Molecular classification of pulmonary sarcomatoid carcinomas suggests new therapeutic opportunities.

Background: Pulmonary sarcomatoid carcinoma (SC) is a rare disease with poor prognosis and with strong inter- and intratumor heterogeneity. However, molecular classification is currently focused on activating MET mutations. We sought to better characterize the molecular diversity of SC using mutational signatures that reflect different mutational processes, such as tobacco-associated adducts (signature 4), BRCA1/BRCA2 deficiency (signature 3), or APOBEC enzyme deamination (signatures 2 and 13).

Patients and methods: Whole-exome sequencing was carried out in 15 SC patients and on data from 10 previously published cases. Hierarchical clustering and consensus non-negative matrix factorization were carried out for samples classification based on mutational signatures.

Results: In the two series, SC distributed between two clusters (C): Csig4 (characterized by signature 4) and Csig2-3-13 (signatures 2, 3, and 13). Csig4 exhibited more frequent MAPK pathway mutations than Csig2-3-13 (pooled series: n = 10/14 versus 2/11, P < 0.05, respectively) and stronger PD-L1 expression (our series: n = 6/9 versus 1/6, P = 0.12). MET alterations were only found in Csig2-3-13 (pooled series: n = 5/11 versus 0/14, P = 0.009), as well as BRCA1/BRCA2 (n = 3/11 versus 0/15), EGFR (n = 1), and IDH1 (n = 1) mutations. Csig2-3-13 patients had better overall survival than Csig4 patients (median: >45 versus 7 months, respectively, P = 0.001).

Conclusions: Our study suggests that SC presents at least two clusters comprising different mutational processes, gene alterations, and PD-L1 expression. New potential treatment possibilities are immune checkpoint inhibitors in Csig4 and specific targeted agents in Csig2-3-13. These findings should encourage clinicians to conduct broad molecular and immunological testing in SC patients beyond MET exon 14 alterations.