Single-nucleotide variants (SNVs) occur throughout the genome and constitute most of the human genetic diversity (Robert and Pelletier, 2018). These genomic variants can alter all steps of gene expression and protein functionality, ultimately modifying the effects of environmental exposures. In the context of drug responses and the safety of commonly prescribed drugs, SNVs are key factors accounting for interindividual variability. Therefore, characterization of populations in terms of the frequency distribution of SNVs relevant to pharmacogenetics provides a powerful approach for evaluating the suitability of its clinical applications in drug selection and dosage optimization, improving therapeutic efficacy while reducing adverse effects.

The American continent is highly heterogeneous and presents the second-greatest number of abundant region-specific alleles (i.e., that are common in one continent but absent in the rest of the world), reflecting their unique evolution (Fedorova et al., 2022). A study of the Consortium of the Ibero-American Network of Pharmacogenetics and Pharmacogenomics (CEIBA-RIBEF) showed that interethnic variability in clinically relevant drug metabolizing enzymes (CYP2D6, CYP2C9, and CYP2C19) in Latin America leads to a lack of correlation between the ‘‘predicted’’ enzyme metabolic capacity and the genotype (Naranjo et al., 2018). Differences in admixture history in Latin American populations have important implications for the frequency distribution of variants associated with drug absorption, distribution, metabolism, and excretion (ADME) responses between and within populations (Céspedes-Garro et al., 2016), making extrapolation of data not suitable. By characterizing the admixture proportions in Latin American countries and the distribution of ADME variants of pharmacogenetic relevance, it should be possible to share health policies and logistic solutions between similar populations. At the individual level, individual ancestry proportions may determine the probability of having a pharmacogenetic relevant genotype (Bonifaz-Pena et al., 2014). Thus, collectively studying SNVs across human populations is of great interest to develop precision medicine.

Cuba is the most populated country in the Caribbean and has a rich genetic heritage. The population is essentially the result of admixture between Spaniards, West Africans, and, to a lesser degree, native Amerindians who inhabited the island. Evidence from historical and anthropological studies indicates that native populations were drastically reduced to a few thousand people within 50 years after 1492 (Guanche, 2020). Studies on the Cuban population, using ancestry informative marker (AIM) panels purposely designed to reveal differences in allele frequencies between Native American, West African, and European populations, have confirmed the small contribution of the Amerindian component to the current Cuban genetic background (Cintado et al., 2009; Diaz-Horta et al., 2010; Marcheco-Teruel et al., 2014; Fortes-Lima et al., 2018). Spanish immigration took place during more than four centuries and consisted mainly of male individuals, who intensely mixed with other ethnic components of the Cuba population (i.e., native and African women). During the XX century, a marked tendency was observed among the Spanish immigrants to establish themselves in the Western–Central areas of the country compared to the eastern region (Guanche, 2020). As a result, insights into Cuban contemporary population and admixture dynamics show different ancestry proportions across provinces and regions of the country (Marcheco-Teruel et al., 2014; Fortes-Lima et al., 2018).

Comparisons to other populations have identified interethnic variability in CYP2D6 and CYP2C9 allele frequencies and metabolic phenotypes among Cubans, Spaniards (Llerena et al., 2014), and Nicaraguans (Llerena et al., 2012). In addition, there is evidence of geographic variation within the Cuban population in allele and genotype frequency of rs1045642 MDR1 when skin color categories are considered (Rodeiro et al., 2022). As complex admixture processes have occurred, we hypothesized that the frequency of pharmacogenetic variants and drug responses may vary within the country. Understanding the genetic heterogeneity and admixture of Cubans between geographical regions should have important implications for the design and interpretation of clinical trials, the implementation of pharmacogenetic tools for drug prescription and dosage adjustment, and the extrapolation of data from other, more homogeneous populations. Despite the increasing number of studies describing genetic biomarkers in Cubans (Sotomayor-Lugo et al., 2022; Camacho et al., 2011; García et al., 2010), information about population allele frequencies from a representative sample of the Cuban population is still limited. In accordance, this study aimed to describe the distribution of 43 SNVs in 25 target pharmacogenes for the Cuban population and in relation to other populations. It was also intended to characterize the distribution of the SNVs by considering the descriptive variables such as place of birth and skin color. This knowledge should provide valuable data for (Robert and Pelletier, 2018) understanding the allele and genotype distribution of variants in clinically relevant pharmacogenes in the country (Fedorova et al., 2022), implementing personalized medicine approaches and pharmacogenetic testing in public health policies, and (Naranjo et al., 2018) establishing public health priorities across the country.

