The role of HPV testing in patients with oropharynx cancer

Issue: September 25, 2011

ByBarbara A. Burtness, MD

There is now compelling evidence that a large subset of oropharynx — tonsil and base of tongue — cancers arise from transforming HPV infection. These cancers have a molecular signature marked by loss of p53 and Rb expression and high p16 expression, which differs from the pattern of p53 mutation, p16 loss and high mutational burden of non-virally associated head and neck cancers, and they are dramatically more treatment responsive than head and neck cancers, which arise from tobacco and alcohol exposure.

The recognition that HPV-associated head and neck cancer is essentially a different disease, which arises through a different mechanism, a different molecular phenotype and a high rate of cure after chemoradiation, has raised the question whether the intensification in treatment that has resulted from decades of study of tobacco- and alcohol-related cancers is necessary for HPV-associated cancers. Indeed, with a 2-year survival in E2399 of 95%, there is little room for improvement with further treatment intensification.

Distinctions in disease types

The ECOG-E1308 trial is the first trial confined to HPV-associated oropharynx cancers. In this study, patients with locally advanced disease receive induction chemotherapy with cisplatin, paclitaxel and cetuximab (Erbitux, Imclone). For those patients with a clinical complete response at the primary site, the definitive therapy of cetuximab and radiation is given to a lower total radiation dose of 54 Gy. The hypothesis underlying this study is that late treatment effects such as swallowing dysfunction are particularly tied to high-dose radiation, and that, after a complete response to chemotherapy in very treatment-responsive cancers, disease control can be obtained with radiation doses conventionally used for subclinical disease.

Barbara Burtness, MD
Barbara Burtness

If there are HPV-associated cancers with intermediate prognosis, as suggested by a recursive partitioning analysis of R0129 in which tobacco exposure and high nodal disease predicted for intermediate risk, response to chemotherapy may be incomplete. Data from E1308 should establish not only whether treatment de-intensification is feasible for complete responders, but also whether chemotherapy response serves as a dynamic biomarker of suitability for treatment de-intensification. The RTOG-R1016 trial also studies whether a less toxic approach to definitive therapy is appropriate for HPV-associated cancer; in this trial, definitive chemoradiation with high-dose cisplatin and an altered fractionation schedule is compared with altered fractionation radiation given with cetuximab.

Importance of biomarker identification

An important question in trials proposing treatment de-intensification is reliable assignment of patients to these studies. That is, which biomarker is best for identifying patients who have favorable prognosis with standard therapy and might be appropriate candidates for trials of treatment de-intensification? The initial reports establishing that HPV-associated cancers occur in the oropharynx and have favorable prognosis utilized HPV in situ hybridization (ISH); however, each of the ongoing national studies described utilizes p16 staining. The reason p16 is such a useful surrogate for HPV association is that when HPV becomes transcriptionally active, the viral E7 oncoprotein enhances degradation of Rb, which usually suppresses p16 expression. In the absence of Rb, p16 content rises. The distinction is aided by the fact that p16 loss is common in tobacco-associated cancers and underscored by the finding that HPV viral load is highest when Rb expression is low and p16 expression high. In clinical terms, immunohistochemistry should show strong and diffuse nuclear and cytoplasmic p16 staining for a case to be considered p16-positive.

Detection of HPV E6 and E7 mRNA by polymerase chain reaction (PCR) is often not feasible in paraffin-embedded clinical material. HPV ISH is another useful surrogate for transcriptionally active, potentially transforming HPV infection, with a correlation to HPV mRNA expression, which is similar to that of p16. Utilizing HPV type-specific PCR for E6/E7, the sensitivity and specificity of HPV16 ISH and p16 staining by immunohistochemistry were determined. Concordance with E6/E7 expression was good for both assays. Sensitivity was 97% for p16 immunohistochemistry vs. 91% for HPV16 ISH, and specificity 82% vs. 91%. For discordant cases, E6/E7 expression was detected in 63% of p16+/HPV ISH-negative cases vs. 50% of p16 negative/HPV ISH-positive cases. However, the absolute number of discordant cases was small. Furthermore, outcome data were not presented, so that the relative performance of the two assays in predicting favorable prognosis after chemoradiation cannot be assessed from these data.

Two large randomized trials have examined outcome by HPV ISH and by p16. In R0129, when oropharynx cancers were categorized as HPV-associated because of positive ISH, the 3-year survival was 82.4% vs. 57.1% among patients with HPV ISH-negative tumors. After adjustment for other clinical variables, those who were HPV ISH-positive had a 58% reduction in the risk of death. When patients were categorized by p16 staining, the p16-positive patients had a 3-year OS of 83.6% compared with 51.3% in the p16-negative patients. After adjustment, this correlated with a 67% reduction in the risk of death. In the TROG 02.02 study, p16 was also predictive of better survival. That study — conducted at non-Western sites — interestingly reported a significant proportion of HPV ISH-negative, p16-positive cancers, of which 79% had evidence of HPV by PCR. The possibility that different populations harbor different types of high-risk HPV has also been raised in the setting of cervical cancer. In one study conducted among women in Jamaica, the most common high-risk HPV types were: HPV 45 (21.7%); HPV 58 (18.8%); and HPV 16 (18.4%), whereas in the initial US reports of HPV-associated oropharynx cancer, the most common types were HPV 16 and 18.

Evidence of favorable outcomes

Thus, at present, we have good evidence that strong p16 immunostaining predicts for favorable outcomes from definitive chemoradiation. This test has many characteristics that might make it suitable for routine clinical use if the current generation of clinical trials leads to different treatment paradigms for HPV-associated oropharynx cancers than for other head and neck cancers. Advantages are that the technique of immunohistochemistry is a routine service for pathology departments; can be performed on formalin-fixed, paraffin-embedded tissue; is not costly; and is associated with low inter-observer variability. Geographic variation in high-risk HPV-type prevalence should not lead to false-negative results with p16 immunostaining. Issues such as the performance of p16 staining relative to HPV ISH, which high-risk HPV types to test for in which populations, and whether dual testing is appropriate when p16 testing is negative, should be resolved by study of the samples collected in the current studies.

Barbara Burtness, MD, is professor of medical oncology; head, translational oncology; chief, head and neck oncology; and co-leader of the Developmental Therapeutics Program at the Fox Chase Cancer Center in Philadelphia. Disclosure: She reports no relevant financial disclosures.

For more information:

Ang KK. N Engl J Med. 2010;363:24-35.

Boyer SN. Cancer Res. 1996;56:4620-4624.

Cmelak AJ. J Clin Oncol. 2007;25:3971-3977.

Fakhry C. J Natl Cancer Inst. 2008;100:261-269.

Gillison ML. J Clin Oncol. 2011;24:5623-5625.

Rischin D. J Clin Oncol. 2010;28:4142-4148.

Shi W. J Clin Oncol. 2009;27:6213-6221.

Watt A. Infect Agent Cancer. 2009;4 Suppl 1:S11.

Weinberger PM. J Clin Oncol. 2006;24:736-747.