex assays, etc. (see ImmunoAssay: A Practical Guide, edited by Brian Law, published by Taylor & Francis, Ltd., 2005 edition).

Exemplary assay formats include enzyme-linked immunosorbent assay (ELISA), radioimmunoassay, fluorescence, chemiluminescence, and fluorescence resonance energy transfer (FRET) or time-resolved-FRET (TR-FRET) immunoassays. Examples of methods for detecting biomarkers include biomarker immunoprecipitation followed by quantitative methods that allow for the resolution of size and peptide levels, such as gel electrophoresis, capillary electrophoresis, planar electrochromatography, and the like.

The method for detecting and/or quantitative detectable label or signal generating material depends on the property of the label. The reaction product catalyzed by a suitable enzyme (wherein the detectable label is an enzyme, see above) can be but is not limited to fluorescence, luminescence or radioactivity, or they can absorb visible light or ultraviolet light. The example of the detector suitable for detecting such detectable label includes but is not limited to X-ray photograph, radioactivity counter, scintillation counter, spectrophotometer, colorimeter, fluorometer, luminometer and densitometer.

Any detection method can be performed by any means that allows for appropriate preparation, processing and analysis of the reaction. This can be performed, for example, in a multi-well assay plate (such as a 96-well or 384-well plate) or using any suitable array or microarray. Liquid reservoirs of various reagents can be prepared manually or automatically, and all subsequent pipetting, dilutions, mixing, distribution, washing, incubation, sample reading, data collection and analysis can be automated using commercially available analytical software, robotics and detection instruments that can detect detectable labels.

Determining biomarker values using gene expression profiling

Measuring mRNA in a biological sample can be used as a surrogate for detecting the corresponding protein levels in the biological sample.Thus, any biomarker or panel of biomarkers described herein can also be detected by detecting the appropriate RNA.

mRNA expression levels are measured by reverse transcription quantitative polymerase chain reaction (RT-PCR followed by qPCR). RT-PCR is used to generate cDNA from mRNA. cDNA can be used in qPCR assays to generate fluorescence as the DNA amplification process progresses. By comparing with a standard curve, qPCR can generate absolute measurements, such as the number of mRNA copies per cell. Northern blots, microarrays, Invader assays, and RT-PCR combined with capillary electrophoresis have all been used to measure mRNA expression levels in samples. See Gene Expression Profiling: Methods and Protocols, Richard A. Shimkets, editor, Humana Press, 2004.

MiRNA molecules are small RNA molecules that do not encode but can regulate gene expression. Any method suitable for measuring mRNA expression levels can be applied to the corresponding miRNA. Recently, many laboratories have investigated the use of miRNA as disease biomarkers. Many diseases involve extensive transcriptional regulation, and it is not surprising that miRNAs can be found as biomarkers. The correlation between miRNA concentration and disease is generally not as clear as the correlation between protein levels and disease, but the value of miRNA biomarkers can be important. Of course, with the differential expression of any RNA during disease, the development of in vitro diagnostic products faces challenges, including the need for miRNAs to survive in diseased cells and be easily extracted for analysis, or for miRNAs to be released into the blood or other matrices, where they must survive long enough to be measured. Protein biomarkers have similar requirements, although many potential protein biomarkers are intentionally secreted in a paracrine manner at sites of disease and function during disease. Many potential protein biomarkers are designed to act outside the cells that synthesize those proteins.

Detection of molecular markers using in vivo molecular imaging techniques

Any of the biomarkers (see Table 1, column 2) can also be used in molecular imaging tests. For example, an imaging agent can be coupled to any of the biomarkers, which can be used to assist in the diagnosis of pancreatic cancer, monitor disease progression/remission or metastasis, monitor disease recurrence, or monitor response to treatment.

In vivo imaging techniques provide non-invasive methods for determining the status of a specific disease within an individual. For example, all parts of the body, or even the entire body, can be viewed as three-dimensional images, providing valuable information about the morphology and structure within the body. Such techniques can be combined with detection of the biomarkers described herein to provide information about an individual's cancer status, particularly pancreatic cancer status.

The use of in vivo molecular imaging techniques has been expanded by various advances in the technology. These advances include the development of new contrast agents or labels, such as radiolabels and/or fluorescent labels, which can provide strong signals inside the body; and the development of new and more powerful imaging technologies that can detect and analyze these signals from outside the body with sufficient sensitivity and accuracy to provide useful information. The contrast agents can be viewed in an appropriate imaging system to provide an image of the part or parts of the body where the contrast agent is located. The contrast agent can be bound or associated with a capture agent, such as an aptamer or antibody, for example, and/or bound or associated with a peptide or protein, or an oligonucleotide (for example, to detect gene expression), or a complex containing any of these substances and one or more macromolecules and/or other particulate forms.

Contrast agents are also characterized by radioactive atoms that can be used for imaging. For scintigraphic studies, suitable radioactive atoms include technetium-99m or iodine-123. Other readily detectable moieties include, for example, spin labels for magnetic resonance imaging (MRI), such as iodine-123, iodine-131, indium-111, fluorine-19, carbon-13, nitrogen-15, oxygen-17, gadolinium, manganese, or iron. Such labels are well known in the art and can be easily selected by those skilled in the art.

Standard imaging techniques include, but are not limited to, magnetic resonance imaging, computed tomography, positron emission tomography (PET), single photon emission computed tomography (SPECT), and the like. For diagnostic in vivo imaging, the type of detection equipment available is a major factor in selecting a given contrast agent, such as the specified radionuclide and specific biomarker for the target (protein, mRNA, etc.). The selected radionuclide typically has a decay type that is detectable by a specified type of equipment. In addition, when selecting a radionuclide for in vivo diagnosis, its half-life should be long enough to allow detection at the time of maximum uptake by the target tissue, but should also be short enough to minimize harmful radiation exposure to the host.

Exemplary imaging techniques include, but are not limited to, PET and SPECT, which are imaging techniques in which a radionuclide is administered systemically (synthetically) or locally to an individual. The absorption of the radiotracer is then measured over time and used to obtain information about the targeted tissue and biomarkers. Due to the high-energy (gamma-ray) emissions of the specific isotopes used and the sensitivity and sophistication of the equipment used to detect them, a two-dimensional distribution of radioactivity can be derived from outside the body.

Commonly used positron-emitting nuclides in PET include, for example, carbon-11, nitrogen-13, oxygen-15, and fluorine-18. Isotopes that decay by electron capture and/or gamma emission are used in SPECT and include, for example, iodine-123 and technetium-99m. An exemplary method for labeling amino acids with technetium-99m is to reduce the pertechnetate ion in the presence of a chelating precursor to form an unstable technetium-99m-precursor complex, which in turn reacts with the metal binding group of a bifunctionally modified chemotactic peptide to form a technetium-99m-chemotactic peptide conjugate.

Antibodies are commonly used in such in vivo imaging diagnostic methods. The preparation and use of antibodies for in vivo diagnosis are well known in the art. Labeled antibodies that specifically bind to any of the biomarkers listed in Table 1, column 2 can be injected into an individual suspected of having a certain type of cancer (e.g., pancreatic cancer), and the disease state of the individual can be diagnosed or evaluated based on the detectability of the specific biomarker used. As described above, the label used is selected based on the imaging modality used. The positioning of the label allows the spread of the cancer to be determined. The amount of the label within an organ or tissue also allows the presence or absence of cancer in that organ or tissue to be determined.

Similarly, aptamers can be used in such in vivo imaging diagnostic methods. For example, an aptamer used to identify a specific biomarker described in Table 1, column 2 (and therefore specifically binding to that specific biomarker) can be appropriately labeled and injected into an individual suspected of having pancreatic cancer, and the pancreatic cancer status of the individual can be diagnosed or evaluated based on the detectability of the specific biomarker. As described above, the label used is selected based on the imaging modality used. The localization of the label allows the spread of the cancer to be determined. The amount of label within an organ or tissue also allows the presence or absence of cancer in that organ or tissue to be determined. Aptamer-directed imaging agents can have unique and advantageous characteristics compared to other imaging agents with respect to tissue penetration, tissue distribution, kinetics, elimination, efficacy, and selectivity.

Such technology can also optionally be carried out with labeled oligonucleotides, for example, by imaging and detecting gene expression with antisense oligonucleotides. These methods are used for in situ hybridization, for example, using fluorescent molecules or radioactive nuclides as labels. Other methods for detecting gene expression include, for example, detecting the activity of reporter genes.

Another common type of imaging technique is optical imaging, in which fluorescent signals within a subject are detected by optical devices outside the subject's body. These signals can be due to actual fluorescence and/or bioluminescence. Improvements in the sensitivity of optical detection devices have increased the use of optical imaging in in vivo diagnostic assays.

In vivo molecular biomarker imaging has increasing applications, including in clinical trials, to more rapidly measure clinical efficacy, for example, in trials of new cancer therapies, and/or to avoid long-term placebo treatment for diseases such as multiple sclerosis, where such long-term treatment may be considered ethically problematic.

For a review of other techniques, see N. Blow, Nature Methods, 6, 465-469, 2009.

Determination of biomarker values using histological/cytological methods

For the evaluation of pancreatic cancer, many tissue samples can be used for histological or cytological methods. Sample selection depends on the location of the primary tumor and the site of metastasis. For example, tissue samples (clamp biopsy, fine needle aspiration (FNA) and/or brush cytology) collected during FNA guided by endoscopic retrograde cholangiopancreatography (ERCP) or endoscopic ultrasound (EUS) can be used for histology. Ascites or peritoneal lavage fluid or pancreatic juice can be used for cytology. Any biomarker identified herein that shows upregulation in pancreatic cancer individuals (see Table 1, column 6) can be used to stain histological samples as an indication of the disease.

In one embodiment, one or more capture reagents specific for the corresponding biomarkers are used for cytological evaluation of pancreatic cell samples and can include one or more of the following: collecting the cell sample, fixing the cell sample, dehydrating, clearing, fixing the cell sample on a microscope slide, permeabilizing the cell sample, analyte retrieval processing, staining, destaining, washing, blocking, and reacting with one or more capture reagents in a buffer solution. In another embodiment, the cell sample is generated from a cell block.

In another embodiment, one or more capture reagents specific for the corresponding biomarkers are used for histological evaluation of pancreatic tissue samples and can include one or more of the following: collecting the tissue sample, fixing the tissue sample, dehydrating, clearing, mounting the tissue sample on a microscope slide, permeabilizing the tissue sample, processing for analyte retrieval, staining, destaining, washing, blocking, rehydrating, and reacting with the one or more capture reagents in a buffered solution. In another embodiment, fixation and dehydration are replaced by freezing.

In another embodiment, one or more aptamers specific for the corresponding biomarkers are reacted with a histological or cytological sample and can serve as a nucleic acid target in a nucleic acid amplification method. Suitable nucleic acid amplification methods include, for example, PCR, q-beta replicase, rolling circle amplification, strand displacement, helicase-dependent amplification, loop-mediated isothermal amplification, ligase chain reaction, and restriction- and circularization-assisted rolling circle amplification.

In one embodiment, one or more capture reagents specific for the corresponding biomarker for histological or cytological evaluation are mixed in a buffer solution, which may contain any of the following components: blocking materials, competitors, detergents, stabilizers, carrier nucleic acids, polyanionic materials, etc.

A "cytology protocol" typically includes sample collection, sample fixation, sample immobilization, and staining. "Cell preparation" may include several processing steps after sample collection, including staining of prepared cells using one or more slow-dissociation-rate aptamers.

Sample collection can include placing the sample directly into an untreated transport container, placing the sample into a transport container containing some type of medium, or placing the sample directly onto a slide (fixing) without any processing or fixing.

Sample fixation can be improved by applying a portion of the collected sample to a slide treated with polylysine, gelatin, or silane. Slides can be prepared by coating a thin, uniform layer of cells on the slide. Care is usually taken to minimize mechanical distortion and drying artifacts. Liquid samples can be processed using the cell block method. Alternatively, liquid samples can be mixed with a fixative solution in a 1:1 ratio at room temperature for approximately 10 minutes.

Cell blocks can be prepared from residual effusions, sputum, urine sediment, gastrointestinal fluids, cell scrapings, or fine needle aspirates. Cells are concentrated or compacted by centrifugation or membrane filtration. Many methods for preparing cell blocks have been developed. Representative methods include fixed precipitation, bacterial agar, or membrane filtration methods. In the fixed precipitation method, the cell pellet is mixed with a fixative such as Bouins solution, picric acid, or buffered formalin, and the mixture is then centrifuged to precipitate the fixed cells. The supernatant is removed and the cell pellet is dried as completely as possible. The pellet is collected and wrapped in lens tissue and then placed in a tissue cassette. The tissue cassette is placed in a jar containing another fixative and processed as a tissue sample. The agar method is very similar to the above method, except that the pellet is removed and dried on a paper towel, then cut in half. The cut surface is placed in a drop of molten agar on a glass slide, and the pellet is then coated with agar, ensuring that no bubbles form in the agar. Allow the agar to harden, then remove any excess agar. Place in a tissue cassette to complete tissue processing. Alternatively, the mass can be suspended directly in 2% liquid agar at 65°C and the sample centrifuged. Allow the agar cell mass to solidify at 4°C for 1 hour. The solid agar can be removed from the centrifuge tube and cut in half. Wrap the agar in filter paper and place in a tissue cassette. From this point on, the processing is the same as the above method. Membrane filtration can be used instead of centrifugation in any of these methods. Any of these methods can be used to produce a "cell block sample."

Cell blocks can be prepared using specialized resins, including Lowicryl resin, LR White, LR Gold, Unicryl, and MonoStep. These resins have low viscosity and can be polymerized at low temperatures and with ultraviolet (UV) light. The embedding method relies on gradually cooling the sample during dehydration, transferring the sample to the resin, and finally polymerizing the cell block at a low temperature using an appropriate UV wavelength.

Cell block sections can be stained with hematoxylin-eosin for examination of cell morphology, while other sections are examined for specific markers.

