Building Sub-Saharan African PBPK Populations Reveals Critical Data Gaps: A Case Study on Aflatoxin B1

Physiologically based pharmacokinetic (PBPK) models allow to simulate the behaviour of compounds in diverse physiological populations. However, the categorization of individuals into distinct populations raises questions regarding the classification criteria. In previous research, simulations of the pharmacokinetics of the mycotoxin aflatoxin B1 (AFB1), were performed in the black South African population, using PBPK modeling. This study investigates the prevalence of clinical CYP450 phenotypes (CYP2B6, CYP2C9, CYP2C19, CYP2D6, CYP3A4/5) across Sub-Saharan Africa (SSA), to determine the feasibility of defining SSA as a single population. SSA was subdivided into Central, East, South and West Africa. The phenotype data were assigned to the different regions and a fifth SSA group was composed of all regions' weighted means. Available data from literature only covered 7.30% of Central, 56.9% of East, 38.9% of South and 62.9% of West Africa, clearly indicating critical data gaps. A pairwise proportion test was performed between the regions on enzyme phenotype data. When achieving statistical significance (p < 0.05), a Cohen's d-test was performed to determine the degree of the difference. Next, per region populations were built using SimCYP starting from the available SSA based SouthAfrican_Population FW_Custom population, supplemented with the phenotype data from literature. Simulations were performed using CYP probe substrates in all populations, and derived PK parameters (C_max, T_max, AUC_ss and CL) were plotted in bar charts. Significant differences between the African regions regarding CYP450 phenotype frequencies were shown for CYP2B6, CYP2C19 and CYP2D6. Limited regional data challenge the representation of SSA populations in these models. The scarce availability of in vivo data for SSA regions restricted the ability to fully validate the developed PBPK populations. However, observed literature data from specific SSA regions provided partial validation, indicating that SSA populations should ideally be modelled at a regional level rather than as a single entity. The findings, emerging from the initial AFB1-focused PBPK work, underscore the need for more extensive and region-specific data to enhance model accuracy and predictive value across SSA.