Faculty Bibliography

Visual inspection of histopathology slides is one of the main methods used by pathologists to assess the stage, type and subtype of lung tumors. Adenocarcinoma (LUAD) and squamous cell carcinoma (LUSC) are the most prevalent subtypes of lung cancer, and their distinction requires visual inspection by an experienced pathologist. In this study, we trained a deep convolutional neural network (inception v3) on whole-slide images obtained from The Cancer Genome Atlas to accurately and automatically classify them into LUAD, LUSC or normal lung tissue. The performance of our method is comparable to that of pathologists, with an average area under the curve (AUC) of 0.97. Our model was validated on independent datasets of frozen tissues, formalin-fixed paraffin-embedded tissues and biopsies. Furthermore, we trained the network to predict the ten most commonly mutated genes in LUAD. We found that six of them-STK11, EGFR, FAT1, SETBP1, KRAS and TP53-can be predicted from pathology images, with AUCs from 0.733 to 0.856 as measured on a held-out population. These findings suggest that deep-learning models can assist pathologists in the detection of cancer subtype or gene mutations. Our approach can be applied to any cancer type, and the code is available at https://github.com/ncoudray/DeepPATH .

Success of immune checkpoint inhibitors in advanced non-small-cell lung cancer (NSCLC) has invigorated their use in the neoadjuvant setting for early-stage disease. However, the cellular and molecular mechanisms of the early immune responses to therapy remain poorly understood. Through an integrated analysis of early-stage NSCLC patients and a Kras mutant mouse model, we show a prevalent programmed cell death 1/programmed cell death 1 ligand 1 (PD-1/PD-L1) axis exemplified by increased intratumoral PD-1+ T cells and PD-L1 expression. Notably, tumor progression was associated with spatiotemporal modulation of the immune microenvironment with dominant immunosuppressive phenotypes at later phases of tumor growth. Importantly, PD-1 inhibition controlled tumor growth, improved overall survival, and reprogrammed tumor-associated lymphoid and myeloid cells. Depletion of T lymphocyte subsets demonstrated synergistic effects of those populations on PD-1 inhibition of tumor growth. Transcriptome analyses revealed T cell subset-specific alterations corresponding to degree of response to the treatment. These results provide insights into temporal evolution of the phenotypic effects of PD-1/PD-L1 activation and inhibition and motivate targeting of this axis early in lung cancer progression.

OBJECTIVES/OBJECTIVE:Pulmonary sarcomatoid carcinoma (PSC) is a rare malignant neoplasm that accounts for a small percentage of non-small-cell lung carcinoma (NSCLC). At least 10% of PSCs has a spindle and/or giant cell component, which is often associated with a poor prognosis. We reviewed the Surveillance, Epidemiology, and End Results (SEER) database for the clinicopathological characteristics and surgical outcomes of PSCs. METHODS:The SEER database (1973-2013) was queried for PSC. A comparison between PSC and other NSCLC patients was performed. Cox regression for overall survival (OS) and logistic regression for node-positive predictors were performed. A propensity-matched (1:2) analysis (including age, gender, grade and stage) among surgically treated cases was done to compare OS in PSC versus other NSCLCs. RESULTS:A total of 955 899 NSCLC patients were identified; of these, 4987 patients had been diagnosed with PSC (0.52%). Men represented 60.9% of cases, with a median age of 68 years. The median size of the tumour was 5 cm and 3.5 cm in PSCs and NSCLCs, respectively (P < 0.001). PSC patients had significantly less Stage I, more high-grade tumours, advanced T stage, N+ disease and M1 disease (P < 0.001). In the PSC cohort, the most significant predictor of N+ disease on multivariate analysis was advanced T stage (P < 0.001). Predictors of OS in Stages I/II PSC on multivariate analysis were advanced age [P < 0.001, hazard ratio (HR) = 1.03], male gender (P = 0.024, HR = 1.25), carcinosarcoma (P = 0.002, HR = 1.76), grade (P = 0.033, HR = 1.81), T stage (P = 0.003, HR = 1.75), N status (P = 0.001, HR = 1.90) and surgical resection (P < 0.001, HR = 0.58). Among matched surgically resected cohorts, a poorer prognosis for OS was evident in PSCs in early stages (I/II) than in other NSCLCs (P = 0.009). CONCLUSIONS:PSC patients present with more advanced stage and with worse survival outcomes than other NSCLC patients. While surgical resection conveys a survival advantage in PSC, this group represents a population at a high risk for relapse and should be evaluated for novel adjuvant therapies.

Purpose In response to advances in the field, the College of American Pathologists (CAP), the International Association for the Study of Lung Cancer (IASLC), and the Association for Molecular Pathology (AMP) recently updated their recommendations for molecular testing for the selection of patients with lung cancer for treatment with targeted tyrosine kinase inhibitors. ASCO has a policy and set of procedures for endorsing clinical practice guidelines that have been developed by other professional organizations. Methods The molecular testing guideline was reviewed for developmental rigor by methodologists. Then an ASCO Expert Panel reviewed the content and the recommendations. Results The ASCO Expert Panel determined that the recommendations from the CAP/IASLC/AMP molecular testing guideline are clear, thorough, and based upon the most relevant scientific evidence. ASCO endorsed the guideline with minor modifications. Recommendations This update clarifies that any sample with adequate cellularity and preservation may be tested and that analytical methods must be able to detect mutation in a sample with as little as 20% cancer cells. It strongly recommends against evaluating epidermal growth factor receptor (EGFR) expression by immunohistochemistry for selection of patients for EGFR-targeted therapy. New for 2017 are recommendations for stand-alone ROS1 testing with additional confirmation testing in all patients with advanced lung adenocarcinoma, and RET, ERBB2 (HER2), KRAS, and MET testing as part of larger panels. ASCO also recommends stand-alone BRAF testing in patients with advanced lung adenocarcinoma. Recommendations are also provided for testing methods for lung cancers that have a nonadenocarcinoma non-small-cell component, for patients with targetable mutations who have relapsed on targeted therapy, and for testing the presence of circulating cell-free DNA. Additional information is available at www.asco.org/thoracic-cancer-guidelines and www.asco.org/guidelineswiki .

Cell death is desirable in cancer cells and undesirable in organs with limited regenerative potential, like the heart. Cell death comes in many forms, but only apoptosis and to a lesser degree necrosis is currently relevant to the clinical imager. Noninvasive imaging of cell death is an attractive option to understand pathophysiology, track disease activity, and evaluate response to intervention. Apoptosis seems to be the most promising target for imaging cell death, because it could be reversible and might be modulated with interventions. Molecular, nuclear, optical, or magnetic resonance imaging-based methods have been developed to identify intermediate steps in the apoptosis cascade. Animal studies show promising results for noninvasive imaging in various cardiovascular diseases. Human studies have shown feasibility, but clinical use is yet inconclusive. Newer technologies offer promise, especially for tracking apoptosis in evaluation of novel therapeutic interventions.