Biomarker or biomarker testing

Despite focused research in conventional therapies, the 5-year survival rate of lung cancer patients remains only 15%. Due to the lack of reliable early diagnostic techniques, most patients are diagnosed with advanced-stage disease and therefore have a poor prognosis. There are more than 100 million current and former smokers in the US who have elevated risk for lung cancer and may benefit from a non-invasive test for early detection. Patients at risk may have subclinical disease for years prior to diagnosis. Therefore the clinical challenge is to develop lung cancer detection methods that are effective during this window of opportunity to increase the rate of early detection and thereby spur early treatment and improve lung cancer patient outcomes. During tumorigenesis, the developing tumor forms a complex cellular microenvironment that produces many soluble mediators, such as proteins and miRNAs, which are important for tumor initiation and growth and thus can serve as lung cancer biomarkers. Because the pulmonary vascular bed comprises approximately 50% of the body’s total vasculature, the peripheral blood is expected to reflect even small variations in lung cancer biomarkers.

Over the past decade, our laboratory has gathered substantial data supporting the analysis of soluble mediators of lung carcinogenesis including inflammatory cytokines, growth and angiogenic factors, and miRNA. Additionally, a considerable literature has developed describing the contribution of these factors to lung carcinogenesis. Based on these findings, in our preliminary studies we have developed a plasma protein-based biomarker panel that accurately distinguished between lung cancer patients and at-risk control subjects. Our predictive model provided 97% sensitivity, 77% specificity, and an area under the ROC of .93 for stage 1 vs. high-risk controls. Because recent studies demonstrated that circulating miRNA can serve as robust and specific diagnostic biomarkers, we developed a panel of 17 miRNAs relevant to lung carcinogenesis. In our preliminary studies utilizing this miRNA panel, we were able to detect miRNAs that are differentially expressed in the plasma of lung cancer patients as compared to high-risk controls. We hypothesize that the proteins and miRNA profiles will each contribute unique information and thus, an integrated panel of protein and miRNA markers may provide higher sensitivity for early lung cancer detection than either platform alone.

Specific aims for the current proposal include: 1) to identify protein profiles in plasma associated with early-stage lung cancer, 2) to identify plasma miRNA profiles associated with early stage lung cancer, and 3) to integrate miRNA and cytokine biomarkers established in Aims 1 and 2 to improve early-stage lung cancer detection. The first and second aims address the clinical problems of patients at risk for lung cancer who present with a suspicious pulmonary nodule of indeterminate significance that has be further evaluated.

Folate/homocysteine (Hcy) metabolism facilitates the methylation of DNA, proteins and carcinogens; nucleic acid synthesis; and glutathione generation. Dysregulation of folate/Hcy metabolism, due to suboptimal vitamin status and/or functional polymorphisms of key enzymes, is an etiologic feature of some cancers, including lung cancer. Several drugs that target enzymes of folate/Hcy metabolism have become mainstays of cancer chemotherapy, most recently pemetrexed (PMX), which targets dihydrofolate reductase (DHFR) and thymidylate synthase (TS) and is used to treat patients with non-small cell lung cancer (NSCLC). Our primary and secondary hypotheses are that folate/Hcy phenotype, perhaps underpinned by genotype, is a major contributor to (i) clinical response to PMX, and (ii) risk of NSCLC itself. We will use archived and newly acquired samples from PMX-treated NSCLC patients to define pre-treatment and in-treatment folate/Hcy phenotypes using LC-MRM/MS technology that can precisely determine the concentrations and relative proportions of key folates (tetrahydrofolate; 5-methyltetrahydrofolate; 5,10-methenyltetrahydrofolate) and Hcy. Genotypes will be determined for approximately twenty folate-related genes. Biostatistical analyses will be used to establish

whether aspects of folate/Hcy phenotype and/or genotype are prospectively or concurrently associated with time to tumor progression, survival time, toxicity, etc. Case-control analysis of genetic and phenotypic data will also be undertaken to determine whether any genetic and/or biomarker variables predict case status. This project has the potential to identify biomarkers and/or genetic factors that can be used to predict or monitor responses of NSCLC patients to PMX, and may provide the basis for the future deployment of personalized medicine strategies in lung cancer treatment.

