Endocrine Journal
Online ISSN : 1348-4540
Print ISSN : 0918-8959
ISSN-L : 0918-8959
ORIGINALS
Likelihood ratio-based integrated personal risk assessment of type 2 diabetes
Noriko SatoNay Chi HtunMakoto DaimonGen TamiyaTakeo KatoIsao KubotaYoshiyuki UenoHidetoshi YamashitaAkira FukaoTakamasa KayamaMasaaki Muramatsu
Author information
JOURNALS FREE ACCESS

2014 Volume 61 Issue 10 Pages 967-988

Details
Abstract

To facilitate personalized health care for multifactorial diseases, risks of genetic and clinical/environmental factors should be assessed together for each individual in an integrated fashion. This approach is possible with the likelihood ratio (LR)-based risk assessment system, as this system can incorporate manifold tests. We examined the usefulness of this system for assessing type 2 diabetes (T2D). Our system employed 29 genetic susceptibility variants, body mass index (BMI), and hypertension as risk factors whose LRs can be estimated from openly available T2D association data for the Japanese population. The pretest probability was set at a sex- and age-appropriate population average of diabetes prevalence. The classification performance of our LR-based risk assessment was compared to that of a non-invasive screening test for diabetes called TOPICS (with score based on age, sex, family history, smoking, BMI, and hypertension) using receiver operating characteristic analysis with a community cohort (n = 1263). The area under the receiver operating characteristic curve (AUC) for the LR-based assessment and TOPICS was 0.707 (95% CI 0.665−0.750) and 0.719 (0.675−0.762), respectively. These AUCs were much higher than that of a genetic risk score constructed using the same genetic susceptibility variants, 0.624 (0.574−0.674). The use of ethnically matched LRs is necessary for proper personal risk assessment. In conclusion, although LR-based integrated risk assessment for T2D still requires additional tests that evaluate other factors, such as risks involved in missing heritability, our results indicate the potential usability of LR-based assessment system and stress the importance of stratified epidemiological investigations in personalized medicine.

Information related to the author
© The Japan Endocrine Society
Previous article Next article
feedback
Top