This study assessed the impact of diabetes mellitus on atherosclerotic lesion formation. Seventy insulin-treated diabetics, 150 non-insulin-treated diabetics, and 607 nondiabetics with chronic anginal syndromes and de novo native coronary stenoses were studied using (1) angiography, and (2) intravascular ultrasound (reference and lesion arterial, lumen, and plaque areas; area stenosis [reference-lesion/reference lumen area]; remodeling index [reference-lesion lumen area/lesion-reference plaque area]; and slope of the regression line relating lumen area to plaque burden [plaque/arterial area]). Despite being diabetic for longer and having similar lumen compromise, insulin-treated patients had (1) less reference plaque (8.3 +/- 3.4 vs 10.5 +/- 4.5 mm2, p = 0.0015), (2) less stenosis plaque (13.0 +/- 4.9 vs 16.9 mm2, p <0.0001), (3) smaller reference arterial areas (17.1 +/- 5.4 vs 19.7 +/- 6.2 mm2, p = 0.0063), and (4) smaller stenosis arterial areas (15.3 +/- 4.9 vs 19.5 +/- 6.5 mm2, p <0.0001) than non-insulin-treated diabetics. With use of multivariate linear regression analysis, insulin use was an independent (and negative) predictor of reference plaque and arterial areas (p = 0.0308 and p = 0.0179) and stenosis plaque and arterial areas (p = 0.0117 and p = 0.0066). This was also true when normalized for body surface area. The remodeling index showed that insulin treatment resulted in an exaggerated impact of plaque accumulation on lumen compromise. This was confirmed by the slope of the regression line relating lumen area to plaque burden. Patients with a longer duration of diabetes who were treated with insulin for > or = 1 year had (paradoxically) less reference segment and stenosis plaque accumulation. Possible explanations include impaired adaptive remodeling and/or arterial (and plaque) shrinkage.