New WHO Reporting System for Lung Cytopathology: Reproducibility Test of the Diagnosis and Usefulness of an Online...

Since no universal cytological classification system for lung cancer has been established, the Japanese Lung Cancer Society and the Japanese Society of Clinical Cytology (JSCC) jointly established and reported four cytological categories: negative for malignancy, atypical cells, suspicious for malignancy, and malignancy. In 2022, the WHO Reporting System for Lung Cytopathology was published. This system presented five cytological classifications, including the four cytological category classifications above and insufficient/inadequate/nondiagnostic. The creation of a classification alone is not practical in actual clinical practice. Thus, we evaluated the reproducibility of the classification through tutorials and identified the issues and problems involved in the wide dissemination of this classification.

Forty-two cases were selected from those used in previously published articles, and diagnosis and tutorial systems were created. The first diagnostic round and tutorial and the second diagnostic round were conducted on the web. Participants were recruited via the JSCC website and emails. Images (×100 and ×400) of the lesions to be diagnosed were categorizing by 4 cytological categories (benign, atypical, suspicious for malignancy, malignant), 7 suggestive pathological diagnoses, and 4 cytological features. The mean correct or incorrect answer rates for the 42 cases and the mean correct response rates for 105 participants were compared between the first and second rounds using McNemar's test and t tests to identify cases with diagnostic difficulties and high tutorial effects.

Comparing the correct response to cytological categories, the results showed that 17 of 42 cases improved significantly. The mean number of correct answers for the four cytological categories increased significantly from 16.0 (38.1%) in the first round to 20.3 (48.3%) in the second round (p < 0.001). For the seven suggestive pathological diagnoses, the mean number of correct answers increased significantly from 20.3 (48.3%) in the first round to 25.1 (59.8%) in the second round (p < 0.001). The mean number of correct responses increased significantly from 40.2 (38%) in the first round to 51.5 (49%) in the second round (p = 0.0147). Four cases were difficult to match even after the tutorial and three cases were highly affected by the tutorial. The most important basis for diagnoses was nuclear findings in the first and second rounds.

Comprehensive tutorials on diagnostic criteria are needed to effectively implement this system globally. In particular, devising ways to appropriately diagnose cancers with mild atypia or without characteristic morphology is important.