Human microbiota research in Africa: a systematic review reveals gaps and priorities for future research

Abstract

Background

The role of the human microbiome in health and disease is an emerging and important area of research; however, there is a concern that African populations are under-represented in human microbiome studies. We, therefore, conducted a systematic survey of African human microbiome studies to provide an overview and identify research gaps. Our secondary objectives were: (i) to determine the number of peer-reviewed publications; (ii) to identify the extent to which the researches focused on diseases identified by the World Health Organization [WHO] State of Health in the African Region Report as being the leading causes of morbidity and mortality in 2018; (iii) to describe the extent and pattern of collaborations between researchers in Africa and the rest of the world; and (iv) to identify leadership and funders of the studies.

Methodology

We systematically searched Medline via PubMed, Scopus, CINAHL, Academic Search Premier, Africa-Wide Information through EBSCOhost, and Web of Science from inception through to 1st April 2020. We included studies that characterized samples from African populations using next-generation sequencing approaches. Two reviewers independently conducted the literature search, title and abstract, and full-text screening, as well as data extraction.

Results

We included 168 studies out of 5515 records retrieved. Most studies were published in PLoS One (13%; 22/168), and samples were collected from 33 of the 54 African countries. The country where most studies were conducted was South Africa (27/168), followed by Kenya (23/168) and Uganda (18/168). 26.8% (45/168) focused on diseases of significant public health concern in Africa. Collaboration between scientists from the United States of America and Africa was most common (96/168). The first and/or last authors of 79.8% of studies were not affiliated with institutions in Africa. Major funders were the United States of America National Institutes of Health (45.2%; 76/168), Bill and Melinda Gates Foundation (17.8%; 30/168), and the European Union (11.9%; 20/168).

Conclusions

There are significant gaps in microbiome research in Africa, especially those focusing on diseases of public health importance. There is a need for local leadership, capacity building, intra-continental collaboration, and national government investment in microbiome research within Africa.

Video Abstract

Microbiome research in Africa

What is known about this topic?	What are the gaps?	What does this study add to our knowledge?
There is an exponential growth of microbiome studies in North America and Europe.	The number of African countries where microbiome studies were conducted is unknown.	Microbiome studies were conducted in 61% of the countries in Africa, with the top three being South Africa, Kenya, and Uganda.
Most of these microbiome studies are dedicated to understanding diseases of public health importance (e.g. cancers, irritable bowel disorder, diabetes, etc.) in these countries.	The extent to which these studies focused on diseases of public health significance in Africa remains uninvestigated.	Only 26.8% (45/168) of the studies focused on diseases of the highest public health importance in Africa, with HIV accounting for 64.4% (29/45).
	The leadership and pattern of collaboration in African human microbiome studies are unknown.	Non-Africans led 79.8% of all the studies, and the most collaborative efforts were between the United States of America and African scientists. There is the need for local leadership, capacity building, intra-continental collaboration, and national government investment in microbiome research within Africa.

Introduction

The human microbiome plays pivotal roles in immune and brain development, nutrition, and metabolism [1, 2]. Imbalances in the gut microbiome have been associated with impairment and diseases of many organ systems [1] including cancers [3, 4], obesity [5], asthma [6, 7], allergy, inflammatory bowel disease, and metabolic diseases [1]. More recent reports have added sickle cell disease [8], brain disorders, and behaviors to the growing list of diseases [9]. Although the causal basis for many microbiome associations is unknown, the microbiome is likely to be key to precision medicine approaches [10].

In order for the microbiome field to contribute effectively to personalized medicine, it is imperative to draw an accurate picture of the human microbiome in health and disease. Almost all research into human health is dependent on context. This is particularly true for microbiome research as gut microbiomes, for example, vary extensively based on geography, age, diet, ethnicity, genetics, disease, medication, climate, and other environmental factors [1]. Consequently, there is an urgent need to characterize the microbiome of as many unique populations as possible.

The microbiomes of western populations have been extensively characterized; however, information regarding the microbiome of residents of Africa is considerably sparser. Microbiome studies extending our understanding of important diseases must be replicated in Africa due to context-specific factors [11]. In particular, environmental determinants may vary [12,13,14,15], and genomic heterogeneity [16] within the human population is more marked compared to other continents. Important environmental exposures include diet, geography, climate, infectious diseases, urbanization, living conditions, and pollution [11,12,13,14]. These variabilities preclude the generalization of microbiome studies conducted in one specific population in Africa to the entire continent. Therefore, the representation of diverse African participants in microbiome studies is a priority.

