P10-01COMPARATIVE EVALUATION OF THREE HPV GENOTYPING ASSAYS ON FFPE SAMPLES OF HEAD AND NECK CANCER

22. HPV and oropharynx / Head and neck cancer
J. Mossong 1, M. Duvivier 1, J. Tapp 1, D. Stieber 1, F. Schmitt 1.
1Laboratoire National de Santé (Luxembourg)

Background / Objectives

HPV is a known risk factor for certain subgroups of HNSCC (head and neck squamous cell carcinoma ) affecting disease prognosis and survival. HPV testing of FFPE (formalin fixed paraffin embedded) samples is becoming a routine in pathological practice, however there is no consensus on which methods to use for identifying HPV. In this study we compared 3 different methods to detect and genotype HPV DNA in archival FFPE samples of HNSCC.


Methods

Samples from 67 patients (mean age 62 years, 51 men, 16 women) diagnosed with HNSCC in 2014 were obtained from the service of pathology of the Laboratoire national de Santé in Luxembourg. DNA was extracted from FFPE cuts using Qiagen kits. HPV testing was conducted on all samples using the Cobas HPV assay (Roche, Switzerland). All positive and a subset of negative samples was also tested by EuroArray HPV (EuroImmune, Germany) and Anyplex II HPV28 (Seegene, Korea) assays according to manufacturers’ instruction. For the Cobas assay, cycle threshold (CT) values indicative of viral concentrations were obtained using the open LC480 software. One sample with non-amplifiable positive control DNA in the Cobas assay was disregarded for further analysis.


Results

Based on the Cobas assay, 24 (36%) and one (1.5%)  of the 66 samples were positive for HPV 16 or other high risk HPV (hrHPV), respectively. All 25 HPV positive samples and a randomly chosen subset of 19 negative samples were retested by Anyplex and EuroArray.  Agreement (cf. Table) between Cobas and EuroArray was higher (kappa 0.78) than between Cobas and and Anyplex (kappa 0.5). Anyplex detected 13 hrHPV and EuroArray 20 hrHPV samples. Both EuroArray and Anyplex detected genotype 66 for the other hrHPV sample detected by cobas. Samples positive by Anyplex and Euroarray assays had significantly lower  CT values in the Cobas assay than negative samples (Anyplex: 30.5 vs 39.9, p<0.0001; EuroArray: 33.6 vs 41.1, p=0.01).

Table: Concordance between EuroArray and Anyplex assays compared to Cobas assay

 
hrHPV genotype +/+ -/- +/- -/+ Agreement Kappa
Euroarray 20 19 0 5 88.6% 0.78 (p<0.0001)
Anyplex 13 19 0 12 72.3% 0.5 (p<0.0001)

 


Conclusion

Depending on the assay, 20-38% of HNSCC samples were found to be positive for hrHPV. Almost all (96%) of the HPV-positive HNSCC cancers  were associated with HPV16. All three methods yielded concordant detection and typing results when viral loads were relatively high. Agreement between methods appears to be related to viral concentration. Although in our study the Cobas assay appears to be the most sensitive method for detecting hrHPV in FFPE samples, further work is needed to establish clinically relevant CT cut-offs.


References