Immunocytochemical detection by a fully automated immunostainer for assisting subclassification of pulmonary tumors on ThinPrep slides in real-life clinical samples: A single-institution experience.

Introduction: This study aims to investigate the feasibility and reliability of ThinPrep slides in detecting the subclassification of lung cancer and develop a process for immunocytochemistry (ICC) with optimized staining steps of an automated immunostainer.

Methods: Cytomorphology and ancillary ICC by automated immunostainer on ThinPrep slides were performed to subclassify 271 cytology cases of pulmonary tumor, which were stained with 2 or more of the following antibodies: p40, p63, thyroid transcription factor-1 (TTF-1), Napsin A, synaptophysin (Syn), and CD56.

Results: The accuracy of cytological subtyping was improved from 67.2% to 92.7% (p < .0001) after ICC. The accuracy of cytomorphology combined with ICC results for lung squamous-cell carcinoma (LUSC), lung adenocarcinomas (LUAD), and small cell carcinoma (SCLC) was 89.5% (51 of 57), 97.8% (90 of 92), and 98.8% (85 of 86), respectively. The sensitivity and specificity of 6 antibodies were as follows: p63 (91.2%, 90.4%) and p40 (84.2%, 95.1%) for LUSC, TTF-1(95.6%, 64.6%) and Napsin A (89.7%, 96.7%) for LUAD and Syn (90.7%, 60.0%) and CD56 (97.7%, 50.0%) for SCLC, respectively. P40 expression on ThinPrep slides had the highest agreement (κ = 0.881) with immunohistochemistry (IHC) results, followed by p63 (κ = 0.873), Napsin A (κ = 0.795), TTF-1 (κ = 0.713), CD56 (κ = 0.576), and Syn (κ = 0.491).

Conclusion: The result of ancillary ICC on ThinPrep slides by fully automated immunostainer was in good agreement with the gold standard in pulmonary tumors subtype and immunoreactivity, objectively achieving accurate subtyping in cytology.