Genetic 10:557-562 (1993) Risk Factors for Atherosclerosis in Twins David L. Duffy, Dianne L. O'Connell, Richard F. Heller, and Nicholas G. Martin Queensland Institute of Medical Research (o.Lo., N. G.M.J, Brisbane; Centre for Clinical Epidemiology and Biostatistics, University of Newcastle (o.O'C., R.F.H.J, Newcastle, Australia We perfonned multivariate genetic analyses of cardiovascular risk factors from two sets of data on US and Australian female twins. Similar models for body- mass index (BMI), serum low density (LDL) and high density (HDL) lipoproteins, including age as a covariate, were fitted successfully to both groups. These suggested that BMI, or genes responsible for a significant proportion of the variance of BMI, explained correlations between lipid sub fractions, as well as those between blood pressure and lipid sub fractions, especially HDL. o 1993 Wiley-Liss, Inc. Key words: heritability, lipoproteins, blood pressure, triglyceride, LISREL INTRODUCTION Two twin data sets examining risk factors for atherosclerosis were provided for analysis at GA W8. We chose to examine the data from the Kaiser-Pennanente Women Twin Registry (NHLBI female twins). In addition, we looked at a twin data set on cardiovascular risk factors kindly provided by Dr. Dianne O'Connell [O'Connell et ai., 1988] describing Australian female adult twins. We thought it interesting to attempt cross-validation of models in these different groups. METHODS The design of both studies is described elsewhere. Preliminary statistical analyses were perfonned using SAS 6.07 [SAS Institute, 1991] and SPSS-PC [Norusis, 1988]. Genetic path models were fitted using LISREL 7.20 [Joreskog & Sorbom, 1990]. The path modelling included phenotypic multivariate regression perfonned in a (forward) stepwise fashion and multivariate genetic factor models. Address reprint requests to Dr. D. Duffy, Epidemiology Unit, Queensland Institute of Medical Research, 300 Herston Road, Herston, Queensland, Australia 4029. 1993 Wiley-Liss, Inc.