A metabolomics model to predict the cardiovascular disorder among the Orang Asli / Nornazliya Mohamad

Introduction: Majority of the Orang Asli rely on the traditional healing method and belief in self therapy despite the introduction of modern medicine. There are lack of medical or clinical reports on their health status as access to the Orang Asli for medical check up is limited due to the low level of trust and misbelief in the modem medicine. Moreover, routine multiple blood sampling for biochemistry tests are difficult to be conducted among them. Metabolomics has been conducted to profile metabolites which are useful to represent the phenotypes of different disease types and stages. The use of global and targeted metabolomics approach enable us to give an information of their metabolism state and therefore provide the insights to their health status. Hence, it was chosen in this study to predict the phenotype of the self claimed healthy Orang Asli. Method: Serum of 83 Orang Asli (OA), 50 healthy volunteers (HT) and 31 patients with history of myocardial infarction (MI) were analysed using LC/MS-Q/TOF. Targeted MS/MS were performed to validate the features of metabolites. Biochemical profiles were determined and recorded for all subjects. Metabolites that were significantly different between MI and HT (p<0.005, >2-fold-changes) were validated using AUROC analysis. Metabolites with AUOO.7 were selected as putative biomarkers, which were further used as variables in building a prediction model (PLSDA) differentiating healthy phenotype and MI. Results and discussions: Our finding revealed that lipids were the major metabolites that significantly different between OA and HT (p<0.005). AUROC analysis had identified twenty-two (22) metabolites as potential biomarkers for myocardial infarction. They include 15(S)-HETE (AUC=0.997), phosphorylcholine (AUC=0.995) and 24,24- Difluoro-25-hydroxy-26,27-dimethylvitamin D3 (AUC=0.976). Prediction results revealed that majority of the OA were clustered with the healthy subjects and therefore have similar metabolic state with the group. This shows that they are healthy as they had claimed and as further illustrated by the conventional biochemical analysis. There were
however, seven (7) OA whose metabolite profiles were clustered with the patients group, highlighting that they had abnormality in lipid metabolism and implied that they-had MI. An evaluation of their biochemistry profiles show abnormalities of cholesterol, triglyceride, HDL and LDL. Conclusion: Metabolomics-PLSDA prediction model was
developed in this study using comprehensive validation approach and it was able to predict the health status of the Orang Asli. This model was developed using the
differential metabolite profiles of the MI and healthy volunteers; and would be useful for categorization of an individual’s phenotype to either group.