The histologic estimation of coronary artery stenoses: Accuracy and the effect of lumen shape

The authors have recently investigated the histologic estimation of coronar y artery stenoses (CAS) to determine its reproducibility and the effect of training on reproducibility. The present study extends this work, examining the accuracy, the sensitivity, and the specificity of the estimation of GA S. Further, the effect of one histologic variable (i.e., arterial lumen sha pe) on the evaluation of CAS is examined. As described previously, 20 rando mly selected Movat-stained coronary artery cross-sections were reviewed thr ee times, at 3-month intervals, by six clinical pathologists (CPs), six pat hology residents (Res), seven anatomic pathologists (APs), and two cardiova scular pathologists (CVPs). Before the third iteration, training in CAS ass essment was provided. In the present study, for comparison with observer es timates, image analysis was used to establish the actual percent CAS and de termine observer accuracy. The results of this study showed, paradoxically, that greater experience did not correlate with greater accuracy: The CPs c onsistently had the highest accuracy scores and the CVPs consistently had t he lowest. Training, however, improved the accuracy scores of all groups. S tenotic arterial cross-sections with residual lumens showing concentric or eccentric polymorphous shapes were consistently underestimated compared to image analysis, while lumens with a eccentric slitlike shape were consisten tly overestimated. (C) 2001 Elsevier Science Inc. All rights reserved.