Introduction

Overweight and obesity, including abdominal obesity,are not uncommon in the Maasai [1], an agro-pastoralist people living in Kenya and Tanzania numbering approxi-mately 1.6 mill [2, 3]. whose cardiometabolic health has been studied frequently for several decades [4, 5] In rural and urban Kenyan Maasai, glucose intolerance was pri-marily associated with increasing visceral adipose tissue in men but increasing body mass index (BMI (kg/m2 ) in women [6]. In rural, Kenyan Maasai, the detrimen-tal effects of abdominal obesity have also been shown as higher insulin resistance, higher first-phase insulin secretion, but lower insulin action compared to other eth-nic groups in rural Kenya [7]. Despite this, we observed low prevalence of diabetes mellitus (DM) of < 2.0% in rural Maasai with an equal distribution between men and women [6, 7]. These results stand in stark contrast toother indigenous populations ≥ 20 years of age such as the Native Americans in the USA and Aboriginal Austra-lians with prevalence rates of ~ 15% and ~ 29%, respec-tively, and with substantial sex difference in prevalence rate in the latter population [8, 9].

    Several studies have investigated sex differences in glucose metabolism in different populations from Den-mark, Australia, Mauritius, and Japan, respectively [10–13], and these have shown that men tend to have a higher fasting plasma glucose while women tend to have higher 2-hour plasma glucose following an oral glucose tolerance test (OGTT). Thus, sex seems to play a role in the physiology of glucose metabolism.

    Incretin hormones, i.e. glucose-dependent insulinotro-pic polypeptide (GIP) and glucagon-like peptide-1 (GLP- 1), are important factors that stimulate nutrient-induced insulin secretion in response to meal intake, including oral glucose [14]. This phenomenon is known as the incretin effect, as oral glucose leads to the release of GIP and GLP-1 from specialized entero-endocrine cells in the gut – in contrast to intravenous glucose which bypasses the gut – with a subsequent blood glucose-lowering effect [15, 16]. Despite their central role in glucose metabolism, no information exists on incretin hormones in relation to oral glucose exposure based on studies in agro-pastoralist populations in Africa, including the Maasai people. Like-wise, glucagon response in relation to glucose homeosta-sis has not been studied in the Maasai.

    Thus, based on sex-specific differences in body com-position, we aimed to study possible differences of the physiological response to an OGTT, with specific focus on incretin hormones in rural Maasai men and women without DM and with a wide range in age and BMI.

Subjects, materials and methods Geographical area and participants

Data were collected in Monduli district of Arusha Region in North-Central Tanzania in June-August 2018 during the cold, dry season. Potential participants of Maasai origin from rural communities (collectively known as Monduli Juu) were informed about the study presented as a diabetes and cardiovascular disease project through a Maasai physician (JS). They were presented with details orally and in writing and signed a consent form or in case of illiteracy gave their thumbprint. Maasai ethnic-ity and age ≥ 15 years were inclusion criteria. Pregnancy, known diabetes or diabetes detected as part of the study, or severe chronic or infectious illness were exclusion cri-teria. To accommodate the study, a makeshift laboratory was set up at the Catholic primary School in Monduli Juu situated approximately 1700 m above sea level.

Data collection

Each participant had to arrive at the study site after an over-night fast on two separate days within a week. A health assessment questionnaire for screening of diabetes and other illnesses (medical history), and pregnancy screening for the women using a human chorionic gonadotropin urine test (Quick-Strip Pregnancy test, Shanghai Eugene Biotech Co., Ltd, China) were performed.

    On Day 1, each participant was measured in light cloth-ing and barefooted, for height using a portable stadiometer (Charder HM200P Portstad, Taichung City, Taiwan), weight using a portable electronic scale (Tanita SC-330s, Tokyo, Japan), waist and hip circumference using a non-stretchable tape measure at the midpoint of the lower palpable rib and the iliac crest, and at the trochanter major, respectively. Mid-upper arm circumference (MUAC) was measured on the dominant arm using a specific MUAC tape measure (Raven Equipment, Dunmow, Essex, UK). BMI and waist-hip ratio were derived.

    After receiving careful instructions on the procedure, each participant subsequently underwent an OGTT while seated. An antecubital vein cannula was inserted for blood sampling, and a fasting blood sample was collected, fol-lowed by ingestion of 75 g of glucose mixed in 300 mL of water within 5 min after the fasting blood sample. Collec-tion of 10–12 mL of blood was done at time points 0, 10, 30, 60, 90 and 120 min. The samples were immediately stored on ice and centrifuged within 10 min. The plasma was sepa-rated and stored on ice until the end of the test day and then moved to a freezer (-20 °C) at Monduli District Hospital. Immediately following the end of the data collection period, the plasma samples were transferred to and stored at Kili-manjaro Clinical Research Institute in Moshi at -80 °C, until they were shipped to Denmark for analysis.