Although non-communicable diseases, including cancers, diabetes, and cardiovascular diseases, have emerged as public health threats in both developed and developing countries, Africa has an additional burden of infectious diseases [17]. Infections account for at least 70% of all deaths on the continent [18], including malaria, tuberculosis, HIV/AIDS, and neglected tropical diseases (Buruli ulcer, trypanosomiasis, schistosomiasis, and guinea worm) [17]. Lower respiratory infections, HIV/AIDS, diarrheal diseases, malaria, preterm birth complications, tuberculosis, neonatal sepsis/infections, stroke, and ischaemic heart diseases are responsible for the highest morbidity and mortality in Africa [19]. Health-related research in Africa, including microbiome-based research, must address the diseases that are of foremost public health importance.

A number of human microbiome studies have been conducted in Africa. Although Brewster and colleagues [14] have provided a survey of microbiome research conducted in Africa, this addressed only gut microbiome studies. Currently, no study has summarized all human microbiome research conducted in Africa in order to identify knowledge gaps and areas for further research. We, therefore, undertook a systematic survey of human microbiome studies involving African participants to provide an overview of and to identify research gaps in the field. Our secondary objectives were: (i) to determine the overall number of peer-reviewed publications; (ii) to identify the extent to which the researches focused on diseases identified by the WHO State of Health in the African Region Report 2018 as being the leading causes of morbidity and mortality [19]; (iii) to provide information on the extent and pattern of collaboration between researchers in Africa and the rest of the world; and  (iv) to identify leadership and the main funders of these studies.

Materials and methods

Search terms and strategy

This review followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [20]. A comprehensive literature search was undertaken from inception through to 1st April 2020 using the following databases: Medline via PubMed, Scopus, ISI Web of Science (Web of Knowledge), and Academic Search Premier, Africa-Wide Information, and CINAHL via EBSCOhost according to the search strategy outlined in (Supplementary Table S1). No filters were applied to any of the searches. All citations were exported into ENDNOTE (X9; Thomson Reuters). The search was independently conducted by two reviewers IA and REA. The reference lists of reviews were searched for eligible papers that were not recovered by the search terms.

Study selection criteria

Studies were included only if they meet all of the following criteria: (i) human studies involving residents of Africa only or as part of a multinational study regardless of age, sex, health status, study design, or care setting; (ii) published in English or French; and (iii) described either bacteria, archaea, fungi, viruses, or parasites identified from any human samples using next-generation sequencing (NGS) including both shotgun metagenomics and targeted amplicon sequencing. Our exclusion criteria were: (i) studies that did not include any human participants from African; (ii) those that utilized publicly available data on African participants; (iii) studies that did not characterize the microbiome; and (iii) studies that did not utilize NGS to characterize the microbiome or those that targeted only specific microorganisms in their analysis.

Screening of studies

Records retrieved from the literature search of the six databases were independently downloaded into ENDNOTE (X9; Thomson Reuters) by two reviewers (IA and REA). These reviewers independently removed duplicates, reviews, commentaries, editorials, notes, news, and opinions. They then screened the title and abstract of residual articles against the inclusion and exclusion criteria. The full texts of the studies that passed this stage were retrieved. The reviewers proceeded to independently review these full texts based on the eligibility criteria. At each stage of the process, the two reviewers compared their results and disagreements were resolved by mutual discussion.

Data extraction and synthesis

Once consensus was reached on which articles to include in the study, IA and REA independently extracted data into a predesigned data extraction table in Microsoft Excel^TM. The data extracted included the country of origin of the samples; techniques used to analyze the microbiome, disease of focus, type of sample, participants metadata (number, age, gender, ethnicity, geographic region of the participants), aims and conclusions of the studies, whether the participants were from rural or urban settings, source of funding for the studies, country location of institutions to which the participating scientists were affiliated, name of the journal, first and last author’s information, and information on data availability. The extracted data were compared for accuracy and merged. IA and REA analyzed the merged data separately, and the results were compared for accuracy.

Under funding, any institute under the National Institute of Health (NIH) and European Union (EU) were captured as NIH and EU, respectively during the analysis. Furthermore, only agencies that directly funded the studies via project-specific grants were captured. Those that indirectly supported the research by providing training grants, scholarships, or fellowships to specific authors were not reported as funders. This is because our objective was to highlight organizations that directly funded human microbiome studies in Africa and we were not able to directly determine whether the funds from these sources were directly invested in the microbiome project reported.

