Genomic profiling of matched tumor samples in patients with rectal cancer undergoing total neoadjuvant therapy.

Abstract

213 Background: Total neoadjuvant therapy (TNT) is increasingly used for locally advanced rectal cancer. However, which patients benefit most from TNT remains to be elucidated. In this study we sequenced a TNT cohort with long-term follow-up to identify genomic correlates of TNT efficacy. Methods: Patients with locally advanced rectal cancer who underwent TNT followed by surgery (Oct 2016-June 2020) were identified at our institution. MSS patients with paired tissues (normal, biopsy, and surgery) with successful DNA extraction and sequencing were analyzed. Retrospective chart review collected patient demographics and clinical outcomes. Genomic analyses were performed with the R maftools package. Descriptive statistics were calculated with Fisher’s exact, Wilcoxon rank sum, and Wilcoxon rank sum exact tests. Univariate survival analyses (UVA) were calculated with Cox proportional hazards regressions. All statistical analyses were performed using R (v4.3.2). Results: 53 patients met inclusion criteria and 32 patients with paired tissues (64 samples) were obtained and successfully sequenced. All patients received chemotherapy, most commonly FOLFOX (81.3%) as part of TNT and a 5-FU based chemoradiation therapy (RT) regimen. Almost all patients received 50.4 Gy (93.8%). Median followup was 57 months (IQR: 47-68). There were 24 pathological responders (R) [complete/partial] and 8 nonresponders (NR) [stable disease/progression]. R vs NR were not statistically different across demographic or treatment characteristics. Local recurrences were infrequent (9.4% 3/32). No covariates were prognostic on UVA survival analyses. However, NR was associated with worse distant metastases free survival (DMFS) (HR = 4.11, CI 1.01-16.8, p = 0.048) on UVA. 5-year DMFS was 83% for R and 50% for NR, and the median DMFS was not reached for R, 41 mo for NR. Tumor mutation burden (TMB) did not change with treatment (pre = 4.5 vs post = 6.2 mutations/Mb; r, 1.7-10.3, p=0.59) and was not different on biopsy between R and NR (4.1 vs 5.9, p=0.09). Pre-treatment, the most mutated gene was APC (81%), [majority nonsense/frameshift (88%)], followed by TP53 (69%), [missense (59%)]. When comparing frequent mutations (n ≥ 5), PIEZO1 was more mutated in NR (63%, 5/8) than R (4% 1/24 patients) (p = 0.023 [multiple testing corrected]). Putative radioresistant KRAS-TP53 co-mutations were uncommon with similar representation in both groups (R 12.5%, 3/24, NR 12.5%, 1/8). Conclusions: Pathologic response positively correlated with DMFS in this cohort with long-term followup. PIEZO1 mutations were negatively prognostic for pathologic NR to TNT. This uncovers a potential new role in chemoRT resistance in addition to a previously reported association with metastatic spread. Further evaluation of PIEZO1 as a prognostic biomarker may help personalize appropriate treatment for patients with locally advanced rectal cancer.