Novel data-driven pathophysiological phenotypes among women with prediabetes in an Indian population

Data on pathophysiological phenotypes for prediabetes in Indian ethnic populations are scarce. We investigated data-driven, pathophysiological phenotypes of prediabetes associated with the incidence of diabetes mellitus.

We characterised pathophysiological prediabetes phenotypes in a survey sample (n = 1455), and its progression to diabetes at one year. Data-driven clusters were identified usingk-means cluster algorithm along with measures of glycaemia, adiposity, dyslipidaemia, beta-cell function, insulin resistance and blood pressure. Logistic regression analysis was employed to assess the risk of diabetes incidence.

Four distinct pathophysiological prediabetes phenotypes were identified: Phenotype 1 (Combined Insulin Resistant and Deficient Prediabetes - CIRDP) [n = 323, 22.2 %]; Phenotype 2 (Severe Insulin Resistant Obese Prediabetes - SIROP) [n = 290, 19.9 %]; Phenotype 3 (Severe Insulin Deficient Prediabetes - SIDP) [n = 544, 37.4 %] and Phenotype 4 (High Blood Pressure Moderate Insulin Deficient Prediabetes - HBPMIDP) [n = 298, 20.5 %]. At one year, the progression to diabetes was highest in Phenotype 1-CIRDP (OR: 6.07, 95 % CI: 3.59-10.65), followed by Phenotype 4-HBPMIDP (OR: 2.05, 95 % CI-1.10, 3.87) compared to Phenotype 3-SIDP after adjusting for age, BMI, educational and marital status.

CIRDP and HBPMIDP had higher progression rates to diabetes at one year compared to SIDP. Eighty percent of the pathophysiological prediabetes phenotypes displayed beta-cell dysfunction.