The prevalence and prognostic impact of mutations promoting chromatin remodelling dysregulation in non-resectable or recurrent/metastatic adenoid cystic carcinoma

Background: Few effective drug therapies exist for patients with recurrent/metastatic (R/M) adenoid cystic carcinoma (ACC). Mutations in genes encoding chromatin remodelling proteins are described in almost 50% of ACC patients (pts). Novel drug therapies are being developed to target these pathways, and may have clinical utility in this sub-group. We sought to classify mutations in chromatin remodelling genes in ACC based on predicted pathogenicity and to determine the impact of chromatin remodelling dysregulation (CRD) on clinical outcomes. Methods: Matched clinical-genomic data from 269 pts with non-resectable or R/M ACC were included in this study. 130 pts were prospectively recruited to an ethically approved study. For these pts, DNA extracted from FFPE tissue was sequenced on a commercially available platform to detect point mutations, indels and copy number variation in 324 genes. In addition, clinical-genomic data from 139 ACC pts were collected from cBioPortal (MetTropism, Cell 2021) for analysis. Mutations were classified as pathogenic using COSMIC, ClinVar and OncoKB. Univariate survival analysis was performed to determine the impact of one or more mutations in chromatin remodelling genes on survival from first recurrence or metastasis. p values determined using Kaplan-Meier and log-rank. Results: In 269 pts with non-resectable or R/M ACC, mutations promoting CRD were identified in ARID1A (11%), CREBBP (5%), EP300 (5%), KMT2C (1%), KMT2D (5%), KDM6A (13%), and SETD2 (3%). CRD mutations were identified in 94/269 (35%) pts, with 27/269 (10%) having 2 or more CRD mutations. For patients in whom CRD mutations were present survival from recurrence was decreased (median OS 4.6 v 8.2 years, HR 1.65 (95% confidence interval (CI) 1.13-2.40), p = 0.01). Analysis of each individual CRD gene identified that association with decreased survival from recurrence was significant for mutations in KDM6A (n = 35; median OS 4.2 vs 6.5 years; HR 2.01 (95% CI 1.25-3.22); p = 0.004), CREBBP (n = 11; median OS 4.2 vs 5.9 years; HR 2.97 (95% CI 1.20-7.38); p = 0.019) and SETD2 (n = 7; median OS 3.7 vs 5.9; HR 2.47 (95% CI 1.002-6.09); p = 0.042). Previous studies have identified NOTCH pathway activation and TP53 loss-of-function as prognostic. In a secondary analysis of pts without NOTCH activation and/or TP53 mutations (n = 202), OS from recurrence was decreased in those with CRD mutations (n = 58 median OS 5.7 vs 9.2 years, HR 1.5 (CI 95% 0.95-2.55), p = 0.07). Conclusions: We have identified a novel prognostic group of ACC pts characterised by mutations promoting CRD, which may have potential therapeutic options. Alterations in EP300/CREBBP/ARID1A may provide a rationale for treatment with CREBBP/EP300 inhibitors currently in clinical development. Significant co-occurrence with NOTCH gain of function may provide a rationale for future combination studies.