Single Base Substitution Mutation Profiling in Non-Matched and Matched Metastases across Various Tumor Histologies and Anatomic Sites

Abstract

<h3>Purpose/Objective(s)</h3> As the landscape of oligometastatic care continues to evolve, it has become clear that not all histologies (or anatomic sites) respond equivalently to systemic therapy or radiation. This study aims to gain more insight into metastatic tumor biology by evaluating single base substitution mutations across various metastatic sites and within matched primary-metastatic patient samples. <h3>Materials/Methods</h3> We evaluated our prospectively collected institutional biorepository for metastatic (Met) tumors which underwent targeted exome sequencing for 1,327 cancer-related genes. The patient cohorts included: 1) non-matched Mets (N-Met; n=794 [707 tumors]) and 2) a subset of matched primary (Pri)-Met (M-Pri-Met; n=207 [390 tumors]). Exome data were enriched for somatic non-silent mutations by filtering out artifacts and known germline and silent mutations. Mutation signatures based on 6 nucleotide base alterations (96 subtypes) were generated and compared across samples followed by hierarchical clustering analysis to assess mutation signature correlation to the COSMIC mutation database. The mutation signatures were assessed across unique histology groups from Prim (n=17) and Met (n=16) sites and compared by Kruskal-Wallis test. Also, counts of individual mutations were related to given mutation signatures. A generalized linear mixed model was used to infer the odds ratio (OR) of mutation incidence in Pri vs Met tumors. <h3>Results</h3> Across all Mets, we found that C>T and G>A alterations were identified in 23% and 21% of tumors, respectively, whereas T>A was <2%. Notable differences were seen in the proportions of mutation signatures (n=9) across Pri histology (range p < 0.03 to 1.0e⁻¹⁶) and among different Met sites within a given histology. For example, our 6^th signature was associated with AID/APOBEC family activity and varied across Pri histology (p = 1.0e⁻¹³) and within Met sites (p = 4.3e⁻⁰⁹). In contrast, our 4^th signature was related to <i>ERCC2</i> mutation, but only differed across Pri histology and not Met site. The median number of mutations related with mutation signatures was 27 (range: 1-1404) with the highest counts in melanoma. Within our M-Pri-Met analysis, several genes were identified to be either private or shared among tumors. For example, within the M-Pri-Met breast (n=93) group, <i>EPHB4</i> mutation was concordant in 58% of samples with a LogOR 1.38 (95% CI 1.2-13.2; p=0.02) in Pri vs Met. In contrast, <i>BRAF</i> mutation was 40% concordant with higher odds of being in Met vs Pri (p=0.04). No significant differences were identified in the lung M-Pri-Met cohort (n=45). Though, individual mutations did not appear to be different in lung histology, we did observe differences in the proportions of mutation signatures within individual patient tumors. <h3>Conclusion</h3> A deeper knowledge of similarities and differences among metastases may aid in biologically-informed metastasis-directed therapy. This analysis highlights diversity in mutation type and incidence among metastases.