The Impact of Ethnicity and Genetic Ancestry on Disease Prevalence and Risk in Colombia

Abstract

Currently, the vast majority of genomic research cohorts are made up of participants with European ancestry. Genomic medicine will only reach its full potential when genomic studies become more broadly representative of global populations. We are working to support the establishment of genomic medicine in developing countries in Latin America via studies of ethnically and ancestrally diverse Colombian populations. The goal of this study was to analyze the effect of ethnicity and genetic ancestry on observed disease prevalence and predicted disease risk in Colombia. Population distributions of Colombia’s three major ethnic groups – Mestizo, Afro-Colombian, and Indigenous – were compared to disease prevalence and socioeconomic indicators. Indigenous and Mestizo ethnicity show the highest correlations with disease prevalence, whereas the effect of Afro-Colombian ethnicity is substantially lower. Mestizo ethnicity is mostly negatively correlated with six high-impact health conditions and positively correlated with seven of eight common cancers; Indigenous ethnicity shows the opposite effect. Malaria prevalence in particular is strongly correlated with ethnicity. Disease prevalence co-varies across geographic regions, consistent with the regional distribution of ethnic groups. Ethnicity is also correlated with regional variation in human development, partially explaining the observed differences in disease prevalence. Patterns of genetic ancestry and admixture for a cohort of 624 individuals from Medellín were compared to disease risk inferred via polygenic risk scores (PRS). African genetic ancestry is most strongly correlated with predicted disease risk, whereas European and Native American ancestry show weaker effects. African ancestry is mostly positively correlated with disease risk, and European ancestry is mostly negatively correlated. The relationships between ethnicity and disease prevalence do not show an overall correspondence with the relationships between ancestry and disease risk. We discuss possible reasons for the divergent health effects of ethnicity and ancestry as well as the implication of our results for the development of precision medicine in Colombia.

A review of ancestrality and admixture in Latin America and the caribbean focusing on native American and African descendant populations

“…Furthermore, a close look at the LAC countries also reveals the different distribution of ancestry proportions in subpopulations and regions, as observed in Cuba (Fortes-Lima et al, 2018), Garifuna population in Honduras (Herrera-Paz et al, 2010), Mexico (Rubi-Castellanos et al, 2009;Silva-Zolezzi et al, 2009), Venezuela (Moura et al, 2015), Colombia (Godinho et al, 2008;Ibarra et al, 2014;Mogollón Olivares et al, 2020;Chande et al, 2021), Bolivia (Heinz et al, 2013;Taboada-Echalar et al, 2013), Paraguay (Simão et al, 2021), Argentina (Godinho et al, 2008), Chile (Eyheramendy et al, 2015) and Brazil (Kehdy et al, 2015;Moura et al, 2015;Mychaleckyj et al, 2017;Secolin et al, 2019;. Also, as observed in several studies, the admixture events were sex biased and occurred between European males and Native American and/or sub-Saharan African females in Cuba (Mendizabal et al, 2008), Jamaica (Deason et al, 2012), Dominican Republic (D'Atanasio et al, 2020), Nicaragua (Nuñez et al, 2010), Belize (Monsalve and Hagelberg, 1997), El Salvador (Lovo-Gómez et al, 2007;Casals et al, 2022), Panama (Migliore et al, 2021), Costa Rica (Carvajal-Carmona et al, 2003), Mexico (Green et al, 2000), Ecuador (González-Andrade et al, 2007), Colombia (Carvajal-Carmona et al, 2003), Brazil (Pena et al, 2011), Uruguay (Sans et al, 2002), and Argentina (Dipierri et al, 1998).…”

“…In Chile, ancestry was inferred from high-density SNP genotype data of 313 individuals from two case-control studies examining hantavirus infection and 22q11 microdeletion syndrome in residents of the country ( Eyheramendy et al, 2015 ). In Colombia, ancestry analysis among 624 individuals from Medellín and 99 individuals from Chocó revealed higher correlations between ancestry and disease prevalence risk by polygenic risk score (PRS) estimates ( Chande et al, 2021 ). In Peru, the Peruvian Genome Project ( Harris et al, 2018 ) and the 12G and 100G-MX Projects ( Romero-Hidalgo et al, 2017 ; Aguilar-Ordoñez et al, 2021 ) generated whole-genome data from Native American populations.…”

Racial Bias Can Confuse AI for Genomic Studies

“…Therefore, careful tests and well-prepared data are needed when applying the conclusions of genomic analyses to minor groups. It has been reported that bias is likely mitigated by carefully choosing appropriate databases and applying data harmonization methods [13], but methods like this are like addressing symptoms but not root causes. To fully address the issue of racial bias in AI models for clinical use, a top-down intervention involving regulatory bodies and publishing standards are claimed to be required [19].…”

“…Although race was regarded as a poor marker of genomics for years, it is now regarded as an incomplete understanding of human genetic variation [10]. Various correlations between race and diseases have been discovered [11][12][13][14], and genomic differences between races have been found to play a role in multiple genome-wide association studies (GWAS) [15][16][17] in various ways.…”

Factors associated with meeting the WHO physical activity recommendations in pregnant Colombian women

“…Concerning race/ethnicity, participants choose one of the following six ethnic groups: (1) Indigenous, (2) Rom , (3) Raizal del Archipiélago de San Andrés y Providencia , (4) Palanquero de San Basilio , (5) Afro-Colombian and (6) none of the above. We focused on the three largest ethnic groups in Colombia (i.e., Afro-Colombian, Indigenous and Mestizo) 25 , 26 . Pregnancy phase that was defined as follows: 1st trimester, 2nd trimester or 3rd trimester.…”