Homologous recombination deficiency signature (HRDsig) in advanced cutaneous melanoma (ACM): A genomic landscape study.

Abstract

9520 Background: HRDsig can aid in identifying tumors with DNA repair deficiencies, guiding PARP inhibitor (PARPi) use. It expands treatment options beyond breast cancer gene (BRCA) mutations enabling personalized therapy. Combining PARPi with targeted therapy or immunotherapy shows promise in overcoming resistance and improving outcomes for treatment-resistant ACM. Methods: Formalin-fixed paraffin-embedded (FFPE) primary tumors and metastatic biopsies from 9,576 ACM cases were analyzed using hybrid capture-based comprehensive genomic profiling (CGP), evaluating all classes of genomic alterations (GA). Central pathology review was conducted for all cases. Microsatellite instability-high (MSIH) status, tumor mutational burden (TMB), genomic ancestry, and mutational signatures were derived from sequencing data. HRDsig was assessed using copy number changes and genomic scars. PD-L1 expression was determined via immunohistochemistry [Dako 22C3; tumor proportion score (TPS)]. Statistical comparisons were made using Fisher’s exact test with Benjamini-Hochberg correction. Results: Among 9,576 ACM cases, 198 (2.1%) were HRDsig positive (HRDsig+). HRDsig+ cases when compared to HRDsig negative (HRDsig-) were older (median age: 69 vs. 67 years; p=0.034), more often female (49.5% vs. 36.2%; p=0.001), and had more GA per tumor (median: 7 vs. 6; p=0.026). HRDsig+ cases were more often of African (5.6% vs. 1.2%; p<0.0001) or American ancestry (8.1% vs. 4.1%; p=0.026) and less often European (84.3% vs. 94.0%; p<0.0001). GA more common in HRDsig+ included IGF1R (5.1% vs. 1.3%; p=0.003), KIT (11.6% vs. 5.6%; p=0.004), KRAS (7.1% vs. 2.7%; p=0.005), NF1 (38.4% vs. 21.4%; p<0.0001), RAD21 (6.7% vs. 3.0%; p=0.037), and TP53 (40.4% vs. 24.3%; p<0.0001). HRDsig- cases exhibited higher TMB (median: 13.8 vs. 6.1; p<0.0001), more frequent TMB >10 mutations/Mb (60.9% vs. 39.9%; p<0.0001), and more UV light exposure trinucleotide signatures (57.3% vs. 36.4%; p<0.0001). GA more common in HRDsig- cases included BRAF (including V600E) (44.6% vs. 23.2%; p<0.0001), CDKN2A (49.2% vs. 36.9%; p=0.003), NRAS (28.0% vs. 11.6%; p<0.0001), and TERT (74.4% vs. 29.8%; p<0.0001). In both positive and negative groups, PD-L1 low-level expression (1-49% TPS) was comparable (43.3%-39.7%), while BRCA1/2 alterations (<2.0%) and MSIH status (0.0%-0.1%) remained rare. Conclusions: HRDsig+ status is rare in ACM but more frequent in non-white genomic ancestries. HRDsig+ ACM cases are less likely to have BRAF GA, more likely to have KIT GA and lower TMB levels. These findings may guide the future development of clinical trials employing combination therapies and PARPi in ACM.