Proposal of a New Scoring Scheme for the Diagnosis of Noninvasive Endocervical Glandular Lesions

The differential diagnosis of endocervical glandular lesions can be very difficult, and the interobserver agreement on borderline cases can be low. We are proposing a new scoring system to aid in the reproducibility of the diagnosis of noninvasive endocervical glandular lesions. The total of 67 diagnostically difficult cases were independently reviewed by five pathologists. After the completion of the first round review, a consensus diagnosis was reached for each lesion by all participants. This consensus diagnosis was used as the reference diagnosis. According to the consensus, the lesions included 21 benign/reactive conditions, 7 endocervical glandular dysplasias, and 39 adenocarcinomas in situ. During the second round review, all cases were assessed using the new scoring scheme, according to which separate scores from 0 to 3 were given to each lesion for: 1) nuclear atypia, 2) stratification, and 3) sum of mitoses/apoptoses (counted in the two most active glands, and the average number used). These three scores were then added to result in the total score (0-3 = benign; 4-5 = endocervical glandular dysplasia; 6-9 = adenocarcinoma in situ). Complete agreement between all observers in the first round review was seen in 35 of 67 cases (52.2%), kappa = 0.565. This agreement improved in the second round with the use of the scoring scheme: 52 of 67 cases (77.6%), kappa = 0.705. If the benign and endocervical glandular dysplasia diagnostic categories were combined, the overall agreement in the second round review would be 63 of 67 cases (94%), meaning that the scheme affords accurate distinction between adenocarcinoma in situ and lesser lesions. We propose applying this new scoring scheme to the diagnosis of noninvasive endocervical glandular lesions to improve interobserver agreement. The use of this scheme will result in more consistency of data in series from different institutions and will allow uniformity on the issue of adenocarcinoma in situ precursor lesions.