Abstract P164: Most Non-invasive Diabetes Risk Scores Identify Type 2 Diabetes Due To Insulin Resistance But Not β-cell Failure: The Africans In America Study

Abstract

Background: Five of the most important Type 2 Diabetes (T2D) prediction scores are: Cambridge, Finnish Diabetes Risk Score (FINDRISC), Rotterdam, Kuwaiti and Omani. These scores are based on the premise that Insulin Resistance (IR) is the primary cause of T2D. Yet, data in Africans suggest β-cell-failure without obesity is the most common cause of T2D. It is unknown whether these scores can identify undiagnosed T2D when the cause is β-cell-failure. Aims: Our aim was to compare in 514 African-born Blacks (age 38±10 (mean±SD); 68% male; BMI 28±5 kg/m 2 ) the ability of Cambridge, FINDRISC, Rotterdam, Kuwait and Omani scores to identify previously undiagnosed T2D due to either IR or β-cell-failure. Methods: Diabetes was diagnosed by OGTT. Insulin resistance was defined by the lowest quartile of the Matsuda index (≤2.97). Beta-cell failure was defined as T2D diagnosis in the absence of IR. Body mass index (BMI), waist circumference (WC) and visceral adipose tissue (VAT) was measured. Area under the Receiver Operator Characteristic curve (AROC) estimated sensitivity, specificity, and criteria for AROC were: 0.70-0.79=acceptable; 0.80-0.89=excellent; >0.90=outstanding. Logistic regression determined odds of T2D at the optimal cut-point for each score. Results: The prevalence of T2D was 7% (38/514). Of the T2D cases, IR occurred in 61% (23/38) and β-cell failure in 39% (15/38). Age did not differ by group, but the IR group had higher BMI (31±5 vs 27±4 kg/m 2 P <0.001) and WC (101±10 vs 91±10 cm P <0.001). This is supported by the higher VAT in the IR group (161±73 vs 106±62 cm 2 P <0.001), an important physiological correlate of insulin resistance. All scores showed acceptable to outstanding predictability for T2D due to IR, ranging from Cambridge at AROC=0.75, to the FINDRISC with AROC=0.92. Only weak to moderate predictability for all scores were shown in β-cell-failure (Cambridge at AROC=0.48 to FINDRISC with AROC=0.63). All models predicted significant odds of T2D due to IR, with FINDRISC≥10 OR=6.8 (95%CI, 2.6, 17.8) P<0 .001; Rotterdam≥7 OR=8.5 (95%CI, 3.1, 23.2) P<0 .001; Cambridge≥23 OR=11.4 (95%CI, 2.6, 50.0) P =0.001) proving most effective. Only the Cambridge score significantly predicted odds of T2D in β-cell-failure (OR=5.7 (95%CI 1.58, 20.54) P =0.001). Conclusions: All 5 diabetes risk scores effectively predicted undiagnosed T2D in Africans when the etiology was IR, but only the Cambridge score predicted T2D when the cause was β-cell failure. The Cambridge risk score rates overweight status, not just obesity status, which might explain why it was successful in predicting T2D in β-cell failure and the others were not. Existing scores which predict T2D might be modified, to effectively assess T2D risk when β-cell failure is the primary cause.