In this section, we studied the frequencies of 43 variants in genes affecting drug metabolism, transport, and pharmacological efficacy in healthy Cuban volunteers. Five out of the six SNVs that showed HWE deviations (Table 1) displayed statistical differences by skin color and four by birthplace. Except for VEGF rs3025039, the markers presented moderate-to-large genetic differentiation in pairwise comparisons within Cubans and with other populations. Hence, population stratification seems to be the most plausible explanation for the observed HWE deviations.

Ancestry analyses in this study revealed admixture diversity among individuals and within group of individuals from the same birthplace and skin color. In general, admixture estimations detected here ranged closely to a previous report on Cubans (Fortes-Lima et al., 2018). An element derived from the population structure described may have pharmacogenetic implications. The European and African components together account for 95% of the genetic diversity in this sample, as compared to the 5% for Amerindian ancestry. Consequently, European and African ancestry will have considerably greater impact on the frequency distribution of variants with pharmacological relevance.

There is a strong relationship between admixture proportions and census categories in the Cuban population, as suggested by Marcheco-Teruel et al. (2014). In agreement with these data, the estimations of European ancestry detected here decreased from self-reported White to Black individuals, and the opposite trend was observed in regard to the African ancestry, while admixed individuals displayed intermediate values for both ancestry components. By place of birth, European ancestry was predominant in the Western and Central regions, while the Eastern region had the highest African contribution, followed by the Western and Central regions.

Genetic differentiation, as described by F_ST pairwise comparisons within the Cuban population, suggested that drug responses may vary individually by skin color categories and regions of the country, with admixture proportions being a key factor. In this line, geographic differences observed in 11 genotype frequencies could also be associated, at least in part, with admixture patterns geographically structured across the island. Estimates of African and European ancestry by birthplace were comparable to reported proportions (Fortes-Lima et al., 2018). The central region showed a genotype distribution similar to that of the Spanish population, while individuals from the Eastern and Western provinces showed intermediate frequencies relative to the parental populations. Eastern provinces had relatively higher African contributions (average 34%) than Western and Central provinces. Data were readily extended to the greater genetic differentiation found in pairwise F_ST comparisons of the Central vs. Eastern regions, showing six markers with moderate divergence.

F_ST pairwise comparisons with other populations indicated genetic similarity to Spanish and Latin American populations. However, interethnic variability was observed. CYP1A1 rs1048943 (CYP1A1*2C) is greatly represented in Latin American populations (Amerindian ancestry), whereas in Africans and Europeans, its frequency is low (Pérez-Morales et al., 2008). Given the small Amerindian component estimated in Cubans, moderate divergence was expected in comparison to Latin American populations (Figure 1). The global allele distribution of rs776746 (CYP3A5*3) increased from 18% to 94% in African to European populations (Fairley et al., 2019), respectively. On average, the frequency of CYP3A5*3 is approximately 80% in Latin Americans, but it was found to be lower in this study (69%), compared to other reports on Chileans (76%) (Roco et al., 2012), Brazilians (73%) (Suarez-Kurtz et al., 2014), and admixed Mexicans (74%) (Fricke-Galindo et al., 2016). On the contrary, rs2740574 (CYP3A4*1B) is found in African populations due to a suggested selection factor against non-African populations involving vitamin D metabolism (Schirmer et al., 2006). The CYP3A4*1B frequency in Cubans was higher than in Ecuadorians (Sinués et al., 2008), admixed Mexicans (Gonzalez-Covarrubias et al., 2019), and Chileans (Roco et al., 2012), but similar to that in Brazilians (Rodrigues-Soares et al., 2018).

