Clin Chem Lab Med 2008;46(1):64–72  2008 by Walter de Gruyter • Berlin • New York. DOI 10.1515/CCLM.2008.007 2007/225 Article in press - uncorrected proof Genetic profiling in healthy subjects from the Stanislas cohort based on 24 polymorphisms: effects on biological variables Eliane Albuisson, Sandy Maumus, Ndeye- Coumba Ndiaye, Be ´ range ` re Marie, Nicolas Jay, Francois Kohler, Ge ´ rard Siest and ¸ Sophie Visvikis-Siest* INSERM U525, Nancy University, Nancy, France Abstract Background: The association of genetic profiles with biological or clinical assessments is not clearly estab- lished especially among apparently healthy subjects. Methods: A multivariate statistical analysis was per- formed on 24 polymorphisms related to the main metabolic pathways involved in cardiovascular dis- eases (CVDs). They were collected among 1551 healthy subjects of the Stanislas cohort to obtain genetic profiles. Association with biological variables was then studied at baseline (t 0 ) and 5 years later (t 5 ). Results: Six genetic clusters were identified with rel- evant profiles and five polymorphisms from the selec- tin, apolipoprotein C3 and lipoprotein lipase genes (SELE-98G/T, APOC3–3175C/G, APOC3–482C/T, APOC3–1100C/T, LPL–93T/G) were sufficiently char- acteristic to associate 99.6% of the subjects with their corresponding cluster. A 5-year follow-up showed that clinical and biological measurements in relation to CVD risk factors already differ with triglyceride (ps0.009 for t 0 and ps0.005 for t 5 ) and high-density lipoprotein cholesterol (ps0.014 for t 0 and ps0.003 for t 5 ) for these previous genetic clusters. Conclusions: This study presents the hypothesis that SELE could be protective, whereas APOC3 could be associated with risk. It remains to be seen whether these polymorphisms will be predictive of CVD events among the selected clusters of different metabolic subtypes after a 10-year follow-up. Clin Chem Lab Med 2008;46:64–72. Keywords: cardiovascular diseases; clusters; multi- variate analysis; polymorphisms; predisposition. Introduction For many years, we have investigated gene-environ- ment interactions in healthy subjects, searching for *Corresponding author: Sophie Visvikis-Siest, INSERM U525, Universite ´ Henri Poincare ´ (Nancy 1), Faculte ´ de Pharmacie, 30 rue Lionnois, 54000 Nancy, France Phone: q33-3-83682184, Fax: q33-3-83321322, E-mail: Sophie.Visvikis-Siest@nancy.inserm.fr Received May 11, 2007; accepted July 31, 2007; previously published online November 18, 2007 relevant genetic markers for cardiovascular risk pre- diction (1–3). A large number of single nucleotide poly- morphisms, primarily based on case control studies, are identified in the literature, and this list must be updated regularly (4–6). Thanks to the Human Genome Project (7) and the International HapMap Consortium, comprehensive websites provide a wealth of readily accessible sequence and polymor- phism data. Proteomics investigators are employing a similar approach. The simultaneous development of systems biology (8) and the accompanying systems literature analysis (9) has opened up a new era that is critical to the success of preventive medicine (10). Above all, cardiovascular diseases (CVDs) are espe- cially complex, with multifactorial determinants in occurrence and aetiology. Factors may or may not correlate, and there is still no clear understanding of the role of genetic polymorphisms as a basis for dif- ferences and contradictions in outcome in genetic- based studies (11, 12). The aim of this study was to identify genetic pro- files among healthy subjects from the Stanislas cohort and then to study the usefulness of this profil- ing as a marker of clinic and/or biological differences or tendencies in the CVD domain. A total of 24 polymorphic genes were selected from the main biological pathways involved in cardiovas- cular risk and disease: lipid metabolism, thrombosis, blood pressure, inflammation, adhesion and homo- cysteine cycle. These genes were measured in a large cohort of healthy families (13) constituted in 1993 and composed of initially apparently healthy subjects. Sta- tistics included a combination of multivariate analysis performed using a similar strategy as described in a previous study (14). Materials and methods Study population The Stanislas cohort was constituted in 1993 and designed to investigate factors related to CVD. Details of the Stanislas cohort have been previously described (1). Individuals were recruited during a detailed check-up examination, and informed consent was obtained from all participants. A total of 1551 individuals (772 men and 779 women), not under treatment with lipid-lowering, antihypertensive or anti- inflammatory drugs, were selected from the cohort. In this relatively young population, with a mean age of 40.5 years (SD 4.6 years), all women were premenopausal. Data collection Personal and medical history, medication, smoking habits and alcohol consumption were recorded using a standard- Brought to you by | Purdue University Libraries Authenticated Download Date | 5/30/15 7:46 PM