RESUMEN
Lung cancer is the leading cause of cancer mortality worldwide, with non-small cell lung cancer (NSCLC) being the most prevalent histology. While immunotherapy with checkpoint inhibitors has shown outstanding results in NSCLC, the precise identification of responders remains a major challenge. Most studies attempting to overcome this handicap have focused on adenocarcinomas or squamous cell carcinomas. Among NSCLC subtypes, the molecular and immune characteristics of lung large cell carcinoma (LCC), which represents 10% of NSCLC cases, are not well defined. We hypothesized that specific molecular aberrations may impact the immune microenvironment in LCC and, consequently, the response to immunotherapy. To that end, it is particularly relevant to thoroughly describe the molecular genotype-immunophenotype association in LCC-to identify robust predictive biomarkers and improve potential benefits from immunotherapy. We established a cohort of 18 early-stage, clinically annotated, LCC cases. Their molecular and immune features were comprehensively characterized by genomic and immune-targeted sequencing panels along with immunohistochemistry of immune cell populations. Unbiased clustering defined two novel subgroups of LCC. Pro-immunogenic tumors accumulated certain molecular alterations, showed higher immune infiltration and upregulated genes involved in potentiating immune responses when compared to pro-tumorigenic samples, which favored tumoral progression. This classification identified a set of biomarkers that could potentially predict response to immunotherapy. These results could improve patient selection and expand potential benefits from immunotherapy.
RESUMEN
BACKGROUND: Tumor mutational burden (TMB) is a recently proposed predictive biomarker for immunotherapy in solid tumors, including non-small cell lung cancer (NSCLC). Available assays for TMB determination differ in horizontal coverage, gene content and algorithms, leading to discrepancies in results, impacting patient selection. A harmonization study of TMB assessment with available assays in a cohort of patients with NSCLC is urgently needed. METHODS: We evaluated the TMB assessment obtained with two marketed next generation sequencing panels: TruSight Oncology 500 (TSO500) and Oncomine Tumor Mutation Load (OTML) versus a reference assay (Foundation One, FO) in 96 NSCLC samples. Additionally, we studied the level of agreement among the three methods with respect to PD-L1 expression in tumors, checked the level of different immune infiltrates versus TMB, and performed an inter-laboratory reproducibility study. Finally, adjusted cut-off values were determined. RESULTS: Both panels showed strong agreement with FO, with concordance correlation coefficients (CCC) of 0.933 (95% CI 0.908 to 0.959) for TSO500 and 0.881 (95% CI 0.840 to 0.922) for OTML. The corresponding CCCs were 0.951 (TSO500-FO) and 0.919 (OTML-FO) in tumors with <1% of cells expressing PD-L1 (PD-L1<1%; N=55), and 0.861 (TSO500-FO) and 0.722 (OTML-FO) in tumors with PD-L1≥1% (N=41). Inter-laboratory reproducibility analyses showed higher reproducibility with TSO500. No significant differences were found in terms of immune infiltration versus TMB. Adjusted cut-off values corresponding to 10 muts/Mb with FO needed to be lowered to 7.847 muts/Mb (TSO500) and 8.380 muts/Mb (OTML) to ensure a sensitivity >88%. With these cut-offs, the positive predictive value was 78.57% (95% CI 67.82 to 89.32) and the negative predictive value was 87.50% (95% CI 77.25 to 97.75) for TSO500, while for OTML they were 73.33% (95% CI 62.14 to 84.52) and 86.11% (95% CI 74.81 to 97.41), respectively. CONCLUSIONS: Both panels exhibited robust analytical performances for TMB assessment, with stronger concordances in patients with negative PD-L1 expression. TSO500 showed a higher inter-laboratory reproducibility. The cut-offs for each assay were lowered to optimal overlap with FO.
