Exploring the prognostic and predictive impact of genomic loss of heterozygosity (LOH) in patients with metastatic colorectal cancer (mCRC)

3576 Background: Genomic LOH consists in the loss of chromosomal regions and is associated with the homologous recombination repair (HRR) system deficiency (dHRR). In ovarian cancer, LOH-high predicts benefit from platinum-based chemotherapy and PARP inhibitors. In mCRC, the role of LOH has been poorly investigated. Methods: An NGS-based assay (FoundationOne CDx; Foundation Medicine, Inc. Cambrige, MA) was used to determine the percentage of genomic LOH and the presence of pathogenetic mutations in HRR-related genes in archival chemo-naïve tumor tissues of mCRC patients included in a real-world registry and in the AtezoTRIBE (NCT03721653) and AVETRIC (NCT04513951) studies. Both these trials assessed the combination of an anti-PDL1 (atezolizumab or avelumab) with the triplet FOLFOXIRI plus bevacizumab or cetuximab. The prespecified cut-off of ≥14.08 for LOH-high described for mCRC (Sokol et al., JCO Precis Oncol 2020) was adopted. Tumors with at least one biallelic alteration in any of the 27 genes involved in the HRR pathway (Riaz et al., Nat Commun 2017) included in the FoundationOne CDx panel (BARD1, BRCA1, BRCA2, BRIP1, MRE11A, NBN, PALB2, RAD51, RAD51B, RAD51C, RAD51D, RAD52, RAD54L, RBBP8, XRCC2, ABL1, ATM, ATR, BAP1, CDK12, DNMT3A, ERCC4, FANCA, FANCC, FANCG, FANCL, PARP1) were defined dHRR. Results: Overall, 196 samples were analysed. None of 7 MSI-H tumors were classified as LOH-high. Fourteen (7%) and 6 (3%) of 189 MSS tumors were classified as LOH-high and dHRR, respectively. In particular, in LOH-high subgroup, 3 (21%) tumors were defined dHRR, while 3 (50%) tumors were classified LOH-high among dHRR tumors. LOH-high tumors were more frequently BRAF mutated (p = 0.019) and dHRR (p = 0.006) compared to LOH-low. Among patients receiving triplet chemotherapy +/- biologic agent alone (N = 55) or with an anti-PDL1 (N = 58), an interaction effect was shown between the effect of the addition of the checkpoint inhibitor and LOH with higher benefit in the LOH-high subgroup (N = 10) in terms of both PFS (p interaction = 0.002) and OS (p interaction < 0.001). In the cohort of patients not receiving the anti-PDL1, longer PFS was observed in patients with LOH-low (N = 125) respect to LOH-high (N = 6) tumors (12.1 vs 5.1 months, HR: 0.11, 95%CI: 0.04-0.26, p < 0.001). No differences in PFS or OS were reported between patients treated with first-line oxaliplatin- (N = 40) vs irinotecan-based doublets (N = 25) or with the triplet FOLFOXIRI (N = 55) vs doublets (N = 65) according to LOH status. No prognostic or predictive impact of HRR deficiency was shown. Conclusions: In MSS mCRC, LOH-high was associated with biallelic alterations in the HRR system, worse prognosis and higher benefit from the addition of anti-PDL1 agents to chemotherapy. Considering the low number of LOH-high tumors in our study, these results deserve confirmation in larger cohorts.