    On Day 2, following an overnight fast, abdominal fat distribution, i.e. visceral adipose tissue (VAT) and subcuta-neous adipose tissue (SAT) was measured using ultrasound-scanning technique (Aquila Basic Unit, Esaote, Pie Medical Equipment, Maastricht, the Netherlands) with a 3.5/5.0 MHz transducer (Probe Article no. 410638 Curved Array HiD probe R40 Pie Medical Equipment, Maastricht, the Netherlands). The measurements were carried out accord- ing to the protocol validated by [17]. Blood pressure was measured using an automatic M10-IT brachial blood pres-sure monitor (Omron, Kyoto, Japan) with the participant sitting in a chair with the cuff placed on the left arm. Three measures were taken with 2 min intervals with the first mea-sure discarded, and an average of the subsequent two mea-sures was recorded.

    Haemoglobin was measured using venous blood on a portable HemoCue® Hb 201+device (HemoCue AB, Ängelholm, Sweden).

    Dietary intake for each participant was estimated based on two interactive 24-hour recall interviews developed to collect dietary information in rural populations living in low and middle-income countries [18].

    As the last measurement, an 8-minute ramped (15–33 cycles) sub-maximal step test stepping up and down a 21.5 cm high step followed by a 2-minute post-test period of sit-ting on a chair, recording heart rate throughout the test and post-test periods was performed in order to estimate aero-bic fitness using a heart rate monitor (Garmin HRM3SS, Olathe, KS, USA) [19].

Laboratory analyses

Plasma concentrations of glucose, insulin and C-peptide were analysed according to the sandwich principle on com-mercially available analysis kits (COBAS 6000 analyser 501 C, Roche, Germany). Plasma concentrations of total GIP, GLP-1 and glucagon were measured using commer-cially available sandwich ELISA kits (cat no: 10-1258-01, 10-1278-01, and 10-1271-01, respectively) (Mercodia, Uppsala, Sweden).

Calculations of AUC and insulin indices

Total area under the curve (tAUC) for the OGTT mea-surements for glucose, insulin, C-peptide, GIP, GLP-1and glucagon, respectively, was calculated with the use of the trapezoidal method [20]. Incremental AUC (iAUC) was calculated by subtracting the fasting plasma results to cap-ture the effect of the OGTT stimulus [21]. Hepatic insulin resistance was assessed by using the homeostasis model analysis – insulin resistance (HOMA-IR) according to the formula by [22], and early phase insulin secretion was cal culated from the first 30 min. of the iAUC. The disposition index (DI) was calculated as early-phase insulin secretion/ HOMA-IR, a measure of the functionality of the pancreas [23]. The Matsuda index was calculated as an assessment of whole-body insulin sensitivity according to the formula by [24]. Insulin/glucagon ratio (IGR) as an indicator of meta-bolic state was based on tAUC, and early phase GIP and GLP-1 secretion was based on iAUC of the first 30 min.

Cut-offs for anthropometric, haemodynamic and biochemical disorders

BMI categories of underweight, overweight and obesity were defined as < 18.5, ≥ 25.0, and ≥ 30.0 kg/m2 , respec-tively [25]. Abdominal obesity was defined as a waist circumference of 94 cm and 80 cm for men and women, respectively, and as waist-hip ratio (WHR) of 0.90 and 0.85 for men and women, respectively [25]. Mid-upper arm cir-cumference (MUAC) of < 24.0 cm was defined as under-weight [26].

    Anaemia was defined as blood haemoglobin < 13.3 g/dL for men and < 12.3 g/dL for women, including an altitude correction factor of + 0.3 g/dL to avoid overestimation of haemoglobin levels [27].

    For hyperglycaemic values, impaired fasting glucose (IFG) was defined as 6.1 to 6.9 mmol/L for the plasma fast-ing value, impaired glucose tolerance (IGT) was defined as < 7.0 mmol/L for fasting plasma value and ≥ 7.8 to < 11.1 mmol/L for the 120 min. value, and DM was defined as ≥ 7.0 mmol/L for the fasting plasma value, or ≥ 11.1 mmol/L for the 120 min. value [28].

