Comprehensive DNA Repair Landscape Analysis Reveals Novel Small Cell Lung Cancer Biology
Introduction
Small cell lung cancer (SCLC) is the most lethal form of lung cancer. The five-year overall survival rate for SCLC patients is only ~7%, a statistic that has not improved for decades. Compared to non-small cell lung cancers, SCLCs highly express DNA repair proteins. Despite this acknowledgement, a comprehensive analysis of the DNA repair machinery within SCLC has not been completed. This study was performed to analyze how heterogeneity in the DNA repair landscape shapes unique SCLC phenotypes.
Methods
We applied a novel DNA repair analysis method to IMPOWER133 (n = 271), MD Anderson GEMINI (n = 85), and SCLC-CellMiner (n = 116) datasets. This single-sample weighted expression (ssWE) method uses transcriptomic data to assess regulation of ten single-strand break repair, double-strand break repair, translesion synthesis, and damage sensing pathways within individual tumors. Importantly, this method considers all pathway effectors and applies a scaling factor to capture effector essentiality to repair completion. Unsupervised k-means clustering was used to group samples following ssWE analysis. Chi-square tests were used to test for significant enrichment patterns between DNA repair and established SCLC subtype assignments. Wilcoxon tests were used to assess differences in mRNA and protein expression across groups.
Results
Unsupervised clustering demonstrated that IMPOWER133 and MDACC GEMINI SCLCs were robustly split into three clusters as defined by their DNA repair phenotypes [DNA Damage Response (DDR) High, Intermediate, Low]. Further analysis demonstrated that expression of DNA damage responsive transcription factors (E2F1, MYBL2, FOXM1) and the intra-S and G2/M checkpoint machinery varied significantly across DDR clusters (DDR High > DDR Intermediate > DDR Low). Importantly, application of our DDR analysis method confirmed that DDR clusters were faithfully recapitulated in SCLC cell line models. Analysis of cell line mRNA sequencing and reverse phase proteomic array (RPPA) data confirmed that cell line DDR clusters exhibited strong differences in DNA repair and cell cycle checkpoint effector expression. Also, we found that DDR clusters were significantly associated with SCLC subtypes (IMPOWER133 Χ2 p<0.001, MDACC Χ2 p<0.001). Most notably, inflamed SCLC (SCLC-I) were significantly under-represented in the DDR High cluster. Given this, we hypothesized that DDR Low tumors exhibited a more “inflamed” phenotype, regardless of subtype. To test this hypothesis, we isolated ASCL1+ SCLC (SCLC-A) tumors, which are considered “immune cold” compared to other subtypes, stratified them by their DDR status, and analyzed expression patterns of key tumor-intrinsic immune genes. Strikingly, we found that splitting treatment naïve SCLC-A tumors by their DDR status identified tumors that appeared more “inflamed” (DDR Low). DDR Low SCLC-A showed increased expression of MHC Class I and inflammatory cytokine genes. Conversely, DDR High SCLC-A tumors exhibited an immune evasive profile. Lastly, these findings were confirmed and further extended through analysis of SCLC cell line DDR clusters.
Conclusion
We provide the first evidence that SCLCs exhibit a spectrum of DNA repair machinery and cell cycle checkpoint pathway expression both within and across subtypes. Additionally, we find that DDR status may be a key determinant of “inflamed” biology independent of the SCLC-I subtype. Moving forward, our results have important implications for SCLC treatment and clinical trial design.