Immunotherapy is used to treat almost all patients with advanced non-small cell lung cancer (NSCLC); however, identifying robust predictive biomarkers remains challenging. Here we show the predictive capacity of integrating medical imaging, histopathologic and genomic features to predict immunotherapy response using a cohort of 247 patients with advanced NSCLC with multimodal baseline data obtained during diagnostic clinical workup, including computed tomography scan images, digitized programmed death ligand-1 immunohistochemistry slides and known outcomes to immunotherapy. Using domain expert annotations, we developed a computational workflow to extract patient-level features and used a machine-learning approach to integrate multimodal features into a risk prediction model. Our multimodal model (area under the curve (AUC) = 0.80, 95% confidence interval (CI) 0.74–0.86) outperformed unimodal measures, including tumor mutational burden (AUC = 0.61, 95% CI 0.52–0.70) and programmed death ligand-1 immunohistochemistry score (AUC = 0.73, 95% CI 0.65–0.81). Our study therefore provides a quantitative rationale for using multimodal features to improve prediction of immunotherapy response in patients with NSCLC using expert-guided machine learning.

No abstract

Background Lineage plasticity, the ability to transdifferentiate among distinct phenotypic identities, facilitates therapeutic resistance in cancer. In lung adenocarcinomas (LUADs), this phenomenon includes small cell and squamous cell (LUSC) histologic transformation in the context of acquired resistance to targeted inhibition of driver mutations. LUAD-to-LUSC transdifferentiation, occurring in up to 9% of EGFR-mutant patients relapsed on osimertinib, is associated with notably poor prognosis. We hypothesized that multi-parameter profiling of the components of mixed histology (LUAD/LUSC) tumors could provide insight into factors licensing lineage plasticity between these histologies. Methods We performed genomic, epigenomics, transcriptomics and protein analyses of microdissected LUAD and LUSC components from mixed histology tumors, pre-/post-transformation tumors and reference non-transformed LUAD and LUSC samples. We validated our findings through genetic manipulation of preclinical models in vitro and in vivo and performed patient-derived xenograft (PDX) treatments to validate potential therapeutic targets in a LUAD PDX model acquiring LUSC features after osimertinib treatment. Results Our data suggest that LUSC transdifferentiation is primarily driven by transcriptional reprogramming rather than mutational events. We observed consistent relative upregulation of PI3K/AKT, MYC and PRC2 pathway genes. Concurrent activation of PI3K/AKT and MYC induced squamous features in EGFR-mutant LUAD preclinical models. Pharmacologic inhibition of EZH1/2 in combination with osimertinib prevented relapse with squamous-features in an EGFR-mutant patient-derived xenograft model, and inhibition of EZH1/2 or PI3K/AKT signaling re-sensitized resistant squamous-like tumors to osimertinib. Conclusions Our findings provide the first comprehensive molecular characterization of LUSC transdifferentiation, suggesting putative drivers and potential therapeutic targets to constrain or prevent lineage plasticity.

No abstract

No abstract