Whether the method is cytological or histological, the sample can be fixed before further processing to prevent degradation. This method is called "fixation" and describes a number of materials and methods that can be used interchangeably. The sample fixation protocol and reagents are optimally selected empirically based on the target to be detected and the specific cell/tissue type to be analyzed. Sample fixation relies on reagents such as ethanol, polyethylene glycol, methanol, formalin, or isopropanol. The sample should be fixed as soon as possible after collection and attachment to the slide. However, the selected fixative can introduce structural changes in various molecular targets, making subsequent detection more difficult. The fixation and immobilization methods and their sequence can change the appearance of the cells, and these changes must be anticipated and recognized by the cytographer. Fixatives can cause certain cell types to shrink and result in the appearance of granules or a network of cytoplasm. Many fixatives work by cross-linking cellular components. This can destroy or alter specific epitopes, create new epitopes, cause molecular associations, and reduce membrane permeability. Formalin fixation is one of the most commonly used cytological/histological methods. Formalin forms methyl bridges between adjacent proteins or within proteins. Precipitation or coagulation can also be used for fixation, with ethanol being a common method. A combination of cross-linking and precipitation can also be used for fixation. Strong fixation methods are best for preserving morphological information, while weaker fixation methods are best for preserving molecular targets.

A typical fixative is 50% absolute ethanol, 2 mM polyethylene glycol (PEG), 1.85% formaldehyde. Variations of this formulation include ethanol (50%-95%), methanol (20%-50%), and formalin (formaldehyde) alone. Another commonly used fixative is 2% PEG 1500, 50% ethanol, and 3% methanol. The slides are placed in the fixative at room temperature for approximately 10-15 minutes, then removed and dried. Once the slides are fixed, they can be rinsed with a buffer solution such as PBS.

Many dyes can be used to differentially highlight and contrast or "stain" cellular, subcellular and tissue features or morphological structures. Hematoylin is used to stain nuclei blue or black. Tangerine G-6 and Eosin Azure both stain the cytoplasm. Tangerine G stains cells containing keratin and glycogen yellow. Eosin Y is used to stain nucleoli, cilia, red blood cells and surface epithelial squamous cells. Romanowsky stain is used on air-dried slides and can be used to enhance replication and distinguish extracellular from intracytoplasmic material.

Staining methods can include treatments that increase the permeability of cells to staining. Treatment of cells with detergents can be used to increase permeability. To increase cell and tissue permeability, fixed samples can be further treated with solvents, saponins, or nonionic detergents. Enzymatic digestion can also improve the accessibility of specific targets in tissue samples.

After staining, the sample is dehydrated using a series of alcohol rinses with increasing alcohol concentrations. The final wash uses xylene or a xylene substitute such as citrus terpenes, which has a refractive index close to that of the coverslip applied to the slide. This final step is called clearing. Once the sample is dehydrated and cleared, a mounting medium is applied. The selected mounting medium has a refractive index close to that of glass and allows the coverslip to adhere to the slide. It also inhibits further drying, shrinking, or fading of the cell sample.

Regardless of the stain or treatment used, final evaluation of the pancreatic cytology specimen is performed by some type of microscopy to allow visual observation of morphology and determination of the presence or absence of markers. Exemplary microscopy methods include bright field microscopy, phase contrast microscopy, fluorescence microscopy, and differential interference contrast microscopy.

If secondary testing of the sample is required after examination, the coverslip can be removed and the slide destained. Destaining involves using the original solvent system used to stain the slide, without the dye, and reversing the original staining procedure. Destaining can also be accomplished by soaking the slide in an acid-alcohol solution until the cells are colorless. Once colorless, the slide is rinsed thoroughly in a water bath and subjected to a secondary staining procedure.

In addition, by using specific molecular reagents, such as antibodies or nucleic acid probes or aptamers, specific molecular differentiation can be combined with cell morphological analysis. This improves the accuracy of diagnostic cytology. Microdissection can be used to separate subsets of cells for additional evaluation, particularly for genetic evaluation of abnormal chromosomes, gene expression or mutations.

Preparation of tissue samples for histological evaluation includes fixation, dehydration, infiltration, embedding and sectioning. The fixing agents used for histology are very similar or identical to those used for cytology, and have the same problem of maintaining morphological characteristics at the expense of molecules such as individual proteins. If the tissue sample is not fixed and dehydrated but is instead frozen and then sectioned when frozen, time can be saved. This is a gentler processing procedure, and more individual markers can be retained. However, freezing is unacceptable for the long-term preservation of tissue samples because the introduction of ice crystals causes the loss of subcellular information. The ice in the frozen tissue sample also hinders the sectioning process to produce extremely thin slices, and therefore can lose some microscope resolution and the image of subcellular structure. Except formalin fixation, osmium tetroxide is also used for fixing and staining phospholipids (membranes).

Dehydration of the tissue is accomplished by continuous washing with increasing concentrations of alcohol. Transparency uses materials that are miscible with alcohol and embedding materials and includes a step-by-step process starting with 50:50 alcohol: clarifier to 100% clarifier (xylene or xylene substitute). Infiltration involves incubating the tissue with a liquid embedding medium (warm wax, nitrocellulose solution), first with a 50:50 embedding medium: clarifier and then with 100% embedding medium. Embedding is accomplished by placing the tissue in a mold or cassette and filling it with a molten embedding medium such as wax, agar or gelatin. The embedding medium is allowed to harden. The hardened tissue sample is then cut into thin sections for staining and subsequent examination.

Prior to staining, tissue sections are dewaxed and rehydrated. Sections are dewaxed with xylene, which may be replaced one or more times, and rehydrated by successive washes in decreasing concentrations of alcohol. Prior to dewaxing, tissue sections may be heat-fixed on glass slides at approximately 80°C for approximately 20 minutes.

Laser capture microdissection allows the isolation of subsets of cells from tissue sections for further analysis.

In cytology, tissue sections or slices can be stained with various stains to enhance the observation of microscopic features. Many commercially available stains can be used to enhance or identify specific features.

To further increase the interaction of molecular reagents with cytological/histological samples, a number of "analyte retrieval" techniques have been developed. The first such technique uses high-temperature heating of fixed samples. This method is also known as heat-induced epitope retrieval or HIER. Many heating techniques have been used, including steam heating, microwaves, high-pressure steam, water baths, and pressure cooking or combinations of these heating methods. Analyte retrieval solutions include, for example, water, citrate, and normal saline buffer. The key to analyte retrieval is the time at high temperature, but lower temperatures for longer times have also been successfully used. Another key to analyte retrieval is the pH of the heated solution. It has been found that low pH provides the best immunostaining, but also produces background that often requires the use of a second tissue section as a negative control. Regardless of the buffer composition, the use of a high pH solution generally yields the most consistent benefits (increased immunostaining without increasing background). Analyte retrieval methods for specific targets are empirically optimized based on the variables of target, time, pH, and buffer composition used for heating. The use of microwave analyte retrieval methods allows for sequential staining of different targets with antibody reagents. However, the time required to obtain the antibody-enzyme complex between staining steps has also been shown to degrade cell membrane analytes. Microwave heating methods have also improved in situ hybridization methods.

To begin the analyte retrieval process, the sections are first dewaxed and hydrated. The slides are then placed in 10mM sodium citrate buffer pH 6.0 in a dish or jar. A representative procedure uses an 1100W microwave, microwaved at 100% power for 2 minutes, then microwaved at 20% power for 18 minutes after ensuring the slides remain covered in liquid. The slides are then cooled in an open container and subsequently rinsed with distilled water. HIER can be used in combination with enzymatic digestion to improve the reactivity of the target to immunochemical reagents.

One such enzymatic digestion protocol uses Proteinase K. Proteinase K is prepared at a concentration of 20 μg/ml in 50 mM Tris base, 1 mM EDTA, 0.5% Triton X-100, pH 8.0 buffer. The method first involves dewaxing the sections in two changes of xylene for 5 minutes each. The samples are then hydrated in two changes of 100% ethanol for 3 minutes each, and in 95% and 80% ethanol for 1 minute each, followed by rinsing in distilled water. The sections are covered with Proteinase K working solution and incubated at 37°C in a humidified chamber for 10-20 minutes (the optimal incubation time can vary depending on the tissue type and degree of fixation). The sections are cooled at room temperature for 10 minutes and then rinsed twice for 2 minutes in PBS Tween 20. If necessary, the sections can be blocked to eliminate potential interference from endogenous compounds and enzymes. The sections are then incubated with the primary antibody appropriately diluted in primary antibody dilution buffer for 1 hour at room temperature or overnight at 4°C. The sections were then rinsed two times for 2 minutes with PBS Tween 20. If required for a specific application, additional blocking could be performed, followed by three more rinses for 2 minutes with PBS Tween 20 before finalizing the immunostaining protocol.

Brief treatment with 1% SDS at room temperature has also been shown to improve immunohistochemical staining. The analyte retrieval method has been applied to slide-mounted sections as well as free-floating sections. Another treatment option is to place the slide in a jar containing citric acid and 0.1% Nonident P40 at pH 6.0 and heat to 95°C. The slide is then washed with a buffer solution such as PBS.

For immunological staining of tissues, nonspecific binding of antibodies to tissue proteins can be blocked by immersing the sections in a protein solution such as serum or nonfat dry milk.

Blocking reactions may involve reducing endogenous biotin levels; eliminating endogenous charge effects; inactivating endogenous nucleases; and/or inactivating endogenous enzymes such as peroxidases and alkaline phosphatase. Endogenous nucleases can be inactivated by degradation with proteinase K; heat treatment; use of chelating agents such as EDTA or EGTA; introduction of vector DNA or RNA; treatment with chaotropic agents such as urea, thiourea, guanidine hydrochloride, guanidine thiocyanate, lithium perchlorate, or diethyl pyrocarbonate. Alkaline phosphatase can be inactivated by treatment with 0.1N HCl for 5 minutes at room temperature or with 1 mM levamisole. Peroxidase activity can be eliminated by treatment with 0.03% hydrogen peroxide. Endogenous biotin can be blocked by immersing slides or sections in an avidin solution (streptavidin, which can replace neutravidin) for at least 15 minutes at room temperature. The slides or sections are then washed in buffer for at least 10 minutes. This step can be repeated at least three times. The slide or section is then immersed in a biotin solution for 10 minutes. This step can be repeated at least three times, using fresh biotin solution each time. The buffer wash procedure is repeated. Blocking protocols should be minimized to prevent damage to the cell or tissue structure or target or targets of interest, but one or more such protocols can be combined to "block" the slide or section prior to reaction with one or more slow-off-rate aptamers. See Basic Medical Histology: the Biology of Cells, Tissues and Organs, authored by Richard G. Kessel, Oxford University Press, 1998.

Determination of biomarker values using mass spectrometry

Many mass spectrometer configurations can be used to detect biomarker values. Several types of mass spectrometers are available or can be manufactured with various configurations. Typically, a mass spectrometer has the following main components: a sample inlet, an ion source, a mass analyzer, a detector, a vacuum system, and an instrument control and data system. The differences in the sample inlet, ion source, and mass analyzer often define the type of instrument and its capabilities. For example, the inlet can be a capillary column liquid chromatography source, or it can be a direct probe or stage, such as used in matrix-assisted laser desorption ionization. Common ion sources include, for example, electrospray, including nanospray and microspray, or matrix-assisted laser desorption ionization. Common mass analyzers include quadrupole mass filters, ion trap mass analyzers, and time-of-flight mass analyzers. Other mass spectrometry methods are well known in the art (see Burlingame et al. Anal. Chem. 70:647R-716R (1998); Kinter and Sherman, New York (2000)).

Protein biomarkers and biomarker values can be detected and measured by any of the following: electrospray ionization mass spectrometry (ESI-MS), ESI-MS/MS, ESI-MS/(MS)n, matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF-MS), surface-enhanced laser desorption/ionization time-of-flight mass spectrometry (SELDI-TOF-MS), surface desorption/ionization on silicon (DIOS), secondary ion mass spectrometry (SIMS), quadrupole time-of-flight (Q-TOF), tandem time-of-flight (TOF/TOF) technology called ultraflex III TOF/TOF, atmospheric pressure chemical ionization mass spectrometry (APCI-MS), APCI-MS/MS, APCI-(MS)N, atmospheric pressure photoionization mass spectrometry (APPI-MS), APPI-MS/MS and APPI-(MS)N, quadrupole mass spectrometry, Fourier transform mass spectrometry (FTMS), quantitative mass spectrometry, and ion trap mass spectrometry.

Sample preparation strategies are used to label and enrich samples prior to mass spectrometry characterization of protein biomarkers and determination of biomarker values. Labeling methods include, but are not limited to, isobaric tagging for relative and absolute quantification (iTRAQ) and stable isotope labeling with amino acids in cell culture (SILAC). Capture reagents used to selectively enrich candidate biomarker protein samples prior to mass spectrometry analysis include, but are not limited to, aptamers, antibodies, nucleic acid probes, chimeras, small molecules, F(ab')2 fragments, single-chain antibody fragments, Fv fragments, single-chain Fv fragments, nucleic acids, lectins, ligand-binding receptors, affybodies, nanobodies, ankyrins, domain antibodies, polymers printed on variable antibody scaffolds (e.g., diabodies), high-affinity multimers, peptide mimetics, peptidomimetics, peptide nucleic acids, threose nucleic acids, hormone receptors, cytokine receptors, and synthetic receptors, as well as modifications and fragments of these substances.

Determining biomarker values using proximity ligation technology

Proximity ligation technology can be used to determine biomarker values. Briefly, a test sample is contacted with a pair of affinity probes, which can be a pair of antibodies or a pair of aptamers, with each member of the pair extending an oligonucleotide. The targets of the affinity probe pair can be two different determinants on a protein or a determinant on each of two different proteins, which can exist as homologous or heterologous multimeric complexes. When the probes bind to the target determinants, the free ends of the oligonucleotide extensions are close enough to hybridize together. Hybridization of the oligonucleotide extensions is promoted by common linker oligonucleotides, which are used to link the oligonucleotide extensions together when they are positioned close enough. Once the oligonucleotide extensions of the probes hybridize, the extended ends are linked together by enzymatic DNA ligation.

In one embodiment, oligonucleotide is extended and comprises the primer site that is used for pcr amplification.In case oligonucleotide extension links together, oligonucleotide forms continuous dna sequence dna, by pcr amplification, it shows the information about the character and amount of target protein, and when target determinant is on two different albumen, about the information of protein-protein interaction.The adjacent position is connected and can provide the highly sensitive and specific mensuration of real-time protein concentration and interaction information by using real-time PCR.Do not have the corresponding oligonucleotide extension of contiguous in conjunction with the probe of the determinant of paying close attention to, and can not be connected or pcr amplification, cause not having signal to produce.

