Evaluation of a combined triple method to detect causative HPV in oral and oropharyngeal squamous cell carcinomas: p16 Immunohistochemistry, Consensus PCR HPV-DNA, and In Situ Hybridization

Abstract

Background: Recent emerging evidences identify Human Papillomavirus (HPV) related Head and Neck squamous cell carcinomas (HN-SCCs) as a separate subgroup among Head and Neck Cancers with different epidemiology, histopathological characteristics, therapeutic response to chemo-radiation treatment and clinical outcome. However, there is not a worldwide consensus on the methods to be used in clinical practice. The endpoint of this study was to demonstrate the reliability of a triple method which combines evaluation of: 1. p16 protein expression by immunohistochemistry (p16-IHC); 2. HPV-DNA genotyping by consensus HPV-DNA PCR methods (Consensus PCR); and 3 viral integration into the host by in situ hybridization method (ISH). This triple method has been applied to HN-SCC originated from oral cavity (OSCC) and oropharynx (OPSCC), the two anatomical sites in which high risk (HR) HPVs have been clearly implicated as etiologic factors. Methylation-Specific PCR (MSP) was performed to study inactivation of p16-CDKN2a locus by epigenetic events. Reliability of multiple methods was measured by Kappa statistics.

Results: All the HN-SCCs confirmed HPV positive by PCR and/or ISH were also p16 positive by IHC, with the latter showing a very high level of sensitivity as single test (100% in both OSCC and OPSCC) but lower specificity level (74% in OSCC and 93% in OPSCC).Concordance analysis between ISH and Consensus PCR showed a faint agreement in OPSCC (κ = 0.38) and a moderate agreement in OSCC (κ = 0.44). Furthermore, the addition of double positive score (ISHpositive and Consensus PCR positive) increased significantly the specificity of HR-HPV detection on formalin-fixed paraffin embedded (FFPE) samples (100% in OSCC and 78.5% in OPSCC), but reduced the sensitivity (33% in OSCC and 60% in OPSCC). The significant reduction of sensitivity by the double method was compensated by a very high sensitivity of p16-IHC detection in the triple approach.

Conclusions: Although HR-HPVs detection is of utmost importance in clinical settings for the Head and Neck Cancer patients, there is no consensus on which to consider the 'golden standard' among the numerous detection methods available either as single test or combinations. Until recently, quantitative E6 RNA PCR has been considered the 'golden standard' since it was demonstrated to have very high accuracy level and very high statistical significance associated with prognostic parameters. In contrast, quantitative E6 DNA PCR has proven to have very high level of accuracy but lesser prognostic association with clinical outcome than the HPV E6 oncoprotein RNA PCR. However, although it is theoretically possible to perform quantitative PCR detection methods also on FFPE samples, they reach the maximum of accuracy on fresh frozen tissue. Furthermore, worldwide diagnostic laboratories have not all the same ability to analyze simultaneously both FFPE and fresh tissues with these quantitative molecular detection methods. Therefore, in the current clinical practice a p16-IHC test is considered as sufficient for HPV diagnostic in accordance with the recently published Head and Neck Cancer international guidelines. Although p16-IHC may serve as a good prognostic indicator, our study clearly demonstrated that it is not satisfactory when used exclusively as the only HPV detecting method. Adding ISH, although known as less sensitive than PCR-based detection methods, has the advantage to preserve the morphological context of HPV-DNA signals in FFPE samples and, thus increase the overall specificity of p16/Consensus PCR combination tests.

Figure 1

IHC expression of p16 protein in representative HPV positive (DNA type 16) and HPV negative OSCCs. P16 is expressed only at basal-parabasal levels in epithelium surrounding OSCC, whereas it is over-expressed in OSCC. The figure in the middle page depicted p16 expression according different HR and LR-HPVs. The figure in the bottom represent a p16 negative OSCC; the p16 under-expression is due to promoter methylation od CDKN2a. Legend. K, cancer samples; M, CDKN2a methylated; U, CDKN2a unmethylated.

Figure 2

IHC expression of p16 protein in OPSCC showed at increasing magnification. Note intense nucleo-cytoplasmic expression (LSAB-HRP, nuclear counterstaining with haematoxylin).

Figure 3

The role of CDKN2a promoter methylation in p16 down-regulation. A) Percentage of CDKN2a promoter methylation in OSCC and in oral epithelia exposed to alcohol and tobacco risk factors. B) Representative examples of Methylation-Specific PCR (MSP) for CDKN2a/INK4a locus (p16) in OSCC. Legend. K, cancer samples; N, normal samples; M, CDKN2a methylated; U, CDKN2a unmethylated.

Figure 4

HR HPV ISH in Head and Neck squamous cell carcinomas. Note the clear nuclear staining with tetrazole blue as punctuate signals and clusters, corresponding to integration and episomic status respectively A, B) Control cases of cervical HSIL. C) Integration and episomal clusters are uniformly distributed in tonsil cancer (OPSCC) harboring type 16 HPV-DNA; note integrative punctuate signals and cluster spot throughout the entire tumor. D) Heterogeneous integration signals in oral cancer harboring type 16 HPV-DNA. (In situ hybridization; tetrazole blue signals stain the viral DNA; nuclear counterstaining with fast red; original magnifications a, b, ×10; c, ×40; d ×60).

Figure 5

ISH techniques for HPV show the morphological context of viral DNA location. Representative case of HPV-DNA positive tumor without virus integration in cancer cells. In spite of positivity for two HPV viruses (HR-HPV 31 and LR-HPV 44 viruses) this case of OSCC showed no integration but cytoplasmic localization in cancer cells as evaluated by two commercially available in situ hybridization techniques; note the cytoplasmic virus location in autolytic cancer cells (All the pictures in this panel were selected from different fields of a single case. A, x40), B, x40), DAKO ISH, brown DAB signals stain the viral DNA; C, x60), D, x60), Ventana ISH, tetrazole blue signals stain the viral DNA).