The utility of p16ink4a expression in uterine cervical biopsies with/without dysplasia at Kenyatta National Hospital

Abstract

Background: Interpretation of intraepithelial neoplasia on hematoxylin& eosin (H&E)-stained cervical tissue sections is subject to substantial rates of discordance among pathologists, resulting in significant impact on patient management.Overexpression ofp16INK4a has been shown to be stronglyassociated with dysplastic and neoplastic epithelium of cervix uteri due to high-riskhuman papillomaviruses (HR-HPV).Therefore,p16INK4a Immunohistochemistry (IHC) provides valuable additional information in the interpretation of cervical histology with resultant improvementin definitive identification of dysplastic lesions and reduction of inter-observer disagreement in conventional histology. Objective: To determine the utility of p16INK4a expression in uterine cervical biopsies with or without dysplasia at Kenyatta National Hospital (KNH). Design: Laboratory-based, retrospective cross-sectional study. Setting: The University of Nairobi (UoN) Anatomical Pathology laboratories at KNH. Study population: Colposcopic biopsy specimens reported as negative for dysplasia, CIN1, CIN2 and CIN3 at KNH’s histology laboratory from June 2011 to June 2013. Main outcome measures: Ninety one previously diagnosed colposcopy biopsies were reevaluated on H&E and analyzed for p16INK4a expression by immunohistochemistry on paraffin wax embedded tissue blocks. Level of agreement in interpretation of cervical biopsies was compared between primary and review histologic results.Results of p16INK4a expression were correlated with age and histo-morphologic findings. Results:The age range was between 21 and 65 years (mean 40.2). On primary H&E-stained reports, 37(40.7%) were CIN2 while 23(25.3%) were CIN1. On review, 32(35.2%) were negative for dysplasia and 24(26.4%) were CIN3. Under half of the cases, 35 (38.5%) showed apositive p16INK4A expression.The diagnostic test on the previous and review histology exam results indicated a significant difference in the paired outcomes, p<0.001. There was a significant difference between the review diagnosis and the primary histology results when comparing high grade cases and disease negative cases (p < 0.001), with kappa of agreement of 0.568. On primary histology results, all (100%) negative for dysplasia and 82% (19/23) of CIN1 cases did not express P16. Almost half (48.6%) of CIN2were negative for p16INK4a. Majority [80% (12/15)] of CIN3 expressed p16INK4a, p < 0.001. On review histology result, all (100%) negative for dysplasia and almost all CIN1 (18/19) cases were nonimmunoreactive, p <0.001. There were 11(69%) CIN2 that stained positive for p16INK4a and all CIN3 expressed p16INK4a except one, 23(95.8%) vs. 1 (4.2%). Conclusion: This study provides strong evidence for the usefulness of p16INK4aimmunostaining in improving diagnostic accuracy of cervical biopsy interpretations, by mitigating the significant interobserver variation in interpretation of cervical biopsies on H&E stain. The highest discordance rates were mainly found in CIN1 and CIN2 categories(the “grey area”), while CIN3 showed lower rates of disagreement. P16INK4a IHC clearly differentiates negative for dysplasia from probable CIN1 on the one hand, and from high grade lesions on the other. This is important because only high grade lesions require further intervention and treatment while negative cases need follow up with screening. Recommendation: Routine p16INK4a Immunohistochemistry should be conducted for all “grey zone” cases, which include CIN1 and CIN2 in order to inform clinical decisions for surgical intervention or monitoring by screening.