Relationships between muscle morphology and insulin sensitivity are improved after adjustment for intra-individual variability in 70-year-old men A. HEDMAN, 1 L. BERGLUND, 1 B. ESSE Â N-GUSTAVSSON, 2 R. RENELAND 1 and H . L I T H E L L 1 1 Department of Public Health and Caring Sciences, Section of Geriatrics, Uppsala University, Uppsala, Sweden 2 Department of Large Animal Clinical Sciences, Faculty of Veterinary Medicine, Swedish University of Agricultural Sciences, Uppsala, Sweden ABSTRACT The purpose of this investigation was to examine to what extent variability in the muscle morphology and insulin sensitivity influence the correlation between them. Reproducibility of muscle character- istics was estimated in duplicate biopsies from the same thigh of 23 subjects from a cohort of 70-year-old men. The coefficient of variation (CV) for different characteristics of muscle morphology was between 11 and 42% in duplicate biopsies. Coefficient of variation for markers of insulin sensitivity ranged between 12 and 39%. The variability reflected by intra-class correlation ranged from 0.23 to 0.60 for muscle morphology and from 0.68 to 0.96 for estimates of insulin sensitivity. The correlation analysis between muscle morphology and insulin resistance was performed in a sample of 515 men from the cohort, correlation coefficients were calculated with (r true ) and without (r) adjustment for intra-individual variation. Insulin sensitivity showed a positive relationship with percentage of type I ®bres (r true  0.33, r  0.21; P < 0.0001) and capillary density (r true  0.43, r  0.21; P < 0.0001) and negative correlations with percentage of type IIB ®bres (r true  ±0.35, r  ±0.24; P < 0.0001). Capillary density was inversely correlated to insulin. Thus, an obvious improvement of the correlation was seen after correcting intra-individual variation. In conclusion, owing to the low degree of reproducibility of muscle morphology variables and insulin sensitivity, implying a noticeable underestimation of correlations, the r-values should be adjusted for within-subject variation in order to demonstrate a more accurate estimate of the strength of the relationships studied. Keywords capillary density, coef®cient of variation, corrected correlation coef®cient, insulin sensitivity, intra-class correlation, muscle ®bre composition. Received 15 December 1998, accepted 28 January 2000 Skeletal muscle is a main site of insulin action and up to 85% of infused glucose during a euglycaemic, hyper- insulinaemic clamp test is disposed in the muscle tissue (DeFronzo et al. 1981). It has been demonstrated that in vivo insulin action is signi®cantly correlated to both the histologically determined proportions of muscle ®bre types and the density of capillary supply to skeletal muscle (Lithell et al. 1981, Krotkiewski et al. 1983, Saltin & Gollnick 1983, Lillioja et al. 1987). There is a relationship between the different types of muscle ®bres and capillarization on the one hand and fasting serum insulin, as well as a direct measurement of insulin sensitivity on the other (Lithell et al. 1981, Krotkiewski et al. 1983, Saltin & Gollnick 1983, Lillioja et al. 1987, Nyholm et al. 1997, Utriainen et al. 1998). The magni- tude of these correlations, however, has been quite vague. Nevertheless, in a large number of studies, a correlation of insulin action to different types of muscle ®bres or capillary supply has not been con®rmed (Isaksson et al. 1993, Hickey et al. 1995, Schalin-Ja Èntti et al. 1995, Kriketos et al. 1996, Zierath et al. 1996). Different results of that kind might be caused by large variation either in measurement of insulin action or in muscle morphology or in both and that the study groups were small. So far, less number of studies have reported the reproducibility in the measurements of Correspondence: Anu Hedman, Department of Public Health and Caring Sciences/Geriatrics, Uppsala University, PO Box 609, S-751 25 Uppsala, Sweden. Acta Physiol Scand 2000, 169, 125±132 Ó 2000 Scandinavian Physiological Society 125