Genetic Ancestry Differences in Tumor Mutation and Expression in Early and Average Onset Colorectal Cancer

ASCO 2022 - Tempus-authored
Authors Rhead, Brooke, Hein, David, BSc, Pouliot,Yannick, Guinney, Justin, P De La Vega, Francisco, Sanford, Nina

Background: The incidence and mortality of early onset colorectal cancer (EOCRC), defined as
CRC diagnosed prior to age 50, are rising, in contrast to declining rates of average onset CRC
(AOCRC). Epidemiologic trends for CRC appear to differ by race/ethnicity, which could be
related in part to underlying differences in tumor mutation and gene expression. Previous
research has used self-reported race/ethnicity categories, which has been shown to be missing
or inaccurate, particularly in highly admixed groups such as Blacks and Hispanics. Hence, we
examined tumor profiles of EOCRC and AOCRC using genetic ancestry.

Methods: Genetic ancestry was inferred from de-identified data of 1,443 and 3,315 patients
with EOCRC and AOCRC, respectively, who underwent tumor profiling with the Tempus xT NGS
648-gene panel and targeted RNAseq. Ancestry informative markers were used to estimate
likelihoods for the five continental groups defined in the 1000 Genomes Project: Africa (AFR),
Americas (AMR), East Asia (EAS), Europe (EUR), and South Asia (SAS). We assessed the
association of genetic ancestry proportions (per 10% increase in each ancestry proportion) with
presence of protein altering somatic mutations and gene expression patterns in key driver genes in CRC.  We also assessed interactions between onset age (EOCRC vs. AOCRC) and genetic ancestry in somatic mutations.

Results: Across all ages, African ancestry was associated with higher odds of mutations in APC
and KRAS and lower odds of BRAF mutations (Table). East Asian ancestry was associated with
higher odds of mutations in NTRK3 and TP53. For EOCRC, the association between KRAS and
African ancestry remained significant (OR 1.07, p=0.04). Furthermore, there was an interaction
(p=0.031) between onset age and race/ethnicity such that the association between African
ancestry and APC mutation was significant in AOCRC (OR=1.09) but not EOCRC (OR=1.00).

Conclusions: The use of genetic ancestry instead of race/ethnicity classes provides a more
quantitative and precise profile of shared genetic background that can underlie biological race
differences in cancer etiology and outcomes.

p.AFR p.AMR p.EAS p.SAS gene OR
2.48E-04 1 1 1 APC 1.05 (AFR)
1.94E-04 0.51 1 1 BRAF 0.91 (AFR)
7.63E-13 1 1 1 KRAS 1.08 (AFR)
1.00E+00 1 0.036 1 NTRK3 1.11 (EAS)
1.00E+00 1 0.036 1 TP53 1.06 (EAS)