Where studies were multinational, we captured only the number of participants from the African cohort. In this situation as well, we listed all the countries involved but highlighted the African countries in bold typeface. Information not specified in the full-text article or its supplementary data were captured as “NA.” Rural/urban designation of the sample’s origin was only indicated when specified in the article using words such as “rural” (rural), “village” (rural), “city” (urban), and “town” (semi-urban or peri-urban or semi-rural). The Human Microbiome Project (HMP) classification of body sites was used to categorize the sample types. The age range was divided into four categories; young children (0 to 5 years), older children (6 to 12 years), adolescents (13 to 17 years), and adults (18 years and above). We determined the article’s accessibility to African researchers by checking if the paper is designated open-access at the journal website or if the journal itself is open-access or if the paper can be obtained from PubMed Central.

To determine the extent to which the studies focused on diseases of high public health importance in Africa, we analyzed the number of studies that focus on any of the following conditions identified in the World Health Organization [WHO] State of Health in the African Region 2018 Report as being in the top 10 causes of morbidity and mortality in Africa: lower respiratory infections, HIV/AIDS, diarrheal diseases, malaria, preterm birth complications, tuberculosis, neonatal sepsis/infections, stroke, and ischemic heart diseases.

Results

Results of the search

The search yielded 5515 records (including three articles from additional sources [hand-searching]) with 3066 remaining after removing duplicates. From these records, 2811 were excluded because of ineligibility, and 255 full-text articles were further assessed for eligibility. After a full-text review, a total of 168 eligible human microbiome studies were obtained. Figure 1 shows the PRISMA flowchart summarizing the steps followed in the selection of the final subset of papers used in the analysis.

Fig. 1

Flow diagram showing the selection of studies according to preferred reporting items for systematic reviews and meta-analysis (PRISMA) guidelines

Human microbiome research publications in Africa

We found 168 published articles that utilized NGS technology to characterize the human microbiome among African participants. Five broad study designs were used, with cross-sectional studies being the most common (46.4%, 78/168) (Tables 1, 2, and 3). Other designs utilized in the studies were case-control (20.8%, 35/168), randomized control trial (14.3%, 24/168), longitudinal (8.9%, 15/168), and cohort design (8.9%, 15/168). One study involved both longitudinal and cross-sectional designs. The majority of the studies (73.2%, 123/168) involved only one sampling time point. The studies were published in 86 different peer-reviewed journals. The most frequent journal of publication was PLoS One (13.1%, 22/168) followed by Scientific Reports (4.8%, 8/168), mBio (3.6%, 6/168), Microbiome (3.0%, 5/168), and PLOS Neglected Tropical Diseases (3.0%, 5/168). More than half of all the studies (67.8%, 114/168) were only published between January 2017 and March 2020 (Fig. 2). A total of 140/168 (83.3%) studies were published as open-access in subscription-based journals or open-access journals or available via PubMed Central and are therefore accessible to researchers based in Africa.

Table 1 Summary of the African Human Gut Microbiome studies characteristics

Table 2 Summary of the African Human Urogenital Microbiome studies characteristics

Table 3 Summary of the African Human Microbiome studies characteristics (other body sites)

Fig. 2

A line plot showing the number of African Human Microbiome research funded by the top four funding agencies over the past 10 years

Distribution of studies across Africa

We analyzed the African countries of sample origin for all 168 eligible papers. The included studies collected samples from participants residing in 33 of the 54 countries in Africa (61%) (Fig. 3). The countries with the highest number of studies were South Africa (16.1%, 27/168), Kenya (13.7%, 23/168), and Uganda (10.7%, 18/168). Tanzania (7.1%, 12/168), Malawi (7.1%, 12/168), and Nigeria (6.5%, 11/168) also had a moderate number of studies conducted in them. The 27 remaining countries had less than ten studies each. Regionally, most of the studies were conducted in East Africa (39.9%, 67/168) followed by Southern (29.8%, 50/168), West (29.2%, 49/168), Central (7.7%, 13/168), and North Africa (6.5%, 11/168). The region with the highest coverage of countries was West Africa, where studies were conducted in 11/15 countries (73%). This was followed by Central Africa 4/7 (57%), East Africa 8/14 (57%), and North Africa 4/7 (57%) and finally Southern Africa 6/11 (55%).

Fig. 3

An African map showing the location, frequency, and body sites investigated in human microbiome studies

Body sites, sample types, methodology, and data archiving

The gastrointestinal tract (GIT) was the most studied body site (52.4%, 88/168, Table 1 and Table S2a), followed by the urogenital tract (24.4%, 41/168, Table 2 and Table S2b) while the eye (1.2%, 2/168, Table 3 and Table S2c) and placenta (0.6%, 1/168, Table 3 and Table S2c) were the least studied sites (Fig. 4). Similarly, the predominant sample type studied was stool (47.6%, 80/168) followed by vaginal samples (16.1%, 27/168). Placenta and fecal membrane samples were the least frequently studied (0.6%, 1/168). A total of 144 studies investigated the bacterial component, while 14 characterized the virome of the human microbiota. One study each focused on only fungi and only protozoa. While two studies investigated both bacteria and viruses, one each focused on bacteria and fungi collectively, and then bacteria and protozoa. Two studies explored viruses, bacteria, and the protozoal component of the microbiota while the remaining two studies investigated helminths in addition to the former three.

