NEW YORK – An international effort led by African institutions and funded by Novartis and GlaxoSmithKline is gathering African population-specific pharmacogenetic data related to tuberculosis and malaria, and leveraging artificial intelligence (AI) to streamline the use of what data currently exists.
The initiative, called Project Africa GRADIENT (Genomic Research Approach for Diversity and Optimizing Therapeutics), eventually aims to help tailor drugs to African populations.
Novartis and GSK established the legal framework for the initiative in 2019 and began funding projects in 2022 with a combined commitment of $3.6 million. It is administered by the South African Medical Research Council and currently supports projects led by nine investigators across South Africa, Mali, and Burkina Faso. The primary goal of the project's current phase is to facilitate research into the genetic diversity of various African regions and investigate how this might impact the effectiveness of therapeutics for tuberculosis and malaria.
"These [infectious diseases] have a very high disease burden in Africa," said Kelly Chibale, professor of organic chemistry at the University of Cape Town and one of the project's nine principal investigators.
Chibale explained that although Africa holds an estimated 18 percent of the world's population, few clinical trials take place on the continent. In addition, of the more than 300 drugs for which the US Food and Drug Administration provides pharmacogenetic advice, only 15 have been studied in African groups, creating a dearth of information and a high unmet need for more and better data.
"Not more than 5 percent of clinical trials have actually been in Africa, so the physiology [and] genetics of Africans are not typically considered," Chibale said.
Chibale's part in the project has been to look for pharmacogenes, or genes that govern drug metabolism. In particular, Chibale focuses on those genes that affect the metabolism of malaria and tuberculosis medicines.
The body naturally processes drugs, breaking them down and eliminating them as with anything else one ingests. The amount of drug that survives these metabolic processes is what actually carries out the drug's function, Chibale explained.
"If the metabolism is slow, you can overdose someone," Chibale said, "whereas with a fast metabolizer, you have to keep giving more [drug] because it will be broken down quickly."
Populations with different genetic backgrounds can have different metabolic gene signatures, making it important to have information from diverse populations when designing drugs.
Chibale pointed to the anti-HIV drug efavirenz, commonly marketed under the brand name Sustiva, among others, as an example where the lack of information on African populations led to public harm.
Efavirenz was introduced to Zimbabwe in 2015 but the patients receiving it quickly stopped taking it due to severe neurological effects that included hallucinations, suicidal ideation, depression, and anxiety. These were traced to a mutation in the CYP2B6 gene that resulted in very slow efavirenz metabolism, which was expressed at a high prevalence among Zimbabweans.
"[People] were being overdosed because they were using the dose that was determined from Caucasian populations that participated in the clinical trial," Chibale said.
With the funding from Project Africa GRADIENT, Chibale has used knowledge graph embeddings –– a machine learning method for filling in missing data based on existing knowledge –– and large language models to identify and prioritize pharmacogenes for more clinically oriented study moving forward.
While Chibale said that he cannot yet disclose specifics of the pharmacogene candidates discovered so far, pending peer-reviewed publication, he believes that the researchers will be able to demonstrate "powerful predictions" with the limited available data.
Concurrently, Chibale has been discussing collaborations with Collen Masimirembwa, professor of clinical pharmacology at the University of Cape Town and another Project Africa GRADIENT investigator.
Masimirembwa heads the Africa Liver Tissue Biorepository Consortium, an effort to build a first-of-its-kind comprehensive liver tissue biorepository for people of African origin. Chibale said that his idea is to monitor drug metabolism in liver fractions and then use that data to generate mathematical models to predict pharmacokinetics in human patients.
"We've actually done this already," Chibale said, "but not using African-derived livers."
Chibale's grant ends in August and in the next phase of his project, he plans to enrich his pharmacogenetic models and expand the pharmacogenomic database that all GRADIENT investigators are contributing to.
