Background: Fine needle aspiration (FNA) cytology is under consideration as an auxiliary preoperative diagnostic technique in the diagnosis of renal masses. However, reports for FNA are contradictory with regard to diagnostic accuracy and applicability.
Objective: To evaluate the diagnostic accuracy and reproducibility of FNA from renal masses.
Design: FNAs performed in-bench (hematoxylin and eosin [H&E] stains) from 66 consecutive renal tumours (58 malignant and 8 benign tumours) were presented twice with a 6-mo interval to five pathologists with little experience in renal cytology. Pathologists were blinded for the results of the first round as well for the surgical specimen. The FNAs were stained for Papanicolaou and Giemsa.
Measurements: Diagnostic accuracy, concordance between smears and surgical specimens, and the generalized kappa for interobserver/intraobserver agreement were calculated.
Results and limitations: The number of nondiagnostic and nonconclusive cases ranged from 5-14% in the first and 3-8% in the second round. Overall accuracy varied between 73-89% and 71-91% for the first and second round, respectively. Sensitivity (72-97%) and positive predictive value (PPV) (93-100%) to classify a malignant tumour in both rounds was high. Sensitivity (25-100%) and PPV (28-100%) to classify a benign tumour was lower with a wide confidence interval. Overall concordance in subtyping ranged from 39-70% in the first, and from 52-74% in the second round. Interobserver agreement ranged from fair (kappa=0.039) to substantial (kappa=0.540) for the different subtypes. The intraobserver agreement (mean kappa=0.357, CI 95%=0.304-0.411) was moderate for all pathologists. The low number of benign tumours in this study precludes sound statements regarding the diagnostic accuracy of FNA to classify benignity.
Conclusion: Despite the lack of experience in renal cytology, all pathologists showed a high diagnostic yield and good overall accuracy in distinguishing between malignant and benign tumours. Concordance in subtyping varied widely among pathologists and was reliable only for clear cell renal cell carcinoma (ccRCC). These results suggest that FNA may have a potential role in the diagnosis of renal tumours although its value in subtyping was limited in our setting.