Fig. 4

Representation of body sites included in microbiome studies conducted in Africa. The most studied diseases are listed for each body site. The total percentage exceeds 100% because eight, three, and one study characterized two, three, and four body sites respectively

The majority of studies characterizing the bacteriome used only 16S rRNA amplicon sequencing (73.8%, 124/168) while 14/168 (8.3%) used only shotgun metagenomic sequencing. Eleven studies (6.5%) used both methods. One study targeted the cpn60 gene in place of the 16S rRNA gene for bacteriome characterization. For virome studies, shotgun metagenomics was used in seven studies while targeted methods including RNA sequencing, phage sequencing, and VirCapSeq-VERT were used in four studies. Targeted sequencing of the ITS1 of 18S rRNA and 5.8S conserved fungal region was used to characterize the mycobiome in one study. One study conducted full-length 16S rRNA sequencing while two utilized both 16S rRNA sequencing and ITS2. One study included both MALDI-TOF culturomics and 16S rRNA sequencing technologies. Two studies failed to specify the method used; however, they were included in the final analysis because they utilized high throughput sequencing. The platform most commonly used was Illumina MiSeq (57.1%, 96/168), followed by Roche 454 pyrosequencer (22.6%, 38/168) and Illumina HiSeq (14.3%, 24/168). Of the 168 studies, 64 (38.1%) did not indicate whether their data are publicly available (Tables S2a-S2c). However, for those that did, most (29.8%, 50/168) deposited their sequence data in the National Center for Biotechnology Institute Sequence Read Archive (NCBI-SRA). Similarly, 20/168 (11.9%) data sets were archived in The European Nucleotide Archive (ENA), 9/168 (5.4%) in GenBank, 5/168 (3.0%) in Metagenomics Rapid Annotation using Subsystems Technology (MG-RAST), 3/168 (1.8%) in Open Science Framework (OSF), and 2/168 (1.2%) in the DNA Data Bank of Japan, and 1/168 (0.6%) in NCBI Gene Expression Omnibus (GEO). Ten studies (6.0%) deposited data in more than one of the repositories mentioned above, while four studies (2.4%) indicated that they would make their data available upon request.

Study participant information

More studies (42.2%, 71/168) investigated adult [≥ 18 years] microbiomes than those of young children [0 to 5 years] (23.8%, 40/168). No study focused on only older children [6 to 12 years old] or adolescents [13 to 17 years] (Fig. 5). However, 31.5% (53/168) of the studies compared the microbiomes of more than one age group. While 51.8% (87/168) of the studies included both males and females, 24.4% (41/168) and 11.3% (19/168) included only females and only males, respectively. The sex of participants was not specified in 19 (11.3%) studies. The two (1.2%) remaining studies included mothers and their infants however, the sex of the infants was undefined. A total of 84.5% (142/168) of the studies did not specify the ethnicity of their participants. While 31% (52/168) of studies focused on participants in rural settings and 4.8% (8/168) investigated microbiomes of urban dwellers, five studies (3%) collected samples from residents in peri-urban communities and 54.7% (92/168) did not specify whether their participants were from rural or urban settings. Only two studies compared microbiomes of participants from rural, urban and semi-urban settings. Eight studies compared rural and urban while one study compared rural and semi-urban residents’ microbiomes. Most studies [60% (101/168)] included less than 100 participants, while 31.5% (59/168) studies enrolled 100 to 499 participants. Six studies included 500 to 999 participants. Only two studies involved 1000 or more people (Fig. 5). In total, Nigeria, the Gambia, Kenya, Malawi, South Africa, and Uganda had the microbiome of more than 1000 residents characterized. Additionally, several of the studies were derived from the same cohort of people [21] and [22,23,24] and [25,26,27,28] and [29, 30] and [31, 32] and [33]. Figure 6 and Figure 7 summarize the gut and urogenital studies in Africa.

