The frequency of targetable somatic genomic alterations is known to differ based on a patient's ancestry; for example, in lung adenocarcinoma EGFR mutations are more common in East Asians than Europeans. Accurately characterizing these differences will help to illuminate differences in the genetic etiology of cancer between populations.


We inferred estimated population ancestry for more than 100,000 de-identified patients from comprehensive genomic profiling (CGP) of tumor specimens assessed in the course of routine clinical care. More than 40,000 germline single nucleotide polymorphisms (SNPs) are sequenced as part of our CGP assay which were also characterized in the publicly available 1000 Genomes data. We used the 1000 Genomes data to train and validate a classifier using these overlapping SNPs to classify individuals into one of five inferred population groups, estimated to be of predominantly African, European, Central and South American, South Asian, or East Asian ethnic origin. We then performed statistical testing to identify differences in tumor somatic genomic properties between the five inferred ancestry groups for each cancer type.


We found that the inference of population ancestry from tumor sequence data and its comparison to somatic mutations recapitulated known differences between populations. Specifically, in lung adenocarcinoma, EGFR mutations were significantly more common in tumor specimens from individuals with inferred East Asian SNP-inferred ancestry (rate=0.52) compared to individuals with inferred European ancestry (rate=0.17, p < 10-119), and KRAS mutations were more common in tumor specimens from individuals with inferred European ancestry (rate=0.40) than individuals with inferred East Asian ancestry (rate=0.14, p < 10-60). In lung adenocarcinoma, we also found that tumor mutational burden (TMB), a biomarker for immunotherapy response, differed significantly between populations. Tumor specimens from patients with inferred East Asian ancestry had the lowest TMB (4.5 mutations/Mb median, 4.7% >20), then from patients with inferred European ancestry (6.3 median, 12.8% >= 20), and inferred African ancestry the highest (9.0 median, 21.2% >=20). Finally, we found that in glioblastoma, TERT promoter mutations were more common in specimens from individuals with inferred European ancestry (rate=0.69) compared to specimens from individuals with inferred East Asian ancestry (rate=0.44). Based on the statistical power provided by this large cohort, novel ancestry based differences in gene alteration rates will be presented across multiple cancer types.


Clear differences in genomics based on SNP-inferred ancestry were observed. In particular, we found that TMB differs significantly between populations in lung adenocarcinoma, suggesting that the likelihood of individuals benefiting from immunotherapy may differ between populations.

Citation Format: Caitlin F. Connelly, Jian Carrot-Zhang, Philip J. Stephens, Garrett M. Frampton. Somatic genome alterations in cancer as compared to inferred patient ancestry [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 1227.