May 1, 2011 (Vol. 31, No. 9)

Vicki Glaser Writer GEN

Search for Novel Biomarkers Detectable in Accessible Bodily Fluids Proves Promising

The best approach to cure cancer is to detect it early, while a tumor is still localized, can be surgically removed, and is more readily treatable with radiation and/or chemotherapeutic and immunotherapeutic agents. Cure rates for early-stage cancers may exceed 80%, whereas once the cancer has spread, survival rates may drop precipitously.

At the 2011 annual meeting of the American Association of Cancer Researchers (AACR) held last month in Florida, David Sidransky, M.D., professor, Johns Hopkins Medical Institutions, and Francisco Esteva, M.D., Ph.D., professor, University of Texas, M.D. Anderson Cancer Center, chaired a symposium that focused on the discovery and development of molecular biomarkers for solid tumors.

The presentations described research aimed at identifying and validating various types of biomarkers that are detectable in readily accessible bodily fluids, such as blood and urine, and have diagnostic and prognostic value to help guide clinical decisions related to cancer risk, as well as the need for more invasive diagnostic procedures. The goal is to detect biosignatures that are more specific and sensitive than existing diagnostic modalities, can detect tumors earlier in the course of disease, and can reduce the need for more invasive and costly biopsies and imaging studies. Molecular biomarkers may also represent cost-effective screening tools in high-risk populations.

Dr. Sidransky led off the session by outlining several challenges in developing and implementing a molecular cancer diagnostic once a tumor-related biomarker has been identified. These include the need for analytical validation of the biomarker to demonstrate high enough accuracy—sufficient specificity and sensitivity—and the importance of identifying the appropriate target population before moving the test forward into a clinical trial setting. Subsequent challenges relate to clinical validation and the need to define a biomarker’s (or panel of biomarkers’) intended use and implications and to understand what the results mean and how best to apply them.

Harvey Pass, M.D., New York University Langone Medical Center and Cancer Center, presented his group’s experience working with SomaLogic to develop an aptamer-based diagnostic to detect malignant mesothelioma in asbestos-exposed individuals. About 27.5 million Americans were occupationally exposed to asbestos between 1940 and 1979.

Mesothelioma has a developmental latency period of 10–40 years. About 3,000 new cases are reported each year in the U.S., but the incidence of pleural mesothelioma will not peak for another 20 years. Median survival of patients with late-stage disease averages 12 months, compared to 48 months for individuals diagnosed with stage 1 mesothelioma; an early-stage diagnosis is made in only about 10% of patients.

The group used SomaLogics’ SOMAmer multiplexed proteomic assay to quantify about 850 proteins simultaneously in 15 µL blood samples. The assay utilizes high-affinity aptamers that bind selectively to a protein target, with slow dissociation rates. They applied the SOMAmer platform to analyze blood samples from three study sites in a prospective case/control study, comparing serum samples from 90 patients with malignant mesothelioma to 80 control samples from asbestos-exposed individuals who did not have mesothelioma.

The technology utilizes fluorescently tagged, biotin-linked aptamers called SOMAmers that bind to the proteins in the sample, forming DNA/protein complexes. These complexes are isolated and processed to select for specific binding. The protein components are then labeled and immobilized. The aptamers are eluted off the proteins, applied to a microarray, and hybridized to complementary moieties for detection and quantification.

The samples were divided into two groups: 75% were used as a training set and 25% as a blinded test set. The study led to the identification of 19 significant biomarkers.

Dr. Pass presented data derived from the application of subsets of these biomarkers to the blinded test set, demonstrating 100% specificity and 80% sensitivity for their ability to distinguish asbestos-exposed controls from mesothelioma cases. The biomarkers were able to detect 15 of 19 stage I/II cases. Ongoing studies are aimed at analytical validation of the biomarker panel using comparative ELISA measurements.

Among women with ovarian cancer, five-year survival rates vary widely from about 90% for stage I disease (localized to the ovary) to 5% for stage IV cancer. At present, about 70% of patients are diagnosed with stage III or IV disease. Measurement of CA125 in the blood is the test currently used to monitor ovarian cancer treatment, follow patients for recurrence, and in some cases screen high-risk individuals to detect early-stage disease.

