Drs. Chavez-Tapia, Mendez-Sanchez, and Uribe argue that our study's lack of a second ultrasonographic evaluation by a different reviewer restricts the accuracy of the diagnosis of NAFLD. We understand that ultrasonography is not a perfect test and may lead to an incorrect diagnosis of NAFLD in many patients. However, as reported in the reference they quoted (1), interobserver variability is much greater than intraobserver variability for evaluation of the pattern and severity of NAFLD. The inclusion of a second reviewer would have decreased the power to detect longitudinal changes of ultrasonographic images. We believe that ultrasonographic interpretation by a single experienced gastroenterologist was a better approach for our longitudinal study.

Dr. Chapez-Tapia and colleagues raised a concern regarding our substitution of BMI for waist circumference for the diagnosis of the metabolic syndrome. We used BMI because we could not obtain waist circumference measurements for the entire study sample. A recent large study (2) found a close correlation between BMI and waist circumference, and similar correlations between BMI and waist circumference and metabolic abnormalities have been identified in Japanese people (3); we believe these reports support our methods. As stated by Dr. Chapez-Tapia and colleagues, the cutoff value of 25 kg/m² may not be optimal for diagnosis of the metabolic syndrome in the Japanese population; however, we adopted this value because it is used as a criterion of obesity for people in East Asia.

We also found that many criteria for the metabolic syndrome were positively associated with the development of NAFLD and negatively associated with regression of the disease, although we did not report these particular observations.

We had no data regarding NASH; however, we appreciated Dr. Kida's and Dr. Sato's extension of our discussion regarding the role of adipocytokines in the pathophysiology of NAFLD.