Double positivity for HPV-DNA/p16 is the biomarker with strongest diagnostic accuracy and prognostic value for human papillomavirus related oropharyngeal cancer patients

Background: The etiologic role of human papillomaviruses (HPV) in oropharyngeal cancer (OPC) is well established. Nevertheless, information on survival differences by anatomic sub-site or treatment remains scarce, and it is still unclear the HPV-relatedness definition with best diagnostic accuracy and prognostic value. Methods: We conducted a retrospective cohort study of all patients diagnosed with a primary OPC in four Catalonian hospitals from 1990 to 2013. Formalin-fixed, paraffin-embedded cancer tissues were subjected to histopathological evaluation, DNA quality control, HPV-DNA detection, and p16/pRb/p53/Cyclin-D1 immunohistochemistry. HPV-DNA positive and a random sample of HPV-DNA negative cases were subjected to HPV-E6*I mRNA detection. Demographic, tobacco/alcohol use, clinical and follow-up data were collected. Multivariate models were used to evaluate factors associated with HPV positivity as defined by four different HPV-relatedness definitions. Proportional-hazards models were used to compare the risk of death and recurrence among HPV-related and non-related OPC. Results: 788 patients yielded a valid HPV-DNA result. The percentage of positive cases was 10.9%, 10.2%, 8.5% https://doi.org/10.1016/j.oraloncology.2018.01.010 Received 19 October 2017; Received in revised form 15 January 2018; Accepted 17 January 2018 ⁎ Corresponding authors at: Cancer Epidemiology Research Program, Catalan Institute of Oncology (ICO), IDIBELL, Av. Gran Via de L'Hospitalet 199-203, 08908 L'Hospitalet de Llobregat, Barcelona, Spain. 1 The authors contributed equally to this work. E-mail addresses: mmena@iconcologia.net (M. Mena), lalemany@iconcologia.net (L. Alemany). Oral Oncology 78 (2018) 137–144 Available online 20 February 2018 1368-8375/ © 2018 Elsevier Ltd. All rights reserved. T and 7.4% for p16, HPV-DNA, HPV-DNA/HPV-E6*ImRNA, and HPV-DNA/p16, respectively. Being nonsmoker or non-drinker was consistently associated across HPV-relatedness definitions with HPV positivity. A suggestion of survival differences between anatomic sub-sites and treatments was observed. Double positivity for HPV-DNA/p16 showed strongest diagnostic accuracy and prognostic value. Conclusions: Double positivity for HPV-DNA/p16, a test that can be easily implemented in the clinical practice, has optimal diagnostic accuracy and prognostic value. Our results have strong clinical implications for patients’ classification and handling and also suggest that not all the HPV-related OPC behave similarly.


RESEARCH HIGHLIGHTS
 Six biomarkers of HPV-relatedness were assessed in788 oropharyngeal cancers  A low HPV attributable fraction in oropharyngeal cancer was observed  Double positivity for HPV-DNA/p16 INK4a showed strongest prognostic value INTRODUCTION About a decade ago the International Agency for Research on Cancer (IARC) established high-risk Human papillomavirus 16(HPV16)as a cause of oropharyngeal carcinoma (OPC) [1]. Since then,increasing amount of information on the role of HPVs in OPC has been generated. The IARC estimates that approximately 29,000 new HPVrelated OPC cases occur every year, corresponding to 31% of the worldwide number of the overall incident OPCcases [2]. These estimates, as well as previous meta-analyses assessing the quantitative contribution of HPV, found high geographic heterogeneity in HPV-attributable fractions (AFs) of OPC, ranging from less than 20% in some world regions, 24% in Southern Europe to more than 60% in North America [3,4]. This low HPV-AF for OPC in Southern Europe has been recently confirmed in tworecent studies conducted by our group [5,6].
HPV-related OPC differs at clinical, epidemiological and molecular level to OPC caused by classic risk factors (i.e. tobacco and alcohol) [7]. The consistent observation of improved survival and better response to treatment of HPV-related OPC has stirred up the state-of-the-art of their management. Indeed,several clinical trialsof deescalation treatmentsare under evaluation, aimingto achievebetter results with less treatment-associated comorbidities [8].However, the biological rationaleunderlyingthese strategies remains poorly understood,and most of schemes are extrapolated from HPV-negative OPC trials. Importantly, around 20% ofHPV-related patients still fail to treatment despite its good prognosis [7].
Diagnosis algorithms for HPV-related OPC are still under development. HPV-DNA detection alone is not sufficient to classify an OPC as HPV-driven since the presence of HPV-DNA could reflect a transient or non-related infection rather than a genuine HPV-driven oncogenic process [9][10][11]. Additionally, the detection of high cellular p16 INK4a expressionby immunohistochemistry (IHC)is the most widely implemented technique in the clinical setting, but is not specific for HPV activity in these tumours [12,13]. Indeed, it has been demonstrated that patients with p16 INK4a high expressionbut HPV-DNA-negative OPC show a significantly less favourable survival than patients with p16 INK4a high expression and HPV-DNA-positive tumours [14,15], indicating that p16 INK4a high expression alone may not accurately classify HPV-related OPC patients.The combination of HPV-DNA detection and p16 INK4a IHC is starting to be recommended to diagnose HPV-related OPCs [15]. Nevertheless, there is still limited information about the accuracy and prognostic value of this combination of biomarkers.
It is imperative to identify the best HPV-relatedness definition for HPV causality and prognosis in OPC. This is a prerequisite to provide a sound approach to study differences in survival of HPV-related OPCby factors such as anatomical subsite [16,17]and bytreatment [18].
In an attempt to elucidate these gaps, weconducted astudy in OPC to assess the association of different HPV-relatedness definitions with patients' overall survival (OS)and progression-free survival (PFS), stratified by anatomical sub-site or treatment.

