In the present study, HMW-adn and HMW-adn/T-adn were negatively correlated with anthropometric variables (BMI, %BF, WC, and WHtR). Furthermore, only WHtR was significantly or marginally significantly correlated with HMW-adn or HMW-adn/T-adn regardless of sex and physique. To the best of our knowledge, this study is the first study regarding the relationship between HMW-adn or HMW-adn/T-adn and anthropometric variables among over 1600 population-based elementary schoolchildren in Japan, where blood collection and the measurement of %BF and WC are not commonly performed in the annual health examination at elementary schools.
In this study, statistically significant differences between boys and girls were found in most anthropometric variables (Table
). For example, %BF among boys was higher than that among girls in the present study, which was consistent with other study results conducted among Japanese children 9–10 years of age
]. A previous study showed that sexual differences in body composition are present very early in life but emerge most dramatically during puberty
]; females enter puberty earlier and undergo a more rapid pubertal transition, whereas boys have a substantially longer growth period. Therefore, we analyzed the data separately for each sex to consider sex differences of anthropometric indices and then examined the relationship of each anthropometric variable to adiponectin levels.
As shown in Table
, anthropometric variables (BMI, %BF, WC, and WHtR) were negatively correlated with HMW-adn and HMW-adn/T-adn. Additionally, HMW-adn and HMW-adn/T-adn were significantly lower in the obesity group than in the non-obesity group regardless of sex (Table
). Even if the definition of obesity in a recent study was applied to this study, HMW-adn and HMW-adn/T-adn levels in the obese group were still significantly lower than those in the non-obese group for both sexes
]. A recent study showed that HMW-adn and HMW-adn/T-adn were negatively correlated with BMI among adults
], while they were reported to be inversely correlated with WC among children
]. Furthermore, some studies reported that HMW-adn and HMW-adn/T-adn levels in obese children were significantly lower than those in non-obese children
]. Therefore, our study results were consistent with other study results.
In the analysis for each physique, the correlation coefficients of adiponectin levels (HMW-adn, T-adn, and HMW-adn/T-adn) with anthropometric variables differed by physique in each sex (Table
); the correlation coefficients in the obesity groups were consistently higher than those in the non-obesity groups. The propensity was similar even if stepwise multiple regression analysis (each anthropometric variable, age, and birth weight were included as independent variables) was performed to analyze the relationship between each adiponectin level (each adiponectin level was log-transformed before analysis) and each anthropometric variable. These results suggested that the impact of changing anthropometric variables on adiponectin levels was stronger among the obese students than among the non-obese students. The adiponectin level was reported to be inversely correlated with visceral adipose tissue area
]. Therefore, the amount of increasing or decreasing visceral adipose tissue observed in relation to changing anthropometric variables in the obese students could be larger than that in the non-obese students, resulting in the physique difference seen in the relationship between adiponectin levels and anthropometric variables. Although further study is needed to elucidate the biological mechanism of this physique difference, it could be necessary to consider physique as well as sex when investigating the relationship between adiponectin levels and anthropometric variables.
Moreover, only WHtR was significantly or marginally significantly correlated with adiponectin levels, including HMW-adn, regardless of sex and physique, as shown in Tables
. Even if stepwise multiple regression analysis (each anthropometric variable, age, and birth weight were included as independent variables) was performed to analyze the relationship between each adiponectin level (each adiponectin level was log-transformed before analysis) and each anthropometric variable, only WHtR was significantly or marginally significantly correlated with each adiponectin level regardless of physique in girls. Previous studies have shown that WHtR was strongly associated with visceral adipose tissue
], while adiponectin was inversely associated with visceral adipose tissue
]. Furthermore, WHtR has been reported to be a stronger predictor of visceral adipose tissue than BMI or WC among adults
]. Therefore, WHtR could be the most useful anthropometric variable for predicting adiponectin levels, which suggests that the monitoring of WHtR is an effective tool for the prevention of MetS, especially among obese students, because the correlation coefficients were consistently higher in the obesity groups than in the non-obesity groups. A recent study reported that WHtR is useful to identify overweight children at higher metabolic and cardiovascular risk
]. In addition, WHtR was reported to be the most significant predictor for total cholesterol, triglyceride, and low-density-lipoprotein cholesterol in Japanese schoolchildren
], while a past study reported that WHtR is a simple and practical anthropometric index for identifying higher metabolic risks in normal and overweight Japanese men and women
]. Moreover, Fujita et al. suggested that BMI, WC, and WHtR are highly accurate indicators of excess abdominal fat in Japanese schoolchildren, demonstrating that anthropometric measures are useful indices for school screening and that WHtR is particularly useful because it is not dependent on age or sex and is easy to use
In our study, the majority of the correlations of adiponectin levels with anthropometric variables for each sex were less than |0.200|, which suggested that the correlations might be weak. In contrast, the majority of the correlations of adiponectin levels with WHtR were more than |0.350| in obese boys and more than |0.500| in obese girls. These results showed that it could be helpful to use WHtR as a surrogate for adiponectin levels among the obese, especially among girls.
In the present study, the blood collection from study subjects was conducted in the morning after eating breakfast, which might have affected the data. However, recent studies showed that the level of circulating adiponectin does not change in response to a high-fat meal or 75 g of oral glucose load for serum glucose
]. Accordingly, the study results could not be affected by postprandial status. The second limitation of this study is the lack of information regarding pubertal stage (Tanner’s stage), although it might be technically difficult to obtain the information from more than 1600 population-based children. A previous study reported that the first sign of puberty was testicular growth (≥ 3 ml) in Japanese boys, attained at a mean age of 10.8 years, and breast development (Tanner stage 2) in girls at a mean age of 10.0 years
]. Because the mean age in the present study participants was 9.3 years old, puberty stage was not likely to have a substantial impact on the results of this study. However, the impact of sex and early pubertal stage on insulin sensitivity and body composition has been reported
], which might affect our study results. Therefore, to verify our study results, pubertal stage should be considered in future research. Thirdly, our study results might be affected by physical activity. However, there was no statistically significant difference between exercise and adiponectin levels for each sex in the present study (data not shown). Fourth, the number of the subjects in the obesity group was very small compared to that in the non-obesity group. Therefore, it might be difficult to compare the correlation coefficients in the obesity group with those in the non-obesity group, because the results in the obesity group could be more affected by outliers. However, our study results were based on Spearman's rank correlation coefficients and the results were similar even when multiple regression analysis (adiponectin level was log-transformed before analysis) was performed. Accordingly, findings in the present study would be not substantially influenced by outliers. Finally, subjects in our study were from one town in Japan. Therefore, it might be difficult to apply the study results to other races, because ethnic differences in serum adiponectin level have been reported