Non-small cell lung cancer (NSCLC)

Background & Significance: Phase I studies with the anti-programmed death-1 antibody (anti-PD-1), nivolumab, in advanced non-small-cell lung cancer (NSCLC) have demonstrated remarkably durable responses in heavily pretreated patients that lasted a median 17 months. However, in these studies very few patients have had tumors available for immune analysis, and basic data on immune measures pre- and post-treatment are not available. This study will move the treatment paradigm forward by assessing the safety and feasibility of preoperative anti-PD-1 in NSCLC and providing detailed information on the effect of PD-1 modulation on the lung tumor immune microenvironment.

Methods: Patients with newly diagnosed stage IB-IIIA NSCLC will receive two doses of nivolumab 3mg/kg IV on day-28 and day-14 prior to planned surgery on day 0. Serial blood samples for immune markers will be obtained. The primary endpoint of this study is the safety and feasibility of preoperative nivolumab in resectable NSCLC. For the exploratory immune analyses, initial tumor biopsy, post-treatment resection specimen, and serial peripheral blood samples will be evaluated for expression of immune markers with the primary comparison being intra-patient, pre- and post-treatment.

Experimental Approach & Statistical Analysis Plan: As this is the first time that nivolumab has been administered pre-operatively in NSCLC, the study will commence with a run-in phase where six patients will be monitored continuously for adverse events through eight weeks following surgery. Overall, the study will continue without pause to a 12-patient expansion phase if none or one of the six patients experiences DLT in the run-in. Detailed statistical plans for the safety and feasibility endpoints and exploratory immune endpoints are described in the main body of this application.

Better methods are required to accurately risk stratify patients with a solitary pulmonary nodule (SPN) who may undergo unnecessary procedures to exclude a cancer diagnosis. 18F-FDG Positron Emission Tomography (PET) is a widely available clinical test used to evaluate concerning SPNs and stage malignant lung nodules. It therefore facilitates clinical assessment but remains an imperfect tool for the diagnosis, localization, and prognostication of SPNs, be they infectious or malignant in nature. Furthermore, it performs well for staging lung cancer, but inaccurately categorizes patients in some cases who receive futile thoracotomies for disseminated disease, or conversely, are undertreated for localized disease. The applicant therefore plans on investigating whether circulating molecular biomarkers can increase the utility of 18F-FDG-PET imaging in clinical practice using statistical models that incorporate in vivo and in vitro diagnostics The concept here is “one-stop shopping,” where a patient can obtain routine molecular imaging and molecular diagnostics from routine blood work during an appointment that in combination increase the accuracy of a clinician’s diagnosis and facilitate the appropriate decision-making process. For this proposal we will develop a model that incorporates a circulating microRNA profile to discriminate infectious from malignant SPNs for patients along with clinical and FDG-PET data and, secondly, identify patients with malignant nodules and detectable circulating tumor cells to determine whether these cells upstage or augment clinical staging or prognosis post-operatively when compared to FDG-PET alone.

Approximately 30% of patients with stage 1 non-small cell lung cancer (NSCLC) will die from recurrent disease despite surgical resection . Currently, there is a lack of reliable molecular predictors being routinely used in the clinic for identifying patients at high risk for developing recurrent disease and therefore may benefit from more aggressive therapies. To date several groups have applied microarray-based methods for global gene expression profiling of frozen tumor tissues in an attempt to identify prognostic ‘signatures’ in patients with early stage NSCLC.  Building upon this, through a meta-analysis of seven independent microarray studies Govindan et al. were able to identify a 64-gene signature that is highly predictive of survival in patients with stage I NSCLC. However, recent advances in high-throughput sequencing have enabled an unbiased view of the transcriptome including novel transcriptional events such as gene fusions, splicing variants, and novel transcripts. This offers an opportunity to incorporate additional markers, which may have eluded previous technologies, but could serve for improving a predictive signature of survival in non-small cell lung cancer (NSCLC). However, despite the potential of RNA-Seq to improve existing predictive signatures, we are still faced with the challenge of translating these findings into a platform where they can be reliably applied to formalin fixed paraffin embedded (FFPE) clinical tissue specimens routinely collected in tertiary referral centers. Therefore, this proposal focuses on refining the 64-gene predictive signature using transcriptome sequencing and developing a NanoString nCounter assay to reliably predict survival of patients with resected stage I NSCLC, using RNA derived from FFPE tumor blocks. To accomplish this we have proposed the following specific aims: (1) use transcriptome sequencing to detect molecular predictors of outcome in patients with stage I NSCLC, (2) validate the technical performance of the NanoString nCounter using an RNA-Seq predictive signature across FFPE tumor blocks of resected stage I NSCLC, and (3) clinically validate the prognostic power of the predictive signature across an independent sample set. Overall, if the proposed aims are met, this research could lead to an improved assay that could be translated into future prospective studies for risk-stratifying stage I NSCLC patients.

