The results obtained after comparing 47 HLA-A and HLA-B allele groups among the LAP and their ancestral populations show that there is widespread variation between the genetic profiles of these admixed or exported populations. In the cluster analysis it is clear that most LAP have substantial Caucasian components, with the exception of some populations such as the Peruvians from Arequipa or the Afro-Brazilians from Paraná. This is in agreement with the uneven process of population replacement and the collapse of many Native American groups that took place throughout the continent.
However, PCO analysis showed that most LAP sit on a wide admixture arch that approaches the ancestral clusters. A few populations fall very close to or in the ancestral clusters, but most are scattered in intermediate regions. Interestingly, population samples that are likely to be a mixture of several LAP, such as those of the USA Hispanic immigrants and the Ibero-American expatriates in Madrid, sit in the center of the distribution. In fact, the heterogeneity of the Hispanic population in the US has been described using other markers [68
], showing differential admixture patterns between areas that have received mostly Mexican immigration and those that are predominantly colonized by Caribbean islanders from Cuba and Puerto Rico. In agreement with this, US Mexicans lie slightly closer to the Amerindian side on the PCO and locate between the Mexican populations from the center and the sample from the northern state of Sinaloa. This further illustrates the heterogeneity of the Mexican populations, where a stronger Caucasian component is preserved in the north of the country [43
], while the US Mexicans are likely to be a combination of northern and southern Mexican populations.
The stronger Caucasian component in some LAP can be attributed to recent European migration [14
], such as that of urban populations from Argentina and some Brazilian populations, or to relatively stronger Caucasian proportions generated at colonial times in areas where Amerindian populations were low at the time of the arrival of Europeans, which is thought to be the case of the Costa Rican Central Valley and the Colombians from Medellin [70
On the Caucasian-SSA axis of admixture, several Brazilian and Cuban populations can be found. It seems that for these populations, Amerindian admixture is very low or absent. This has been noted by others [72
] and it is argued that a dual admixture model is more likely to describe the patterns seen in these populations as opposed to a triple admixture model identified for other LAP. Although not included in our analysis because of the lack of molecular HLA data, serological HLA data from Panama [73
] and Puerto Rico [74
] suggest that these populations are likely to join this group, whereas the data from Uruguay suggest that its major population would cluster with the strong Caucasian component group [75
Our study is limited by both the availability of population data and the need to use HLA allele group data for comparison as opposed to high-resolution allele frequencies or haplotype frequencies. It is likely that an analysis of high-resolution frequencies would give finer results, but it would seriously diminish the number of populations that can be included in the analysis. However, the use of 47 allele groups from the most polymorphic genes in the human genome gives robustness to the analysis.
The effect of ethnicity on complications after HSCT has been suspected for many years [76
] but some studies have not shown such association [78
]. Hence, there is growing interest in unraveling the genetic-ethnic component of GVHD in HLA-compatible HSCT. Currently, there is a project within the International Histocompatibility Working Group that aims at analyzing the risk of GVHD after HSCT in unrelated donor pairs according to the ethnic origin of both patients and donors, based on previous findings in sibling transplantations in isolated and general populations of certain countries [10
]. Preliminary results in a cohort of unrelated transplants showed that Hispanic pairs have high risks of mortality and acute GVHD (grades III-IV) only second to African American pairs. Moreover, Hispanic-Hispanic pairs had the highest risk of relapse [79
]. Both analyses were carried out having Asian/Pacific (mostly Japanese) ethnically matched pairs as the reference group. These findings suggest that ethnic heterogeneity in the Hispanic population may be playing a role on the risk of complications after HSCT, and the complexity of the admixture patterns illustrated in this study and others is likely to account for much of this variation. Also, ethnicity has been associated with other complications after HSCT such as chronic GVHD [80
]. Moreover, an increased risk of complications has been reported specifically for Hispanic groups in North America when compared to other ethnicities in terms of survival [81
] and treatment failure [82
It is likely that the evidence for differential outcome in different ethnic groups could be explained, at least in part, by differences in allele frequencies in genes that are relevant to the immune response and that show variable interethnic polymorphism, such as the cytokine genes [83
]. Moreover, polymorphisms in other genes such as those that intervene in drug metabolism or drug targets may play a role in the way patients from different ethnicities respond to treatment in HSCT, especially in admixed populations [84
LAP show widespread variation in their genetic profiles, and this complicates genetic association studies made in these populations. There is noticeable variation not only between regions and countries, but also between areas of the same country [43
]. Furthermore, the presence of minority populations of different ethnic composition adds to the complexity of population stratification in Latin America. Additionally, many populations remain to be studied. If an ethnic component is to be used as one prognostic factor affecting the risk of complications after HSCT, the application of this concept in Latin American populations will have to take into account the great diversity found among the different populations derived from this region and the different population subgroups generated by different admixture histories. Consequently, there is need of a more detailed understanding of the genetic profiles of the LAP, in order to be able to accurately stratify genetic risk in HSCT.
It is also important that a better definition of individual ancestry in LAP is reached in view of the evident limitations of both self-reported [87
] and researcher-assigned ethnicities [41
]. To this purpose, the use of a more objective assignment based on ancestry markers [69
] is likely to increase the accuracy of the information derived from these studies. Hopefully, a finer characterization of the risk of complications after HSCT in LAP will help foresee these complications and increase the access and success of transplantation in these populations.