Concordance of p16INK4a and E6*I mRNA among HPV-DNA positive oropharyngeal, laryngeal and oral cavity carcinomas from the ICO international study

Background: Tests or test algorithms for diagnosing HPV-driven oral cavity and laryngeal head and neck carcinomas (HNC) have not been yet validated, and the differences among oral cavity and laryngeal sites have not been comprehensively evaluated. We aimed to assess the utility of a diagnostic algorithm for the detection of HPV-driven oral cavity (OCC), oropharyngeal (OPC) and laryngeal (LC) carcinomas using HPV-DNA testing followed by p16INK4a immunohistochemistry, taking E6*I mRNA detection as the reference standard. Methods: Formalin-fixed paraffin-embedded OCC, OPC, and LC carcinomas were collected from pathology archives in 29 countries. All samples were subjected to histopathological evaluation, DNA quality control, and HPV-DNA detection. All HPV-DNA-positive samples (including 78 OCC, 257 OPC, and 51 LC out of 3680 HNC with valid HPV-DNA results) were also tested for p16INK4a immunohistochemistry and E6*I mRNA. Three different cutoffs of nuclear and cytoplasmic staining were evaluated for p16INK4a: (a) >25%, (b) >50%, and (c) ≥70%. The concordance of p16INK4a and E6*I mRNA among HPV-DNA-positive OCC, OPC, and LC cases was assessed. Results: A total of 78 OCC, 257 OPC, and 51 LC were HPV-DNA-positive and further tested for p16INK4a and E6*I mRNA. The percentage of concordance between p16INK4a (cutoff ≥ 70%) and E6*I mRNA among HPV-DNA-positive OCC, OPC, and LC cases was 79.5% (95% CI 69.9–89.1%), 82.1% (95% CI 77.2–87.0%), and 56.9% (95% CI 42.3–71.4%), respectively. A p16INK4a cutoff of >50% improved the concordance although the improvement was not statistically significant. For most anatomical locations and p16INK4a cutoffs, the percentage of discordant cases was higher for HPV16- than HPV-non16-positive cases. Conclusions: The diagnostic algorithm of HPV-DNA testing followed by p16INK4a immunohistochemistry might be helpful in the diagnosis of HPV-driven OCC and OPC, but not LC. A different p16INK4a expression pattern was observed in those cases HPV-DNA-positive for types other than HPV16, as compared to HPV16-positive cases. Our study provides new insights into the use HPV-DNA, p16INK4a, and HPV-E6*I mRNA for diagnosing an HPV-driven HNC, including the optimal HPV test or p16INK4a cutoffs to be used. More studies are warranted to clarify the role of p16INK4a and HPV status in both OPC and non-OPC HNC