The aforementioned assays allow for the detection of biomarker values useful in methods for diagnosing pancreatic cancer, wherein the method comprises detecting at least N biomarker values in a biological sample from an individual, each of the at least N biomarker values corresponding to a biomarker selected from the group of biomarkers provided in Table 1, Column 2, wherein, as described in detail below, classification of the biomarker values indicates whether the individual has pancreatic cancer. While certain of the pancreatic cancer biomarkers described herein can be used individually to detect and diagnose pancreatic cancer, the methods described herein are also useful for grouping multiple subsets of pancreatic cancer biomarkers, each of which can be used as a panel of three or more biomarkers. Thus, various embodiments of the present application provide combinations comprising N biomarkers, wherein N is at least three biomarkers. In other embodiments, N is selected from any number between 2 and 65 biomarkers. It should be understood that N can be selected from any number within any of the aforementioned ranges, as well as similar but higher order ranges. According to any of the methods described herein, biomarker values can be detected and classified individually, or they can be detected and classified together, for example, in a multiplex assay format.

In another aspect, the present invention provides a method for detecting the absence of pancreatic cancer, the method comprising detecting at least N biomarker values in a biological sample from an individual, each of the at least N biomarker values corresponding to a biomarker selected from the group of biomarkers provided in Table 1, Column 2, wherein, as described in detail below, classification of the biomarker values indicates the absence of pancreatic cancer in the individual. While certain of the pancreatic cancer biomarkers described herein can be used alone to detect and diagnose the absence of pancreatic cancer, the methods described herein are also useful for grouping multiple subsets of pancreatic cancer biomarkers, each of which can be used as a panel of three or more biomarkers. Thus, various embodiments of the present application provide a combination comprising N biomarkers, wherein N is at least three biomarkers. In other embodiments, N is selected from any number between 2 and 65 biomarkers. It should be understood that N can be selected from any number within any of the aforementioned ranges, as well as similar but higher order ranges. According to any of the methods described herein, biomarker values can be detected and classified individually, or they can be detected and classified together, for example, in a multiplex assay format.

Biomarker classification and disease score calculation

The biomarker "signature" of a given diagnostic test comprises a collection of markers, each of which has different levels in the population of interest. In this regard, different levels can refer to different means of marker levels for individuals in two or more groups, or different variances in two or more groups, or a combination of both. In the simplest form of a diagnostic test, these markers can be used to assign an unknown sample from an individual to one of two groups: a disease group or a non-disease group. Assigning a sample to one of two or more groups is called classification, and the procedure used to achieve this assignment is called a classifier or classification method. Classification methods can also be referred to as scoring methods. There are many classification methods that can be used to construct a diagnostic classifier from a collection of biomarker values. Generally, classification methods are most easily performed using supervised learning techniques, in which a data set is collected using samples obtained from individuals in two (or more, for multiple classification states) different groups that one wishes to distinguish. Because the category (group or population) to which each sample belongs is known in advance for each sample, the classification method can be trained to obtain the desired classification response. Unsupervised learning techniques can also be used to generate diagnostic classifiers.

Common methods for developing diagnostic classifiers include decision trees; bagging, boosting, forests and random forests; rule inference based learning; Parzen Windows; linear models; logic; neural network methods; unsupervised clustering; K-means; hierarchical ascending/descending; semi-supervised learning; prototype methods; nearest neighbor sampling; kernel density estimation; support vector machine; hidden Markov model; Boltzmann Learning; and classifiers can be combined simply or in a way to minimize a specific objective function. For a review, see, e.g., Pattern Classification, R.O. Duda, et al., editors, John Wiley & Sons, 2nd edition, 2001; also see The Elements of Statistical Learning - Data Mining, Inference, and Prediction, T. Hastie, et al., editors, Springer Science+Business Media, LLC, 2nd edition, 2009; all of which are incorporated herein by reference in their entirety.

In order to produce a classifier with supervised learning technology, a sample set called training data is obtained. In the case of a diagnostic test, training data includes samples of different groups (categories) that will be assigned later from unknown samples. For example, samples collected from the individual of a control group and the individual of a specific disease group can form training data to develop a classifier that can classify unknown samples (or, more particularly, the individual from which the sample is derived) as suffering from the disease or without the disease. Developing a classifier from training data is known as training the classifier. The specific details of classifier training depend on the nature of supervised learning technology. As an example, the example of training a naive Bayesian classifier is described below (see, for example, Pattern Classification, R.O.Duda, et al., editors, John Wiley & Sons, 2nd edition, 2001; also see The Elements of Statistical Learning-Data Mining, Inference, and Prediction, T.Hastie, et al., editors, Springer Science+Business Media, LLC, 2nd edition, 2009).

Because there are typically many more potential biomarker values in the training set than samples, care must be taken to avoid overfitting. Overfitting occurs when the statistical model describes random error or noise rather than the underlying relationship. Overfitting can be avoided in various ways, including, for example, limiting the number of markers used in developing the classifier, assuming independence of marker responses, limiting the complexity of the underlying statistical model used, and ensuring that the underlying statistical model fits the data.

An illustrative example of developing a diagnostic test using a collection of biomarkers involves applying a naive Bayesian classifier, a simple probabilistic classifier based on Bayes' theorem, with strictly independent treatment of the biomarkers. Each biomarker is described by a class-dependent probability density function (pdf) for the RFU values or logRFU (relative fluorescence units) values measured in each class. The joint pdf for the collection of markers in a class is assumed to be the product of the individual class-dependence pdfs for each biomarker. Training a naive Bayesian classifier in this context means assigning parameters ("parameterization") to characterize the class-dependence pdfs. Any potential model for the class-dependence pdfs can be used, but the model should generally fit the data observed in the training set.

Specifically, the class-dependent probability of measuring a value x _i for biomarker i in a disease class is written as p( _xi | d), and the overall naive Bayesian probability of observing n markers with a value is written as where each x _i is the measured biomarker level expressed in RFU or log RFU. Class assignment for an unknown is facilitated by calculating the probability of having a measured disease compared to the probability of not having the disease (control) for the same measured value. The ratio of these probabilities is calculated from the class-dependent pdf by applying Bayes' theorem, i.e., where p(d) is the prevalence of the disease in the population suitable for testing. Taking the logarithm of both sides of this ratio and substituting the naive Bayesian class-dependent probability from above yields a value known as the log likelihood ratio, which simply represents the log likelihood of not having a particular disease compared to having the disease, and is essentially composed of the sum of the individual log likelihood ratios for n individual biomarkers. In its simplest form, an unknown sample (or, more specifically, the individual from whom the sample was derived) is classified as not having the disease if the ratio is greater than 0, and as having the disease if the ratio is less than 0.

In an exemplary embodiment, the class-dependent biomarker pdfs p( _xi |c) and p( _xi |d) are assumed to be normally distributed or log-normally distributed in the measured RFU values xi, i.e., and for _xi with similar expression. Parameterization of the model requires estimating two parameters of each class-dependent pdf from the training data, the mean μ and the variance δ ² . This can be achieved in many ways, including, for example, maximum likelihood estimation, least squares estimation, and any other method known to those skilled in the art. Substituting the normal distribution into the log-likelihood ratio defined above, the following expression is obtained:

Once the set of μ and δ ² has been defined for each pdf for each class from the training data, and the disease incidence in the population is determined, the Bayesian classifier is fully determined and can be used to classify unknown samples with measured values.

The performance of a naive Bayes classifier depends on the number and quality of biomarkers used to construct and train the classifier. As defined in Example 3 below, individual biomarkers are ranked according to their KS-distance (Kolmogorov-Smirnov). If the classifier performance metric is defined as the area under the receiver operating characteristic curve (AUC), a perfect classifier scores 1, while a random classifier scores an average of 0.5. The KS-distance between two sets A and B of size n and m is defined as the value Dn _,m = sup _x |FA _,n (x)-FB _,m (x)|, which is the maximum difference between the two empirical cumulative distribution functions (cdfs). The empirical cdf of a set A of n observations _Xi is defined as where is the indicator function, which is equal to 1 if _Xi < x and equal to 0 otherwise. By definition, this value lies between 0 and 1, where a KS-distance of 1 indicates that the empirical distributions do not overlap.

The addition of subsequent markers with good KS distances (e.g., >0.3) often improves classification performance if the subsequent markers are independent of the first marker. Using sensitivity plus specificity as the classifier score, a variant of the greedy algorithm will directly produce many high-scoring classifiers. (Greedy algorithms are those that apply a metaheuristic to the problem, making local optimization choices at each stage to find the overall optimum.)

The algorithm used here is described in detail in Example 4. Briefly, all single-analyte classifiers are generated from a table of potential biomarkers and added to a list. Next, all possible second analytes are added to each stored single-analyte classifier, and a predetermined number of the best-scoring pairs, for example, one thousand, are stored in a new list. This new list of the best two-marker classifiers is used to develop all possible three-marker classifiers, and the best one thousand are again stored. This process continues until the scores plateau or begin to deteriorate with the addition of additional markers. The high-scoring classifiers that remain after convergence can be evaluated for their expected performance for the intended application. For example, in one diagnostic application, a classifier with high sensitivity and moderate specificity may be more desirable than one with moderate sensitivity and high specificity. In another diagnostic application, a classifier with high specificity and moderate sensitivity may be more desirable. The desired performance level is typically selected based on the trade-off that must be made between the number of false positives and false negatives that can be tolerated for a particular diagnostic application. This trade-off typically depends on the medical consequences of false positive or false negative errors.

Various other techniques are known in the art and can be used to generate a number of potential classifiers from a list of biomarkers using a naive Bayesian classifier. In one embodiment, a so-called genetic algorithm can be used to combine different markers using the fitness score defined above. Genetic algorithms are particularly well suited for developing large and diverse populations of potential classifiers. In another embodiment, so-called ant colony optimization can be used to generate a collection of classifiers. Other strategies known in the art can also be used, including, for example, other evolutionary strategies as well as simulated annealing and other random search methods. Metaheuristic methods such as harmony search can also be used.

Exemplary embodiments use any number of the pancreatic cancer biomarkers listed in Table 1, column 2, in various combinations to create a diagnostic test for detecting pancreatic cancer (see Example 2 for a detailed description of how these biomarkers were identified). In one embodiment, a method for diagnosing pancreatic cancer uses a naive Bayesian classification method in conjunction with any number of the pancreatic cancer biomarkers listed in Table 1, column 2. In an illustrative example (Example 3), the simplest test for detecting pancreatic cancer from GI and normal control populations can be constructed using a single biomarker such as CTSB, which is differentially expressed in pancreatic cancer with a KS-distance of 0.52. Using the parameters μc _,i , σc _{,i, μd ,i,} and σd _,i for CTSB from Table 16 and the equation for log-likelihood described above, a diagnostic test with an AUC of 0.79 can be generated, as shown in Table 15. The ROC curve for this test is shown in Figure 2.

For example, the addition of biomarker C5a, which has a KS-distance of 0.40, significantly improved classifier performance to an AUC of 0.85. Note that the score of a classifier constructed from two biomarkers is not a simple sum of the KS-distances; when combining biomarkers, when the KS-distances are not additive, many weaker markers are used to achieve the same level of performance as a strong marker. For example, the addition of a third marker, C5, increased classifier performance to an AUC of 0.88. The addition of additional biomarkers such as CCL18, CSF1R, KLK7, ETHE1, C5-C6, KLK8, and VEGFA generated a series of pancreatic cancer tests, summarized in Table 15 and shown as a series of ROC curves in Figure 3. The classifier score as a function of the number of analytes used in classifier construction is shown in Figure 4. The AUC of this exemplary 10-marker classifier was 0.91.

The markers listed in Table 1, column 2, can be combined in many ways to generate a classifier for diagnosing pancreatic cancer. In some embodiments, the panel of biomarkers consists of different numbers of analytes, depending on the specific diagnostic performance criteria selected. For example, certain combinations of biomarkers may produce a more sensitive (or more specific) test than other combinations.

Once the panel is defined to include a specific set of biomarkers from Table 1, column 2, and a classifier is constructed from the training data set, the definition of the diagnostic test is complete. In one embodiment, the procedure for classifying an unknown sample is shown in FIG1A . In another embodiment, the procedure for classifying an unknown sample is shown in FIG1B . The biological sample is appropriately diluted and then subjected to one or more assays to generate relevant quantitative biomarker levels for classification. The measured biomarker levels serve as input to a classification method, which outputs a classification for the sample and an optional score reflecting the confidence of the class assignment.

Table 1 identifies 65 biomarkers that can be used to diagnose pancreatic cancer. This is surprisingly high compared to what is typically found in biomarker discovery attempts, likely due to the scale of the study, which encompassed over 800 proteins measured in hundreds of individual samples, in some cases at concentrations in the low femtomolar range. Presumably, the large number of biomarkers discovered reflects diverse biochemical pathways involved in tumor biology and the body's response to the presence of a tumor; each pathway and process involves numerous proteins. The results suggest that no single protein from a small set of proteins is uniquely informative about such complex processes; rather, multiple proteins are implicated in related processes, such as apoptosis or extracellular matrix repair.

Given the numerous biomarkers identified in this study, it is anticipated that a large number of high-performing classifiers could be derived that could be used in a variety of diagnostic approaches. To test this concept, tens of thousands of classifiers were evaluated using the biomarkers in Table 1. As described in Example 4, many subsets of the biomarkers shown in Table 1 can be combined to generate useful classifiers. For example, descriptions of classifiers containing one, two, and three biomarkers for detecting pancreatic cancer are provided. As described in Example 4, all classifiers constructed using the biomarkers in Table 1 performed significantly better than classifiers constructed using "no markers."

The performance of the classifiers obtained by randomly excluding some of the markers in Table 1 was also tested. Random exclusion produced smaller subsets from which classifiers were constructed. As described in Example 4, Part 3, the classifiers constructed from the random subsets of markers in Table 1 performed similarly to the best classifier constructed using the complete list of markers in Table 1.

The performance of a 10-marker classifier obtained by excluding the "best" individual markers from the 10-marker pool was also tested. As described in Example 4, Part 3, classifiers constructed without the "best" markers from Table 1 also performed well. Many subsets of the biomarkers listed in Table 1 performed near optimally, even after removing the top 15 markers listed in the table. This suggests that the performance characteristics of any particular classifier may not be due to some small core group of biomarkers, and that disease processes may affect many biochemical pathways, which alter the expression levels of many proteins.

The results of Example 4 suggest several possible conclusions: First, the identification of a large number of biomarkers allows them to be aggregated into a vast number of classifiers that provide similarly high performance. Second, classifiers can be constructed such that specific biomarkers can be substituted for other biomarkers in a manner that reflects the redundancy inherent in the complexity of the underlying disease process. That is, the information about the disease contributed by any individual biomarker identified in Table 1 overlaps with the information contributed by other biomarkers, and thus a specific biomarker or panel of biomarkers in Table 1 need not be included in any classifier.