The development of non-invasive diagnostic tools to detect lung cancer at an early stage is the subject of active investigations because of its enormous potential impact on the number-one cause of cancer-related deaths. We have recently discovered by shotgun proteomics profiling 164 candidate biomarkers that are differentially expressed in stage 1 lung cancer tissue compared to non-malignant matched controls tissues and found in the circulation. Validation of these candidates by traditional methods using western blot or ELISA is a slow, low throughput, and expensive process. In this application, we propose to take advantage of the individual’s immune response to capture and detect autoantibodies directed against these 164 candidate protein targets identified by shotgun proteomics. To achieve this goal, we intend (1) to establish immune-based assays for the detection and quantitation  of lung cancer autoantibodies using an indirect ELISA; (2) to validate the diagnostic performance of our autoantibody signature in blood samples from individuals with early-stage lung cancer or matched nodule controls; and (3) to determine whether the diagnostic performance of our panel of autoantibody signatures augments the diagnostic accuracy of known predictive models for the evaluation of lung nodules. These studies may lead to the development and early validation of a new molecular diagnostic tool for lung cancer.

Recently, the National Lung Screening Trial reported a 20% reduction in lung cancer mortality and lung cancer chemoprevention trials targeting the arachidonic acid pathway have demonstrated decreases in lung cancer associate markers .These studies highlight possibilities for future reductions in lung cancer mortality through early detection and chemoprevention. Our group has previously identified smoking- and lung cancer-specific gene expression alterations in cytologically normal airway epithelial cells that can serve s a clinically relevant biomarker for the early detection of lung cancer. We propose to extend our studies of the field of injury by profiling  these cells from high-risk smokers with preinvasive lesions (dysplasia) and matched controls without lesions by RNA-Seq. We will characterize the molecular alterations (coding, noncoding, and novel RNA expression and alterative splicing) associated with the field of injury in premalignancy and identify whether these differences are similar in both current and former smokers. Next, we will develop molecular signatures that predict subsequent progression or regression of premalignant lesions as well as changes that occur as a result of progression or regression. Collectively, these signatures will aid in the stratification of high-risk smokers for chemoprevention trials, serve as intermediate markers of efficacy in these trials, and identify new targets for chemoprevention. Finally, we will explore whether these molecular signatures re associated with occult lung cancer or future lung cancer development in independent cohorts as common alterations may indicate early events in lung carcinogenesis and have additional utility as biomarkers in the early detection of lung cancer.