Study design and population
We designeda retrospective cohort study of all patients diagnosed with a primary OPC in four hospitals of Catalonia from 1990 to 2013 (Catalan Institute of Oncology-ICO-Hospital Universitari de Bellvitge, Hospital de Sant Pau, Hospital del Mar and Hospital ParcTaulí).Protocols were approved by the ethics committee of each participating hospitals.
Cancer cases were identified from medical records/pathology reports of the centres of origin. We included cases that fulfilled the following criteria: to be diagnosed with primary invasive cancer of the oropharynx (any histology; codes from the International From all eligible cases, we reviewed medical records of the patients and accessed information on demographics, smoking and alcohol consumption, clinical and follow-up data; andformalin-fixed paraffin embedded (FFPE) tumour samples from the diagnosis previous to treatment when available.
In order to assess potential carryover HPV contamination at the local level, we additionally includeda set of control samples selected by local investigators (5% of the number of cases evaluated, corresponding to tissue samples of patients with diagnoses non-related with HPVprocessed in the same laboratory).

FFPE Blocks Processing and Histopathological Evaluation
All specimens processing was centralized at ICO. FFPE blocks were re-embedded whenever necessary. First and last sections were used for histopathological evaluationafter hematoxylin and eosin (H&E) staining. Two in-between sections were used for HPV-DNA testing, genotypingand E6*ImRNA detection; four additional slides were obtained to assess expression of cellular proteins by IHC. A block was classified as "adequate" for HPV testing if invasive cancer was observed in the two H&E stained sections of the specimen. Pathology review was performed blind with respect to the original local diagnosis and followed a pre-established algorithm for diagnostic consensus involving three pathologists, as reported elsewhere [5]. Pathological classification was based on the World Health Organization pathological criteria for head and neck cancer [19].
FFPE blocks were processed under strict conditions of pre/post polymerase chain reaction (physical separation), and blank paraffin blocks were systematically tested in parallel to serve as sentinels for contamination as previously published [20].

HPV-DNA Detection and Genotyping
The detailed methods used for HPV-DNA detection and genotyping have been    (3), undifferentiated carcinoma (4) and neuroendocrine carcinoma (3), and all of them were primary tumors. The tonsil was the most common anatomical sub-site (40.0%). After adjusting for significant co-variates, HPV-related patients were significantly more likely to be non-smokers and non-drinkers andto have a SCC of the tonsil, consistently across the four HPV-relatedness definitions analyzed. Association of HPV-positivity and female gender was observed in all univariate but none multivariate analyses.
As described in

DISCUSSION
Mounting evidence supports the etiologic role of oncogenic HPVs in certain OPCs and the potential implications in the management of HPV-related patients. Our knowledge remains however incomplete regarding differences in prognosis by anatomic sub-site or treatment, or about the differential performance in terms of diagnostic accuracy and prognostic values between HPV-related biomarkers that can be easily implemented in the clinical setting.
To the best of our knowledge, this study represents the first attempt to address jointly all these issues in a large retrospective series of unselected patients.In an era of deescalation clinical trials,this information is crucial in order to unequivocally identify patients who can really benefit from de-escalate protocolsand to avoidworsening their outcomes.
The epidemiology of HPV-related OPC in our cohort differed in some aspects from what is observed in other high-income countries.HPV-AFs were slightly higherin women than in men, as has already been observed in other series [5], in contrast with what is observed in the United Statesin cohorts from the same time periods [24],. This discrepancy may reflect distinct temporal, geographical,and sociodemographic trendsin population exposure toboth tobaccouse and/or oral HPV infection, leading to arapid shift in the epidemiology of HPV-relatedOPC.
We examined the HPV-diagnostic accuracy of several biomarkers with a previously validated robust and comprehensive methodology [5]. In line with our previous results [5] and a recent meta-analysis           Panel "b" showed Kaplan-Meier curve for patients who underwent induction chemotherapy followed by chemo-radiotherapy or bioradiotherapy. Panel "c" showed Kaplan-Meier curve for patients who underwent cisplatin-radiotherapy.
Panel "d" showed Kaplan-Meier curve for patients who underwent cetuximabradiotherapy. Improved OS was not observed on panel "d".