Introduction
The burden of type 2 diabetes is globally high, with sub-Saharan African migrants in Europe being almost three times more likely to have type 2 diabetes than the European host population [
1]. In the sub-Saharan African region, type 2 diabetes prevalence has been increasing rapidly over recent decades and is projected to increase most in this region over all world regions [
2].
A combination of genetics and health-related behaviours are thought to cause type 2 diabetes by contributing to increased insulin resistance and reduced beta cell function. These result in impaired fasting blood glucose (IFBG) and, ultimately, type 2 diabetes [
3,
4]. Interestingly, increased insulin resistance is initially compensated by elevated beta cell function [
5,
6]. However, once insulin resistance increases to a level where beta cell function can no longer compensate, IFBG develops [
6].
The pathogenesis of type 2 diabetes in sub-Saharan Africans is still poorly understood [
3]. The importance of insulin resistance and beta cell failure to produce insulin in the development of IFBG and type 2 diabetes seems to differ among sub-Saharan African populations compared with European populations [
7‐
12]. Studies have shown that insulin resistance is more severe and beta cell insulin production is elevated in African populations with normal glucose tolerance compared with European populations [
8,
9,
11]. Furthermore, during the progression of normal glucose tolerance to IFBG and type 2 diabetes, insulin production has been found to decrease more rapidly in African populations than in European populations [
10,
11]. However, most of these studies have been performed in African-Americans [
8,
9,
11,
12] (who are more genetically admixed compared with sub-Saharan African populations) or on a small sample size [
7,
10].
The high burden of type 2 diabetes among African migrants in Europe, and the steep increase in type 2 diabetes in urbanising Africa, indicate that environmental exposures, such as changes in health-related behaviours, and resulting changes in anthropometrics might play a role in a disruption of balance between beta cell function and insulin sensitivity in these individuals [
13]. However, data on the role of the environment in the development of insulin resistance and beta cell dysfunction among Africans are limited. Furthermore, there are indications that the determinants of insulin resistance and beta cell dysfunction at play among Africans may differ from European populations [
14]. However, it is unclear which determinants (e.g. sociodemographic factors, anthropometrics or health-related behaviours) are driving insulin resistance and beta cell dysfunction in African populations [
15].
The Research on Obesity and Diabetes among African Migrants (RODAM) study found differences in IFBG and type 2 diabetes prevalence among Ghanaians that were resident in multiple geographical locations. While type 2 diabetes prevalence was highest among Ghanaians in Berlin (12.8%), followed by Amsterdam (11.2%), London (9.4%), urban Ghana (9.3%) and rural Ghana (4.9%), IFBG prevalence was highest among Ghanaians in Amsterdam (27.5%), followed by London (14.7%), urban Ghana (11.9%), Berlin (11.2%) and rural Ghana (11.2%) [
16]. The relatively homogenous but geographically distinct population of the RODAM study offers an important opportunity to investigate the role of environment and health-related behaviours in insulin resistance and beta cell dysfunction among this sub-Saharan African population.
We therefore used data from the RODAM study as a means to: (1) assess the contribution of insulin resistance and beta cell dysfunction to differences in IFBG between sub-Saharan Africans in different geographical locations; and (2) investigate which sociodemographic, anthropometric and health-related behaviour determinants contribute to insulin resistance and beta cell dysfunction among this population.
Discussion
The findings from this study demonstrate that, compared with Ghanaians in rural Ghana, the prevalence and age- and sex-adjusted odds for IFBG were higher for those in urban Ghana, Amsterdam and London but were similar between those in rural Ghana and Berlin. Insulin resistance accounted for the differences in IFBG between locations of residence, except for Amsterdam, while beta cell dysfunction did not. BMI and waist circumference were most strongly associated with insulin resistance in Ghanaians. Exceeding the guideline for alcohol consumption and smoking were the two determinants that showed the strongest positive associations with beta cell dysfunction.
