Endocrine Abstracts

Background: Ketosis-prone type 2 diabetes (KPD) is characterized by diabetic ketoacidosis (DKA) or unprovoked ketosis without the typical phenotype of autoimmune type 1 diabetes.

Aims: To describe the phenotype of individuals presenting with inaugural DKA or ketosis without β-cell autoimmunity, and to identify predictors of insulin independence.

Methods: This single-center retrospective study included all inpatients aged ≥ 18 years, presenting with inaugural DKA or ketosis between 2010 and 2022. Demographic, clinical, and biological data were collected at baseline (during hospitalization) and last follow-up visit. Patients were classified into Aβ categories: presence (A+) or absence (A-) of β-cell autoimmunity and insulin dependence (β-) or independence (β+) at last follow-up visit. Predictors of insulin independence in the A- group were identified using backward multivariate logistic regression analysis, including all variables with P < 0.200 in univariate analysis. Data are reported as medians [P5-P95] or odds ratios (OR) with [95% confidence interval]. Receiver operating characteristic (ROC) curve analysis was performed to identify the optimal fasting C-peptide (CP) cutoff at baseline predicting for insulin independence.

Results: A total of 144 patients were included in the study, of whom 47% were in the A- group. Age and BMI at baseline were higher in the A- group compared to the A+ group (46 [26-68] vs. 30 [18-59] years, P < 0.001; 28.2 [18.4-38.2] vs. 22.0 [17.8-30.7] kg/m², respectively). Males and Sub-Saharan Africans represented 78% and 35% of the A- group, compared to 61% ( P = 0.032) and 9% ( P = 0.001) of the A+ group, respectively. At last follow-up visit, 39% of patients in the A- group were insulin independent (A-β+). No difference was found between the A-β+ and A-β- groups, except for fewer microvascular complications at baseline in the A-β+ group than in the A-β- group (0% vs. 18%, P = 0.037). Variables included in the multivariate model were overweight (OR 2.42 [0.79-7.48], P = 0.122), hypertension (OR 2.22 [0.77-6.42], P = 0.140), overall micro- and macrovascular complications (OR 7.12 [6.13-52.6], P = 0.098), and fasting CP level at baseline (OR 5.86 [0.79-43.4], P = 0.084). In the multivariate analysis, fasting CP level at baseline was predictive of insulin independence (OR 11.5 [1.02-131.0], P = 0.048). The fasting CP cutoff at baseline with best sensitivity (76%) and specificity (83%) for predicting insulin independence was 0.15 nmol/l.

Discussion: KPD is not uncommon among individuals presenting with inaugural DKA or ketosis at our center. Patients with KPD have a different phenotype from individuals with autoimmune diabetes. In our study, fasting CP level at baseline was an independent predictor of insulin independence in individuals with KPD.