918 ADDED VALUE OF NOVEL BIOMARKERS IN TYPE 2 DIABETES PREDICTION

Abstract

Objective: To summarize the literature assessing the added value of novel biomarkers for predicting the occurrence of type 2 diabetes Design: Systematic review and qualitative synthesis. Data Sources: MEDLINE and EMBASE databases searched from January 2000 to July 2011. Study selection: Published studies in English assessing the improvement of type 2 diabetes risk prediction subsequent to adding novel biomarkers to traditional risk factors in prediction models. Data Extraction: Independent, duplicate identification of studies, data extraction and quality assessment conducted by two reviewers using standardized forms, with discrepancies resolved by consensus or a third reviewer. Extracted data was on study methods and metrics used for evaluating the incremental predictive values of novel biomarkers. Results: We included 28 publications from 25 studies that assessed the incremental predictive ability of novel biomarkers. All the studies reported a change in the area under the receiver-operating characteristic curve, which was modest, ranging from 0 to 0.1, with claims of statistically significant improvement in eight studies. The net reclassification index was evaluated in seven studies, and ranged from -0.03% to 10.2% after inclusion of genetic markers in four studies (statistically significant in two cases), and from -0.5 % to 11.8% after inclusion of non-genetic markers in three studies (non-significant). The integrated discrimination index (0.0015 to 2.4) was reported in five studies, being statistical significant in only three of these. Conclusions: Currently known novel biomarkers do not significantly improve T2DM risk prediction above and beyond the ability of known traditional risk factors.