However, as Christine Coticchia, Ph.D., research fellow at Children’s Hospital Boston and Harvard Medical School, pointed out, CA125 is relatively nonspecific for ovarian cancer and uninformative in a substantial percentage of patients. Dr. Coticchia and colleagues are studying a combination of two matrix metalloproteases (MMPs)—MMP-2 and MMP-9—in urine for their utility as biomarkers to predict the presence of ovarian cancer in women with normal CA125 levels.

Zymographic analysis of urine to measure MMP levels showed significant differences in MMP-2+MMP-9 levels in ovarian cancer versus control samples. Higher levels of each biomarker alone correlated with a greater likelihood of ovarian cancer, but the combination fingerprint of MMP-2+MMP-9 yielded greater diagnostic accuracy—which improved even more when multiplexed with age, with the three variables together yielding an area under the curve (AUC) of 0.82.


Principle of SomaLogic’s multiplex SOMAmer affinity assay. (A) Binding. SOMAmers and samples are mixed in 96-well microwell plates and allowed to bind. Cognate and noncognate SOMAmer-target protein complexes form. Free SOMAmer and protein are also present. (B–H) Schematic sequence of assay steps leading to quantitative readout of target proteins. (B) SOMAmer-protein binding: DNA-based SOMAmer molecules (gold, blue, and green) have unique shapes selected to bind to a specific protein. SOMAmers contain biotin (B), a photo-cleavable linker (L), and a fluorescent tag at the 59 end. Most SOMAmers (gold and green) bind to cognate proteins (red), but some (blue) form noncognate complexes. (C) Catch-1: SOMAmers are captured onto a bead coated with streptavidin (SA) which binds biotin. Uncomplexed proteins are washed away. (D) Proteins are tagged with NHS-biotin. (E) Photocleavage and kinetic challenge: UV light (hn) cleaves the linker and SOMAmers are released from beads, leaving biotin on bead. Samples are challenged with anionic competitor (dextran sulfate). Noncognate complexes (blue SOMAmer) preferentially dissociate. (F) Catch-2 SOMAmer-protein complexes are captured onto new avidin-coated beads by protein biotin tag. Free SOMAmers are washed away. (G) SOMAmers are released from complexes into solution at high pH. (H) Remaining SOMAmers are quantified by hybridization to microarray containing single-stranded DNA probes complementary to SOMAmer DNA sequence, which form a double-stranded helix. Hybridized SOMA-mers are detected by fluorescent tags when the array is scanned. [Reprinted from PLoS ONE]

Lung Cancer Protection

Charles Birse, Ph.D., associate director at Celera, talked about serum biomarker panels that detect lung cancer in never-smokers. He noted that about 20% of lung cancers occur in never-smokers, and that figure is expected to increase. About 62% of these tumors are adenocarcinomas and 18% squamous cell carcinomas, compared to 19% and 53%, respectively, among smokers.

Dr. Birse envisions serum biomarkers being used as an adjunctive diagnostic tool either applied before a CT scan to determine the need for imaging or post-CT to stratify patients with evidence of a pulmonary nodule according to their risk for a malignant lesion and to assess the need to perform a lung biopsy.

Dr. Birse and colleagues analyzed tumor tissue and cell lines using mass spectrometry to identify cell surface and secreted proteins differentially expressed in cancer versus healthy samples. The team then prioritized the markers identified from the MS data and validated them using ELISA immunoassays. They studied a biomarker panel in more than 600 specimens from patients (smokers) with non-small-cell lung cancer (NSCLC) and control subjects, which they divided into a set of training serum samples and a set of test samples, using the data to develop an algorithm for lung cancer detection.

A subsequent independent, case/control, validation study using a six-marker model was performed in 80 never-smokers, 40 with lung cancer and 40 matched controls. The study yielded promising results and demonstrated the ability of the blood test to distinguish samples from the smoker cohort with malignancy (all stages), with an AUC for the training set of 0.877 and AUC for the test set of samples of 0.868. When the model was applied to the never-smoker patient population, the algorithm was again able to discriminate the malignant cases with an AUC of 0.906, at 83% sensitivity and 83% specificity.

