P-354: Whole genome sequencing reveals significant genetic admixture in multiple myeloma patients, impacting assessment of etiology

Introduction: In the U.S. population, having African ancestry (AA) conveys a higher risk of multiple myeloma (MM) than white ancestry (WA), however, published whole genomic sequencing (WGS) data is predominantly from WA populations. The Polyethnic-1000 (P1000) is a multi-institutional initiative with the New York Genome Center, investigating cancers having a higher prevalence in AA populations. We hypothesize that self-described race and ethnicity, according to limited proscribed categories, are insufficient to delineate biological contribution to MM development. Unique biological insights may require comprehensive definition of genetic origin, considering heterogeneity within historically defined groups.

Methods: We applied Admixture, a composition profile based on SNPs corresponding to geographical reference populations, to both targeted sequencing (MSK-Heme-IMPACT) and WGS. Admixture estimates proportion per patient from 5 super populations (African [AFR], American [AMR], European [EUR], East Asian, South Asian), with WGS allowing resolution of 23 subpopulations. From patients self-identifying as Black or Hispanic, 111 samples had IMPACT and 101 WGS. Results were compared with the CoMMpass dataset and WA WGS, to a total WGS cohort of 1221.

Results: Genomic complexity hidden by self-reported race and ethnicity was revealed by genetically determined Admixture. IMPACT data estimated that while 42 (38%) had AFR super family contribution ≥0.9, 37 (33%) had 0.5-0.8 and 32 (29%) had < 0.5. 24% self-identifying as Hispanic had ≥0.5 AFR.

From WGS, 46/101 had ≥ 0.25 from at least 2/23 subpopulations (consistent with grandparents) with 94 having ≥0.125 from different subpopulations (consistent with great-grandparents). From those with ≥0.25 from different populations, 33 were within AFR while 11 were across super-populations.

Hierarchical clustering analysis of the P1000 samples together with 134 CoMMpass samples having AFR > 0.1 produced 3 main patient clusters; 2 with predominantly AFR ancestry, clustering by proportion of AFR contribution, while 1 cluster was highly admixed with EUR and AMR. Clustering based on subpopulations was highly analogous.

Considering the entire WGS cohort produced 6 clusters; 5 had a predominant super population, with the majority of AFR collapsed into 1 cluster, while 1 cluster was heterogenous in composition. Self-described WA patients also had significant admixture revealed.

Conclusions: Self-reported race and ethnicity don’t allow consideration of the significant variability of genetic admixture present in our patients, with likely insufficient granularity on inherited risk of MM. Our genomic datasets could benefit from increasing AA and Hispanic representation. Clustering analysis should consider both the entire cohort and separately analyze non-WA samples, to allow adequate resolution of genetic origin. Ongoing studies will incorporate genetic admixture alongside somatic genomic assessment to accurately investigate progression risk from precursor disease to MM.