Previous studies have reported about CYP allele frequencies in Cubans (Llerena et al., 2014; Rodrigues‐Soares et al., 2020). In individuals with European ancestry, rs35742686 (CYP2D6*3) is found at a low frequency (Zhou et al., 2017), as well in Latin Americans (Naranjo et al., 2018), and it is virtually absent in Chileans (Roco et al., 2012). In agreement with these data, CYP2D6*3 was not detected in the present sample cohort. CYP2D6*10 is related to Asian ancestry but is also distinctive of African populations together with CYP2D6*17. However, a higher CYP2D6 *10 allele frequency was reported in this study (18%) and Brazilians (Suarez-Kurtz et al., 2012a). The six variant alleles studied in the CYP2C subfamily ranged closely to reports on Brazilians (Suarez-Kurtz et al., 2012b) and CYP2C9*2 and *3 to Venezuelans (Flores-Gutiérrez et al., 2017). CYP2C19*3 is less frequent than CYP2C19*2 worldwide and has its highest frequency in Asians (Zhou et al., 2017). It is almost non-existent in Latin America (Suarez-Kurtz et al., 2012b), it is not present in Chileans (Roco et al., 2012), and it was also not detected in this Cuban cohort. One of the highest frequencies of CYP2C19*2 and *3 is found in Venezuela, associated with their Amerindian component (Castro de Guerra et al., 2013).

Frequency estimates of EPHX1 rs2234922 and AHR rs2066853 are 8% and 15% in Mexicans (Pérez-Morales et al., 2011), respectively, and 8% for AHR rs2066853 in Brazilians (Abnet et al., 2007), whereas we observed lower frequencies for both SNVs. The estimated frequency of rs1045642 (ABCB1) in this study (37%, A allele) was intermediate to the frequencies reported for Spaniards (48%) and African Americans (16%) but similar to those observed in Brazilians (39%) (Scheiner et al., 2010) and in a previous Cuban study (36.5%), with comparable demographical variables (Rodeiro et al., 2022).

The frequency of the GSTT1 null genotype in Cubans was closer to that observed in Spanish than to Africans (Kasthurinaidu et al., 2015), with an F_ST value in Cuba vs. Africa comparison near the moderate differentiation threshold (F_ST = 0.045). Results are consistent with those of other studies stating that the GSTM1 null genotype is more frequent than GSTT1 worldwide (Nakanishi et al., 2022). The prevalence of both deletions was similar to admixed Mexicans (Palma-Cano et al., 2017) and Chileans (Roco et al., 2012) but lower than in Brazilians (Magno et al., 2009) and Colombians (Ramírez et al., 2019).

Variants and genotypes (germline DNA) do not depend on disease status or evolution, in contrast to somatic mutations; thus, its clinical use can be considered with strong confidence as predictive biomarkers. Several SNVs in genes related to DNA repair processes and growth factors have clinical annotations on PharmGKB with different levels of evidence (Whirl‐Carrillo et al., 2021). Therefore, validating the usefulness of the recommended markers would promote a safe and cost-effective use of drugs in the clinical management of patients.