Allen Taylor, Ph.D., research fellow at Fred Hutchinson Cancer Research Center, presented studies aimed at detecting autoantibodies to angioprotein-like protein 2 (ANGPTL3), a growth factor involved in regulating angiogenesis that is elevated in the serum of a mouse model of NSCLC and in plasma collected from newly diagnosed patients with NSCLC. Circulating autoantibodies to tumor antigens such as ANGPTL3 may be produced during the early stages of tumor development and could, therefore, contribute to early diagnosis, even before tumor antigens are detectable in serum.

The study showed that the circulating concentration of and autoantibody reactivity to ANGPTL3 were significantly elevated in the sera of patients with NSCLC that was collected up to 12 months before their diagnosis, compared to healthy controls. These measures were also significantly elevated in the sera of newly diagnosed NSCLC patients compared to control samples.

Dr. Taylor presented data showing an inverse correlation between autoantibody reactivity to ANGPTL3 and antigen concentration in individual serum samples. He proposed that autoantibodies represent an alternative type of biomarker that could be used in combination with tumor antigen detection to improve the predictive performance of a test for early detection of NSCLC.

Men are commonly screened for prostate cancer risk using measurements of prostate specific antigen (PSA); however, the PSA test suffers from relatively low specificity and sensitivity as a diagnostic tool. Most men with an elevated PSA test result are found not to have cancer. According to the U.S. National Cancer Institute (NCI), only 25–35% of men who have a biopsy performed as follow-up to an elevated PSA level have prostate cancer.

Scott Tomlins, M.D., Ph.D., a resident at the University of Michigan Health System, presented work done in collaboration with Gen-Probe, University of California at San Diego Medical Center, Université Laval, and Dianon Systems to develop a urine-based test to detect the TMPRSS2:ERG gene fusion, which is present in about 50% of prostate cancers, and which Dr. Tomlins describes as the most specific prostate cancer biomarker currently identified.

Urinary TMPRSS2:ERG scores have been shown to correlate with positive findings of clinically significant prostate cancer on biopsy and prostatectomy. The scores are significantly higher in samples from patients with cancerous versus benign prostate lesions.

Dr. Tomlins concluded that this new biomarker may improve on the predictive power of the PCA3 assay, which detects the messenger RNA of prostate cancer gene 3 (PCA3) in male urine after a digital rectal examination. PCA3 assay results are used to guide decisions to perform a biopsy following a positive PSA test.

Urinary TMPRSS2:ERG in combination with PCA3 can enhance the utility of serum PSA for predicting prostate cancer and the presence of clinically relevant cancer on biopsy. TMPRSS2:ERG has not yet been evaluated for its utility as a screening tool.

Pancreatic Cancer

More than 80% of patients with pancreatic cancer are diagnosed with either locally advanced or metastatic disease that is associated with a poor survival rate. Researchers at Duke University compared the secretomes of pancreatic cancer samples versus normal pancreatic cells with the aim of simultaneously identifying differentially secreted proteins and developing the reagents needed to detect them. They used an iterative selection process known as Systematic Evolution of Ligands by Exponential Enrichment (SELEX) to identify protein-binding RNA aptamer that can differentiate between the sera of pancreatic cancer and samples from healthy patients by binding to molecules selectively secreted by cancerous cells.

Partha Ray, Ph.D., post-doctoral fellow, presented data that led to the identification of several RNA aptamers that differentially bind the secretome of pancreatic cancer cells compared to normal pancreatic cells. One of these aptamers, M9-5, demonstrated specific binding in 20 of 22 patients with pancreatic cancer and 0 of 20 controls. He also showed that M9-5 binding decreases in samples from patients with reduced tumor burden and proposed the utility of this biomarker to monitor response to therapy.

The researchers identified the target for M9-5 as cyclophilin B, a secreted protein that had not previously been associated with pancreatic cancer. A cyclophilin B ELISA can detect quantifiable differences between sera from patients with or without pancreatic cancer.

Previous articleResearchers Find Ewing’s Sarcoma Candidate Blocks Oncogenic Gene in Prostate Cancer
Next articleTo Battle Super-Bugs, Antibiotic R&D Needs to Be Re-Incentivized