It is speculated that the increasing burden of IFBG and type 2 diabetes in African migrants and urbanising Africa might be driven by environmental exposures that disturb the balance between beta cell function and insulin sensitivity [
3,
13]. To our knowledge, this is the first study that has looked at differences in insulin resistance and beta cell function among a sub-Saharan African population living in different geographical locations. Previous research by Osei et al demonstrated greater insulin resistance in African-Americans compared with Nigerians [
30]. Here we found that levels of insulin resistance varied between Ghanaians living in different geographical locations, even within Europe. Insulin resistance accounted for most of the differences in IFBG between locations. In contrast, after inclusion of inverse HOMA-B in an age- and sex-adjusted logistic regression model, the odds for IFBG were increased in multiple geographical locations compared with rural Ghana. This finding is consistent with a study in Koreans that found no association between lower HOMA-B and IFBG [
31]. In the current study, the lack of contribution of HOMA-B to differences in IFBG between locations and the weak association between HOMA-B and IFBG may be suggestive of compensation by the beta cells to overcome the reduced insulin sensitivity that accompanies this condition. Hence, in IFBG, environmental and associated health-related behaviour changes seem to impact insulin resistance more than beta cell dysfunction.
In sub-Saharan Africans, the prediabetes phase (defined as IFBG in this study) was suggested to be characterised by insulin resistance rather than failure of insulin secretion; per SD increase of HOMA-IR the odds for IFBG were 3.26 compared with 1.13 for 1 SD increase in HOMA-B. In the corresponding calculations of attributable risk for IFBG, we found that if the entire study population had a HOMA-IR that was 1 SD higher, there would be between 63% and 75% more individuals with IFBG. In comparison, if HOMA-B was 1 SD higher in the entire population, between 5% and 17% more individuals would have IFBG (Table
2). This is in line with studies in African-Americans that found more severe insulin resistance in individuals with normal glucose tolerance or prediabetes, the latter being defined as IFBG or impaired glucose tolerance (IGT) [
8,
9,
11,
12]. On the other hand, studies among sub-Saharan Africans with type 2 diabetes have found variable insulin resistance in these individuals compared with individuals without type 2 diabetes, but fast declines in insulin secretion after diagnosis with type 2 diabetes [
7,
10]. According to the model by Gibson [
13], sub-Saharan Africans in a normal blood glucose state present an equilibrium characterised by both higher insulin resistance and higher beta cell function. When insulin resistance becomes more severe, initially beta cell function compensates for this change. However, when this beta cell compensation fails, type 2 diabetes develops. Longitudinal studies are needed to fully elucidate the progression from normal blood glucose state to impaired glucose tolerance to type 2 diabetes among sub-Saharan Africans.
The determinants that contributed to insulin resistance among Ghanaians included anthropometric indices that were also identified by Amoah et al in relation to IFBG in a sample of 36 Ghanaians residing in urban Ghana [
32]. Additionally we found that alcohol consumption exceeding the recommendations and smoking contributed to more beta cell dysfunction. The negative associations of beta cell dysfunction with BMI and waist circumference reinforce the notion that compensation by beta cells for more severe insulin resistance might play a role in IFBG. Furthermore, a family history of diabetes was found to contribute to increased insulin resistance severity and somewhat less beta cell dysfunction. This apparent contradiction could again result from an increased compensatory beta cell function among those with more severe insulin resistance.
Lifestyle interventions, including either physical activity alone or combined with dietary advice aiming at weight loss, have been shown to significantly reduce the progression of IGT to type 2 diabetes [
33]. While pharmacological intervention with, for example, metformin also targets insulin resistance, previous studies do not demonstrate it to be more effective than lifestyle interventions in terms of primary prevention [
33,
34]. The reduction of BMI could be fruitful as a means to reduce the high burden of IFBG and subsequent type 2 diabetes among urban Africans and African migrants in Europe. Despite the observed associations of smoking and excessive alcohol consumption with beta cell dysfunction, public health interventions may not be worthwhile considering the relatively low prevalence of alcohol consumption and smoking in the population studied.