An exemplary embodiment uses a naive Bayesian classifier constructed from the data in Table 16 to classify unknown samples. The procedure is illustrated in Figures 1A and 1B. In one embodiment, the biological sample is optionally diluted and subjected to a multiplex aptamer assay. The data from the assay is normalized and calibrated as described in Example 3, and the resulting biomarker levels are used as input to the Bayesian classification scheme. Log-likelihood ratios are calculated individually for each measured biomarker and then summed to produce a final classification score, which is also referred to as a diagnostic score. The resulting assignments and overall classification scores can be reported. Optionally, the individual log-likelihood risk factors calculated for each biomarker level can also be reported. Details of the classification score calculation are shown in Example 3.

Reagent test kit

Any combination of the biomarkers listed in Table 1, Column 2 (and additional biomedical information) can be detected using a suitable kit, such as a kit for performing the methods disclosed herein. In addition, any kit can contain one or more detectable labels described herein, such as fluorescent moieties, etc.

As further described herein, in one embodiment, the kit comprises: (a) one or more capture reagents (e.g., at least one aptamer or antibody) for detecting one or more biomarkers in a biological sample, wherein the biomarkers include any of the biomarkers listed in Table 1, column 2, and optionally (b) one or more software or computer program products for classifying an individual from whom the biological sample was obtained as having or not having pancreatic cancer, or determining the likelihood that the individual has pancreatic cancer. Alternatively, in addition to one or more computer program products, one or more instructions for manually performing the above steps can be provided.

The combination of a solid support with a corresponding capture reagent and signal-generating material is referred to herein as a "detection device" or "kit." The kit may also include instructions for using the device and reagents, processing samples, and analyzing data. In addition, the kit may be used in conjunction with a computer system or software to analyze and report the results of the biological sample analysis.

The kit may also contain one or more reagents (such as solubilization buffer, detergent, washing agent or buffer) to process the biological sample. Any kit described herein may also include, for example, buffer, blocking agent, mass spectrometry matrix material, antibody capture agent, positive control sample, negative control sample, software and information such as protocol, guidance and reference data.

In one aspect, the present invention provides a kit for analyzing pancreatic cancer status. The kit comprises PCR primers for one or more biomarkers selected from Table 1, column 2. The kit may also comprise instructions for use and correlation of the biomarkers with pancreatic cancer. The kit may also comprise a DNA array containing a complement of one or more biomarkers selected from Table 1, column 2, reagents and/or enzymes for amplifying or isolating sample DNA. The kit may also comprise reagents for real-time PCR, such as TaqMan probes and/or primers, and enzymes.

For example, a kit can include: (a) reagents comprising at least a capture reagent for quantifying one or more biomarkers in a test sample, wherein the biomarkers include the biomarkers listed in Table 1, column 2, or any other biomarker or biomarker panel described herein; and optionally, (b) one or more algorithms or computer programs for performing the following steps: comparing the amount of each biomarker quantified in the test sample to one or more predetermined cutoff values, assigning a score to each biomarker quantified based on the comparison, combining the assigned scores for each biomarker quantified to obtain a total score, comparing the total score to a predetermined score, and determining whether an individual has pancreatic cancer using the comparison. Alternatively, in addition to one or more algorithms or computer programs, one or more instructions for manually performing the above steps can be provided.

Computer methods and software

Once a biomarker or biomarker panel is selected, a method for diagnosing an individual may include the following steps: 1) collecting or otherwise obtaining a biological sample; 2) performing an analytical method to detect and measure the biomarker or biomarker panel in the biological sample; 3) performing any data normalization or standardization required by the method for collecting biomarker values; 4) calculating a marker score; 5) combining the marker scores to obtain an overall diagnostic score; and 6) reporting a diagnostic score for the individual. In this method, the diagnostic score can be a single numerical value determined from the sum of all marker calculations, which is compared to a preset threshold value indicating the presence or absence of a disease. Alternatively, the diagnostic score can be a series of bars, each representing a biomarker value, and the response pattern can be compared to a preset pattern to determine the presence or absence of a disease.

At least some embodiments of the methods described herein can be implemented using a computer. An example of a computer system 100 is shown in FIG6 . Referring to FIG6 , system 100 is shown to include hardware components electrically coupled via bus 108, including a processor 101, an input device 102, an output device 103, a storage device 104, a computer-readable storage medium reader 105a, a communication system 106, a processing acceleration (e.g., a DSP or a special-purpose processor) 107, and a memory 109. The computer-readable storage medium reader 105a is further coupled to a computer-readable storage medium 105b, which collectively represents remote, local, fixed, and/or removable storage devices plus storage media, memory, etc., to temporarily and/or permanently contain computer-readable information. This can include storage device 104, memory 109, and/or any other such storage-capable system 100 resources. System 100 also includes software components (shown currently located in working memory 191), including an operating system 192 and other code 193, such as programs, data, etc.

About Fig. 6, system 100 has extensive flexibility and configurability.Therefore, for example, single computer structure (single architecture) can be used to complete one or more servers, which can be further configured according to the scheme, scheme variation, expansion etc. of current expectation. However, it should be understood by those skilled in the art that implementation scheme can be better utilized according to more specific application requirements. For example, one or more system elements can be performed as sub-element (as in communication system 106) in system 100 parts. It is also possible to use customized hardware and/or specific elements to perform in hardware, software or hardware and software. In addition, although it is possible to use other computing devices such as network input/output device (not shown) to be connected, it should be understood that it is also possible to utilize wired, wireless, modem and/or other connections or multiple connections with other computing devices.

In one aspect, the system can include a database containing signatures of biomarkers characteristic of pancreatic cancer. The biomarker data (or biomarker information) can be used as input to a computer for use as part of a computer-implemented method. The biomarker data can include data described herein.

In one aspect, the system further includes one or more devices to provide input data to the one or more processors.

The system also includes a memory for storing a data set of hierarchical data elements.

In another aspect, the means for providing input data includes a detector to detect characteristics of the data element, such as a mass spectrometer or a gene chip reader.

The system may also include a database management system.User requests or queries may be formatted in an appropriate language understood by the database management system, which processes the query to extract relevant information from the database of training sets.

The system can be connected to a network that connects a network server and one or more clients. The network can be a local area network (LAN) or a wide area network (WAN) as known in the art. Preferably, the server includes the hardware required to run a computer program product (such as software) to access database data to process user requests.

Described system can comprise operating system (as UNIX or Linux), to carry out the order from database management system.On the one hand, operating system can be run on global communication network, as run on the Internet, and utilizes global communication network server to connect such network.

The system may include one or more devices comprising a graphical display interface including interface elements such as buttons, pull-down menus, scroll bars, and fields for text input, which are common elements of graphical user interfaces known in the art. Requests logged in the user interface may be transmitted to an application in the system for formatting to search for relevant information in one or more system databases. User login requests or queries may be formulated in any suitable database language.

The graphical user interface may be generated by graphical user interface code as part of an operating system and may be used to input data and/or display the inputted data. The results of the processed data may be displayed on the interface, printed on a printer in communication with the system, stored in a storage device, and/or uploaded to a network or provided in the form of a computer-readable medium.

The system can communicate with an input device to provide data about the data element to the system (e.g., expression value). In one aspect, the input device can include a gene expression profiling system, including, for example, a mass spectrometer, a gene chip, or an array reader.

Methods and apparatus for analyzing pancreatic cancer biomarker information according to various embodiments can be implemented in any suitable manner, such as using a computer program running on a computer system. Conventional computer systems comprising a processor and random access memory, such as a remotely accessible application server, network server, personal computer, or workstation, can be used. Other computer system components may include storage devices or information storage systems, such as mass storage systems, and user interfaces, such as conventional monitors, keyboards, and tracking devices. The computer system can be a stand-alone system or part of a computer network comprising a server and one or more databases.

The pancreatic cancer biomarker analysis system can provide comprehensive data analysis functions and operations, such as data collection, processing, analysis, reporting, and/or diagnosis. For example, in one embodiment, a computer system can execute a computer program that can receive, store, search, analyze, and report information related to pancreatic cancer biomarkers. The computer program can include multiple modules that perform various operations or operations, such as a processing module that processes raw data and generates supplemental data, and an analysis module that analyzes the raw data and supplemental data to generate a pancreatic cancer status and/or diagnosis. Diagnosing pancreatic cancer status can include generating or collecting any other information, including additional biomedical information, about the individual's condition relative to the disease, identifying whether further testing is necessary, or otherwise evaluating the individual's health status.

With respect to FIG. 7 , an example of a method utilizing a computer according to the principles of the disclosed embodiments can be seen. In FIG. 7 , a flow chart 3000 is shown. In block 3004 , biomarker information for an individual can be retrieved. The biomarker information can be retrieved from a computer database, for example, after testing a biological sample from the individual. The biomarker information can include biomarker values, each of which corresponds to one of at least N biomarkers selected from the group consisting of the biomarkers provided in column 2 of Table 1 , where N = 2-65. In block 3008 , the computer can be configured to classify each biomarker value. In block 3012 , a likelihood that the individual has pancreatic cancer can be determined based on the multiple classifications. This indication can be output to a display or other display device for viewing by a person. Thus, for example, the indication can be displayed on a computer monitor screen or other output device.

Now referring to FIG. 8 , an alternative method utilizing a computer according to another embodiment is illustrated by flowchart 3200. At block 3204, biomarker information for an individual can be retrieved using a computer. The biomarker information includes biomarker values corresponding to biomarkers selected from the group of biomarkers provided in column 2 of Table 1. At block 3208, the biomarker values can be classified by the computer. Furthermore, at block 3212, an indication can be made regarding the likelihood that the individual has pancreatic cancer based on the classification. This indication can be output to a display or other display device for viewing by a person. Thus, for example, the indication can be displayed on a computer monitor screen or other output device.

Some embodiments described herein may be implemented to include a computer program product. The computer program product may include a computer-readable medium having computer-readable program code embodied in the medium, such that an application program can be executed on a computer having a database.

As used herein, a "computer program product" means an organized set of instructions in the form of natural or programming language statements, embodied on a physical medium of any nature (e.g., written, electronic, magnetic, optical, or other nature), and usable with a computer or other automated data processing system. Such programming language statements, when executed by a computer or data processing system, cause the computer or data processing system to function in accordance with the specific content of the statements. Computer program products include, but are not limited to, source code and object code contained in a computer-readable medium and/or programs in tests or databases. In addition, a computer program product that allows a computing system or data processing device to function in a preselected manner may be provided in a variety of forms, including but not limited to original source code, assembly code, object code, machine language, encrypted or compressed forms of the foregoing, and any and all equivalents.

In one aspect, the present invention provides a computer program product that indicates the likelihood of pancreatic cancer. The computer program product includes a computer-readable medium containing program code, the program code executable by a processor of a computing device or system, the program code comprising: code for retrieving data attributed to a biological sample from an individual, wherein the data includes biomarker values, each of the biomarker values corresponding to one of at least N biomarkers in the biological sample selected from the group of biomarkers provided in Table 1, Col. 2, where N = 2-65; and code for executing a classification method that indicates the pancreatic cancer status of the individual as a function of the biomarker values.

In another aspect, the present invention provides a computer program product that indicates the likelihood of pancreatic cancer. The computer program product includes a computer-readable medium containing program code, the program code executable by a processor of a computing device or system, the program code comprising: code for retrieving data attributed to a biological sample from an individual, wherein the data includes biomarker values corresponding to biomarkers in the biological sample selected from the group of biomarkers provided in Table 1, Col. 2; and code for executing a classification method that indicates pancreatic cancer status for the individual as a function of the biomarker values.

Although various embodiments of the method or apparatus of the present invention have been described, it should be understood that the embodiments can be executed by code coupled to a computer, such as code on a computer or that can be logged by a computer. For example, software and databases can be used to perform many of the above methods. Therefore, in addition to the embodiments performed by hardware, it should also be noted that these embodiments can be implemented by using such products, which include computer-usable media with computer-readable program code contained therein, which allow the functions disclosed herein to be exercised. Therefore, it is expected that the embodiments can also be considered to be protected by this patent in the form of its program code, etc. In addition, the embodiments can be embodied as code stored in virtually any type of computer-readable memory, including but not limited to RAM, ROM, magnetic media, optical media or magneto-optical media. More generally, the embodiments can be implemented in software or hardware or any combination thereof, including but not limited to software running on a general-purpose processor, microcode, PLA or ASIC.

It is also contemplated that the embodiments described herein can be implemented as computer signals embodied in carrier waves, as well as signals transmitted via transmission media (e.g., electrical and optical signals). Thus, the various types of information described above can be formatted in a structure, such as a data structure, and transmitted as an electrical signal via a transmission medium, or stored on a computer-readable medium.

It should also be noted that many structures, materials and clauses listed herein can be listed as modes for performing functions or steps for performing functions. Therefore, it should be understood that such language is entitled to cover all of these structures, materials or clauses disclosed in this specification and their equivalents, including contents incorporated herein by reference.

The biomarker identification process, the use of the biomarkers disclosed herein, and various methods for determining biomarker values are described above in detail with respect to pancreatic cancer. However, the application of the process, the use of the identified biomarkers, and the methods for determining biomarker values can be applied to other specific types of cancer, cancer in general, any other disease or medical condition, or to identify individuals who may or may not benefit from adjunctive medical treatment. References to pancreatic cancer herein should be understood to include other types of cancer, cancer in general, or any other disease or medical condition, except when the context clearly indicates a specific outcome related to pancreatic cancer.

Example

The following examples are for illustrative purposes only and are not intended to limit the scope of this application, which is defined by the appended claims. All examples described herein were performed using conventional standard techniques known to those skilled in the art. Conventional molecular biology techniques described in the following examples can be performed as described in standard laboratory manuals, such as Sambrook et al., Molecular Cloning: A Laboratory Manual, 3rd ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y., (2001).

Example 1. Multiple aptamer analysis of samples

This example describes a multiplex aptamer assay used to analyze samples and controls to identify the biomarkers listed in Table 1, column 2 (see Figure 9) and to identify the cancer biomarkers listed in Table 19. For the pancreatic cancer, lung cancer, and mesothelioma studies, the multiplex assay used 823 aptamers, each unique to a specific target.

In this method, the pipette tip is changed with each solution addition.

