Clinically Relevant Genomic Findings in Routine Oncology Practice: A 31-Gene Pan-Cancer Analysis Benchmarked Against MSK-IMPACT 50K
Petr Alekseevich Shatalov, Elena Alekseevna Kirillova, Egor Mikhailovich Veselovsky, Alexey Aleksandrovich Traspov, Maria Pavlovna Raygorodskaya, Maria Alexandrovna Revkova, Daria Valerievna Bagdasarova, Irina Alexeevna Leukhina, Anna Petrovna Shinkarkina, Alena Valerevna Murzaeva, Victoria Sergeevna Surkova, Nikita Maksimovich Filimonov, Maria Konstantinovna Trukhanova, Andrei Dmitrievich Kaprin, Petr Viktorovich ShegaiBackground/Objectives: Real-world genomic profiling can identify clinically relevant alterations, but observed frequencies and therapeutic relevance depend on assay design, diagnostic resolution, and cohort composition. We characterized a routine-practice solid-tumor cohort and benchmarked genomic patterns against MSK-IMPACT 50K. Methods: This retrospective study included 3018 patients and 3187 sequencing sample identifiers analyzed by targeted panels or exome sequencing at the National Medical Research Radiological Centre in 2021–2026. Diagnoses were harmonized using OncoTree. The principal analysis used a prespecified 31-gene single-nucleotide variant (SNV)/small insertion/deletion (indel) set restricted to OncoKB Oncogenic or Likely Oncogenic variants. Potential therapeutic relevance was assessed using an organ-level European Society for Medical Oncology Scale for Clinical Actionability of molecular Targets (ESCAT) tier I–II framework; microsatellite instability (MSI), missense-only tumor mutational burden (TMB), and homologous recombination deficiency (HRD)-sum were analyzed in assay-specific subsets. Results: Major organ-associated findings included KRAS (17.9%) and EGFR (14.0%) in lung tumors, PIK3CA in breast tumors (23.6%), KRAS in pancreatic (70.2%) and bowel tumors (35.8%), BRAF in skin tumors (36.4%), and PTEN in uterine tumors (39.3%). Selected frequencies differed from MSK-IMPACT. In breast cancer, PIK3CA frequency was lower in the dedicated triple-negative breast cancer (TNBC) cohort than in the remaining cohort (15.7% vs. 25.0%), but this explained only part of the benchmark difference. Among 3001 unique organ-mapped patients, 625 (20.8%) had at least one potential ESCAT tier I–II SNV/indel or MSI finding; 558 (18.6%) had tier I, 53 (1.8%) tier II, and 14 (0.5%) were classified as I/II as the highest category. Conclusions: Routine sequencing reproduced recognizable tumor-site-associated patterns while revealing quantitative differences from MSK-IMPACT. Approximately one fifth of organ-mapped patients had a potential ESCAT tier I–II finding, representing molecular therapeutic-information yield rather than confirmed treatment eligibility or benefit.