To illustrate this statement, 23% of the global population carries the EGFR rs712829 T allele. Similarly, the frequency for Hispanics is 21%, with notable ethnic variations. This variant is less common in Peruvians, but reaches 28% in Puerto Ricans (Fairley et al., 2019) and 23% in Mexicans (Torres-Jasso et al., 2015). A higher frequency was observed in Cubans (32%), closer to the frequencies found in Europeans and Iberians. Regarding VEGF rs3025039, Hispanics have the highest frequency for the T variant allele, particularly among Mexicans and Peruvians. However, these variants were less common in the Cuban sample, and a low frequency is also observed in Iberians, Colombians (Fairley et al., 2019), and Brazilians (Carvalho et al., 2021). This is important in the context of cancer susceptibility, side effects, and response rate to EGFR and VEGF-targeting drugs. The variant allele of EGFR correlates with increased receptor expression (Heist and Christiani, 2009), whereas the VEGF variant is associated with significantly lower VEGF plasma levels (Sibertin-Blanc et al., 2015), which would make carriers suitable for EGFR-target therapies and less appropriate for VEGF-target therapies. Meanwhile, moderate differentiation to Africans was observed for rs1042522 (TP53), which had a 63% prevalence in Cubans. Studies in Latin America agree on the predominance of the TP53 Arg72 allele (Rodrigues-Soares et al., 2018; Olloquequi et al., 2022). Few variants in DNA repair enzymes have been characterized in Latin American populations. The frequency of rs13181 (ERCC2), rs861539 (XRCC3), and rs25487 (XRCC1) in Cubans resembled those of Mexicans (Pérez-Morales et al., 2011) and Brazilians (Duarte et al., 2005).

Attention has been drawn to variants in methylation enzymes or methyl recognition binding proteins (Majchrzak-Celińska et al., 2019). SNVs in epigenetic writer–reader–eraser enzymes had noticeable differences in prevalence among Africans, Europeans (Iberians), and Latin Americans. Hispanics have the highest frequency of MBD5 rs16828708 (Fairley et al., 2019). Its frequency in Cubans matched that of the Spanish, displaying moderate divergence from Latin Americans (Figure 1). Allele variants in MBD2/4 and EZH1/2 enzymes showed moderate differentiation in F_ST pairwise comparison to Africa. According to the 1000 Genomes Project database, the frequencies observed here are closer to Spaniards than to Africans. The MGMT enzyme is of special interest; its expression is ubiquitous in normal cells, with considerable variation in the activity in the same tissue of different individuals, and it is frequently overexpressed in tumors vs. non-neoplastic tissues (Christmann et al., 2011). Genome-wide association studies have revealed that non-coding SNVs of MGMT affect gene expression (Huang et al., 2018). The F_ST values indicated low divergence from Spain and Latin America, but several markers showed large-to-moderate genetic differentiation to Africans. The F_ST statistic provides an approach to estimating the genetic divergence of pharmacogenetic variants and its potential impact on clinical response to drugs in different populations or strata within a population (Suarez-Kurtz and de Araújo, 2022). It also supports that the distribution of genetic variants among Cubans is influenced by the large variability in ancestry proportions.

The cost-effectiveness of pharmacogenetic testing has been demonstrated, supporting its integration into clinical practice (Morris et al., 2022; Plumpton et al., 2019). The pharmacogenes described here have been widely studied for their usefulness in predicting and tailoring efficient therapeutic responses (Zhao et al., 2021) and have clinical annotations on the PharmGKB Knowledge Base (Altman, 2007), with variable levels of evidence for current clinical guidelines. The prevalence of variants in these pharmacogenes in the Cuban population justifies the implementation of pharmacogenetic tools. While the use of AIM panels to infer ancestry in pharmacogenetic studies is recommended, the use of descriptive variables like skin color categories and region of birth may also be informative in different contexts. For instance, it may explain how admixture diversity may be translated into different prevalence of genetic variants among skin color categories and country regions. This information should be taken into account when designing clinical trials and public health policies related to pharmacogenetic testing. As an illustrative example, important pharmacogenetic variants, such as CYP3A4*1B and CYP3A5*3, related to immunosuppressive drug responses, were found to have high divergence between self-identified White/Black individuals and between Central and Eastern provinces. Among the human CYP3A enzymes, CYP3A4 and CYP3A5 are considered the most important in drug metabolism. Approximately half of the medications that are metabolized by P450 are CYP3A substrates (Saiz-Rodríguez et al., 2020). Thus, the present results provided evidence for its applications in public health priorities across the island. A study limitation is that recruitment was most successful in Western provinces, mainly in Havana. Future comprehensive studies should promote the participation of individuals from all provinces in the Central and Eastern regions.

In summary, this data was a contribution to pharmacogenetic characterization of Cuban population and to the development of precision medicine in the country.