Inhibitors of the VEGF pathway, including the VEGF antibody bevacizumab (BV) and multitargeted VEGFR tyrosine kinase inhibitors such as vandetanib, have in some cases improved progression-free survival (PFS) and/or overall survival (OS) in patients with advanced NSCLC. While benefits in the overall NSCLC population have thus far been modest, it is clear that a subset of patients have dramatic benefit, while others may experience no benefit or even life-threatening toxicities. Furthermore, virtually all patients who respond inevitably develop therapeutic resistance. Thus, the development of markers predictive of response or resistance for VEGF inhibitors is critically needed to advance NSCLC treatment. There are currently no validated predictive markers for VEGF inhibitors, and the development of such markers is made more challenging by the impact of both tumor- and host-derived circulating factors on tumor angiogenesis. Useful predictive markers will likely need to reflect factors from both sources. We hypothesize that by comprehensively profiling angiogenesis-related CAFs and other peptides in the blood, as well as germline variations in angiogenesis-related genes, non-invasive predictive markers for identifying which patients benefit from VEGF inhibitors (alone or with chemotherapy) can be developed and validated.

The investigators of this proposal have a longstanding collaboration to develop predictive markers for VEGF inhibitors and, along with other investigators, have identified three promising and potentially complementary approaches using blood-based markers that will be tested in this grant. We have also played leadership roles in major randomized clinical trials of VEGF inhibitors, which have enabled us to leverage a unique set of resources to address these questions.

In Aim 1, we will use multiplexed CAF profiling to identify high-priority candidate predictive markers for differential benefit from VEGF inhibitors with chemotherapy compared to chemotherapy alone using two independent sets of samples from randomized phase II trials: one testing bevacizumab (BV) with chemotherapy (pemetrexed/carboplatin or Pem/C) vs. Pem/C alone (BATTLE-FL) trial, and another testing the VEGFR/EGFR tyrosine kinase inhibitor vandetanib (VAN) plus carboplatin/paclitaxel (VCP) vs. CP, in first-line NSCLC trial. We will also test CAF markers of benefit for VAN vs. standard therapy with erlotinib using specimens from a recently completed randomized phase III trial (ZEST).

In Aim 2, we will use MALDI-TOF (matrix-assisted laser desorption/ionization time-of-flight) mass spectroscopy to develop a serum proteomic classifier for benefit for VEGF inhibitors, using samples from the same trials as in Aim 1. We previously used a similar approach to develop a serum proteomic marker of benefit (Veristrat) for EGFR inhibitors.

Finally, in Aim 3, we will validate a signature of SNPs from three angiogenesis-related genes that were found to be predictive of benefit for BV with CP (BCP) vs. CP in the ECOG4599 trial. This will be tested in the same three trials of VEGF inhibitors. It is worth noting that our interest is not in developing markers for a particular drug, but rather to identify markers and pathways relevant to predicting response or resistance to inhibitors of the VEGF pathway, a validated therapeutic target in NSCLC. Our data thus far suggest there will be overlap in markers for BV and VEGFR TKIs but it will be important to determine whether this is true, particularly as more potent and specific inhibitors are tested.

Impact. This project has the potential to address a critical unmet need in NSCLC by producing validated, non-invasive blood-based predictive markers for identifying patients more likely to benefit from VEGF inhibitors, and can provide critical insights into mechanisms of resistance to guide future combination regimens.