Furthermore, unless otherwise stated, most solution transfers and wash additions used a 96-well head from a Beckman Biomek FxP. Unless otherwise stated, manual pipetting procedures used a 12-channel P200 Pipetteman (Rainin Instruments, LLC, Oakland, CA). A custom buffer, called SB17, was prepared in-house and contained 40 mM HEPES, 100 mM NaCl, 5 mM KCl, 5 mM MgCl2, 1 mM EDTA, pH 7.5. A custom buffer, called SB18, was prepared in-house and contained 40 mM HEPES, 100 mM NaCl, 5 mM KCl, 5 mM MgCl2, pH 7.5. All steps were performed at room temperature unless otherwise stated.

1. Preparation of Aptamer Stock Solution

For aptamers without a photocleavable biotin linker, custom aptamer stocks for 10%, 1%, and 0.03% plasma were prepared using the appropriate photocleavable biotinylated primers at 8x concentration in 1x SB17, 0.05% Tween-20, resulting in a primer concentration that is 3-fold that of the relevant aptamer. Primers hybridized to all or part of the corresponding aptamer.

Each of the three 8x aptamer solutions was diluted 1:4 into 1xSB17, 0.05% Tween-20 (1500 μL of the 8x stock solution was diluted into 4500 μL of 1xSB17, 0.05% Tween-20) to achieve a 2x concentration. Each diluted aptamer master mix was then distributed into four 2mL screw-capped test tubes, 1500 μL per test tube, and heated to 95°C for 5 minutes, followed by incubation at 37°C for 15 minutes. After incubation, the four 2mL test tubes corresponding to the specific aptamer master mix were combined into a reagent trough, and 55 μL of the 2x aptamer mixture (for all three mixtures) was manually pipetted into a 96-well Hybaid plate and sealed with foil. The end result was three 96-well foil-sealed Hybaid plates. The concentration of each aptamer was 0.5 nM.

2. Assay Sample Preparation

A frozen aliquot of 100% plasma stored at -80°C was placed in a 25°C water bath for 10 minutes. The thawed sample was placed on ice, vortexed gently (setting 4) for 8 seconds, and then placed back on ice.

At 4°C, 16 μL of sample was transferred to a 96-well Hybaid plate using a 50 μL 8-channel multichannel pipette to prepare a 20% sample solution, with each well containing 64 μL of the appropriate sample dilution (for plasma, 0.8x SB18, 0.05% Tween-20, 2 μM Z-block_2, 0.6 mM MgCl ₂ ). This plate was kept on ice until the next sample dilution step was started.

To begin sample and aptamer balance, the 20% sample plate was briefly centrifuged and placed on a Beckman FX, where it was mixed by pipetting up and down with a 96-well pipette. A 2% sample was then prepared by diluting 10 μL of the 20% sample into 90 μL of 1xSB17, 0.05% Tween-20. Then, 6 μL of the 2% sample obtained was diluted into 194 μL of 1xSB17, 0.05% Tween-20 to prepare a 0.06% sample plate. Dilution was performed on a Beckman Biomek FxP. After each transfer, the solution was mixed by pipetting up and down. Then, 3 sample dilution plates were transferred to their respective aptamer solutions by adding 55 μL of sample to 55 μL of the appropriate 2x aptamer mixture. Sample and aptamer solution were mixed on a robot by pipetting up and down.

3. Sample equilibrium binding

The sample/aptamer plate was sealed with foil and placed in a 37°C incubator for 3.5 hours before proceeding to the Catch 1 step.

4. Preparation of Catch 2 Bead Plates

A 5.5 mL aliquot of MyOne (Invitrogen Corp., Carlsbad, CA) streptavidin C1 beads (10 mg/mL) was washed twice with an equal volume of 20 mM NaOH (incubation for 5 minutes each wash), washed three times with an equal volume of 1x SB17, 0.05% Tween-20, and resuspended in 5.5 mL of 1x SB17, 0.05% Tween-20. 50 μL of this solution was manually pipetted into each well of a 96-well Hybaid plate using a 12-span multichannel pipette. The plate was then covered with foil and stored at 4°C for use in the assay.

5. Preparation of Catch 1 Bead Plates

Three 0.45 μm Millipore HV plates (Durapore membrane, Cat# MAHVN4550) were equilibrated with 100 μL of 1x SB17, 0.05% Tween-20 for at least 10 minutes. The equilibration buffer was then filtered through the plates, and 133.3 μL of a 7.5% streptavidin-agarose bead slurry (in 1x SB17, 0.05% Tween-20) was added to each well. To keep the streptavidin-agarose beads suspended as they were transferred to the filter plates, the bead solution was manually mixed 15 times using a 200 μL, 12-channel pipette. After dispensing the beads into the three filter plates, vacuum was applied to remove the bead supernatant. Finally, the beads were washed in the filter plate with 200 μL of 1× SB17, 0.05% Tween-20 and then resuspended in 200 μL of 1× SB17, 0.05% Tween- 20. The bottom of the filter plate was blotted and the plate was stored for assay.

6. Load Cytomat

The cytomat was loaded with all tips, plates, all reagents in wells (except the NHS-biotin reagent, which was prepared just prior to addition to the plate), 3 prepared Catch 1 filter plates, and 1 prepared MyOne plate.

7.Catch 1

After a 3.5 hour equilibration period, the sample/aptamer plate was removed from the incubator, centrifuged for approximately 1 minute, the foil removed, and placed on the Beckman Biomek FxP platform. The Beckman Biomek FxP program was started. Unless otherwise noted, all subsequent steps in Catch 1 were performed by the Beckman Biomek FxP robot. Within this program, vacuum was applied to the Catch 1 filter plate to remove the bead supernatant. 100 microliters of each of the 10%, 1% and 0.03% equilibrium binding reactions were added to their respective Catch 1 filter plates, and each plate was mixed for 10 minutes using an on-deck orbital shaker at 800 rpm.

Unbound solution was removed by vacuum filtration. The Catch 1 beads were washed with 190 μL of 100 μM biotin in 1x SB17, 0.05% Tween-20, followed by 190 μL of 1x SB17, 0.05% Tween-20 by dispensing the solution and immediately applying vacuum to filter the solution through the plate.

Then, 190 μL of 1× SB17, 0.05% Tween-20 was added to the Catch 1 plate. The plate was blotted dry using a blot station to remove droplets and then incubated at 800 rpm, 25° C. for 10 minutes using an orbital shaker.

The robot removes this wash solution by vacuum filtration and the bottom of the filter plate is blotted dry using a ready-to-use blot device to remove droplets.

8. Tagging

The NHS-PEO4-biotin aliquot was thawed at 37°C for 6 minutes and then diluted 1:100 with labeling buffer (SB17 0.05% Tween-20 at pH=7.25). The NHS-PEO4-biotin reagent was dissolved in anhydrous DMSO to a concentration of 100mM and stored frozen at -20°C. With the assistance of a robot, the diluted NHS-PEO4-biotin reagent was manually added to the ready-to-use tank, and the robot program was manually restarted to distribute 100 μL of NHS-PEO4-biotin to each well of each Catch 1 filter plate. This solution was incubated with Catch 1 beads on an orbital shaker at 800rpm for 5 minutes.

9. Kinetic Challenge and Photolysis

The labeling reaction was stopped by adding 150 μL of 20 mM glycine in 1x SB17, 0.05% Tween-20 to the Catch 1 plate while still containing the NHS tag. The plate was then incubated on an orbital shaker at 800 rpm for 1 minute. The NHS-tag/glycine solution was removed by vacuum filtration. 190 μL of 20 mM glycine (1x SB17, 0.05% Tween-20) was then added to each plate and incubated on an orbital shaker at 800 rpm for 1 minute before being removed by vacuum filtration.

190 μL of 1x SB17, 0.05% Tween-20 was added to each plate and removed by vacuum filtration.

The wells of the Catch 1 plate were then washed three times by adding 190 μL of 1x SB17, 0.05% Tween-20, placing the plate on an orbital shaker at 800 rpm for 1 minute, and then vacuum filtering. After the final wash, the plate was placed on top of a 1 mL deep-well plate and removed from the platform. The Catch 1 plate was centrifuged at 1000 rpm for 1 minute to remove as much extraneous volume as possible from the agarose beads prior to elution.

The plate was placed back onto the Beckman Biomek FxP and 85 μL of 1x SB17, 10 mM _DxSO4 in 0.05% Tween-20 was added to each well of the filter plate.

The filter plate was removed from the platform and placed on a Variomag Thermoshaker (Thermo Fisher Scientific, Inc., Waltham, MA) and irradiated under a BlackRay (Ted Pella, Inc., Redding, CA) light source while shaking at 800 rpm for 10 minutes.

The photocleaved solution was eluted sequentially from each Catch 1 plate into the same deep-well plate by first placing the 10% Catch 1 filter plate on top of a 1 mL deep-well plate and centrifuging at 1000 rpm for 1 minute. Then, the 1% and 0.03% Catch 1 plates were centrifuged sequentially into the same deep-well plate.

10. Catch 2 Bead Capture

A 1 mL deep well block containing the combined Catch 1 eluate was placed on the platform of the Beckman Biomek FxP for Catch 2.

The robot transferred all photocleavage eluates from the 1 mL deep-well plate to a Hybaid plate containing previously prepared Catch 2 MyOne magnetic beads (after removal of the MyOne buffer by magnetic separation).

The solution was incubated at 25°C with shaking at 1350 rpm for 5 minutes on a Variomag Thermoshaker (Thermo Fisher Scientific, Inc., Waltham, MA).

The robot transfers the plate to a ready-to-use magnetic separator. The plate is incubated on the magnet for 90 seconds, and then the supernatant is removed and discarded.

11. Wash with 30% glycerol at 37°C

The Catch 2 plate was moved to a ready-to-use incubator shaker, and 75 μL of 1x SB17, 0.05% Tween-20 was transferred to each well. The plate was mixed at 1350 rpm at 37°C for 1 minute to resuspend and warm the beads. 75 μL of 60% glycerol was transferred to each well of the Catch 2 plate at 37°C, and the plate was mixed at 1350 rpm at 37°C for another minute. The plate was robotically transferred to a 37°C magnetic separator, where it was incubated on a magnet for 2 minutes, after which the supernatant was robotically removed and discarded. These washes were repeated two more times.

After the third 30% glycerol wash was removed from the Catch 2 beads, 150 μL of 1x SB17, 0.05% Tween-20 was added to each well and incubated at 37°C with shaking at 1350 rpm for 1 minute before removal by magnetic separation on a 37°C magnet.

The Catch 2 beads were washed one final time with 150 μL of 1× SB19, 0.05% Tween-20 at 25° C. with shaking at 1350 rpm for 1 minute before magnetic separation.

12. Catch 2 Bead Elution and Neutralization

The aptamers were eluted from the Catch 2 beads by adding 105 μL of 100 mM CAPSO containing 1 M NaCl, 0.05% Tween-20 to each well. The beads were incubated with this solution for 5 minutes with shaking at 1350 rpm.

The Catch 2 plate was then placed on a magnetic separator for 90 seconds, and 90 μL of the eluate was transferred to a new 96-well plate containing 10 μL per well of 500 mM HCl, 500 mM HEPES, 0.05% Tween-20. After transfer, the solution was mechanically mixed by pipetting 90 μL up and down 5 times.

13. Hybridization

Beckman Biomek FxP transferred 20 μL of the neutralized Catch 2 eluate to a fresh Hybaid plate, and 5 μL of 10x Agilent Block containing a 10x spike hybridization control was added to each well. Then, 25 μL of 2xAgilent Hybridization Buffer was manually pipetted into each well of the plate containing the neutralized sample and blocking buffer, and the solution was mixed by slowly pipetting 25 μL up and down 15 times to avoid excessive foam formation. The plate was centrifuged at 1000 rpm for 1 minute.

A gasket slide was placed in an Agilent hybridization chamber, and 40 μL of each sample containing hybridization and blocking solution was manually pipetted into each gasket. An 8-channel variable multichannel pipette was used to minimize bubble formation. A custom Agilent microarray slide (Agilent Technologies, Inc., Santa Clara, CA) was then slowly lowered onto the gasket slide with the barcode facing upward (see the Agilent manual for detailed description).

The upper part of the hybridization chamber is placed on the slide/backing sandwich structure, and the clamping bracket is placed on the entire device. The devices are clamped by tightening the screws.

Visually inspect each slide/backing slide sandwich to ensure that solution bubbles can move freely within the sample. If bubbles are not free, tap the hybridization chamber assembly to release any bubbles near the gasket.

The assembled hybridization chamber was incubated in an Agilent hybridization oven at 60°C with rotation at 20 rpm for 19 hours.

14. Post-hybridization Washing

Place approximately 400 mL of Agilent Wash Buffer 1 in each of two separate glass staining dishes. Place one staining dish on a magnetic stir plate and place a slide rack and stir bar in the buffer.

Prepare a staining dish for Agilent Wash 2 by placing a stir bar in an empty glass staining dish.

A fourth glass staining dish was set aside for a final acetonitrile wash.

Disassemble each of the six hybridization chambers. Remove each slide/backing sandwich from its hybridization chamber and immerse it in a staining dish containing Wash 1. Use a pair of tweezers to pry the slide/backing sandwich apart while still submerging the microarray slide. Quickly transfer the slide to a slide rack in a staining dish containing Wash 1 on a magnetic stir plate.

The slide rack was slowly raised and lowered 5 times. The magnetic stirrer was turned on at a low setting and the slides were incubated for 5 minutes.

With 1 minute remaining in Wash 1, add Wash Buffer 2, preheated to 37°C in the incubator, to the second prepared staining dish. Quickly transfer the slide rack to Wash Buffer 2 and remove any excess buffer from the bottom of the rack by scraping it against the top of the staining dish. Slowly raise and lower the rack five times. Turn on the magnetic stirrer at a low setting and incubate the slides for 5 minutes.

Slowly remove the slide rack from Wash 2. It takes approximately 15 seconds to remove the slides from the solution.

When 1 minute is left in Wash 2, add acetonitrile (ACN) to the fourth staining dish. Transfer the slide rack to the acetonitrile staining dish. Slowly raise and lower the slide rack 5 times. Turn on the magnetic stirrer at a low setting and incubate the slides for 5 minutes.

Slowly remove the slide rack from the ACN staining dish and place it on absorbent paper. Quickly dry the bottom edge of the slide and place the slide in a clean slide box.

15. Microarray Imaging

The microarray slides were placed in an Agilent scanner slide container and loaded into the Agilent microarray scanner according to the manufacturer's instructions.

The slides were imaged in the Cy3 channel at 5 μm resolution with 100% PMT setting and XRD option 0.05. The resulting tiff images were processed using Agilent Feature Extraction Software version 10.5.