Fig. 5

African map with pie charts showing the age categories and the number of participants included in human microbiome studies per country. The size of the divisions within the pie charts corresponds to the proportion of studies that included each age category (young children (0 to 5 years), older children (6 to 17 years), and adults (≥ 18 years)). The size of the pie chart represents the cumulative number of participants from all studies conducted in the country

Fig. 6

African map with pie charts showing the age categories and the number of participants included in human gut microbiome studies per country. The size of the divisions within the pie charts corresponds to the proportion of studies that included each age category (young children (0 to 5 years), older children (6 to 17 years), and adults (≥ 18 years)). The size of the pie chart represents the cumulative number of participants from all studies conducted in the country

Fig. 7

African map with pie charts showing the age categories and the number of participants included in human urogenital microbiome studies per country. The size of the divisions within the pie charts corresponds to the proportion of studies that included each age category (adolescents (13 to 17 years) and adults (≥ 18 years)). The size of the pie chart represents the cumulative number of participants from all studies conducted in the country

Diseases of focus of the studies

To identify the extent to which the studies focused on diseases of major public health importance in Africa, we analyzed the diseases of focus. Of the 168 eligible studies, 38.1% (64/168) did not focus on any specific disease (Fig. 4). Of the remaining 61.9% (104/168) that investigated the microbiome in the context of a specific disease, 45 studies focused on the top nine diseases responsible for the highest morbidity and mortality in Africa. They are as follows: lower respiratory infections (4), HIV/AIDS (29), diarrheal diseases (6), malaria (2), preterm birth complications (1), tuberculosis (1), and neonatal sepsis/infections (2). Other diseases that were frequently studied included malnutrition (8/104), bacterial vaginosis (5/104), obesity only (2/104), diabetes only (2/104), obesity and diabetes (1/104), and metabolic syndrome (1/104). Under neglected tropical diseases, only one study investigated Buruli ulcer, two each focused on trachoma and schistosomiasis and four on other parasitic infections (helminths and blastocystis). Fifteen studies examined non-communicable diseases (cancers, anemia, atopic dermatitis, environmental enteric dysfunction, and toxic blood metal levels).

Intercontinental and intra-continental collaborations among study co-authors

We analyzed the countries of institutional affiliation of all authors on each manuscript in order to understand the extent and pattern of collaborations between researchers in Africa and the rest of the world. For within-country collaborations, 17 studies had all the collaborating scientists based within the same country [Egypt (6), USA (5), South Africa (3), France (2), and Germany (1)]. Out of these, Egypt and South Africa were the only African countries where the collaborating scientists were from the same country. Furthermore, seven of the studies that involved researchers collaborating from more than one country did not include any African scientists as an author (Table 4). Asian countries whose scientists collaborated with African scientists included China, India, Bangladesh, Indonesia, Thailand, and Vietnam. Scientists from South America who collaborated with African scientists were based in Colombia, Brazil, Puerto Rico, Venezuela, and Chile. A total of 85.7% (144/168) of the studies involved intercontinental collaborations between one African country and one or more non-African countries (Fig. 8). Among these studies, the most significant collaborative efforts were between scientists in the USA and African countries, mainly South Africa (13/168), Uganda (12/168), Kenya (10/168), and Malawi (10/168). Intercontinental collaboration was also common between African scientists and researchers based in the UK, Canada, and the Netherlands.

Table 4 Different types of collaborations in the African Microbiome studies (intra-continental, collaborations from the same country, and between non-African countries)

Fig. 8

Heatmap of intercontinental collaborations between African countries and non-African countries

Leadership in microbiome studies

To determine the extent to which these studies were led by African scientists, we analyzed the countries of institutional affiliations of the first (Fig. 9A), and the senior (last) authors of the studies (Fig. 9B) as proxies. Among first authors with a single country of institutional affiliation, 43.5% (73/168) were from the USA, 6.5% (11/168) from South Africa, 4.8% (8/168) from Canada, and France and 3% (5/168) from Germany. A total of 12.5% (21/168) were affiliated with institutions in more than one country. Out of these 21 studies, the first authors of 13/21 were affiliated to both an African institution and a non-African institution while 8/21 were affiliated to two institutions from different non-African countries. Only one study had the first author affiliated to institutions in two African countries (South Africa and Zimbabwe).

Fig. 9

Pie charts showing the percentage of affiliations per country for the first author (A) and for the last author (B)

South Africa 6.5% (11/168), Egypt 3.6% (6/168), and Nigeria 1.8% (3/168) were the only African countries that had a scientist with a single African country of institutional affiliation as the first author.

In contrast, other first authors with affiliations to institutions in Africa (Nigeria, Kenya, Mali, Botswana, Malawi, Morocco, Zambia, and Niger) also concurrently held affiliations to institutions in non-African countries mainly the USA and the UK. Regionally, the majority of these authors were from Southern and East Africa (5/13 from Southern Africa and 4/13 from East Africa compared to 3/13 from West Africa, one from North, and none from Central Africa).