Cancer therapy needs to be better personalized to individual patients, each of whom carries a unique spectrum of mutations, inherited and/or acquired. One of the best cases for this is lung cancer, where molecular subgrouping patients and targeting their mutations has shown promise. While many groups have searched fervently for additional tumor-acquired mutations that are associated with drug sensitivity and resistance, few have emerged since the discovery of the initial mutations, such as EGFR. In contrast, we have applied our expertise in microRNAs (miRNAs), their biologic function and interaction with oncogenes, and as a result have discovered a new class of inherited, germ-line mutations that appear to have strong potential as companion diagnostics. These mutations are in a relatively unexplored region of the genome, the 3’ untranslated regions (UTRs), once thought to be junk DNA. The first such discovered biomarker, the “KRAS-variant,” is an inherited 3’UTR mutation in KRAS, which affects microRNA (miRNA) binding and KRAS pathway expression, as well as tumor biology. The KRAS-variant was first shown to be a genetic marker for an increased risk of developing non-small cell lung cancer (NSCLC). In addition, our data indicate that this variant represents an inherited form of an activated KRAS allele, one of the most important human oncogenes. Several groups have found that the KRAS-variant acts as a biomarker of poor outcomes across multiple cancers, including head and neck cancer, ovarian cancer, and colon cancer. We as well as others have shown that this biomarker is predictive of response to specific cancer treatments, and is not just prognostic of aggressive tumor biology. For example, the KRAS-variant predicts resistance to cisplatin in ovarian and head and neck cancer, resistance to cetuximab in combination with irinotecan chemotherapy, but sensitivity to cetuximab when delivered alone. Furthermore, our preliminary data in this proposal indicate that the KRAS-variant is a powerful companion diagnostic predicting poor response to specific agents, and excellent response to others. These findings, combined with evidence that this allele of KRAS can be targeted using siRNA and miRNA strategies that we have pioneered, leads us to believe that this KRAS-3’UTR mutation is a powerful companion diagnostic in lung cancer, and ultimately may be a critical new target for cancer therapy.

The genotype-directed use of EGFR TKIs in EGFR-mutant lung cancer patients has fundamentally changed the face of the disease with improved response to therapy, preserved quality of life, and prolonged progression-free survival. However, tumors that initially respond become resistant after ~ 1 year (i.e., acquired resistance, AR). The biology of AR is incompletely understood and is largely inferred from pre-clinical models. Our group at Mass General implemented a wide-scale program to perform repeat biopsies on EGFR-mutant patients with AR to first-generation EGFR inhibitors (gefitinib, erlotinib). This translational research paradigm has been instrumental in gaining insight into AR and has led to novel observations such as PIK3CA and BRAF mutations as mechanisms of AR and the abundant frequency of phenotypic changes at the time of AR like small cell lung cancer and epithelial-to-mesenchymal transitions. The repeat biopsy results have directly helped our patients as well, by informing choices about the most relevant clinical trials or alternative treatments. Currently, a new cohort of EGFR TKIs is entering the clinic: 2^nd-generation irreversible pan-HER inhibitors and 3^rd-generation T790M-specific inhibitors. Little is known about how AR develops to these newer therapies or how the biology of EGFR-mutant cancers evolves over time after exposure to sequential EGFR blockers.

Our proposal aims to discover the mechanisms of AR to next-generation EGFR TKIs and to understand the evolution of EGFR-mutant cancers as they face the selective pressure of sequential types of EGFR-directed therapies. To accomplish this we will collaborate across institutions to amass a large number of pre-treatment and post-treatment paired patient biopsies which we will interrogate via allele-specific genotyping as well as whole genome sequencing to define mechanisms of AR. In addition we will utilize an innovative technique of taking tumor tissue directly from a biopsy and deriving patient-specific in vitro and in vivo models in the laboratory. Expansion of these models will yield a larger supply of AR cancer material to aid descriptive and functional research about the biology of AR. In addition, establishment of pre-treatment patient-specific models will allow us to verify whether the mechanism of AR derived in the lab through ex-vivo therapy mirrors the eventual clinical mode of AR.