Example 2. Biomarker Identification

Potential pancreatic cancer biomarkers were identified for the diagnosis of pancreatic cancer in asymptomatic individuals and symptomatic individuals with acute or chronic pancreatitis (or both), pancreatic obstruction, GERD, gallstones, or abnormal imaging that was later found to be benign (collectively referred to as GI and normal controls). Enrollment criteria for this study were age 18 years or older, the ability to provide informed consent, and the provision of a plasma sample and a documented diagnosis of pancreatic cancer or benign findings. For cases, blood samples were collected prior to treatment or surgery and subsequently diagnosed with pancreatic cancer. Exclusion criteria included prior diagnosis or treatment of cancer (excluding squamous cell carcinoma of the skin) within 5 years of the blood draw. Plasma samples were collected from two different sites and included 143 pancreatic cancer samples and 115 control samples. The RFU values for 823 analytes in each of these 258 samples were measured and reported using the multiplex aptamer affinity assay described in Example 1. Because plasma samples were obtained from two independent studies and sites under similar protocols, site-specific differences were tested prior to biomarker discovery analysis.

Each case and control population was compared separately by generating a class-dependent cumulative distribution function (cdf) for each of the 823 analytes. The KS-distance (Kolmokorov-Smilov statistic) between the values of two sets of samples is a nonparametric measure of the degree to which the empirical distribution of values from one set (set A) differs from the distribution of values from another set (set B). For any value of a threshold value, T, some proportion of values from set A is less than T, and some proportion of values from set B is less than T. The KS-distance measures the maximum (unsigned) difference between the proportions of values selected from the two sets for any T.

This set of potential biomarkers can be used to construct a classifier that assigns samples to control or disease groups. In fact, many such classifiers are generated from these sets of biomarkers, and the frequency of any biomarker for a good scoring classifier is determined. Those biomarkers that appear most frequently in the top scoring classifiers are the most useful for generating diagnostic tests. In this example, a Bayesian classifier is used to develop the classification space, but many other supervised learning techniques can be used for this purpose. The fitness of the score of any individual classifier is judged using the area under the receiver operating characteristic curve (AUC) of the Bayesian surface classifier, assuming a disease prevalence of 0.5. This scoring metric varies from 0 to 1, with 1 being a perfect classifier. Details of constructing a Bayesian classifier from biomarker population measurements are described in Example 3.

Using the 65 analytes in Table 1, a total of 973 10-analyte classifiers were found with an AUC of 0.90 for diagnosing pancreatic cancer versus control. From this set of classifiers, a total of 11 biomarkers were found to be present in 30% or more of the high-scoring classifiers. Table 13 provides a list of these potential biomarkers, and Figure 10 is a frequency plot of the identified biomarkers.

Example 3. Naive Bayesian Classification of Pancreatic Cancer

From the list of biomarkers identified as useful for distinguishing pancreatic cancer from controls, a set of 10 biomarkers was selected and a naive Bayes classifier was constructed, as shown in Table 16. The class-dependent probability density functions (pdfs), p(xi| _c ) and p( _xi |d), were modeled as log-normal distribution functions, where _xi is the log of the measured RFU value for biomarker i, c and d refer to the control and disease populations, respectively. The function is characterized by a mean μ and a variance ^σ2 . The parameters of the pdfs for the 10 biomarkers are listed in Table 16, and examples of the original data and model fitted with normal pdfs are shown in Figure 5. As shown in Figure 5, the underlying assumptions appear to fit the data very well.

The naive Bayes classification for such a model is given by the following equation, where P(d) is the disease prevalence in the population suitable for testing, and n = 10. Each term in the sum is the log likelihood ratio of a single marker, and the overall log likelihood ratio of a sample without the disease of interest (i.e., pancreatic cancer in this case) to one with the disease is the simple sum of these individual terms plus a term representing the disease prevalence. For simplicity, we assume that p(d) = 0.5, so that

Given the unknown sample measurements in log(RFU) for each of the 10 biomarkers of 6.3, 9.3, 8.7, 10.8, 7.4, 11.4, 11.7, 9.0, 8.0, and 7.3, the classification calculations are detailed in Table 16. The individual components comprising the log likelihood ratios for the disease versus control categories are tabulated and can be calculated from the values of the parameters and in Table 16. The sum of the individual log likelihood ratios is -3.044, or a likelihood of 21 for individuals without disease versus individuals with disease, where likelihood e ^3.044 = 21. The first three biomarker values have likelihoods that are more consistent with the disease group (log likelihood > 0), but the remaining seven biomarkers are all found to consistently favor the control group. Multiplying the likelihoods gives the same result as above; the likelihood that the unknown sample does not have disease is 21. In fact, this sample is from the control population in the training set.

Example 4. Greedy Algorithm for Selecting Biomarker Panels for Classifiers

This example describes the selection of biomarkers from Table 1 to form a panel that can be used as a classifier in any of the methods described herein. A subset of the biomarkers in Table 1 was selected to construct a classifier with good performance. This method was also used to determine which potential markers to include as biomarkers in Example 2.

The measure of classifier performance used here is AUC; a performance of 0.5 is the baseline expected value for a random (coin toss) classifier, a score between 0.0 and 0.5 for classifiers that perform worse than random, and a score between 0.5 and 1.0 for classifiers that perform better than random. A perfect classifier without error has a sensitivity of 1.0 and a specificity of 1.0. The method of Example 4 can be used for other conventional measures of performance, such as the F-measure, the sum of sensitivity and specificity, or the product of sensitivity and specificity. In particular, it may be desirable to weight sensitivity and specificity differently, thereby selecting those classifiers that have higher specificity at the expense of some sensitivity, or selecting those classifiers that have higher sensitivity at the expense of some specificity. Because the methods described herein only involve one measure of "performance," any weighting scheme that results in a single performance measure can be used. Different applications will have different benefits for true positive and true negative findings, as well as different costs associated with false positive and false negative findings. For example, screening asymptomatic high-risk individuals and differentiating pancreatic cancer from benign GI conditions generally do not have the same optimal trade-off between specificity and sensitivity. The different demands of the two tests generally require different weightings for positive and negative misclassifications, which are reflected in the performance measure. Changing the performance measure generally changes the exact subset of markers selected from column 2 of Table 1 for a given data set.

For the Bayesian approach to distinguishing pancreatic cancer samples from control samples described in Example 3, the classifier was fully parameterized by the distribution of biomarkers in disease and benign training samples, and the list of biomarkers was selected from Table 1; that is, given a set of training data, the subset of included markers was selected to determine the classifier in a one-to-one manner.

The greedy approach used here is used to retrieve the best marker subsets from Table 1. For a small number of markers or classifiers with fewer markers, each possible marker subset is enumerated and the performance of the classifier constructed with this particular marker set is evaluated (see Example 4, Part 2). (This approach is known in the field of statistics as "best subset selection"; see, for example, Hastie et al.). However, for the classifiers described herein, the number of combinations of multiple markers may be very large, and it is not feasible to evaluate each possible set of 10 markers, as 30,045,015 possible combinations can be generated from a list of only 30 total analytes. Because it is impractical to search through each marker subset, a single best subset may not be found; however, by using this approach, many excellent subsets have been found, and in many cases, any of these subsets may represent the best subset.

Instead of evaluating each possible marker set, a "greedy" forward stepwise approach can be performed (see, for example, Dabney AR, Storey JD (2007) Optimality Driven Nearest Centroid Classification from Genomic Data. PLoS ONE 2 (10): e1002. doi: 10.1371/journal.pone.0001002). Using this approach, a classifier starts with the best single marker (based on the KS-distance of each marker) and grows at each step by sequentially trying each member of the marker set that is not currently in the classifier in the marker list. A marker that is best combined with the existing classifier is added to the classifier. Repeat until further improvement in performance is no longer achieved. Unfortunately, this approach may miss valuable marker combinations, and some individual markers are not all selected before the method terminates.

The greedy method used here is a modification of the aforementioned forward stepwise method. In order to expand the search, instead of keeping only a single candidate classifier (marker subset) at each step, a list of candidate classifiers is kept. This list is seeded with each single marker subset (using each marker in the table itself). The list is gradually expanded by deriving new classifiers (marker subsets) from the classifiers currently in the list and adding them to the list. Each marker subset currently on the list is extended by adding any marker from Table 1 that is not part of the classifier (whose addition to the subset does not copy the existing subset) (these are called "allowed markers"). Each existing marker subset is extended by each allowed marker from the list. Obviously, this method will eventually produce every possible subset, and the list will run out of space. Therefore, all generated classifiers are only kept until the list is less than a certain predetermined size (usually enough to keep all three marker subsets). Once the list reaches the predetermined size limit, it becomes elite; that is, only those classifiers that show a certain level of performance are kept on the list, and the other classifiers fall to the end of the list and are discarded. This is achieved by maintaining a list sorted by classifier performance; inserting new classifiers that are at least as good as the worst classifier currently on the list forces the exclusion of the current underachiever. A further implementation detail is that the list is completely permuted at each generation step; thus, each classifier on the list has the same number of tokens, and at each step the number of tokens per classifier is increased by one.

Because this approach uses different combinations of markers to generate a list of candidate classifiers, one has asked whether classifiers can be combined to avoid the errors that may be generated by the best single classifier or by a small number of groups of the best classifiers. Such "ensemble" and "committee of experts" methods are well known in the fields of statistics and machine learning and include, for example, "averaging," "voting," "stacking," "bagging," and "boosting" (see, for example, Hastie et al.). These combinations of simple classifiers provide a way to reduce the variance in classification due to noise in any particular set of markers by including information from several different classifiers and, therefore, a larger set of markers from the biomarker table, effectively averaging across classifiers. An example of the usefulness of this approach is that it can prevent outliers in a single marker from negatively impacting the classification of a single sample. The need to measure a larger number of signals may be impractical in conventional "one marker at a time" antibody assays, but there is no downside for fully multiplexed aptamer assays. These techniques benefit from a broader biomarker table and use multiple sources of information about the disease process to provide more robust classification.

The biomarkers selected in Table 1 gave classifiers that performed better than classifiers constructed with "non-markers," ie, proteins with signals that did not meet the criteria included in Table 1 (as described in Example 2).

For classifiers containing only one, two, and three markers, all possible classifiers obtained using the biomarkers of Table 1 were enumerated and the performance distribution examined in comparison with classifiers constructed from similar tables of randomly selected non-marker signals.

In Figure 11, AUC is used as a measure of performance; a performance of 0.5 is the baseline expected value for a random (coin toss) classifier. The histogram of classifier performance is compared with a histogram of the performance of a similar exhaustive classifier constructed from a "non-labeled" table of 65 non-labeled signals; the 65 signals were randomly selected from aptamers that did not demonstrate differential signal between the control and disease populations.

Figure 11 shows a histogram of the performance of all possible single-, two-, and three-marker classifiers constructed from the biomarker parameters of Table 14 for biomarkers that can distinguish between a control group and pancreatic cancer, and compares these classifiers with all possible single-, two-, and three-marker classifiers constructed using 65 "non-marker" aptamer RFU signals. Figure 11A shows a histogram of the performance of the single-marker classifier, Figure 11B shows a histogram of the performance of the two-marker classifier, and Figure 11C shows a histogram of the performance of the three-marker classifier.

In Figure 11, the solid line is a bar graph showing the classifier performance of all single-, two-, and three-marker classifiers using the biomarker data for GI and normal controls and pancreatic cancer in Table 14. The dashed line is a bar graph showing the classifier performance of all single-, two-, and three-marker classifiers using the control and pancreatic cancer data but using a random set of non-marker signals.

Classifiers constructed from the markers listed in Table 1 formed a distinct histogram, separating well from classifiers constructed using signals from "non-markers" for all single-, two-, and three-marker comparisons. The performance and AUC scores of classifiers constructed from the biomarkers in Table 1 also increased faster with the number of markers than those constructed from non-markers, with the separation between marker and non-marker classifiers increasing as the number of markers per classifier increased. All classifiers constructed using the biomarkers listed in Table 14 performed better than classifiers constructed using "non-markers."

The distribution of classifier performance shows that there are many possible multi-marker classifiers that can be derived from the set of analytes in Table 1. Although some biomarkers are inherently better than others, as demonstrated by the distribution of classifier scores and AUCs for single analytes, it is desirable to determine whether such biomarkers are required to build a high-performing classifier. To make this determination, the behavior of classifier performance is examined by removing a certain number of the best biomarkers. Figure 12 compares the performance of a classifier built using the full list of biomarkers in Table 1 with the performance of a classifier built using a subset of the biomarkers from Table 1 that excludes the top biomarkers.

Figure 12 demonstrates that classifiers constructed without the optimal markers perform well, suggesting that classifier performance is not determined by a small core set of markers and that changes in the underlying processes associated with the disease are reflected in the activity of many proteins. Many subsets of biomarkers in Table 1 performed close to optimal, even after removing the top 15 of the 65 markers in Table 1. After removing the 15 top markers from Table 1 (ranked by KS-distance), classifier performance increased with the number of markers selected from the table to reach an AUC of almost 0.87, close to the performance of the best classifier score of 0.91 selected from the complete list of biomarkers.

Finally, Figure 16 illustrates the ROC performance of a representative classifier constructed from the parameter list in Table 14 according to Example 3. A five-analyte classifier was constructed using CTSB, C5a, C5, CCL18, and CSF1R. Figure 16A illustrates the performance of the model, assuming independence of these markers, as in Example 3, and Figure 16B illustrates the empirical ROC curve generated from the study data set used to define the parameters in Table 14. As demonstrated by the AUC, it can be seen that the performance for a given number of selected markers is qualitatively consistent, and quantitative agreement is generally good, although model calculations tend to overestimate classifier performance. This is consistent with the view that the information contributed by any particular biomarker regarding the disease process is redundant with that contributed by the other biomarkers provided in Table 1, which assumes complete independence. Thus, Figure 16 demonstrates that Table 1, combined with the methods described in Example 3, allows for the construction and evaluation of numerous classifiers that can be used to distinguish pancreatic cancer from controls.

Example 5. Incorporation of CA19-9

Cancer-associated antigen 19-9 (CA19-9) is a known serum marker for pancreatic cancer. The reported sensitivity and specificity of CA19-9 for pancreatic cancer are 80-90%, respectively. However, the accuracy of CA19-9 in identifying patients with small, surgically resectable cancers is limited. The specificity of CA19-9 is also limited; CA19-9 is often elevated in patients with various benign pancreatic and biliary disorders.

