Multiple factors are required for the prediction of metabolic syndrome

Abstract

We read with great interest the paper by Xia et al., 1 who reported an association between the metabolic syndrome (MetS) and liver enzymes by the following two methods. After setting optimal cut-off values for liver enzymes in MetS by receiver-operating characteristic (ROC) curve analysis, logistic regression analysis was conducted to predict MetS by each liver enzyme. They concluded that MetS was significantly associated with increased serum levels of alanine aminotransferase (ALT) and c-glutamyl transpeptidase (GGT), even after adjustment for age, body mass index (BMI), and lifestyle factors, such as smoking and alcohol consumption.Receiver-operating characteristic curve analysis, a univariate analysis, is often used, 2 although multivariate analysis should also be used to evaluate the relative contribution of each liver enzyme to MetS. Xia et al. 1 did not use all the three liver enzymes simultaneously as independent variables in their analysis illustrated in table 3.I recently addressed this issue using data not described in recent studies. 3,4 In 2008, a total of 2537 male workers aged 34-63 years participated in this study. Patients with fasting plasma glucose levels of under 140 mg ⁄ dL were selected, because the validity of the homeostasis model assessment for insulin resistance (HOMA-IR) decreases at plasma glucose levels exceeding 140 mg ⁄ dL or higher. Patients who showed positive hepatitis type B antigen and ⁄ or positive hepatitis type C antibody were also excluded. Patients with diabetes, hypertension, dyslipidaemia, hyperuricaemia, coronary, and ⁄ or cerebrovascular diseases were excluded. Finally, 2178 patients, who also attended a health examination in 2010, were included in the analysis. Similar to the statistical method selected by Xia et al., I also conducted a logistic regression analysis to assess the predictive ability of several factors simultaneously for MetS. I used BMI, age, serum aspartate aminotransferase (AST), serum ALT, serum GGT, the HOMA-IR, and uric acid as independent variables.A logistic regression analysis (Wald method) revealed the odds ratios (95% confidence intervals) of age, BMI, uric acid, Log-transformed value of the HOMA-IR, Log-transformed GGT, and no habitual drinking in 2008 for the presence of metabolic syndrome. 0.70 (0.49-0.99), respectively.Only serum GGT was selected among the liver enzymes as a significant predictor of MetS in my study. As significant correlations among the three liver enzymes exist, especially between ALT and AST (r = 0.757, P < 0.001), attention must also be paid to the multicollinearity of independent variables. 5 Although the prediction of MetS should be conducted using several related factors simultaneously as independent variables to determine relative significances, I understand that the strategy used by Xia et al. (table 3) was not a wrong strategy to determine the contribution of each liver enzyme to the prediction of MetS. REFERENCES 1. Xia MF, Yan HM, Lin HD et al. Elevation of liver enzymes within the normal limits and metabolic sy...

Elevation of liver enzymes within the normal limits and metabolic syndrome: reply

“…Even after adjustment for other liver enzyme levels, ALT and GGT were still significantly associated with MS in our study, whereas Kawada 1 showed that GGT was the only liver enzyme significantly associated with MS. One possible explanation for the difference may be the study population. The study of Kawada was a prospective community‐based cohort focusing on the risk factors for the new onset of MS, with 2 years of follow up, and individuals with diabetes, hypertension, dyslipidaemia, hyperuricaemia and coronary and/or cerebrovascular disease were excluded from the study 1 . However, in our study we enrolled all participants with and without current MS, even though it was a cross‐sectional study 2 .…”

“…

We appreciate the thoughtful comments by Kawada. 1 The purpose of our study 2 was to detect the association between liver enzymes and metabolic syndrome (MS) and, further, to determine the optimal cutoff values for liver enzymes in MS in 1503 individuals from a physical examination centre. We concluded that both aminotransferase (ALT) and c-glutamyl transpeptidase (GGT) were significantly associated with MS in our study.

…”