Interobserver variability and accuracy of p16/Ki-67 dual immunocytochemical staining on conventional cervical smears

Background p16/Ki-67 dual immunocytochemical staining (DS) has been proven as a sensitive and specific test for triage of HPV positive women with good reproducibility and accuracy. However, implementation of the test into an organized screening program (OSP) is not easy. The aims of this study were to compare the performance and agreement of DS results among three Slovenian cytopathological laboratories involved in the national OSP, and to define cases where staining results can be difficult to interpret. Methods Cervical smears were obtained for DS from 129 women referred to colposcopy. Smears were evaluated blindly in three laboratories by a cytotechnologist and a cytopathologist after initial training. Results were positive, suspicious, negative or inadequate. Five characteristics of DS staining were recorded. After primary evaluation, an extensive expert-led additional training was undertaken, including a discussion of difficult cases and a practical exam. Smears were re-evaluated and results compared to primary evaluation. Results After the additional training, the overall percentage of agreement among the three laboratories increased from 77.5 to 89.9% and kappa increased from 0.70 to 0.86. Sensitivity for CIN2+ increased in two laboratories, to 90.5 and 85.7%, without the loss of specificity (75.8%). In one laboratory, the sensitivity slightly decreased from 90.5 to 88.9%, but the specificity increased from 63.6 to 68.2%. Difficult cases had significantly less DS cells, weak intensity of p16 staining, suboptimal cell morphology and background staining compared to positive cases. Conclusion Additional expert-led training and discussion of difficult cases are necessary for accurate interpretation of DS in laboratories involved in OSP. The most difficult cases were those with single stained cells and weak p16 staining. Training protocol for safe implementation of p16/Ki-67 DS in OSP is proposed.