The present study demonstrated a strong, linear relationship between waist circumference and insulin resistance as expressed by HOMA-IR in a large population of Japanese men. Both geometric means and prevalence odds of elevated HOMA-IR were progressively greater in proportion to the size of waist circumference. The increase was evident even in men with the average size of waist, i.e., 80–84 cm. It was also found that 85 cm of waist circumference was an optimal cutoff for predicting insulin resistance. The finding adds to evidence for optimality of the cutoff for waist circumference proposed for Japanese men.
Obesity has been known to be positively related to insulin resistance. Increased secretion of free fatty acids, inflammatory cytokines and decreased secretion of adiponectin are molecules mediating obesity and insulin resistance [18
]. Few studies have directly addressed the relationship between waist circumference and insulin resistance or hyperinsulinemia [20
]. A small cross-sectional study reported a linear increase in the prevalence of hyperinsulinemia across the deciles of waist circumference in 185 healthy men in Canada [20
]. In a cross-sectional study of 2746 volunteers aged 18–72 years, including 798 men, waist circumference was strongly correlated with HOMA-IR [21
]. The present study was the ever largest study of men examining the relation between waist circumference and HOMA-IR. It is also notable that even waist circumference of 80–84 cm was associated with an evident increase in the prevalence odds of insulin resistance. The finding is in agreement with the notion that obesity-related risk is present at much lower levels of obesity in Asians as compared with Caucasians [22
Insulin resistance is an obesity-related condition preceding the development of impaired glucose tolerance and type 2 diabetes. Insulin resistance, through suppression of glucose uptake in skeletal muscle and increase in hepatic glucose production, causes hyperglycemia [1
]. Insulin resistance expressed by HOMA-IR is well in agreement with that evaluated directly by the euglycemic clamp method [15
], and is generally accepted as a valid method in epidemiological surveys [23
]. However, there is no clear cutoff for the definition of insulin resistance based on HOMA-IR. Insulin resistance based on HOMA-IR has been defined differently in different studies. For instance, insulin resistance was defined as HOMA-IR of ≥ 3.80, which corresponded to the 90th percentile in healthy subjects, in Spain [24
], and a value of 4.00 or greater was used in a Swedish study [21
]. HOMA-IR ≥ 1.73 was used in a Japanese study [25
]. We defined insulin resistance arbitrarily as HOMA-IR greater than the 80th percentile (≥ 2.00) in examining the relationship with waist circumference. Repeated analyses using HOMA-IR cutoffs of 1.50 and 2.50 showed almost the same results in terms of the prevalence odds (data not shown). Different values of HOMA-IR were used in the ROC curve analysis of searching for an optimal cutoff for waist circumference. It should be noted that the optimal cutoff point was consistently 85 cm for any values of HOMA-IR. However, the discordance between abdominal obesity (≥ 85 cm) and insulin resistance accounted for no less than 33%. This value was almost equal to or slightly greater than those reported for the discordance between metabolic syndrome and insulin resistance among Americans and Turks [26
]. Thus, although waist circumference was found to be strongly related to insulin resistance, it is unlikely that Japanese men have less confounders in the relationship under study as compared with other ethnicities.
More emphasis has been placed on waist circumference as an estimate of visceral adiposity rather than body mass index in the prevention of obesity-related diseases. In the present study population, waist circumference and body mass index were highly correlated with each other (Pearson correlation coefficient 0.86), and body mass index was as strongly associated with HOMA-IR as waist circumference. For example, an increase of one SD in body mass index was associated with an 42.0% increase in HOMA-IR while the corresponding value for waist circumference was 43.7%.
A strength of the present study was that the study population was very large and relatively homogeneous in terms of the social background. There were several weaknesses to be discussed. The present study was based on estimated measures for both visceral adiposity and insulin resistance, and these estimates necessarily suffered some inaccuracy which may have attenuated the association between the two. Interpretation of causality is difficult in cross-sectional studies. It is possible that insulin resistance or related conditions may increase visceral adiposity. Furthermore, inflammation and atherosclerosis are linked to insulin resistance, and these conditions may confound the relationship under study. Our study subjects were not representative of middle-aged Japanese men. Age was limited to a small range, and the subjects were those who had remained in the SDF until the age of 50 years on average. It is possible that the study subjects differed from the general population with respect to both waist circumference and insulin resistance, although body mass index in the study population did not differ from that of the general population. In the National Nutrition Survey in 2000, means of body mass index for men aged 40s and 50s were 23.5 and 23.6, respectively [27
]. It is also a limitation that the study did not include women. Women, particularly middle-aged ones, are very few in the SDF. Finally, it may be argued that inclusion of men with oral medication for diabetes may have distorted the association between waist circumference and HOMA-IR. However, such men still had higher values of HOMA-IR as compared with the others (geometric means 2.02 versus 1.17).