The affiliations of the last authors followed a similar pattern (Fig. 9B); 42.3% (71/168) were from the USA, followed by Canada 6% (10/168), France 6% (10/168), South Africa 4.8% (8/168) and Germany, Switzerland and the UK (3.6% (6/168) each). Thirteen percent (22/168) were affiliated to institutions in more than one country, mainly the USA, the UK, and Australia. Similar to the observation made with the first authors, South Africa (8/168), Nigeria (2/168), and Egypt (6/168), and, in this case, Morocco (1/168) were the only African countries that had a scientist with a single African country of institutional affiliation as the last author.

African researchers from South Africa, Zimbabwe, Kenya, Malawi, Uganda, and The Gambia were also simultaneously affiliated to institutions in other countries outside the African continent, mainly the USA and the UK. Regionally, the majority of these authors were from Southern and East Africa (11/17 from Southern Africa and 4/17 from East Africa compared to 2/17 from West Africa and none from North and Central Africa). Using the first and last authors as proxies for the leadership of studies, we found that 79.8% of all the studies had first and/or last authors affiliated to institutions outside Africa.

Funding

We analyzed the agencies that directly supported the studies by awarding research grants. We found that funding from USA sources predominated (Fig. 2), with more than 70% of the studies partially or fully funded by American governmental institutions, foundations, and agencies. These included the National Institute of Health (NIH) 76/168 (45.2%), Bill and Melinda Gates Foundation 30/168 (17.8%), the United States National Science Foundation 8/168 (4.8%), and Blood Systems Research Institutes 5/168 (3.0%). This was followed by the European Union 20/168 (11.9%), through the European and Developing Countries Clinical Trials Partnership (EDCTP) (4/20), European Research Council (5/20), European Union Regional development fund (2/20), European Union’s Seventh Framework program (3/20) and other European Union agencies (6/20). Other funding sources included the Wellcome Trust (UK) 14/168 (8.3%), and the Canadian Institute of Health Research (CIHR) 8/168 (4.8%). It is noteworthy that South Africa is the only African country to have funded a published microbiome study on the continent.

Discussion

We conducted a systematic survey of studies that utilized NGS to characterize the human microbiome of residents of Africa. Our results revealed that up to 1^st April 2020, 168 published studies utilized NGS to characterize the human microbiome of African participants. Of the 61.9% (104) of studies that examined the microbiome in the context of disease, less than half (43.3%, 45/104) focused on diseases that are responsible for the highest morbidity and mortality in Africa with HIV/AIDS accounting for 29/45 studies alone. With regard to collaboration, partnerships between the USA and African scientists were most common. However, the leadership of these studies (first and last authorship) was mainly assumed by the American scientists.

African human microbiome publications

With the advances in NGS extending over a decade, it is interesting that half of all the studies were only published within the past 3 years (Fig. 2). It is, however, disappointing that only 168 studies investigated the African human microbiome using this technology. Considering that Africa is made up of 54 independent countries with extremely diverse genetic backgrounds and cultures and is also the second most populous continent with a population of 1.3 billion (2018 estimates [34]), Africa is under-represented in the global microbiome literature. More than half of the studies involved less than 100 participants further suggestive of reduced coverage. Additionally, several of the publications were derived from the same cohort of people [21] and [22,23,24] and [25,26,27,28] and [29, 30] and [31, 32] and [33], which further reduces the diversity and coverage of African people included in microbiome studies. Since the literature search extended only to April 2020, the numbers reflected for this year are lower.

Most studies (73.2%) involved a single sampling time point. Cross-sectional designs are appropriate for studies that aim to describe the microbiome signatures associated with a particular outcome of interest [35]. However, owing to high within-subject and between-subject variability and the influence of environmental factors, longitudinal study designs with multiple temporally-separated sampling points are recommended for more robust and reproducible results [35]. Cross-sectional designs were common probably because of the following factors: budgetary constraints, invasiveness of sampling procedure, participant compliance to study protocol, and availability of samples in the case of retrospective studies [35]. With regard to budgetary constraints, multiplexing techniques [36, 37] allow multiple samples from the same or even different origins to be processed and sequenced together. This technique substantially reduces sequencing costs.

