Exploring the Landscape of Latent Autoimmune Diabetes and Maturity Onset Diabetes of the Young in Africa: A Scoping Review

Inclusion and exclusion criteria

Studies on LADA and/or MODY patients in African countries were considered for inclusion. Studies with prospective, cross-sectional, retrospective, interventional designs or dissertations and grey literature published in English or French were included. Systematic and narrative reviews on LADA and/or MODY were excluded, however, there reference lists were hand-searched to identify original research articles. Exclusion criteria was studies conducted among non-Africans and studies whose data sets could not be extracted.

Search strategy

The search was conducted using broad search terms to increase number of articles retrieved. Search terms used were LADA, MODY, autoimmune diabetes, prevalence, morbidity, mortality, characteristics, latent autoimmune diabetes in adults, maturity onset diabetes in adults, Africa, African countries either singly or in combination using Boolean connectors (“AND”, “OR”, “AND OR”). Articles were searched across various databases including EMBASE, MEDLINE, PubMed, Scopus, Google Scholar, Web of Science, and African Journal Online (AJOL).

Screening articles for inclusion

Microsoft excel and Endnote X8 were used to manage the search results. The first screening step involved assessing the articles titles for conformity with inclusion criteria. Abstracts of shortlisted articles were screened to select articles with relevant information and full text of relevant articles downloaded. An in-house data extraction sheet was used to chart shortlisted articles. Information extracted included first author, year of publication, country of study, study design, source population, age range, gender, risk factors/inclusion criteria, clinical parameters, biochemical parameters, prevalence, and main findings. Details of extracted information can be accessed through Extended data table S1 and S2: DOI: 10.6084/m9.figshare.22592893.

Description of articles included in the final analysis

A total of 40 articles (26 on LADA and 14 on MODY) were retrieved. After further screening, 17 papers on LADA and seven papers on MODY were excluded from final analysis for various reasons (Figures 1a and 1b) respectively. Sixteen papers (nine on LADA and seven on MODY) satisfied the inclusion criteria. The summary of final analysis of the 16 papers is presented in Extended data tables S1 and S2 DOI: 10.6084/m9.figshare.22592893. The papers were published between 2000 and 2022.

Figure 1a. The flow chat used to assess LADA articles for inclusion in the final analysis.

The screening steps included identification, screening, and selection of eligible articles for inclusion.

Figure 1b. The flow chat followed during assessment of MODY articles for inclusion in the final analysis.

Mapping

Nigeria was the most represented country in terms of LADA research, accounting for three of the identified research articles^10–12 the rest of the countries contributed one research paper each; South Africa,¹³ Tanzania¹⁴ Ghana,¹⁵ Tunisia.¹⁶ Sudan¹⁷ and Kenya.¹⁸ For MODY, four papers were identified from Tunisia^19–22 and one each from Morocco,²³ South Africa²⁴ and Sudan²⁵ that had evaluated the prevalence of MODY in those countries (Figure 2).

Figure 2. Summary of distribution of papers included in the final analysis.

Nine papers were analyzed for LADA and seven for MODY.

The map representation of the studies based on the countries from which they were conducted has been shown in Figures 3a and 3b.

Figure 3a. Representation of the various countries from which the LADA studies analyzed were conducted.

Figure 3b. Representation of the various countries from which the MODY studies analyzed were conducted.

Study designs

Cross sectional study design was adopted by all the nine papers on LADA and six out of the seven papers on MODY. One paper on MODY²³ adopted a mixed study design approach with a prospective participant recruitment with a cross-sectional analysis approach. The widespread use of cross-sectional study design is informed by the relatively lower cost and shorter time involved. Thus, the cost and duration of data collection appeared to be important determinants of the choice of a study design.²⁶ The downside of this approach, however, is the high risk of bias since the data give a snapshot of event at a particular point in time, without considering time dependent factors that could confound the study outcomes.²⁷ hence findings from such studies must be interpreted with caution.²⁸

Sample size and inclusion criteria

Participants included in the LADA studies were enrolled in hospitals, with most studies adopting a purposive sample collection strategy. The lowest sample size was 47 participants¹⁷ while the highest sample size was 235.¹⁰ In MODY studies, the sample sizes ranged between 11²⁰ to 1643.²⁴ The low sample sizes in these studies could be partly due to the high cost of diagnosis. Participants were sampled from patients who were undergoing diabetes type II management and screened for LADA.¹ The purposive sampling approach applied in these is mainly employed when screening general population would be costly.²⁹ Due to the smaller sample sizes used in the LADA studies, statistical power of the reported result may be put to question. Consequently, the interpretation of findings of such studies should be made with caution. MODY on the other hand is a genetic disease and familial studies recruiting patients linked to confirmed cases are preferred. MODY is an autosomal dominant disorder inherited disorder, justifying the need to identify individuals related to confirmed cases for enrollment. The south African study,²⁴ however screened the general population to identify MODY cases. Taken together, it can be said that both the inclusion criteria and sample sizes recorded in the various studies in this analysis were based on convenience and cost minimization, the danger of which is reduced statistical power of the study findings.

