In lung cancer patients, the only available diagnostic material often comes from biopsy or from cytological samples obtained by fine needle aspiration (FNA). There is a lack of easily detectable cytomorphological features for rapid on-site evaluation (ROSE) to orient lung cancer diagnosis towards a specific tumor histotype. We studied the cytological features evaluated on site to define tumor histotype and to establish the number of specimens to be taken. Cytological specimens from 273 consecutive patients were analyzed with ROSE: bronchoscopy with transbronchial needle aspiration (TBNA) had been performed in 72 patients and with endobronchial ultrasound (EBUS)-TBNA in 201. Cytomorphological features were correlated with the final diagnosis and diagnostic accuracy was measured. Analysis of the different cytomorphological parameters showed that the best sensitivity and specificity were obtained for adenocarcinoma by combining the presence of nucleoli and small/medium cell clusters, and for squamous cell carcinoma by considering the presence of necrosis ≥50% and large cell clusters. For small cell carcinoma, the best diagnostic accuracy was obtained by combining moderate necrosis (<50%) and the presence of single cells. Overall accuracy ranged from 90% to 97%. We showed that it was possible to establish the histotype of the most frequent lung cancers during ROSE using only a few easily identifiable cytomorphological parameters. An accurate diagnosis during ROSE could help endoscopists to decide how many tumor samples must be taken, e.g.a higher number of samples is needed for the biomolecular characterization of adenocarcinoma, whereas one sample may be sufficient for squamous cell carcinoma.
Keywords: ROSE; cytomorphological features; histotype; lung cancer.