Insulin sensitivity in women: a comparison among values derived from intravenous glucose tolerance tests with different sampling frequency, oral glucose tolerance test or fasting.

OBJECTIVE: To determine the correlation between insulin sensitivity (S(I)) obtained by the minimal model method applied to a frequently sampled (n=33) intravenous glucose tolerance test (FSIGT(33)), and values obtained by reduced FSIGTs, oral glucose tolerance test (OGTT), or fasting.
DESIGN: Retrospective analysis on tests performed in prospective studies.
METHODS: A total of 78 FSIGT(33), and 59 OGTT were performed in non-diabetic women of which 10 were young cyclic females in the early follicular menstrual phase, 10 were young non-obese subjects with polycystic ovary syndrome (PCOS) and 30 were in post-menopause. Some of these individuals were investigated both prior to and during specified treatments. FSIGT(33) was transformed into FSIGT(22) and FSIGT(12) by removing samples from the analysis. Values of SI derived from reduced FSIGTs or calculations performed on glucose and insulin values observed in fasting conditions and/or during OGTT were related to those of FSIGT(33).
RESULTS: S(I) values derived from FSIGT(33) were highly correlated with those derived from FSIGT(22) (r=0.965) or FSIGT(12) (r=0.955), but were only weakly correlated with those derived from fasting or OGTT calculations (r below 0.5). Between-group (PCOS vs normal) or within-group (prior to and during treatment) comparisons showed that reduced FSIGTs were only slightly less powerful than FSIGT(33) in detecting differences in S(I).
CONCLUSIONS: In non-diabetic women, reduced FSIGTs but not calculations based on fasting or OGTT values may be used in place of FSIGT(33) to document S(I) and its variation.