We thank Massulli and Vaccaro (1) for their interest and comments regarding our article (2). Moreover, we are pleased to hear that they found results similar to ours in a population-based study of Italian men. They reported that, like Korean smokers, Italian smokers had higher triglycerides and lower HDL cholesterol levels than those who had never smoked. They also showed that smoking is not associated with high fasting glucose or high blood pressure, which is similar to our findings. It is a general belief that insulin resistance is the main mechanism underlying the development of metabolic syndrome. Therefore, they tested the association between smoking and insulin resistance using the homeostasis model assessment of insulin resistance index (HOMA-IR). Contrary to their expectation, they could not find an association with HOMA-IR and they suggested that smoking-associated dyslipidemia is mediated by mechanisms other than insulin resistance.
We agree with their suggestions; however, we would like to comment on some points that must be considered. First, both their study and our own used cross-sectional observational data. As we mentioned in our article, the cross-sectional observational design has inherent limitations. Patients with type 2 diabetes and hypertension are more likely to be taking medicines that influence insulin sensitivity. Furthermore, the lifestyles, diet, and other behavioral factors that can influence insulin sensitivity may have differed. Second, HOMA-IR and fasting insulin values have an inherent limitation for predicting insulin resistance. Third, previous cohort data, which investigated temporal associations to identify causal relationships, have demonstrated that smoking increases the risks of diseases such as type 2 diabetes (3, 4), which are known to have insulin resistance as their underlying mechanism. From these findings, although we agree with their suggestion, we cannot be totally confident that the association between smoking and metabolic syndrome is not mediated by insulin resistance. Further well-designed study of the temporal relationships is needed to evaluate this hypothesis.
