Basal and Postglucagon C-Peptide Levels in Ethiopians with Diabetes

From May to July 1993, 56 subjects with type 1 diabetes and 97 subjects with type 2 diabetes (37 on insulin, 53 on oral agents, and 7 on diet alone) from the Diabetes Clinic of the Black Lion Hospital, Addis Ababa University, Addis Ababa, Ethiopia were included in the study. All subjects with either type 1 or type 2 diabetes who volunteered to participate in the study and were available for laboratory examination were included. None took a morning dose of insulin or oral agents on the day of examination.

Based on the WHO criteria (2,3), subjects with acute onset of classical symptoms who required insulin therapy from the onset to control hyperglycemia or had an episode of ketoacidosis during the course of their illness were classified as type 1 diabetic subjects. The rest were classified as type 2 diabetic subjects.

In addition, 50 nondiabetic subjects were included as control subjects. These were healthy medical student volunteers (n=6) or outpatients in the hospital being treated for minor medical problems (n=44).

After obtaining informed consent from each subject, demographic information was collected by personal interview. Additional clinical information was collected from medical records.

Blood pressure, height, weight, and waist and hip circumferences were measured. Previously described techniques for measuring waist and hip circumferences were used (20). BMI and waist-to-hip ratio (WHR) were calculated.

A sample of 5–10 ml venous blood was withdrawn from each subject in the fasting state for BCP measurement. In 86 subjects (22 type 1 diabetic, 53 type 2 diabetic, and 11 control subjects), the glucagon test was performed by injecting 1 mg glucagon intravenously, followed 6 min later by a second blood-draw of 5–10 ml from a separate venous site. The selection of subjects for the glucagon test was based simply on their agreement to have the test done.

After centrifugation of the blood samples, the sera were isolated and then frozen at −20°C. In September 1993, the transportation of the sera to Germany was performed using dry ice-filled containers at temperatures reaching −70°C. Within <24 h, the specimens reached their destination and were kept frozen at −20°C.

The C-peptide assays were performed within 1–2 months of receipt at the Endocrine Laboratory of the Center for Internal Medicine, University of Leipzig, Leipzig, Germany, using radioimmunoassay kits from Diagnostic Systems Laboratories (Webster, TX). Assays were performed in duplicate using the 4-h assay protocol. The assays were run in several batches with deliberate mixing of groups. In those subjects in whom the glucagon test was performed, both the basal and postglucagon sera of each patient were simultaneously analyzed in the same assay run. The performance characteristics of the assay were: minimum detection limit of 0.003 nmol/l, intra-assay variability of <10%, and interassay variability of <9%.

Quantitative data were summarized by means, SEs, and percentiles, whereas frequencies and percents summarized the categorical data. Group comparison with respect to quantitative and categorical data were performed using the Wilcoxon’s rank-sum test and χ² test, respectively. The cutoff values of 0.2 nmol/l for BCP and 0.32 nmol/l for PGCP were used in assessing concordance with the clinical classification of diabetes. These values, which have been shown by Gjessing et al. (5) to be good discriminators between type 1 and type 2 diabetes, were selected after demonstrating their agreement with our own receiver-operator curve data.

In the group with type 2 diabetes, there was a female preponderance, whereas in the group with type 1 diabetes and the control group, there was a male preponderance. Subjects with type 2 diabetes were on average older than those with type 1 diabetes and control subjects. Type 2 diabetic subjects were relatively more obese (higher BMI) and had a greater degree of upper-body obesity (higher WHR) than type 1 diabetic or control subjects. In all three groups, women were more obese (higher BMI) but had a lesser degree of upper-body obesity (lower WHR) than men (data not shown). The mean systolic and diastolic blood pressure readings were higher in type 2 diabetic subjects than type 1 diabetic or control subjects. Of the 97 subjects with type 2 diabetes, 37 were on insulin, 53 were on oral agents, and 7 were exclusively on diet. Comparison of socioeconomic variables such as occupation and educational level between control and diabetic subjects (both groups) did not reveal any significant differences (data not shown), suggesting that they had similar nutritional status (Table 1).

Comparison of baseline characteristics of the 86 subjects who had the glucagon test with the 117 who did not revealed no significant differences, with the exception of age in the type 2 diabetic group, in which those who had the glucagon test were younger (48.6 ±1.46 vs. 53.7 ±1.7 years, P=0.04).

The distribution of BCP values in the three groups of subjects is shown in Fig. 1A. Whereas 96% of control subjects and 93% of subjects with type 2 diabetes had BCP levels ≥0.2 nmol/l, only 16% of those with type 1 diabetes attained this level.

Figure 1B shows the PGCP levels and Fig. 1C shows the C-peptide increments in the 86 subjects who had the glucagon test performed. All control subjects, 92% of type 2 diabetic subjects, and 14% of type 1 diabetic subjects had PGCP levels ≥0.32 nmol/l. With the exception of a single control subject, all subjects with PGCP levels ≥0.32 nmol/l had BCP values ≥0.2 nmol/l (data not shown). In type 2 diabetic subjects, BCP level was positively correlated with BMI and duration of diabetes but had no correlation with age, sex, and systolic blood pressure (data not shown).