Chibale also plans to use information from a database created by the Human Heredity and Health in Africa (H3Africa) consortium to help enrich his models. Although H3Africa created the publicly available H3Africa Data Archive as part of its initial data collection effort, Chibale said that the database he intends to use, while constructed by H3Africa, is something else.
"Most of [that] data is not publicly available yet," he said.
In a joint email statement from both companies, GSK and Novartis said that "the existing databases that Project Africa GRADIENT investigators are using are not diverse enough. The next step for the GRADIENT Joint Steering Committee is therefore to decide whether and how to support the development of infrastructure for unified biobanks/databases to incorporate newly generated data."
Marlo Möller, professor of biology and human genetics at the South African Medical Research Council Centre for Tuberculosis Research, is another of the nine investigators. She leads a project focused on evaluating genetic diversity as the contributing factor to variability in exposure and/or response to first- and second-line drugs used to treat tuberculosis.
Her project is funded through 2025, at which point she expects to have matched genetic and pharmacokinetic data on slightly more than 1,500 individuals.
"The combined datasets will be available for additional data mining and training in genetics and [pharmacokinetic] analyses," she said.
Möller said that some groups outside of Project Africa GRADIENT are already using the pharmacogenetic variants that her group has identified in their own projects, although this work is relatively early and Möller declined to name specific collaborators until it is published. Möller's work for Project Africa GRADIENT has so far taken place all in South Africa but she said that she plans to eventually also conduct pharmacokinetic analyses in Namibia.
Möller said that the collaborations made possible through Project Africa GRADIENT will outlast the project's initial funding.
"Through Project GRADIENT, we met Özlem Tastan Bishop from Rhodes University, and we successfully applied for an ARUA-The Guild Clusters of Excellence proposal, [through which], we will be eligible to apply for additional funding," she said.
Tastan Bishop, director of the Research Unit in Bioinformatics at Rhodes University and another Project Africa GRADIENT investigator, used her funding to establish the HUMA-PGx (HUman Mutation Analysis-PharmacoGenomics) pharmacogenomic database, which integrates pharmacogenomic, drug, variant, sequence, and structural information into a single source for comprehensive pharmacogenomics data analysis.
"HUMA-PGx will be an open-source resource after article publication," she said.
Tastan Bishop added that her GRADIENT funding concludes at the end of the year, and she is actively seeking additional funding.
"The initial two-year grant has been extremely helpful in kick-starting our work," she said, "but to achieve our long-term goals, we require extended funding, ideally for five years."
While the companies credited the information-sharing network established by the GRADIENT investigators over the course of this project as a key success, Chibale said that much of the groundwork was laid out by H3Africa.
"They've been able to develop data sharing mechanisms within their consortium and appropriate agreements with the communities that provide samples," the spokesperson said, which has resulted in a "blueprint" for others to follow.
Collecting and sharing data between GRADIENT-funded groups has helped to "firm up" genomics data infrastructure in the region, the spokesperson said, noting that systemic obstacles to further development remain in place, such as resource limitations and the "decentralized and inconsistent nature of data collection, storage, and analysis" throughout the areas in which project investigators work.
In fact, many stakeholders within Africa's genomics space have been working to overcome these challenges. A network of centers of excellence is spreading across the continent, which foreign investment has increased dramatically over the last few years.
Although individual projects are funded for different periods of time, project Africa GRADIENT as a whole is currently in its second year of funding, and Novartis and GSK said that additional funding will be considered by the Joint Steering Committee at the end of this phase.
"There is a hope that the database being built could be leveraged to attract funding for new projects in the future, and depending on the analysis of the results from the current projects, the initiative could be expanded into other disease areas beyond malaria and [tuberculosis]," the companies said via their spokesperson.
A frequently cited complaint of many African researchers is that a reliance on foreign funding often accompanies a loss of control over a given project's focus and direction. Despite GRADIENT funding coming from Switzerland-based Novartis and UK-based GSK, Chibale said that Project Africa GRADIENT is a good example of the growing number of African-led genomics initiatives taking place across the continent.
"There will always be international partnerships," he said. "It's a question of who leads."