Diagnostic utility and pitfalls of intraoperative pulmonary imprint cytology based on final pathological diagnoses

Objective This study aimed to determine the reliability of imprint cytology (IC) for intraoperative diagnosis of pulmonary lesions. Methods We reviewed 113 cases of pulmonary lesion resection for which a scratch imprint was made intraoperatively. We divided the specimens into two groups (benign and malignant) and compared the scratch IC‐based diagnoses against the final histopathological diagnoses in each group for concordance. We also analysed those cases in which the scratch IC preparation was classified as inadequate. Results The sensitivity, specificity, positive and negative predictive values, and accuracy of IC diagnoses among the patient cohort were 87.7% (72/82), 100% (7/7), 100% (72/72), 41.2% (7/17) and 88.8% (79/89), respectively. IC yielded some false‐negative results in terms of malignancy, although most of these imprints were of early cancer or cancer with mild cytological atypia. Five (41.6%) of 12 lesions for which the imprint was deemed inadequate were diagnosed histologically as granulomas with caseous necrosis. Conclusion IC‐based diagnoses of pulmonary lesions as malignant corresponded well with the final histopathological diagnoses, but IC‐based diagnoses of negative (ie, without malignant cells) were not as reliable. Thus, pathologists should recognise the limitations of IC, especially for identifying malignant lesions. Also, the possibility of latent bacterial infection in a granuloma with caseous necrosis indicates that an IC preparation deemed inadequate for diagnosis should not be ignored. Imprint cytology correlated well with the final histopathological assessment for malignant diagnoses, however, it missed infections such as TB when caseous necrosis was not recognised and the specimen was reported as negative or non‐diagnostic. This has implications for observing health and safety regulations for frozen sections even on non‐diagnostic imprints.