Statistics

Continuous background variables are presented as mean (SD) and summary measures, AUCs and indices are pre-sented as mean (SD), or if skewed, data were log-transformed to determine p-value and then calculated as geometric mean (95% CI). Sex differences between variables were deter-mined by Student’s t-test. For an assessment of the role of exposure variables for the outcome differences of AUCs for men and women, multivariable linear regression analyses were conducted using sex, age and BMI as fixed indepen-dent variables. The impact of BMI as an exposure variable on the outcome variables (glucose, insulin, C-peptide, GIP, GLP-1, and glucagon) was greater than the impact of WC, WHR, VAT, SAT or VAT/SAT ratio and therefore chosen as the anthropometric exposure variable.

Ethical considerations

Ethical approval was obtained locally from Tumaini Uni-versity, Moshi (Certificate no. 507), and from the National Institute of Medical Research in Tanzania (Ref. no. NIMR/ HQ/R.8a/Vol IX/2806). All procedures were conducted in accordance with the Declaration of Helsinki and partici-pants were treated as prescribed in the Tanzanian national

Data and resources availability

The data obtained and analysed in the current study are available on reasonable request.

Results

Number of eligible study participants and background characteristics

In total, 74 Maasai volunteered to participate in the study. We excluded one with known DM, 5 due to pregnancy, and 10 (1 male, 9 females) due to missing a 2-h blood test value. Thus, 58 individuals (29 men, 29 women) completed the study. Mean age was 34.8 (SD 11.9) (range 17–65) years, and mean BMI was 20.3 (SD 2.9) (range 14.0-30.9) kg/m2 . Background characteristics by sex are presented in Table 1.

OGTT-derived plasma concentrations

Based on the fasting test and OGTT, one male participant had IFG, and three participants (1 male, 2 females) had IGT, respectively. Fasting plasma glucose concentrations in men and women were significantly different at 5.2 (SD 0.6) vs. 4.9 (SD 0.5) mmol/L, respectively (p=0.031). Based on the OGTT, there was a significant difference in geometric mean plasma GLP-1 with higher concentrations in women (8.5 (95%: 6.5; 11.0) pmol/L) than men (5.1 (95%:4.1; 6.5) pmol/L) (p=0.005) at 120 min. For tAUC, GLP-1 was sig-nificantly higher in women (1336 (95% CI: 1048; 1704) pmol/L x min) compared to men (870 (95% CI: 690; 1097) pmol/L x min) (p=0.011). All other tAUC comparisons were statistically non-significant. Comparisons by iAUC did not alter any of the results (data not shown). All results based on the OGTT are presented in Table 2, and tAUCs are graphically presented in Fig. 1.

    When tAUCs were adjusted for sex, age and BMI in regression analyses, borderline significant differences between men (reference group) and women were seen for insulin (11 724, 95% CI: -394; 23 841) nmol/L x min (p=0.058), C-peptide (29 529, 95% CI: -4989; 64 047) nmol/L x min (p=0.092), and for GLP-1 (464, 95% CI: -38; 966) pmol/L x min (p=0.069), respectively. Results for glucose, GIP and glucagon were statistically insignificant (data not shown). Replacing BMI with WC, WHR, VAT, SAT or VAT/SAT ratio when adjusting the regression analy-ses of tAUCs did not alter the results (see Supplementary Table). Furthermore, omitting individuals having pre-DM (n=4) from the adjusted regression analyses did not alter the results (data not shown).

Under/overweight, hypertension and anaemia

Underweight was found in fourteen (6 males, 8 females) and eight (3 males, 5 females) participants by BMI and MUAC, respectively. Based on BMI, one male partici-pant was overweight, and one male participant was obese. Abdominal obesity was found in five (2 males, 3 females) and one female participant by WC and WHR, respectively. Nine participants had hypertension (5 males, 4 females). One female participant had anaemia.

Discussion

In this study, based on a standard 75 g OGTT, we demon-strated that in traditionally living Maasai, men had higher mean fasting plasma glucose compared to women, while there was a tendency towards higher mean 120 min. plasma insulin and C-peptide in women compared to men. At 120 min. of the OGTT, mean plasma GLP-1 was higher in women compared to men. When calculating AUC (total or incremental), women had higher mean plasma GLP-1 con centrations. Furthermore, adjusted AUC insulin, C-peptide and GLP-1 plasma levels were borderline significant, with higher values for women.

    The higher fasting plasma glucose in Maasai men sup-ports similar results in studies carried out in different popu-lation groups, i.e. Mauritius, Australia, Denmark, and Japan, respectively [10–13], even though we did not find higher fasting insulin concentrations or higher HOMA-IR in men.

