Warfarin anticoagulation in patients with cardiovascular disease in sub-Saharan Africa



Asiimwe, Innocent ORCID: 0000-0002-1196-1822
(2021) Warfarin anticoagulation in patients with cardiovascular disease in sub-Saharan Africa. PhD thesis, University of Liverpool.

[img] Text
201005255_Aug2021.pdf - Unspecified

Download (23MB) | Preview

Abstract

Background Warfarin is the most widely prescribed oral anticoagulant in sub-Saharan Africa. Its use is however limited by a narrow therapeutic window and a wide inter-patient variability, among other challenges which all lead to poor anticoagulation quality. To tackle some of these challenges, the National Institute of Health Research Global Health Research Group on Warfarin anticoagulation in PATients in Sub‐SaHaran Africa (War-PATH) was formed. As part of the War-PATH project, this PhD aimed to develop clinical and pharmacogenetic dosing models to help improve anticoagulation control in patients from sub-Saharan Africa. Methods The literature was systematically searched to identify warfarin dosing algorithms and the genetic factors affecting warfarin dosing in black-Africans. Based on the literature search and clinical expertise, demographic, clinical, environmental and genetic factors were identified for potential inclusion in clinical and pharmacogenetic dosing models. Using cohorts of patients prescribed warfarin from South Africa and Uganda recruited between June 2018 to July 2019 (n = 364) and multivariable regression methods, a clinical dosing model was developed and externally validated in a cohort recruited from August 2019 to March 2020 (n = 270). To ensure the robustness of the employed regression techniques, a machine learning exercise was conducted in which ordinary least squares regression, the most commonly applied regression technique, was compared with twenty other techniques including non-linear least squares, artificial neural networks, and support vector regression. To further identify genetic factors important in predicting warfarin dose, a genome-wide association study was conducted. Following this, a pharmacogenetic model was developed using non-linear least squares and plasma concentrations of warfarin and its metabolites were measured to evaluate the impact of genetic factors on warfarin pharmacokinetics. Results Four predictors (age, weight, target International Normalized Ratio range, and HIV status) were included in the final clinical model while the pharmacogenetic model included nine predictors (the four included in the clinical model together with simvastatin/amiodarone status, compound CYP2C9 genotype status (number of reduced function alleles *3, *5, *6, *8, and *11), VKORC1 rs9923231, CYP4F2 rs2108622, and rs12777823). Based on the percentage of patients at high risk of suboptimal anticoagulation, our developed models were better than common clinical practice (fixed dose-initiation with 35 mg/week), with the pharmacogenetic model performing better than the clinical model. In our Sub-Saharan cohort, we observed that simpler techniques such as ordinary least squares regression perform similarly to more complex supervised machine learning techniques. In further work, the genome-wide association study (n = 548) identified 236 nominally significant associated SNPs (p ≤ 1 × 10−5). However, no SNP passed the genome-wide significance threshold (p ≤ 5 × 10−8), with the most significant SNP having a p-value of 6.7 × 10−8. Additionally, due to having low minor allele frequencies in the study population, none of the CYP2C9, VKORC1, and CYP2C gene cluster SNPs, previously identified to be of importance in patients of black-African ancestry, were nominally significant. Lastly, patients with CYP2C9 reduced function alleles (*3, *5 *6, *8 and *11) had an increased (S):(R)-warfarin ratio (adjusted p = 2.1 × 10−11), which was consistent with decreased (S)-warfarin metabolism. Conclusion We have developed the first warfarin dose-initiation clinical and pharmacogenetic models for black-African patients in Uganda and South Africa, and these will be tested for clinical utility before wide-scale implementation. Based on a systematic evaluation of the literature, very few studies have been undertaken in sub-Saharan Africa, which increases the urgency of conducting more work (including a well-powered genome-wide association study and developing a polygenic dosing model) in this region to further evaluate the factors determining warfarin response and utilize them to achieve more optimized dosing.

Item Type: Thesis (PhD)
Divisions: Faculty of Health and Life Sciences
Faculty of Health and Life Sciences > Institute of Systems, Molecular and Integrative Biology
Depositing User: Symplectic Admin
Date Deposited: 08 Feb 2022 15:04
Last Modified: 18 Jan 2023 21:32
DOI: 10.17638/03134701
Supervisors:
URI: https://livrepository.liverpool.ac.uk/id/eprint/3134701