Small cell lung cancer (SCLC) exhibits aggressive clinical behavior and poor outcome and response to therapy. Moreover, there are no effective strategies for early detection of SCLC. Therefore, there is an urgent need to identify new strategies and markers for SCLC detection. We recently characterized the molecular airway field cancerization (FC) in non-small cell lung carcinoma (NSCLC) patients and identified airway profiles that varied spatially (n=1165) and with time (n=1395) following surgery. Moreover, we characterized the adjacent airway FC in NSCLC (n=1606 genes), a subset of which (n=299) significantly distinguished airways of smokers with lung cancer from those of healthy smokers and another (n=208) separated airways of different NSCLC subtypes. Although we previously showed that SCLCs and adjacent airways are characterized by higher level of allelic losses compared to those of NSCLC cases, the global airway FC in SCLC and its relevance to detection has not been explored. We hypothesize that the molecular airway FC in SCLC is different compared to that in NSCLC or benign lung disease and comprises biomarkers specific for early detection of SCLC. We will 1) perform global expression profiling of the airway FC and tumors or nodules in patients with SCLC, NSCLC, and benign disease for identification of airway markers specific to SCLC and 2) test identified airway mRNA markers for SCLC detection in bronchial and nasal compartments in patients with suspect lung cancer. We anticipate that our proposal, due to its unique resources and approaches, will identify novel markers for SCLC detection.

Lung cancer remains the leading cause of cancer death in the world for both men and women. Prevention studies have largely been ineffective. A chemoprevention trial using the prostacyclin analogue iloprost demonstrated improvement in endobronchial histology in former smokers. An ongoing trial is investigating lung cancer chemoprevention with pioglitazone, which decreases lung cancer risk in diabetics. The effects of iloprost and pioglitazone are mediated by PPARγ and its downstream targets. We hypothesize that response to iloprost and pioglitazone depends on expression of PPARγ and that measures of prostacyclin and PPARγ activity will correlate with histology. We will measure a PPARγ expression in biopsy tissues to determine if expression can predict response to iloprost. We will assess methylation and acetylation in tissues with loss of PPARγ expression, which could identify a potential role for demethylation or deacetylation agents in creating a response to chemoprevention. We will measure activity of iloprost and pioglitazone through QPCR of downstream targets. Previously generated tissues from mouse models of lung cancer chemoprevention will be used to correlate markers with human samples. miRNA have the potential to be important markers but there is little information on miRNA associated with lung cancer chemoprevention. We identify potential miRNA markers of iloprost and pioglitazone activity in dysplastic cell lines treated with iloprost or pioglitazone and in tissues from mouse models of chemoprevention. Data from this project will inform future clinical trials using iloprost and pioglitazone, improve targeted lung cancer chemoprevention, and identify relevant miRNA markers for future chemoprevention studies.

Whole genome pooled retroviral shRNA screens were performed in NSCLC cell lines A549 and NCI-H460. Out of nearly 18,000 genes, the top hit whose knockdown most reproducibly showed cytoxicity was PSMA1, a core subunit of the 20S proteasome. The combination of radiotherapy and PSMA1 knockdown significantly enhanced tumor control in a mouse xenograft model. We observed that proteasome inhibition results in increases in nuclear 1κBa, which in turn decreases NF-κB binding to the promoters of Fanconi Anemia/Homologous Recombination (HR) genes including FANCD2 and BRCA1. This reduces the expression and subsequent availability of these DNA repair proteins for recruitment to double strand breaks (DSBs) induced by radiation. These mechanistic studies provided the basis for the proposed hypothesis-driven translational research.

The central hypothesis is that high levels of HR-mediated repair of radiation-induced DNA DSBs are associated with poor tumor control, which may be improved with proteasome inhibitors. First, CT-guided fractionated radiotherapy ± bortezomib will be tested in diverse genetically engineered mouse models of NSCLC to identify mutations and other biomarkers associated with responses to radiation and/or proteasome inhibitors. Immunofluorescence and immunohistochemistry against proteins, identified in prior mechanistic studies, as well as gene expression profiling (GEP), will be performed. Then, the biomarkers will be examined in archived specimens from patients treated with radiotherapy, to assess associations with outcomes. DASL will be used to perform GEP studies using formalin-fixed, paraffin-embedded specimens. The goal will be to identify pathway-based expression signatures that predict poor outcomes with radiotherapy and improved outcomes with addition of proteasome inhibitors.