The latter is mainly an indication of hepatic insulin resis-tance [29]. Thus, hepatic insulin resistance did not seem to be affected by higher central fat accumulation (WC, WHR, and VAT/SAT ratio) among men in the current study popula-tion. Even though 120 min. glucose levels were not differ-ent between the sexes, there was a tendency towards higher 120 min. insulin and C-peptide concentrations in women. This was supported by AUC concentrations for insulin and C-peptide adjusted for sex and BMI, indicating that over-all, during a standard OGTT and at the 120 min. time-point, higher insulin/C-peptide concentrations in women is a result of inter-sex physiological differences. Women seem to exhibit a greater insulin secretion capacity than men [30]. This is possibly due to differences in height and thereby lean body mass, i.e. there is a higher glucose-to-muscle mass ratio in women compared to men [31], and therefore an increased need for insulin in women. As there was no difference in early phase insulin or C-peptide secretion (first 30 min. of the OGTT) between men and women, the higher 120 min. insulin level could be a compensatory mechanism for a lower peripheral insulin sensitivity in women com-pared to men, which conflicts with previous studies show ing higher peripheral insulin sensitivity in women, albeit under non-physiological conditions (clamp studies) due to enhanced glucose uptake by skeletal muscle in women [32, 33]. Nonetheless, our results of whole-body insulin sensitiv-ity as calculated through the Matsuda index did not support sex differences.

    The incretin hormones GIP and GLP-1 stimulate insu-lin secretion in a glucose-dependent manner [34, 35].

Therefore, enhanced OGTT-derived insulin concentration, which we observed in women, may also reflect sex differ-ences in GIP and GLP-1 secretion. While we found no dif-ferences in GIP secretion, plasma concentrations of GLP-1 at 120 min. and tAUC GLP-1 were higher in women. Sex differences in incretin hormone response during the 75 g OGTT between men and women have previously been reported in individuals with normal glucose tolerance as well as hyperglycaemia, albeit with different results [36, 37]. In a twin study discordant for type 2 diabetes including obese Danish individuals with an average age of 60 years, the authors showed that plasma GIP and GLP-1 concentra-tions were higher in women following a 180 min. OGTT [36], while young Japanese individuals with normoglycae-mia (average age 25 years) exhibited higher GIP concen-trations among men following a 120 min OGTT [37]. The discrepancy between the incretin response to an OGTT, including the current findings, could be explained by differ-ences in age, ethnicity and body composition. Furthermore, in an epidemiological study including a sub-group of 774 Danish individuals with normoglycaemia, women had sub-stantially higher iAUC at 30 and 120 min. than men when adjusted for age and BMI [38].

    Plausible physiological explanations for the higher GLP-1 plasma concentrations in women could be differ-ences in gastric emptying and/or caused by sex-specific

hormones. As for the former, little information is available in normoglycaemic individuals, and the available evidence shows conflicting results in difference of gastric emptying rate between men and women [39, 40]. Sex hormones, on the other hand, have been shown to play a role in relation to not only GLP-1 secretion, but also insulin and glucagon secretion. Estrogens act on the pancreas to increase insulinsecretion, while they decrease glucagon secretion [39, 40]. Furthermore, enteroendocrine L-cells express receptors for estrogen and for progesterone, and they secrete more GLP-1 in response to these two steroids [41].

    It is of note that Nielsen et al. [41]. recently reported a similar OGTT-derived incretin effect in Africans from Tanzania comparing groups of individuals with DM (diag-nosed from OGTT or HbA1c) or without DM. The authors speculate that the lack of difference could be due to the rela-tively young age of the participants (average age < 50 years) and therefore possibly at an early stage in their diabetes dis-ease. However, a difference in the assays utilized compared to the current study could also explain the “negative” find-ing in the study by Nielsen et al. [41].

    When added together, the favourable incretin profile in women (higher 2-h as well as tAUC GLP-1 concentrations) together with increased insulin/C-peptide concentrations (borderline significant) do not result in a lower plasma glu-cose tAUC compared to men. This could be due to the fact that by giving the same glucose load to men and women, the relative glucose load was higher in women due to their lower weight (~10 kg difference), and thereby lower lean body mass. Thus, women are better at handling the blood glucose, probably as a consequence of higher GLP-1 concentrations.