Most (83.3%) of the studies were either published in open access journals or as open access articles in subscription-based journals or freely accessible through PubMed Central. This may be due to the open-access revolution that has gained ground in the scientific world [38]. Furthermore, the open access publishing policies adopted by the top funders (NIH, Bill and Melinda Gates Foundation, and Wellcome Trust) of the studies may also explain this observation [39]. Open-access publishing of studies conducted in Africa is crucial because the majority of libraries in African universities struggle to afford expensive subscriptions to prominent publishing companies. Although journal access initiatives such as WHO Health InterNetwork Access to Research Initiative (HINARI) [40, 41] allow access to some of these subscription-based journals, some vital research articles remain behind a paywall. Open access publishing will, therefore, improve access to studies conducted on the continent to researchers, students, and the general public. Access to research already conducted in Africa will inform, equip and encourage African scientists to engage in microbiome research. It will also encourage intra-continental collaboration by increasing the visibility of African researchers who already have the capacity to undertake microbiome research.

Similarly, 61.9% of the studies mentioned storing their sequence data in publicly available repositories mainly NCBI-SRA. The increase in data archiving for public access is fuelled in part by funder [39] and journal requirements [42]. This will allow the secondary use of the data by other researchers, particularly those in Africa who may not have the funding, capacity, and facilities to generate such data. The preference for NCBI may be influenced by the fact that most of the studies were led by America scientists who may be more familiar with NCBI-SRA than the other repositories.

The countries where most studies were conducted were in East and Southern Africa. This may be influenced by the fact that most of the first and last authors who had multiple affiliations (from both African and non-African institutions) were from East and Southern Africa. Therefore, these scientists have more opportunities through their North American/European affiliations to foster collaborations outside Africa and also secure funding for microbiome studies in these specific regions of the continent. Another reason for the over-representation of Eastern and Southern Africans in the microbiome studies may be the higher prevalence of HIV in these parts of Africa (20 million in Eastern and Southern Africa compared to 6 million in West, Central, and North Africa collectively in 2018 [43]. As a high proportion of studies focused on HIV/AIDS (29/168 compared to less than 10 for any other disease), it follows that more of such studies will be situated in these two regions to permit the recruitment of required large numbers. However, Africans have widely different genetic and cultural backgrounds [16] and this diversity may affect their microbiomes [1, 35, 44]. This variability argues for broader coverage of residents of Africa from all regions in microbiome studies.

Most of the studies reported very little metadata related to participants. For instance, 54.7% of the studies did not specify whether participants are from rural or urban areas. Other studies mentioned the hospitals where the patients were recruited without specifying any further details about the location of residence of the participants themselves. This specification is important because Africans in cities are increasingly adopting western diets and lifestyles compared to those in rural areas [12, 45]. This change in lifestyle can confound microbiome associations found in studies and must, therefore, be accounted for. Indeed, Lokmer et al. found that Cameroonians along an urbanization gradient differed by diet, habitat, and socio-cultural conditions, and this affected their gut and salivary microbiomes [46]. This difference further underscores the importance of collecting as much metadata as possible for microbiome studies.

Ethnicity information was not collected in 84.5% of the studies. Ethnicity may directly impact the microbiome, but more importantly, it is frequently strongly associated with a specific culture, lifestyle, and diet, which in turn affect the microbiome [12, 47, 48]. Failure to collect this information may be because ethnicity is not always easy to define. Also, in studies in localized geographic areas, ethnicity may be relatively homogenous and therefore not the focus of the research. Furthermore, ethnicity may be confounded by the increasing frequency of inter-marriage. Additional metadata that would add value to studies include disease status, medication exposure, family history, socio-economic status, and lifestyle (diet, smoking, alcohol consumption, physical activity) [49].

Most of the studies (124/168) utilized 16S rRNA metagenomic sequencing to profile the bacterial component of the microbiome. This limits the number of studies that have looked at the fungi, viral, and eukaryotic components of the African microbiome [50]. These other components are also important in health and disease [51, 52] and therefore warrant attention. The extensive use of 16S rRNA metagenomic sequencing limits the resolution of microbial profiles to genus level [50]. It also fails to provide the genomic as well as functional contexts of the bacteria identified [50]. Decreasing the cost of shotgun metagenomic sequencing and simplifying bioinformatic analysis techniques will tip the scale toward this superior methodology.

African human microbiome studies focusing on diseases of significant public health concern

Apart from HIV/AIDS, which was the focus of 29 studies, few studies focused on diseases among the top 10 diseases of public health importance in Africa. Human microbiome studies focusing on diseases including malaria, diarrheal diseases, pneumonia, tuberculosis were limited. This may be due to the perceived relative low contribution of the microbiome to each of these diseases. However, the role of the microbiome in these conditions cannot be completely ruled out as limited research has been conducted in these areas. Metabolic diseases including obesity and diabetes that are mediated by the microbiome were also sparsely studied. These conditions are also highly prevalent in Africa and warrant microbiome-based investigation [17, 53]. While 38.1% of studies did not characterize the microbiome in the context of any particular disease, research on healthy individuals are important to establish what the “normal” or “healthy” microbiome is for comparative purposes. Additionally, although certain conditions such as bacterial vaginosis are not part of the top 10 diseases of public health importance, they are still relevant health issues in Africa, particularly for reproductive health outcomes which are a focus of the United Nation’s Sustainable Development goals.

