Body Weight Change and Type 2 Diabetes

Abstract

Listen

Translate article

To the Editor: Weight gain in early adulthood is a well-known predictor of type 2 diabetes mellitus, regardless of initial obesity status. However, the quantitative association between weight gain and diabetes has not been fully established. In particular, it remains unclear as to whether there is a threshold of weight gain above which diabetes risk will increase (similar to body mass index [BMI] >25 kg/m2 as a criterion for high risk of diabetes).1 We explored the dose-response association between weight gain from early adulthood and risk of type 2 diabetes mellitus using a meta-analytic approach. We conducted electronic literature searches for studies that prospectively investigated relative risk (RR) of incident type 2 diabetes mellitus in relation to weight changes compared with weight maintenance, using MEDLINE (from 1950 to January 2013) and EMBASE (from 1974 to January 2013). Studies were included if (1) all participants were derived from the general population 25 years or older; (2) periods of assessment of weight change did not extend beyond the diagnosis of diabetes; and (3) initial weight was assessed when participants were 18 to 24 years old. From a total of 8253 articles retrieved from the study search, we found seven eligible studies2–8 consisting of 342,703 participants and 15,397 cases (eTable, eFigure, https://links.lww.com/EDE/A704). For each study, we plotted logarithms of relative risk (lnRR) for diabetes in weight change groups against mean weight change (expressed as the midpoint of the upper and lower boundaries for data on weight change) and compared this with a reference group of people who maintained their weight. In the highest and lowest weight change categories, we estimated the midpoint by assuming that the breadth of weight change in these categories was equal to that of their closest category. If the analyzed studies expressed weight change in kilograms,2,4,5,8 we standardized the weight change into BMI units using data on height reported by other studies that reported on the same study population. Linear and spline regression analyses were conducted to describe the quantitative association, using the inverse of the square of the standard error for lnRR as the study weight. The Figure shows linear and spline regression curves for type 2 diabetes mellitus risk by BMI change. The linear regression model had high goodness of fit for the risk of diabetes (adjusted R2 = 0.81), with the result that a 1 kg/m2 increase in BMI was associated with an 18% (95% confidence interval = 14–22%) elevated risk of diabetes. The spline regression model also had high goodness of fit (adjusted R2 = 0.87) and was better than the linear model (P < 0.001). In the spline model, the risk of diabetes was slightly accelerated between the two knots (3.7 and 6.6 kg/m2 of BMI gain). However, there was no apparent threshold above which diabetes risk was dramatically elevated.FIGURE: Regression of relative risk (RR) of type 2 diabetes mellitus on change in body mass index (BMI). Solid curve indicates RR based on a restricted cubic-spline regression of BMI gain, with knots represented by white circles. Area of each black circle shows its statistical weight.These results suggest a strong linear relationship between the amount of weight gain in adulthood and subsequent risk of type 2 diabetes mellitus. There was no specific level of weight gain from early adulthood for detecting those at high risk of diabetes, unlike criteria for overweight or obesity. One major study limitation is that information on initial weight depended on self-report in all but one included study.4 Another limitation is that the reason for weight change was unknown. Weight loss might be intentional or it could be due to an illness that in turn might affect diabetes risk. Despite these limitations, our results suggest that weight gain predicts future diabetes risk in a continuous manner. This suggests that there is no safe level of weight gain with regard to an increased risk of diabetes. ACKNOWLEDGMENTS We thank Satomi Fukuya and Mami Haga for their excellent secretarial work. Satoru Kodama Department of Health Management Center Mito Kyodo General Hospital Ibaraki, Japan Department of Hematology, Endocrinology and Metabolism Niigata University Faculty of Medicine Niigata, Japan Chika Horikawa Sakiko Yoshizawa Department of Hematology, Endocrinology and Metabolism Niigata University Faculty of Medicine Niigata, Japan Kazuya Fujihara Department of Internal Medicine University of Tsukuba Institute of Clinical Medicine Ibaraki, Japan Yoko Yachi Department of Hematology, Endocrinology and Metabolism Niigata University Faculty of Medicine Niigata, Japan Shiro Tanaka Department of Pharmacoepidemiology Graduate School of Medicine and Public Health Kyoto University Kyoto, Japan Akiko Suzuki Osamu Hanyu Department of Hematology, Endocrinology and Metabolism Niigata University Faculty of Medicine Niigata, Japan Hiroaki Yagyu Department of Internal Medicine University of Tsukuba Institute of Clinical Medicine Ibaraki, Japan Hirohito Sone Department of Hematology, Endocrinology and Metabolism Niigata University Faculty of Medicine Niigata, Japan [email protected]