Candidate malaria susceptibility/protective SNPs in hospital and population-based studies: the effect of sub-structuring.

Background: Populations of East Africa including Sudan, exhibit some of the highest indices of genetic diversity in the continent and worldwide. The current study aims to address the possible impact of population structure and population stratification on the outcome of case-control association-analysis of malaria candidate-genes in different Sudanese populations, where the pronounced genetic heterogeneity becomes a source of concern for the potential effect on the studies outcome.

Methods: A total of 72 SNPs were genotyped using the Sequenom iPLEX Gold assay in 449 DNA samples that included; cases and controls from two village populations, malaria patients and out-patients from the area of Sinnar and additional controls consisting of healthy Nilo-Saharan speaking individuals. The population substructure was estimated using the Structure 2.2 programme.

Results & Discussion: The Hardy-Weinberg Equilibrium values were generally within expectation in Hausa and Massalit. However, in the Sinnar area there was a notable excess of homozygosity, which was attributed to the Whalund effect arising from population amalgamation within the sample. The programme STRUCTURE revealed a division of both Hausa and Massalit into two substructures with the partition in Hausa more pronounced than in Massalit; In Sinnar there was no defined substructure. More than 25 of the 72 SNPs assayed were informative in all areas. Some important SNPs were not differentially distributed between malaria cases and controls, including SNPs in CD36 and NOS2. A number of SNPs showed significant p-values for differences in distribution of genotypes between cases and controls including: rs1805015 (in IL4R1) (P = 0.001), rs17047661 (in CR1) (P = 0.02) and rs1800750 (TNF-376)(P = 0.01) in the hospital samples; rs1050828 (G6PD+202) (P = 0.02) and rs1800896 (IL10-1082) (P = 0.04) in Massalit and rs2243250 (IL4-589) (P = 0.04) in Hausa.

Conclusions: The difference in population structure partly accounts for some of these significant associations, and the strength of association proved to be sensitive to all levels of sub-structuring whether in the hospital or population-based study.