The extent and pattern of collaboration with researchers in Africa and the rest of the world

African scientists collaborated most commonly with American scientists on microbiome research projects with the latter often assuming leadership. Reasons for this observation are not known but could be speculated. One factor may be that the American partners were the principal investigators of the grants funding the studies. They may also have conducted the laboratory investigations, data analysis, and drafting of the manuscripts. The African collaborator’s primary role may have only been recruitment and sample collection [54]. The practice of scientists from the global north using African scientists as conduits to obtain samples, then shipping them away without building the capacity of their African partners or directly benefiting the continent is commonly known as “helicopter research”. To address this phenomenon, the H3Africa consortium ethics working group developed a guideline in 2018 on the ethical handling of genomic samples from Africa [55]. It calls for investigators from the global north to build the capacity of their African collaborators to equip them to work independently post projects [55]. The guideline also invites western researchers to allow for substantial intellectual contribution from African scientists on studies that draw on samples recovered from the continent [55]. For this guideline to effectively combat “helicopter research,” funding agencies could specify local capacity building as a condition for awarding grants to western scientists who partner with African scientists. Institutional review boards in Africa could consider making capacity building a requirement in studies that involve international collaboration.

Additionally, African governments must recognize the importance of research and invest in microbiome studies. Apart from South Africa, through the Department of Science and Technology, no other African country directly funded any of the microbiome research projects identified here. This factor may also contribute to African scientists’ inability to initiate and therefore lead microbiome studies. Intra-continental collaboration within Africa was almost non-existent, possibly hampered by lack of funding and language barrier. Similar findings were made by Boshoff, who investigated intra-regional research collaboration among countries within the Southern African Development Community (SADC) [56]. This author found only 3% and 5% of intra-regional and continental collaboration respectively in contrast to 47% inter-continental collaboration with high-income countries [56]. Onyancha et al. also observed a similar pattern for research collaborations in sub-Saharan Africa [57] where intra-continental collaboration was minimal compared to inter-continental north-south partnerships. To encourage intra-continental collaborations, Onyancha recommended regional conferences as well as student and staff exchanges [57]. However, these exchanges will have a limited impact if researchers cannot access funding to conduct projects. Lack of pathways to independent funding necessitates outside collaboration and is therefore likely to be a key limitation for African leadership on articles and grants. Access to independent funding streams is the most important factor that should be tackled to address low African leadership of microbiome studies. High reagent costs associated with microbiome studies in Africa also frequently result in the shipping of samples out of the region. Microbiome research in Africa would be greatly improved by efforts to reduce the cost per sample for assays such as 16S amplicon sequencing.

International collaborations with non-African partners followed colonial ties [54] as well as commonality of a national language, with African scientists from Francophone countries collaborating with French scientists, while English-speaking western countries partnered with Anglophone African researchers. An African scientist who collaborates with a western scientist increases his/her chances of securing funds for research, and this may explain the preference for international collaboration. Indeed, several funding agencies specifically make international collaboration, usually with a partner from the funder’s own country, a requirement for funding. This requirement further discourages intra-continental collaboration.

Computational resources to handle bioinformatics analysis are also scarce on the continent, making inter-continental partnerships important. However, the H3Africa Consortium [58] through its subsidiary, the H3ABioNet [59] has launched many initiatives to build capacities for African scientists to lead and conduct microbiome research in Africa. Additional efforts [59], including workshops by other agencies, are being made to further build bioinformatics capacity in Africa [11].

Conclusion

Residents in Africa are under-represented in human microbiome studies. There is a need to build capacity for microbiome research in Africa, increase collaboration among scientists within Africa, and ensure equitable partnerships with international collaborators. African governments and research funding agencies should identify microbiome research as a priority area for investigation and funding.

Limitations

Certain studies utilized the same cohort resulting in multiple counting of the same individuals. Funding information was sometimes difficult to extract as some authors did not clearly distinguish personal funding from project funds. Some African researchers may receive internal funding from within their research institutions, which may not be captured in our review. African scientists may travel abroad for educational purposes, and during this period may be affiliated with non-African institutions Although they may still return to Africa in leadership positions, this could not be assessed in this review. The number of studies that focused on priority health care areas of Africa may be underestimated due to the exclusion of publications that did not employ NGS technology.