There is recent evidence that metformin, a drug commonly used to treat diabetes, has anticancer effects and can significantly improve the rate of pathological response in breast cancers treated with chemotherapy. We propose to review clinical records of lung cancer patients treated over the past 25 years at Johns Hopkins to determine whether metformin treatment is associated with improved clinical response to chemotherapy for lung cancer. Because our preliminary laboratory work suggests that lung cancer cells with mutations of the LKB1 gene are particularly sensitive to metformin, we will secondarily determine whether mutations of the LKB1 gene (which is mutated in up to 40% of lung adenocarcinomas) are associated with a higher frequency of response to chemotherapy among these patients treated with metformin. In a second specific aim, we will conduct studies using xenografts of human lung cancers in the laboratory to further examine possible links between LKB1 mutations and response to metformin, with or without paclitaxel. Establishing such a link could lead to improved chemotherapy (using metformin in combination with conventional chemotherapy drugs) and to development of a biomarker that can identify patients with cancers who might have improved response to chemotherapy with metformin.

The most effective curative modality we have for lung cancer is still surgery, but, even with optimum surgery, many patients relapse and die of their lung cancer, and this fraction is improved only slightly by adjuvant chemotherapy, which is nonetheless toxic for everybody. Thus, we need to identify prognostic biomarkers (to define which patients are likely to relapse after surgery) and predictive biomarkers (to define who will benefit from adjuvant therapy, and what specific type of adjuvant therapy should be used on each particular patient). In this proposal, collaborators from two different lung cancer SPOREs in three different universities will use two sets of high information-content analyses of protein and RNA expression already acquired on a cohort of early-stage lung cancer patients who did not receive adjuvant chemotherapy and in a panel of cell lines rigorously characterized for sensitivity to platinum doublets, and will add to that data funded by this proposal on a new cohort of patients treated with adjuvant chemotherapy to generate our classifier. This classifier will then be reduced to a small number of RNA or protein assays and tested in a larger set of resected patients with and without adjuvant chemotherapy.

Pulmonary nodules found on imaging studies are a significant diagnostic problem. Most lesions are indeterminate, but because of the concern for lung cancer all patients require further evaluation with resultant radiation risk, significant cost, and delays in diagnosis. We believe that the development of a serum test for lung cancer would have a significant impact on patient care and outcomes. We recently identified a novel panel of 25 serum autoantibodies associated with non-small cell lung cancer (NSCLC) and constructed a protein microarray containing the autoantigens. In a pilot study, we tested the microarray with human sera and developed a classification algorithm that incorporated microarray test results with nodule size. This algorithm had an overall sensitivity of 81% and specificity of 83% for the diagnosis of lung cancer. We now propose to further develop this panel and hypothesize that an autoantibody signature along with nodule size will be able to determine which patients have lung cancer.

Lung cancer is the leading cause of death for men and women in the United States. Screening methods for the early detection of lung cancer must be developed to circumvent the phenomenon that sees the majority of patients being diagnosed with advanced, and thus incurable, lung cancer. While computerized tomography was demonstrated to reduce mortality by 20% in a high-risk smoker patient cohort, further examination of this method, complementary screening methods, and tools to screen less at-risk populations are needed to make the detection of early-stage lung cancer most beneficial. Using the expertise of our multidisciplinary clinical lung cancer team (Drs. Kelly, Gandara, Cooke, Yoneda, and Murin) with leaders in metabolomics and lipid biology (Dr. Fiehn) and glycomics (Drs. Miyamoto and Lebrilla), we aim to identify mass spectroscopy-detectible biomarkers of early lung cancer. The advent of “omics” technologies will allow us to expand the current pool of DNA and protein biomarkers and facilitate our vision of a systems-biology screening and/or diagnostic tool based on “biomarker panels.” Our approach begins with a discovery phase that will identify biomarkers in tumor and blood with the aim of finding blood-based biomarkers that most closely reflect tumor biology. Next we will proceed to the testing phase to determine the performance of these biomarkers in four independent cohorts (two lung cancer cohorts, healthy controls, and benign nodules). Biomarkers that meet our pre-specified criteria will require further evaluation that is beyond the scope of this proposal.

