Diabetic ketoacidosis (DKA) has been reported in subjects who lack the clinical characteristics of type 1 diabetes (1–3). In a preliminary analysis of the “types” of diabetes in patients presenting with DKA, we found that Hispanic patients had a significantly higher proportion with type 2 diabetes when compared with Caucasians and African Americans (1).
We performed a prospective analysis to compare demographic and clinical characteristics among ketosis-prone indigent subjects belonging to these three ethnic groups. We interviewed 271 consecutive patients at the time of admission for DKA over a 3-year period. Fasting serum C-peptide and glucose levels were measured in all patients after resolution of the ketoacidosis. Pearson’s χ2 test or one-way ANOVA were used, as appropriate, to evaluate group differences. Fasting serum C-peptide levels have been used to distinguish subjects with preserved β-cell function from those with absent β-cell function. We used a cutoff level of 0.33 nmol/l to separate these groups. This serum C-peptide concentration is widely accepted as a cutoff value in the literature (4), and we confirmed this by using receiver operator curve analysis in comparison with the area under the curve for C-peptide response to glucagon stimulation (3). A multivariate analysis was also performed to evaluate factors predictive of fasting C-peptide ≥0.33 nmol/l.
Of the 271 subjects admitted with DKA, 44% were African American, 40% Hispanic, and 16% Caucasian. The proportion of subjects admitted for DKA associated with new-onset diabetes was very similar among all three ethnic groups: 27–28%. However, only 44% of the Hispanic subjects were admitted with DKA secondary to noncompliance with the prescribed treatment for diabetes, as compared with 61% in the African Americans and 57% in the Caucasians (P = 0.01).
The Hispanic group had a significantly higher C-peptide level, 0.41 ± 0.35 nmol/l, compared with 0.25 ± 0.45 in the African American and 0.24 ± 0.32 in the Caucasian groups (P = 0.007). A significantly higher proportion of Hispanics (56%) compared with African Americans (29%) and Caucasians (32%) had a fasting plasma C-peptide level ≥0.33 nmol/l. The C-peptide–to–glucose ratios were 0.038 ± 0.021, 0.02 ± 0.029, and 0.024 ± 0.034 nmol/mmol, respectively, for the Hispanic, African-American, and Caucasian groups (P = 0.0004). In the multivariate analysis, Hispanic ethnicity (odds ratio 3.92, 95% CI: 1.96–8.12), duration of known diabetes <6 months (3.69, 1.57–8.76), and BMI ≥30 kg/m2 (5.70, 2.61–13.04) were significant predictors of fasting plasma C-peptide ≥0.33 nmol/l.
In summary, this prospective analysis of ketosis-prone diabetes shows that, compared with Caucasian and African- American patients, Hispanic patients are more likely to have better preserved β-cell functional reserve, as assessed by a fasting serum C-peptide concentration ≥0.33 nmol/l and by the C-peptide–to–glucose ratio. These differences suggest that there is a higher frequency of ketosis-prone type 2 diabetes among Hispanics than among Caucasians and African Americans in this cohort of indigent subjects. Ethnic comparisons of β-cell function and insulin sensitivity in ketosis-prone diabetes are needed to better understand this syndrome.