Squamous dysplasia of the uterine cervix: tissue sampling-related diagnostic considerations in 600 consecutive biopsies.

Despite the technological advances in colposcopic techniques, there continues to be a 10% to 20% discordance rate between the colposcopic findings and the histological diagnoses on the resultant biopsies. One of the many factors to which this may be theoretically attributable is related to sampling error from the paraffin-embedded tissue block. In this study, we evaluated the clinical efficacy of routinely obtaining 6 sections in cervical biopsies, using the frequency with which dysplastic lesions would be missed with various levels of sectioning as the sole benchmark determinant of clinical efficacy. Our database was searched for all cervical biopsies in which a diagnosis of squamous dysplastic lesion was made for the period February 1, 2006 to April 28, 2006. All cases were processed in 6-level sectioning, which entails cutting and staining (hematoxylin-eosin) 6 consecutive sections from the paraffin block without preserving or discarding any intervening unstained sections. The first level at which a diagnosis of dysplastic lesion could be unequivocally made by a gynecologic pathologist was determined. Six hundred consecutive biopsies from 404 patients were reviewed. For the whole cohort, the average level at which a dysplastic lesion was unequivocally diagnosable was 1.9 (median, 1). Three hundred fifty-seven (59.5%), 97 (16.2%), 41 (6.8%), 55 (9.2%), 34 (5.7%), and 16 (2.6%) of the 600 lesions were diagnosable at levels 1, 2, 3, 4, 5, and 6, respectively. The cervical intraepithelial neoplasia (CIN) 2 and 3 (n = 89) was, on average, diagnosable at an earlier level (1.35) compared with CIN 1 (level, 2.025; P < 0.001 [n = 511]). Indeed, 79.8% of the CIN 2-3 cases were diagnosable at level 1, as compared with 56% of the CIN 1 cases (P < 0.001); 87, 38, 52, 32, and 16 cases of CIN 1 and 10, 3, 3, 2, and 0 cases of CIN 2-3 were diagnosable at levels 2, 3, 4, 5, and 6, respectively. Therefore, if sectioning were limited to 3 levels, 17.5% (105/600) of all dysplastic lesions would have been missed, including 19.6% (100/511) of CIN 1 and 5.6% (5/89) of CIN 2-3. Because not more than 3 levels are routinely evaluated in most laboratories, our findings suggest that sampling error is indeed at least 1 significant factor contributing to colposcopic/histological discrepancies. Using our clinical efficacy standard, when no pathologic findings are initially identified in a colposcopic-directed biopsy, at least 5 levels (a priori or in recuts) are required to ensure a 100% diagnostic accuracy for CIN 2-3.