The overarching concept guiding our work is that combinations of biomarkers and imaging characteristics will be most robust in predicting which patients with lung nodules should receive an aggressive diagnostic and/or therapeutic approach and which patients should be monitored for nodule stability over time. Our objectives are to improve two biomarker tests that already have shown significant promise-- sputum chromosomal aneusomy by fluorescence in situ hybridization (CA-FISH) and the analysis of multiple serum analytes by a highly multiplexed aptamer-based assay developed by SomaLogic, Inc.-- and then to compare test performance with predictive models based on nodule and patient characteristics. These objectives are based on two separate hypotheses: 1. The current CA-FISH assay panel can be further improved by the development of two sub-panels, one for detection of squamous cell carcinoma of the lung (SCC) and the second for the detection of adenocarcinoma of the lung (AC). 2. The current aptamer-based analysis can be further improved by developing separate assays specifically designed for the detection of AC, SCC, K-ras, and EGFR-mutated lung tumors. The assays will then be applied to sputum and blood specimens from the Pittsburgh SPORE Pittsburgh Lung Screening Study (PLuSS) and Colorado SPORE Lung Nodule cohorts, in concert with analysis of the CT characteristics of the corresponding lung nodules, including size, volume, density, location, or airway involvement, to compare test performances and to assess whether combining tests would be valuable.

Evaluating suspicious pulmonary nodules in the setting of a lung cancer diagnostic work-up is a common clinical management challenge. High resolution multi-detector computed tomography visualizes many noncalcified lesions that could represent either potentially lethal cancers or benign processes. Although the vast majority of these nodules are benign, non-neoplastic lesions, a subset will be malignant and in these cases, patients could benefit from immediate treatment. A non-invasive test that maintains a high sensitivity and specificity across the spectrum of clinical presentations of pulmonary nodules would be of tremendous clinical benefit by reducing the number unnecessary invasive procedures and minimizing the time between detection and definitive treatment. The overall goal for the proposed studies is to develop a convenient and low-cost serum test that will, either alone or in combination with clinical features, accurately classify indeterminate nodules and direct proper clinical management. Our approach to develop such a test will be to profile and identify which proteins detected in the serum at the time of the detection of these nodules were immunogenic and induced the production of circulating autoantibodies. Differences in the autoantibody profile between the malignant and non-malignant groups will then be further evaluated against additional clinical specimens from both our institution and two pre-specified collaborator institutions to identify the optimal test for this urgent clinical need. Once completed, we anticipate this test will aid clinicians in deciding the optimal treatment for patients found to have suspicious nodules and improve the current standard of care.

Melanoma has the highest propensity to metastasize to the brain once the disease has spread, while non-small cell lung cancer (NSCLC) has the highest incidence of brain metastases among all cancers. Approximately 50,000 patients develop brain metastases from these diseases every year. Patients with brain metastases have exceedingly limited therapeutic options since they are typically excluded from clinical trials. Recently a number of new systemic immune therapies, such as nivolumab (BMS-936558) and lambrolizumab (MK-3475), two inhibitors of PD-1 (an immune inhibitory molecule), have led to dramatic responses in metastatic melanoma (MM) and NSCLC. However, these therapies have not been studied in patients with untreated brain metastases.  Little is known about immune cells in the brain, and whether the immune system can be stimulated to reject cancer cells. Moreover, little is known about what tumor characteristics are associated with response to MK-3475 in brain metastases. This proposal includes innovative correlative biomarker studies incorporated into a clinical trial uniquely designed to study the activity of MK-3475 in patients with limited, small brain metastases in MM and NSCLC. This drug carries the hope to provide prolonged responses similar to responses seen in other metastatic sites, and might enable patients to avoid radiation therapy, and improve their outcome. As part of this clinical trial, we will collect brain and other metastatic specimens from patients enrolled and study the expression of immune inhibitory molecules utilizing a novel quantitative assay and algorithms that we have developed in preliminary work. These studies will facilitate selection of patients who are more likely to respond to this class of drugs.