PAGE. Abstracts of the Annual Meeting of the Population Approach Group in Europe.
PAGE 25 (2016) Abstr 5848 [www.page-meeting.org/?abstract=5848]
Oral: Growth modelling
Nick Holford (1), Tim Kenealy (2)
(1) Department of Pharmacology and Clinical Pharmacology, University of Auckland, New Zealand (2) Department of General Practice, University of Auckland, New Zealand
Objectives: The most common consequence of obesity is the development of Type 2 diabetes. Body mass index (BMI) is widely used to quantify obesity but it is a size metric without a clear biological basis. Fat mass, on the other hand, is anatomically and physiologically identifiable. Increased fat has been suspected as the causal factor for developing Type 2 diabetes and decreased survival [1, 2]. The objectives of this study are:
Methods: A cohort of 21,002 adult pre-diabetic patients were followed in a New Zealand general practice setting. Total body weight (TBW), height (HT), sex, ethnicity and current smoking habit were recorded at each visit until a diagnosis of Type 2 diabetes was made or the study ended. Fat free mass and fat mass were predicted from weight, height and sex . The clinical outcome event was the time to diagnosis of Type 2 diabetes. Mixed effect joint models of TBW, HT and time to event  were used for model selection and parameter estimation with NONMEM 7.3.
Results: There was a significant trend for TBW (-0.46 kg/y) and HT (-0.12 cm/y) to decrease with age. Significant covariates for baseline TBW were sex, smoking, and race and for rate of decline covariates were sex and smoking. The rate of loss of TBW was 13% faster in women and 33% faster in women smokers. Smaller effects were seen with HT. Predicted fat mass tended to increase in young adults before declining. The baseline hazard of developing Type 2 diabetes had a Gompertz distribution. Explanatory factors were sex, smoking, race and body size. The time course of fat mass was a significant predictor of the hazard.
Conclusions: We have developed a mixed effects model for growth in pre-diabetic humans. The time course of TBW and HT can predict changes in size metrics such as BMI and fat mass in this high risk population. The time to event model provides quantitative evidence for the lipotoxicity theory  of the pathogenesis of Type 2 diabetes. In a drug development setting these models can quantify the benefits of therapeutic goals based on weight reduction and be used to simulate more effective clinical trials .