The extent to which CA 19-9 is elevated in pancreatic cancer is associated with long-term prognosis. In addition, in patients who appear to have potentially resectable disease, the magnitude of CA 19-9 levels can also help predict the presence of radiographically occult metastatic disease. Serial monitoring of CA 19-9 levels can be used to follow up patients after potentially curative surgery and patients receiving chemotherapy for advanced disease. Elevated CA 19-9 levels typically precede the appearance of radiographic recurrence of disease, but confirmation of disease progression should be performed by imaging studies and/or biopsy. Detection of biomarker levels in combination with CA 19-9 can improve the sensitivity, specificity, and/or AUC for detecting pancreatic cancer (or other pancreatic cancer-related uses) compared to CA 19-9 alone.

Elevated CA19-9 levels are considered to be 35-40 U/ml in serum.

We received clinical CA19-9 measurements for a subset of the training samples. Of the original 100 cases and 69 controls, we had CA19-9 measurements for 99 cases and 52 controls. Therefore, we trained a new set of random forest models on this subset of samples using the subset of SOMAmers in Table 1. We also trained a new classifier that incorporated the CA19-9 measurements with our SOMAmer set (the merged set).

The classifier performance of the three different methods (SOMAmer, CA19-9, and the combined panel) is shown in Figure 13. The SOMAmer panel and CA19-9 performed similarly, but when the two were combined into a single classifier, the performance improved significantly. For a specificity of 100%, the SOMAmer panel and CA19-9 had a sensitivity just under 50%, while the combined classifier had a sensitivity of approximately 75%.

Further analysis showed that when CA19-9 was included in the classifier, the number of SOMAmers required for the same relative performance was reduced. Figure 14 shows the performance of random forest classifiers using CA19-9 and one or two additional SOMAmers. The left graph shows the performance of the model trained with CA19-9 and HAMP, while the right graph shows the performance of CA19-9, HAMP, and CTSB.

Example 6. Clinical Biomarker Panel

A random forest classifier was constructed from a panel of selected biomarkers that are likely to be most suitable for clinical diagnostic testing. Unlike the model selected by the naive Bayes greedy forward algorithm, the random forest classifier does not assume that the biomarkers are randomly distributed. Therefore, this model can utilize biomarkers from Table 1 that are not effective in the naive Bayes classifier.

The panel was selected using a backward elimination method using the Gini importance metric provided by the random forest classifier. Gini importance is a measure of the effectiveness of a biomarker in correctly classifying samples in the training set. This measure of biomarker importance can be used to eliminate markers that are less important to the performance of the classifier. The backward elimination method begins by building a random forest classifier that includes all 65 biomarkers in Table 1. The least important biomarker is then eliminated, and a new model is constructed using the remaining biomarkers. This process continues until only a single biomarker remains.

The final panel selected provided the best balance between maximal AUC and minimum number of markers in the model. The 10-biomarker panel that met these criteria consisted of the following analytes: APOA1, CTSB, C2, MMP7, HAMP, TFPI, C5, c5a, SFRP1, and ETHE1. A plot of the ROC curve for this biomarker panel is shown in FIG15 . The figure shows two possible decision cutoffs, indicated by arrows: a symptomatic cutoff, where a sensitivity of 84% or more can be achieved with at least 80% specificity; and an asymptomatic cutoff, where a specificity of 97.5% can be achieved with at least 60% sensitivity.

Example 7. Biomarkers for diagnosing cancer

Identification of potential biomarkers for general diagnosis of cancer. Case and control samples from three different types of cancer (pancreatic cancer, lung cancer, and mesothelioma) were evaluated. Inclusion criteria were age at least 18 years and signed informed consent at the collection site. Both cases and controls were excluded from known malignancies other than the cancer under consideration.

Pancreatic cancer. Case and control samples were obtained as described in Example 2.

Lung cancer. Case and control samples were obtained from three academic cancer center biorepositories and one commercial biorepository to identify potential markers for differential diagnosis of non-small cell lung cancer (NSCLC) from controls of high-risk smokers and individuals with benign lung nodules. The study consisted of 978 samples collected from smokers and patients with benign nodules, as well as 320 individuals diagnosed with NSCLC.

Pleural mesothelioma.

Case and control samples were obtained from an academic cancer center database to identify potential markers for differentiating malignant pleural mesothelioma from patients with a history of asbestos exposure or benign lung disease, including suspicious radiological findings later diagnosed as non-malignant. The study consisted of 30 samples collected from individuals exposed to asbestos and 41 samples collected from patients with mesothelioma.

The final list of cancer biomarkers was identified by combining the sets of biomarkers considered in each of the three different cancer studies. A Bayesian classifier using biomarker sets of increasing size was successfully constructed using a greedy algorithm (as described in more detail in Section 7.2 of this Example). Sets (or panels) of biomarkers that can be used to generally diagnose cancer in the cancer type were compiled as a function of set (or panel) size and their performance was analyzed. This analysis resulted in a list of 10 cancer biomarkers, shown in Table 19, each of which was present in at least one of these continuous marker sets, the size of which ranged from 3 to 10 markers. As an illustrative example, we describe the generation of a specific panel of 10 cancer biomarkers, as shown in Table 32.

7.1 Naive Bayesian Classification of Cancer

As described in Section 7.2 of this Example, a panel of 10 potential cancer biomarkers was selected from the list of biomarkers in Table 1 using a greedy algorithm for biomarker selection. A different naive Bayes classifier was constructed for each of these three. The class-dependent probability density functions (pdfs), p( _xi |c) and p( _xi |d), were modeled as log-normal distribution functions, where _xi is the log of the measured RFU value for biomarker i, c and d refer to the control and disease populations, respectively. The function is characterized by a mean μ and a variance ^σ² . The parameters of the pdfs for the three models consisting of 10 potential biomarkers are listed in Table 31.

The naive Bayes classification for such a model is given by the following equation, where P(d) is the disease prevalence in the population suitable for testing, and n = 10. Each term in the sum is the log likelihood ratio of a single marker, and the total log likelihood ratio of a sample without the disease of interest (i.e., in this case, each specific disease from the three different cancer types) to one with the disease is simply the sum of these individual terms plus a term representing the disease prevalence. For simplicity, we assume p(d) = 0.5, so that

Given the unknown sample measurements in log(RFU) for each of the 10 biomarkers of 10.1, 8.9, 8.8, 8.8, 9.1, 7.3, 8.2, 9.5, 6.7, and 7.7, the classification calculations are detailed in Table 32. The individual components comprising the log likelihood ratios for the disease versus control categories are tabulated and can be calculated from the values of the parameters and in Table 31. The sum of the individual log likelihood ratios is -4.568, or a likelihood of 96 for individuals without disease versus those with disease, where likelihood e ^4.568 = 96. Only one biomarker value had a likelihood that was more consistent with the disease group (log likelihood > 0), but the remaining nine biomarkers were all found to consistently favor the control group. Multiplying the likelihoods gives the same result as above; the likelihood that the unknown sample does not have disease is 96. In fact, this sample is from the control population in the NSCLC training set.

7.2 Greedy Algorithm for Selecting Cancer Biomarker Panels for Classifiers

Part 1

A subset of the biomarkers in Table 1 were selected to construct potential classifiers that can be used to determine which can be used as general cancer biomarkers to detect cancer.

Given the set of markers, a different model was trained for each of the three cancer studies, so a comprehensive measure of performance was needed to select a set of biomarkers that could simultaneously classify many different types of cancer. The measure of classifier performance used here was the average of the area under the ROC curve for all naive Bayes classifiers. The ROC curve plots the true positive rate (sensitivity) of a single classifier against the false positive rate (1-specificity). The area under the ROC curve (AUC) ranges from 0-1.0, where an AUC of 1.0 corresponds to perfect classification and an AUC of 0.5 corresponds to a random (coin toss) classifier. Other conventional measures of performance such as the F-measure or the sum or product of sensitivity and specificity can be applied. In particular, one may want to weight sensitivity and specificity differently to select those classifiers with higher specificity at the expense of some sensitivity, or to select those classifiers with higher sensitivity at the expense of specificity. We chose to use AUC because it covers all combinations of sensitivity and specificity in a single measure. Different applications will have different benefits for true positive and true negative discoveries, and will have different costs associated with false positive and false negative discoveries.Changing the performance measure may change the exact subset of markers that is selected for a given set of data.

For the Bayesian approach to distinguishing cancer samples from control samples described in Section 7.1 of this Example, the classifier was fully parameterized by the distribution of biomarkers in each of the three cancer studies, and the list of biomarkers was selected from Table 19. That is, given a set of training data, the included subset of markers was selected to determine the classifier in a one-to-one manner.

The greedy approach used here is used to retrieve the best marker subsets from Table 1. For a small number of markers or classifiers with fewer markers, each possible marker subset is enumerated and the performance of the classifier constructed with this particular marker set is evaluated (see Example 4, Part 2). (This approach is known in the field of statistics as "best subset selection"; see, for example, Hastie et al.). However, for the classifiers described herein, the number of combinations of multiple markers may be very large, and it is not feasible to evaluate each possible set of 10 markers, as 30,045,015 possible combinations can be generated from a list of only 30 total analytes. Because it is impractical to search through each marker subset, a single best subset may not be found; however, by using this approach, many excellent subsets have been found, and in many cases, any of these subsets may represent the best subset.

Instead of evaluating each possible marker set, a "greedy" forward stepwise approach can be performed (see, for example, Dabney AR, Storey JD (2007) Optimality Driven Nearest Centroid Classification from Genomic Data. PLoS ONE 2 (10): e1002. doi: 10.1371/journal.pone.0001002). Using this approach, a classifier begins with the best single marker (based on the KS-distance of each marker), and grows at each step by successively trying each member of the marker set that is not currently in the classifier in the marker list. A marker that is best combined with the existing classifier is added to the classifier. Repeat until no further improvement in performance is achieved. Unfortunately, this approach may miss valuable marker combinations, and some individual markers were not all selected before the method terminated.

The greedy method used here is a processing of the aforementioned forward stepwise method. In order to expand the search, instead of keeping only a single marker subset in each step, a list of candidate marker sets is kept. This list is seeded with a list of single markers. The list is gradually expanded by deriving new marker subsets from the marker subsets currently in the list and adding them to the list. Each marker subset currently on the list is extended by adding any marker from Table 1 that is not part of the classifier (which is added to the subset without duplicating the existing subset) (these are called "allowed markers"). Each new set of markers is defined, and a set of classifiers consisting of one each for each cancer study is trained using these markers, and the overall performance is measured by the average AUC of all three studies. In order to avoid potential overfitting, the AUC of each cancer study model is calculated by a 10-fold cross validation method. Each existing marker subset is extended by each allowed marker from the list. Obviously, this method will eventually produce every possible subset, and the list will run out of space. Therefore, all generated marker sets are only kept when the list is less than some predetermined size. Once the list reaches a predetermined size limit, it becomes elitist; that is, only those classifier sets that show a certain level of performance remain on the list, and the other classifiers fall to the end of the list and are discarded. This is achieved by maintaining the list sorted in order of classifier set performance; inserting new tag sets whose classifiers are at least as good as the worst classifier currently on the list, so as to force the elimination of the end under the currently implemented classifier set. A further implementation detail is that the list is completely permuted at each generation step; therefore, each tag set on the list has the same number of tags, and at each step the number of tags per classifier is increased by one.

In one embodiment, the set (or panel) of biomarkers that can be used to construct a classifier for diagnosing general cancer versus non-cancer is based on the average AUC for a particular combination of biomarkers used in the classification scheme. We identified a number of biomarker combinations derived from the markers in Table 19 that were able to effectively classify different cancer samples from controls. Representative panels are shown in Tables 22-29, which show a range of 100 different panels of 3-10 biomarkers, with the average validation (CV) AUC for each panel shown. The total number of occurrences of each marker in each of these panels is shown at the bottom of each table.

The biomarkers selected in Table 19 produced classifiers that performed better than the classifier built with "non-markers". In Figure 17 we show the performance of our 10-biomarker classifier compared to other possible classifiers.

Figure 17A shows the distribution of mean AUCs for classifiers constructed from a randomly sampled set of 10 "non-markers" taken from the entire set of 10 present in all three studies, excluding the 10 markers in Table 19. The performance of the 10 potential cancer biomarkers is shown as vertical dashed lines. This plot clearly shows that the performance of these 10 potential biomarkers far exceeds the distribution of other marker combinations.

Figure 17B shows a similar distribution to Figure 17A, however, the randomly sampled set was limited to the 55 biomarkers from Table 1 that were not selected by the greedy biomarker selection method of the 10-analyte classifier. This figure demonstrates that the 10 markers selected by the greedy algorithm represent a subset of biomarkers that generalizes to types of cancer much better than the classifier built with the remaining 55 biomarkers.

Finally, Figure 18 shows the classifier ROC curves for each of the three cancer research classifiers. The above embodiments and examples are shown as examples only. No particular embodiment, example, or element of a particular embodiment or example is considered to be a critical, required, or essential element or feature of any claim. In addition, no element described herein is required to implement the appended claims unless expressly described as "essential" or "critical." Various changes, modifications, substitutions, and other variations may be made to the disclosed embodiments without departing from the scope of the present application as defined by the appended claims. The description, including the figures and examples, is illustrative and not restrictive, and all such modifications and substitutions are intended to be included within the scope of this application. Therefore, the scope of this application should be defined by the appended claims and their legal equivalents, rather than by the above examples. For example, the steps listed in any method or process claim may be performed in any feasible order and are not limited to the order in any embodiment, example, or claim. In addition, in any of the above methods, one or more biomarkers of Table 1 or Table 19 may be specifically excluded as individual biomarkers or as biomarkers from any group.

