New WHO System Boosts Ability to Classify Conjunctival Lesions
By Lynda Seminara
Selected By: Richard K. Parrish II, MD
Journal Highlights
American Journal of Ophthalmology, March 2021
Download PDF
Accurate diagnosis of conjunctival melanocytic intraepithelial lesions (CMILs) is important both prognostically and therapeutically. Milman et al. looked at interobserver agreement and accuracy of the new World Health Organization (WHO) classification system for CMILs and compared the findings with two commonly used classification schemes. They found that accuracy was comparable for the three systems. Interobserver agreement for distinguishing high- and low-grade lesions was greatest for the WHO system.
For this study, the authors reviewed pathology and other records for patients who underwent a primary biopsy procedure for conjunctival primary acquired melanosis (PAM) at Wills Eye Hospital from 1974 to 2002 and had follow-up for at least three years. Collected data included demographics and clinical findings such as disease course. The authors created virtual digital histopathology slides from actual slides to ensure uniformity of appearance for the 12 ophthalmic pathologists who assessed them. Three classifications systems were applied for each slide: WHO (fourth edition), PAM, and C-MIN (conjunctival melanocytic intraepithelial neoplasia).
Overall, 64 patients (83 primary excision procedures) had tissue that was adequate for histopathologic evaluation. Interobserver agreement for differentiating low- and high-grade lesions was 81% for WHO, 76% for PAM, and 67% for C-MIN. With all three systems, low-grade lesions were the most difficult type to interpret. Average accuracy for identifying lesions with recurrence potential was 83% for WHO and C-MIN and 81% for PAM. Assessment by C-MIN took slightly longer than the other systems.
Although the new WHO scheme is viable in the context of pathology, the authors emphasized the clinical importance of identifying specific types of conjunctival epithelial acquired pigmentation, as outlined in the PAM classification system. With additional refinements in digital pathology, artificial intelligence, and molecular genetics, the authors stated that they are confident that “an integrated morphologic–molecular classification system will emerge that will further improve our ability to accurately diagnose these challenging lesions.”
The original article can be found here.