Ties that bind: p16 as a prognostic biomarker and the need for high-accuracy human papillomavirus testing.

Abstract

Squamous cell carcinoma of the head and neck (SCCHN) is comprised of two distinct entities: tobacco-associated, human papillomavirus (HPV) –negative SCCHN, and HPV-positive SCCHN. HPV-positive tumors arise primarily in the oropharynx and have a favorable prognosis when treated with chemotherapy, radiation, surgery, or chemoradiotherapy. p16 immunohistochemistry (IHC), which is widely used as a surrogate marker of HPV, and direct detection of HPV genetic material (eg, using in situ hybridization [ISH] or polymerase chain reaction [PCR]) are powerful, well-established prognostic biomarkers for oropharyngeal (OP) SCCHN. HPVpositive SCCHN shows absence of alterations of the tumor suppressors TP53 and CDKN2A because p53 and p16 proteins are already inactivated by HPV E6/E7 viral proteins. Genetically, HPV-positive SCCHN tumors closely resemble cervical cancers. By contrast, HPV-negative tumors arise throughout the upper aerodigestive tract and have a poor prognosis. Such tumors frequently harbor TP53 mutations and CDKN2A/p16 loss, and on average show significantly higher levels of epidermal growth factor receptor expression. Genetically, HPV-negative SCCHN tumors closely resemble lung and esophageal squamous cell carcinomas. Similarly, when focusing exclusively on p16-expressing tumors, it is apparent that we are still dealing with the same two distinct entities: HPV-positive and HPV-negative, p16-positive tumors. This pool of p16-positive tumors includes, first, all tumors that are etiologically related to HPV (p16 is universally expressed in such tumors), and second, 10% to 12.5% of HPV-negative tumors, including both oropharynx and nonoropharynx tumors, which may activate the cell cycle in a p16-independent manner (eg, Rb mutations/ CCND1 amplification). HPV-negative, p16-expressing tumors show the genetic fingerprint of HPV-negative tumors, for example, TP53 mutations and high levels of epidermal growth factor receptor expression. The relative incidence of these two biologically distinct entities depends on the anatomic origin of the tumor and the overall rate of HPV positivity of SCCHN, which differs between geographic regions. Accordingly, the performance characteristics of p16 IHC depend on the context: in OP tumors in the United States, given a high HPV incidence and pretest probability of approximately 70%, p16 IHC performs well as an HPV surrogate with a positive predictive value of 92.7% and a negative predictive value of 92.8%. By contrast, in non-OP tumors with a low pretest probability, p16 performs poorly, with a higher fraction of p16-positive/HPV-negative tumors; for example, for SCCHN originating in the oral cavity, the positive predictive value of p16 IHC as an HPV-surrogate marker is only 41.3%. Thus, use of p16 IHC as a surrogate marker for HPV in the oral cavity, larynx, hypopharynx, and other non-OP areas should be discouraged. Given the strong prognostic implications of both p16 IHC and HPV status for OP tumors, Chung et al, in the article that accompanies this editorial, pose an important follow-up question: What are the prognostic implications of p16 positivity and HPV-ISH positivity in non-OP SCCHN? Using tumor material from three Radiation Therapy Oncology Group (RTOG) studies (RTOG 0129, 0234, and 0522), the authors performed p16-IHC and DNA-based HPV ISH on 297 non-OP tumors. A surprisingly large fraction was positive: 26.3% of oral cavity tumors, 16.4% of hypopharynx tumors, and 17.1% of larynx tumors by p16 IHC, and 14.6%, 5.3%, and 7.9% by HPV ISH. Importantly, p16 positivity was twice as common as HPV-ISH positivity (19.3% v 9.4% overall). p16-positive status correlated significantly with improved progression-free and overall survival with impressive hazard ratios of 0.65 and 0.57, respectively. HPV ISH showed a similar trend toward improved progression-free and overall survival, but did not reach statistical significance, likely because of the smaller number of HPV-ISH–positive samples. To understand the relationship of p16 and HPV in this study, it is pivotal to examine the methodologic limitations, specifically the lack of high-accuracy HPV testing (ie, E6/E7 detection by PCR, or RNAbased ISH). Assessment of HPV by DNA-based HPV ISH is prone to false negatives in 13% to 41% of tests, resulting in a lower than ideal sensitivity. In other words, a sizable number of samples in this study were likely labeled as p16 positive/HPV negative, although they were actually HPV positive. This contaminates the HPV-ISH– negative/p16-positive group and complicates analysis—especially when trying to assess p16 as an independent prognostic marker. Overall, there are three important conclusions that can be derived from the article by Chung et al. First, the observed incidence of p16 positivity and HPV positivity in non-OP SCCHN samples was unexpectedly high. Overall, 19.3% were p16-IHC positive, and 9.4% were HPV-ISH positive. By contrast, the Cancer Genome Atlas reported an incidence of 5.4% HPV positivity in non-OP tumors, and Lingen et al reported an incidence of 5.9% for oral cavity tumors. Zuo et al, in another cohort of chemoradiotherapy-treated patients, reported 18.4% of JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L