Table 1: Cancer biomarkers

Table 1: Continued from previous page

Table 2: A panel of biomarkers

Table 2: Continued from previous page

Table 3: Two biomarker panels

Table 3: Continued from previous page

Table 3: Continued from previous page

Table 4: Panel of three biomarkers

Table 4: Continued from previous page

Table 4: Continued from previous page

Table 5: Panel of four biomarkers

Table 5: Continued from previous page

Table 5: Continued from previous page

Table 6: Panel of five biomarkers

Table 6: Continued from previous page

Table 6: Continued from previous page

Table 7: Panel of six biomarkers

Table 7: Continued from previous page

Table 7: Continued from previous page

Table 7: Continued from previous page

Table 7: Continued from previous page

Table 8: Panel of seven biomarkers

Table 8: Continued from previous page

Table 8: Continued from previous page

Table 8: Continued from previous page

Table 8: Continued from previous page

Table 9: Panel of eight biomarkers

Table 9: Continued from previous page

Table 9: Continued from previous page

Table 9: Continued from previous page

Table 9: Continued from previous page

Table 10: Panel of nine biomarkers

Table 10: Continued from previous page

Table 10: Continued from previous page

Table 10: Continued from previous page

Table 10: Continued from previous page

Table 11: Panel of ten biomarkers

Table 11: Continued from previous page

Table 11: Continued from previous page

Table 11: Continued from previous page

Table 11: Continued from previous page

Table 12: Marker counts in the biomarker panel

Table 12: Continued from previous page

biomarkers 3 4 5 6 7 8 9 10 NRP1 44 20 12 5 2 2 2 1 PLAT 48 54 92 123 143 145 165 177 SERPINA5 5 0 0 0 0 0 0 0 SERPINF2 5 3 3 0 0 0 0 0 SGTA 3 2 1 0 0 0 0 0 TFPI 100 60 51 46 57 70 91 111 THBS2 4 0 0 0 0 0 0 0 THBS4 66 110 146 193 243 276 334 354 TIMP1 twenty two 2 0 0 0 0 0 0 TNFRSF18 8 9 3 0 1 3 2 2 TNFRSF1B 20 12 8 6 4 1 0 0 top1 6 3 0 0 0 0 0 0 VEGFA 16 33 47 51 75 142 268 455 VEGFC 5 4 2 1 0 0 0 0

Table 13: Analytes in the 10-marker classifier

CTSB C5a ETHE1 CSF1R CCL18 C5 KLK7 VEGFA KIT THBS4 LTF

Table 14: Parameters from the Naive Bayes classifier training set

biomarkers CSF1R 10.712 10.995 0.398 0.399 CTSB 8.836 9.398 0.287 0.621 IL1RL1 9.702 10.189 0.533 0.780 GDF11 8.889 8.578 0.291 0.379 ETHE1 7.373 7.443 0.119 0.121 CCL23 8.795 8.975 0.312 0.329 FGFR3 6.992 7.166 0.178 0.225 KIT 9.770 9.623 0.287 0.318 FSTL3 8.787 9.029 0.290 0.374 THBS2 7.481 7.922 0.270 0.633 SERPINF2 9.264 9.175 0.115 0.162 TNFRSF1B 10.748 11.028 0.380 0.452 TNFRSF18 12.308 12.279 0.139 0.168 BMP6 7.958 8.138 0.142 0.239 GFRA1 7.324 7.465 0.182 0.200 CRP 11.965 12.304 0.735 0.233 SERPINA5 10.309 10.101 0.3000 0.419 KLKB1 11.802 11.666 0.159 0.211 APOE 8.081 8.314 0.406 0.656 SFRP1 7.096 7.219 0.221 0.309 C2 11.506 11.611 0.100 0.132 CKM 7.313 7.192 0.154 0.116 TFPI 10.179 10.490 0.261 0.352 INSR 8.480 8.633 0.224 0.255 NID2 8.595 8.806 0.213 0.384 HAMP 10.424 11.079 0.788 0.617 MDK 8.034 8.495 0.570 0.578 CDK5-CDK5R1 6.937 6.994 0.108 0.111 NID1 9.771 9.941 0.213 0.357 VEGFC 7.454 7.540 0.118 0.126 C9 11.911 12.076 0.234 0.233 LTF 10.120 9.870 0.442 0.419 IL12A-IL12B 7.311 7.273 0.052 0.057 C5 9.485 9.603 0.119 0.143 IL18R1 7.643 7.845 0.186 0.475 CCL18 11.320 11.616 0.477 0.398 FEGFA 8.532 8.601 0.170 0.134 IDUA 8.428 8.694 0.366 0.558 top1 6.892 6.842 0.088 0.091 C5-C6 6.506 6.593 0.133 0.144 TIMP1 9.815 10.148 0.264 0.430 C5a 11.354 11.606 0.254 0.246 THBS4 10.013 0.794 0.359 0.400 ENTPD1 7.225 7.299 0.110 0.103 LBP 9.102 9.489 0.439 0.548 KLK3-SERPINA3 9.0134 9.287 01.353 0.422 MCM2 7.794 7.975 0.226 0.359 SGTA 5.920 5.883 0.060 0.079 ESM1 9.715 9.919 0.330 0.476

Table 14: Continued from previous page

biomarkers PLAT 8.517 8.838 0.461 0.502 KLK7 8.322 7.989 0.321 0.391 CCL23 7.909 8.097 0.227 0.267 ACP5 10.198 10.436 0.292 0.343 NRP1 8.832 9.047 0.243 0.256 MMP7 9.084 9.574 0.437 0.706 ACY1 9.898 10.411 0.628 0.919 ALPL 10.577 10.290 0.377 0.417 IL11RA 7.312 7.213 0.110 0.107 APOA1 9.701 9.480 0.171 0.295 CKB-CKM 7.506 7.025 0.653 0.479 KLK8 7.361 7.421 0.100 0.178 AHSG 11.914 11.826 0.133 0.167 HINT1 5.835 5.793 0.086 0.104 MRC1 9.628 9.995 0.370 0.490 FCGR3B 10.920 11.145 0.255 0.269

Table 15: AUCs for exemplary combinations of biomarkers

# AUC 1 CTSB 0.791 2 CTSB C5a 0.853 3 CTSB C5a C5 0.880 4 CTSB C5a C5 CCL18 0.890 5 CTSB C5a C5 CCL18 CSF1R 0.895 6 CTSB C5a C5 CCL18 CSF1R KLK7 0.895 7 CTSB C5a C5 CCL18 CSF1R KLK7 ETHE1 0.906 8 CTSB C5a C5 CCL18 CSF1R KLK7 ETHE1 C5-C6 0.902 9 CTSB C5a C5 CCL18 CSF1R KLK7 ETHE1 C5-C6 KLK8 0.903 10 CTSB C5a C5 CCL18 CSF1R KLK7 ETHE1 C5-C6 KLK8 VEGFA 0.913

Table 16: Calculations from the Naive Bayes Classifier Training Set

Table 17: Clinical characteristics of the training group

Table 18: 10 Biomarker Classifier Proteins

Table 19: Biomarkers of general cancers

ACY1 APOA1 C5 CCL23 CKB-CKM CKM ENTPD1 GDF11 HAMP HINT1 KIT KLK3-SERPINA3 LBP SERPINF2 THBS2 TIMP1 C9 FSTL3 IL12A-IL12B CDK5-CDK5R1 CCL23

Table 20: A panel of biomarkers

Table 21: Two-biomarker panel

Table 21: Continued from previous page

Table 21: Continued from previous page

Table 22: Panel of three biomarkers

Table 22: Continued from previous page

Table 22: Continued from previous page

Table 23: Panel of Four Biomarkers

Table 23: Continued from previous page

Table 23: Continued from previous page

Table 24: Panel of five biomarkers

Table 24: Continued from previous page

Table 24: Continued from previous page

Table 25: Panel of six biomarkers

Table 25: Continued from previous page

Table 25: Continued from previous page

Table 25: Continued from previous page

Table 25: Continued from previous page

Table 26: Panel of seven biomarkers

Table 26: Continued from previous page

Table 26: Continued from previous page

Table 26: Continued from previous page

Table 26: Continued from previous page

Table 27: Panel of Eight Biomarkers

Table 27: Continued from previous page

Table 27: Continued from previous page

Table 27: Continued from previous page

Table 27: Continued from previous page

Table 28: Panel of Nine Biomarkers

Table 28: Continued from previous page

Table 28: Continued from previous page

Table 28: Continued from previous page

Table 28: Continued from previous page

Table 29: Panel of ten biomarkers

Table 29: Continued from previous page

Table 29: Continued from previous page

Table 29: Continued from previous page

Table 29: Continued from previous page

Table 30: Counts of markers in the biomarker panel

Table 31: Parameters derived from the Naive Bayes classifier for the cancer dataset

Table 32: Calculations from the Naive Bayes Classifier Training Set

Claims (32)

1. Use of the capture agent in the preparation of a kit for diagnosing pancreatic cancer in an individual or determining the likelihood of pancreatic cancer in an individual by means of: Provide a biomarker set comprising at least three biomarker proteins listed in Table 1, wherein two of the at least three biomarker proteins are biomarker combinations selected from CTSB and THBS4, CTSB and C5, and CTSB and C2. In biological samples from individuals, the levels of biomarkers corresponding to at least three biomarker proteins in the group are measured, wherein the levels of said biomarkers provide an indication of the likelihood that the individual has or does not have pancreatic cancer. The trapping agent is specific to the at least three biomarker proteins. 2. The use of claim 1, wherein the diagnosis includes the differential diagnosis of pancreatic cancer from benign disease conditions. 3. The use of claim 2, wherein the benign disease condition is selected from pancreatitis and gastrointestinal disorders. 4. The use of claim 1, wherein the individual has an abdominal mass. 5. Use of the capture agent in the preparation of a kit for screening individuals at high risk of asymptomatic pancreatic cancer by means of: Provide a biomarker set comprising at least three biomarker proteins listed in Table 1, wherein two of the at least three biomarker proteins are biomarker combinations selected from CTSB and THBS4, CTSB and C5, and CTSB and C2. In biological samples from individuals, the levels of biomarkers corresponding to at least three biomarker proteins in the group are measured, wherein the levels of said biomarker proteins provide an indication of the likelihood that the individual has or does not have pancreatic cancer. The trapping agent is specific to the at least three biomarker proteins. 6. The use of claim 1 or 5, wherein measuring the level of the biomarker protein comprises performing an in vitro assay. 7. The use of claim 6, wherein the in vitro assay comprises at least one capture reagent corresponding to each of the biomarkers, and further comprises selecting the at least one capture reagent from aptamers and antibodies. 8. The use of claim 6, wherein the in vitro assay is selected from immunoassays, aptamer-based assays, histological or cytological assays. 9. The use of claim 1 or 5, wherein the biological sample is selected from whole blood, plasma, serum, and pancreatic juice. 10. The use of claim 9, wherein the biological sample is plasma. 11. The use according to any one of claims 1-5, wherein the biological sample is pancreatic tissue, and wherein the biomarker value is derived from histological or cytological analysis of the pancreatic tissue. 12. A device for indicating the likelihood of an individual having pancreatic cancer, said device comprising: A device provides a biomarker group comprising at least three biomarker proteins listed in Table 1, wherein two of the at least three biomarker proteins are biomarker combinations selected from CTSB and THBS4, CTSB and C5, and CTSB and C2. A device for retrieving biomarker information of an individual on a computer, wherein the biomarker information includes biomarker values, each corresponding to at least three biomarker proteins, wherein the biomarker values are obtained by measuring the levels of the at least three biomarker proteins in the group; A device for classifying each of the biomarker values using a computer; and A device that indicates the likelihood of an individual having pancreatic cancer based on multiple classifications. 13. The device of claim 12, wherein indicating the likelihood that the individual has pancreatic cancer includes displaying the likelihood on a computer display. 14. A device for indicating the likelihood of an individual having pancreatic cancer, said device comprising: A computer-readable medium containing program code executable by a processor of a computing device or system, the program code comprising: A module for retrieving data attributed to biological samples from said individual, wherein said data includes levels of at least three biomarker proteins selected from Table 1, wherein two of said at least three biomarker proteins are combinations of biomarkers selected from CTSB and THBS4, CTSB and C5, and CTSB and C2, wherein the levels of said biomarker proteins are measured in said biological sample; and A module that performs a classification method that indicates the pancreatic disease status of the individual as a function of the levels of the at least three biomarker proteins. 15. The device of claim 14, wherein the classification method uses a probability density function. 16. The device of claim 15, wherein the classification method uses two or more categories. 17. The use according to any one of claims 1-5, wherein the levels of at least four biomarker proteins are measured. 18. The use according to any one of claims 1-5, wherein the levels of at least five biomarker proteins are measured. 19. The use according to any one of claims 1-5, wherein the levels of at least six biomarker proteins are measured. 20. The use according to any one of claims 1-5, wherein the levels of at least seven biomarker proteins are measured. 21. The use according to any one of claims 1-5, wherein the levels of at least eight biomarker proteins are measured. 22. The use according to any one of claims 1-5, wherein the levels of at least nine biomarker proteins are measured. 23. The use according to any one of claims 1-5, wherein the individual is at high risk of pancreatic cancer due to smoking, drinking alcohol, or a family history of pancreatic cancer. 24. The use of any one of claims 1-5, wherein the biomarker is selected from Table 18. 25. The use according to any one of claims 1-5, further comprising the biomarker CA19-9. 26. The use according to any one of claims 1-5, wherein: (i) When the biomarker group contains a combination of biomarkers CTSB and THBS4, the biomarker group also contains C5, GDF11, CCL23, KLK3, SERPINA3, MMP7, C2 and/or CRP; (ii) When the biomarker group comprises a combination of biomarkers CTSB and C5, the biomarker group also comprises THBS4, GDF11, CCL23, KLK3, SERPINA3, MMP7, C2, and/or CRP; or (iii) When the biomarker group contains a combination of biomarkers CTSB and C2, the biomarker group also contains C5, THBS4, GDF11, CCL23, KLK3, SERPINA3, MMP7 and/or CRP. 27. The use according to any one of claims 1-5, wherein the individual is a person. 28. The use of any one of claims 1-5, wherein the individual is classified as having or not having pancreatic cancer, or the likelihood of the individual having pancreatic cancer is determined based on the level of the biomarker and at least one additional biomedical information corresponding to the individual. 29. The use of claim 28, wherein the at least one additional biomedical information is independently selected from... (a) Information corresponding to the individual's weight changes, (b) Information corresponding to the race of the individual. (c) Information corresponding to the gender of the individual. (d) Information corresponding to the individual's drinking history, (e) Information corresponding to the individual's occupational history, (f) Information corresponding to the individual's family history of pancreatic cancer or other cancers. (g) Information corresponding to the individual's clinical symptoms, (h) Information corresponding to other laboratory tests, and (i) Information corresponding to the individual's exposure to known carcinogens. 30. The use of claim 28, wherein the at least one additional biomedical information is independently selected from: -Information corresponding to the presence or absence of a pancreatic mass or other abdominal mass. -Information on the presence or absence of at least one genetic marker in the individual that is associated with a high risk of cancer in the individual or a member of the individual's family, and - Information corresponding to the gene expression values of the individual. 31. The use of claim 28, wherein the at least one additional biomedical information is selected from information on the presence or absence of at least one genetic marker corresponding to a high risk of pancreatic cancer in the individual or a family member of the individual. 32. The use of claim 28, wherein the at least one additional biomedical information is selected from information corresponding to the individual's smoking history.