    There is solid evidence that the composition of dietary intake has an impact on glucose homeostasis, including postprandial and glucose challenge tests [42]. In the current study, the composition of the habitual dietary intake (aver-age of two 24-h recall interviews) at the macronutrient level was similar among Maasai men and women, including very high relative protein intake in both sexes (~ 2 g · day− 1 · kg bw− 1). Aerobic fitness – a proxy for physical activity– is also related to glucose homeostasis and insulin sensitiv-ity, and the average fitness levels (~ 16% higher in men) were moderate for both sexes and within normal level of difference between men and women, i.e. ~15–25% for this parameter [43, 44].

Limitations of the study

The study had several limitations, which could have affected the results of the OGTT. Apart from securing a state of over-night fasting, we could not secure standardization of the dietary intake for 3 days prior to the OGTT as recommended. Furthermore, we had to allow the study participants to walk slowly to a nearby restroom during the OGTT if requested. Regarding insulin indices, we acknowledge that HOMA-IR as well as the Matsuda index are indirect surrogates derived from the OGTT rather than clamp-based measures. Due to the “field conditions”, we were only able to collect simple anthropometric and estimated aerobic fitness data. Thus, neither body composition nor fitness could be measured with high precision. Finally, the results presented should be interpreted with caution and need replication in other inde-pendent datasets among Maasai or other agro-pastoralist groups with similar lifestyle and -conditions.

    In conclusion, in rural Maasai men and women of wide ranges of age (17–65 years) and BMI (14.0–30.9 kg/m2 ) and without having DM, fasting and OGTT-derived plasma glucose concentrations showed a higher average fasting concentration in men, while women exhibited higher aver-age 120 min. insulin, 120 min. GLP-1 and tAUC GLP-1 concentrations. A larger sample size and inclusion of more sophisticated methodologies should be included in future studies to investigate sex-specific biology in this unique population known to have a low prevalence of DM despite growing obesity at the population level.

Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s00592-0 26-02710-8.

Acknowledgements Personal thanks: We highly appreciate the will-ingness of the Maasai to participate in the current study, and we are indebted to Ms. Mary Himiri for organising logistics of this study and specifically for collecting dietary intake data. The authors acknowl-edge the collection of blood samples by University of Copenhagen stu-dents at the time of the study, Anne Schou, Frederik Andersen, Andreas Clipét-Jensen, and Katrine Gotthelf Hansen. Finally, we thank district medical officer, at Monduli District Hospital, Dr. Edward Lengai, for endorsing the study.

Author contributions Dirk Lund Christensen: writing – original draft, visualisation, methodology, formal analysis, conceptualisation, proj-ect administration, funding acquisition, investigation, supervision. Cathrine Olesen Emborg: writing – review & editing, visualisation, methodology, investigation. Kaushik Laxmidas Ramaiya: writing – re-view & editing, visualisation, project administration. Venance Philip Maro: writing – review & editing, visualisation, project administra-tion. Ib Christian Bygbjerg: writing – review & editing, visualisation, methodology. Joseph Sironga: writing – review & editing, visualisa-tion, methodology, investigation. Jens Juul Holst: writing – review & editing, visualisation, methodology, Kajiru Kilonzo: writing – review & editing, visualisation, investigation. Bolette Hartmann: writing – review & editing, visualisation, methodology, validation. Flemming Dela: writing – review & editing, visualisation, methodology. Steen Larsen: writing – review & editing, visualisation, conceptualisation, methodology, investigation. Jørn Wulff Helge: writing – review & ed-iting, visualisation, methodology, formal analysis, conceptualisation, funding acquisition.

Funding Open access funding provided by Copenhagen University. The study was funded by the Augustinus Foundation, the Dr. Thor-vald Madsen’s Grant for the Advancement of Medical Science, and the North-East Diabetes Trust. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writ-ing of the manuscript, or in the decision to publish the results.

Declarations

Conflict of interest The authors declare the following financial in-terests/personal relationships, which may be considered as potential competing interests: D.L.C. is supported by funding from the World Diabetes Foundation (Project no. WDF06-0221). K.L.R. is a board member of the World Diabetes Foundation and is currently supported by funding from the World Diabetes Foundation (Project no. WDF06- 0221). J.J.H. is on the Advisory Panel for Novo Nordisk A/S. and is a consultant for Novo Nordisk A/S, AstraZeneca, Fractyl Health, Inc., MSD Life Science Foundation, and Structure Therapeutics, Inc. B.H. is an employee of Bainan Biotech. The remaining authors declare no potential conflicts of interest relevant to this study.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.o rg/licenses/by/4.0/.

This article is excerpted from the 《Acta Diabetologica》 by Wound World.