We found that insulin resistance accounted for the differences in IFBG between all geographical locations except Amsterdam. Among Ghanaians in Amsterdam, the odds for IFBG remained three times higher compared with rural Ghana when adjusted for insulin resistance and beta cell function. An excessively high prevalence of IFBG among Ghanaians in Amsterdam has been reported previously [
35]. In addition, a very high prevalence of IFBG has also been observed among the host Dutch population [
36]. We found that Ghanaians in Amsterdam had more severe insulin resistance compared with all other sites, and lower beta cell function compared with London, urban Ghana and Berlin (however, the difference in beta cell function was only significant compared with Berlin). This may imply that the failure of beta cell function to compensate for the more severe insulin resistance in Ghanaians in Amsterdam had already set in. In post hoc analyses (data not shown), waist circumference was the only modifiable risk factor associated with IFBG in Amsterdam, while in the other locations waist circumference was not associated with IFBG. Hence, there is a potential for body fat distribution and, perhaps, the ratio of subcutaneous:visceral body fat to play a role in the difference in IFBG prevalence in Ghanaians in Amsterdam compared with those in other geographical locations [
37]. However, this requires further study.
It is unclear why there was a relatively low prevalence of IFBG in Berlin compared with other European sites. Previously, in the same cohort, we found Berlin to have a marginally higher prevalence of type 2 diabetes compared with other European sites and Ghana [
16]. It is possible that IFBG progresses more rapidly into type 2 diabetes in this location than in other sites. Further research is required to determine the progression from normal glucose to IFBG and subsequent type 2 diabetes in different geographical contexts.
Strengths and limitations
This was the first study to compare IFBG, HOMA-IR and HOMA-B in a large sample of relatively homogenous sub-Saharan Africans, i.e. all stemming from Ghana, resident in distinct geographical locations, using standardised approaches. We have confirmed the importance of insulin resistance as opposed to beta cell dysfunction in sub-Saharan Africans in general. Additionally, we have shown that insulin resistance is important in the context of urbanisation and migration and have studied the contribution of a range of determinants to this condition. These findings provide a foundation for future cardiovascular outcome studies, as well as studies into the progression of IFBG to type 2 diabetes in sub-Saharan Africans.
The reliance on fasting blood samples in this study only allowed us to use IFBG as an endpoint for prediabetes; thus we were not able to investigate the associations between geographical differences and determinants of IGT. While IGT and IFBG identify different groups of individuals [
4], both are considered high risk groups for type 2 diabetes [
38].
As a study strength, comparisons between geographical locations were unlikely to be affected by site differences as the method for fasting glucose determination was standardised between locations. Additionally, all blood samples were collected and transported according to the same protocols in each of the research locations. Biochemical analyses were performed centrally, in Charité–University Medicine Berlin, Berlin, Germany, to avoid intra-laboratory bias.
We used the geographical differences as a means to study the role of insulin resistance and beta cell dysfunction in IFBG in sub-Saharan Africans. Because of the cross-sectional nature of the data we cannot make any statements about causality. However, reverse causality between insulin resistance and IFBG, as well as between the determinants and insulin resistance, is not very likely. Nevertheless, longitudinal data are needed to study the role of environment on insulin resistance and beta cell dysfunction in the progression from normal glucose control toward IFBG and, ultimately, type 2 diabetes.
Furthermore, we used self-reported data for the identification of determinants, such as diet and physical activity. Reporting bias, such as socially desirable answering, might have influenced assessment of these determinants [
39,
40]. Further in-depth investigations are needed to study the role of these health-related behaviours in IFBG among sub-Saharan Africans.
Acknowledgements
The authors are very grateful to the advisory board members for their valuable support in shaping the methods, to the research assistants, interviewers and other staff of the five research locations who have taken part in gathering the data and, most of all, to the Ghanaian volunteers participating in this project. We thank N. Huckauf and C. Kalischke from the Department of Endocrinology and Metabolism, Charité University Medicine (Berlin, Germany) for their excellent support in biochemical characterisation of the participants and data handling. We gratefully acknowledge J. van Straalen from the Department of Clinical Chemistry, Academic Medical Centre (Amsterdam, the Netherlands) for his valuable support with standardisation of the laboratory procedures, and the Academic Medical Center (AMC) Biobank for support in biobank management and storage of collected samples.