Age ranges and sex

Age and sex are independent variables that can potentially confound dependent variables. Thus, age or sex stratification of data during analysis is important in age or sex dependent diseases. For instance, LADA is most common among adults suggesting that the researchers would be more interested in enrolling adult participants. Indeed, LADA studies recruited participants who are 30 years to 85 years. On the contrary, the onset of MODY can be as early as <5 years, explaining why studies in MODY recruited participants from five years.^3,23 Both LADA and MODY are not sex-linked disorders, hence, the sex of participants was not considered as an inclusion criterion. However, patterns of disease in male and female participants could provide some insights into association of the disease with a particular sex and shape the design of interventions.

Clinical parameters

A questionnaire was used to collect clinical information. Clinical parameters such as duration of diabetes, weight, and height were collected alongside patient’s demographic data including age and sex. BMI, used as a surrogate marker for obesity, was calculated to determine the prevalence of obesity among the participants. In addition, some studies collected data on diabetes related complications.^10,14 The MODY studies collected information about age, sex, and BMI. The current diabetes treatment was also recorded for some of the studies.^22–24 Clinical parameters can be used as prognostic markers for disease outcome. Thus, analysis of these parameters is important for understanding disease trajectory. Moreover, these markers can be used to answer questions such as the link between LADA and BMI and establish the cutoff age for LADA diagnosis.

A similar approach as that used in LADA studies to collect clinical parameters was adopted for MODY, underscoring the importance of clinical data in prevalence studies. In addition to regular clinical data, information about current treatment is also key for MODY as these would not only influence the clinical presentation of the disease but also help to evaluate the performance of management strategies. It is also worth noting that some types of MODY patients may develop diabetes related complications. Hence subsequent studies should screen for such complications during participants enrollment.

Laboratory parameters

Measurement of biochemical laboratory parameters was made in all the 16 studies reviewed. The HBA1C and fasting blood glucose levels were measured for all participants and four studies^10,12,13,18 measured the C-peptide levels for diagnosis of diabetes. The pancreatic antibody, anti-GAD65 was used as a diagnostic marker for LADA diagnosis.

Baseline measurement of HBA1C was performed prior to patient enrollment into the MODY studies. Negative pancreatic antibody was screened prior to participant enrollment in six studies apart from one.²¹ Normally, the diagnosis of LADA is based on several biochemical markers including at least a positive pancreatic antibody and low to normal c-peptide levels.¹ The reactivity of GAD-65 antibody has been established to indicate LADA positivity.³⁰ However, this antibody is not specific to LADA, its presence may also indicate type I diabetes. There is therefore need for subsequent studies to uncover more specific markers for LADA diagnosis.

Prevalence

The prevalence of LADA in selected African countries where the studies were performed ranged from 1.8 to 18%. The Nigerian study reported a prevalence of 14% of LADA among patients who were on treatment for type II diabetes.¹⁰ The mean BMI of LADA positive participant was 27.6, which exceeded the normal BMI range of 18.5 to 24.9. However, the studies could not find a link between the elevated BMI levels to LADA due to the possible confounding effect of type II diabetes patients.¹ Contrary to the high BMI levels reported here, another study conducted in northern Nigeria,¹² where LADA prevalence was found to be 10.5%, participants did not attain high BMI levels. Other confounding factors that were not accounted for by study design or adjusted for during data analysis might influence the variation in BMI levels. Another study from Nigeria reported a prevalence of 11.9%¹⁰ based on anti-GAD65 seroprevalence. Compared to the Ghana study, which reported a prevalence of 11.7%,¹⁵ there was no significant difference in prevalence rate. The Tanzanian study¹⁴ reported a 5.1% prevalence of LADA among individuals type II diabetes. An evaluation of the association of LADA and diabetes related complications was conducted and found no significant association. A longitudinal study design to follow the participants over time would have more suited to explore the associations.

The study conducted in Kenya reported a moderately lower prevalence rate of 5.7% (n=124).¹⁸ The study recruited patients with clinically defined type 2 diabetes. Similarly, the south African study reported a low prevalence rate of 2.5% (n=80).¹³ LADA in Sudan was the lowest at 1.8% (n=15) among all studies analyzed.¹⁷

For MODY studies, prevalence was assessed based on presence of mutations associated with the diseases. For instance, study conducted in Tunisia²¹ used molecular tools to screen mutations in HNF1A, GCK, HNF4A and INS genes. Mutations in GCK and HNF4A genes were found among 13.05% (n=23) of study participants. Another study in Tunisia by Ref. 20 performed sequencing of 27 genes previously linked with presence of MODY and found two mutation in HNF1A and two others in GCK. However, the presence of these mutations does not imply the presence of MODY, rather, they predispose individuals to MODY. The prevalence data must therefore be interpreted carefully to show the risk of developing the disease but not necessarily the presence of the disease.

It is therefore important to carefully interpret this data. Recently, Ref. 22 screened for GCK and HNF1A genes and reported 1 (n=17) variant in GCK associated with MODY pathogenesis. It is important to note that the studies conducted in Tunisia did not identify genes associated with MODY in most of the confirmed MODY cases. Thus, the involvement of other novel genes not screened in the Tunisia study.¹⁹ The Moroccan study²³ reported 10% (n=22) prevalence of GCK and HNF1A mutations, while the rest of the patients did not carry mutations in any of the genes screened despite the high level of HBA1C. Community screening for MODY in the South African study²⁴ found a 5.9% (n=1643) prevalence of HNF1A mutations and only two participants had GCK mutations.