Of the 97 subjects with type 2 diabetes, 37 were on insulin therapy, whereas 60 were managed without insulin. As compared with the non-insulin-treated subgroup, the insulin-treated subgroup had lower BCP levels, lower PGCP levels, as well as a lower C-peptide increment (Table 2).

In both the type 1 and type 2 diabetic groups, subjects with BCP levels <0.2 nmol/l had lower mean BMI than those with higher levels. All six type 2 diabetic subjects with BCP levels <0.2 nmol/l were on insulin therapy (Table 3).

To our knowledge, this is the first large study to analyze the basal and stimulated C-peptide levels in groups of type 1 diabetic, type 2 diabetic, and control subjects in an African population. Our findings demonstrate that C-peptide levels clearly distinguish, albeit not completely, between type 1 and type 2 diabetic subjects. Multivariate analysis revealed that sex was not correlated with C-peptide levels; therefore, it is unlikely that the differences in C-peptide levels among the groups are due to differences in sex distribution.

The fact that virtually all subjects with PGCP values ≥0.32 nmol/l also had BCP values ≥0.2 nmol/l indicates the equivalence of both cutoffs in differentiating between type 1 and type 2 diabetes. Using these cutoffs, however, there was discordance between C-peptide levels and type of diabetes in 14–16% of type 1 diabetic subjects and 7–8% of type 2 diabetic subjects. The most likely explanation for the discordance is misclassification using the WHO clinical criteria. In particular, all of the six subjects with type 2 diabetes having BCP <0.2 nmol/l, who were on insulin and had lower BMI values, may have been misclassified.

The six type 2 diabetic subjects having BCP levels <0.2 nmol/l had a lower mean BMI value compared with that of subjects with higher BCP levels, and all were on insulin therapy. Moreover, in another study in which we tested for antibodies against GAD (21), 2 of those same 6 subjects were positive, whereas only 1 of 80 subjects with BCP ≥0.2 nmol/l was positive. These findings are in agreement with observations made elsewhere regarding the existence of a subgroup of type 2 diabetes that exhibits several features of type 1 diabetes (6,22), which has been named latent autoimmune diabetes in adults (23).

The mean BCP level in type 2 diabetic subjects was only marginally higher (though statistically significant) than that of control subjects. Only 4 of the 97 subjects with type 2 diabetes (4%) had C-peptide values above the upper limit of normal (1.3 nmol/l). These findings suggest that hyperinsulinemia, as estimated by C-peptide measurements, may not be a major feature of type 2 diabetes in Ethiopians. In contrast, results from the developed world as well as from other ethnic groups (24,15,26) and African-Americans (27) demonstrate that type 2 diabetic subjects are characterized by hyperinsulinemia. On the other hand, studies from Africa (13,14,27,28) indicate that African blacks (with and without diabetes) have lower levels of C-peptide when compared with Caucasians, Indians (East), and African-Americans. Other studies conducted in different parts of Africa, including Ethiopia (15,16,17,19), demonstrate no convincing evidence of hyperinsulinemia in type 2 diabetes. In the current study, the low prevalence of obesity in type 2 diabetic subjects (average BMI <25 kg/m²) may be a major reason for the lack of hyperinsulinemia.

Our study also demonstrated that insulin-requiring type 2 diabetic subjects have significantly lower BCP, PGCP, and C-peptide increments after glucagon than those on either dietary treatment alone or oral agents. These findings are supportive of numerous reports in the literature (8,9,10,11,12,15,29) and suggest that determination of basal or stimulated C-peptide levels, in concert with clinical criteria, may prove useful in the early identification of subjects with type 2 diabetes who will require insulin therapy.

In agreement with our findings, previous studies of several populations, including Ethiopia, have demonstrated that subjects with type 2 diabetes have higher BMI and/or WHR compared with type 1 diabetic or control subjects (30,31,32,33). Thus, as in other populations, obesity, in particular upper-body obesity, appears to be a risk factor for the development of type 2 diabetes in Ethiopians.

In summary, our study of 203 Ethiopian subjects allows us to make the following conclusions: 1) in concert with clinical criteria, BCP and PGCP levels are good discriminators between type 1 and type 2 diabetes in Ethiopians and may also be useful in identifying those subjects with type 2 diabetes who will require insulin therapy; 2) in contrast to studies elsewhere, type 2 diabetic subjects in Ethiopia are not hyperinsulinemic (as gauged by C-peptide levels); and 3) type 2 diabetic subjects with low BCP levels exhibit features of type 1 diabetes and may have been misclassified by WHO criteria.

We are grateful to the nursing staff of the Diabetes Clinic, Black Lion Hospital, Addis Ababa, Ethiopia, for their assistance during the recruiting and examination of subjects and to the laboratory staff of the Endocrine Laboratory at the University of Leipzig, Germany, especially Drs. Kellner and Grossman, for their assistance in the determination of C-peptide levels.

Part of the data in this study were originally presented as part of the doctoral thesis (Dr. Med., Germany) of E.S.S. in 1995. In addition, some of the data were presented as an abstract at the Scientific Sessions of the 58th Annual Meeting of the American Diabetes Association in Chicago in June 1998.