CDKN2A deletion in p16‐negative/HPV‐positive head and neck squamous cell carcinoma: Highlighting the molecular basis behind the limitation of relying on p16 immunohistochemistry as a surrogate marker for HPV involvement

Abstract

Aims

Immunohistochemistry (IHC) for p16 is widely used as a surrogate marker for HPV involvement in oropharyngeal squamous cell carcinoma (OPSCC), among other tumours. Confirming HPV status in OPSCC is critical, as HPV‐positive tumours have better overall survival and may require de‐escalated therapy compared to HPV‐independent OPSCC in the future. However, discordance exists between p16 and HPV, and direct HPV testing is occasionally required to ensure an accurate diagnosis. The aim of this study is to highlight the genomic basis behind the limitation of relying on p16 IHC as a surrogate marker for HPV involvement.

Methods and Results

Through a multi‐institutional collaboration, this case series compiled four patients with a ‘false’ negative p16 staining pattern in HPV‐positive non‐keratinizing head and neck squamous cell carcinoma. All cases demonstrated minimal to no p16 IHC staining and were positive for HPV by direct RNA in situ hybridization. Through CDKN2A fluorescence in situ hybridization testing, three patients demonstrated a homozygous deletion of CDKN2A and one demonstrated a heterozygous deletion.

Conclusions

This series highlights the genomic basis for the ‘false’ negative p16 results, raising awareness of a significant diagnostic pitfall while emphasizing the importance of careful consideration of clinicopathologic parameters in the clinical workup of these cases.

We present a case series of HPV‐positive non‐keratinizing squamous cell carcinoma of the head and neck with discordant negative p16 staining due to CDKN2A deletion. These findings highlight the genomic basis behind the diagnostic pitfall of reliance on p16 immunohistochemistry as a surrogate marker of HPV and raise practice‐changing considerations for routine clinical practice.

Abbreviations

FISH

fluorescence in situ hybridization

HN

head and neck

IHC

immunohistochemistry

ISH

in situ hybridization

OPSCC

oropharyngeal squamous cell carcinoma

SCC

squamous cell carcinoma

Introduction

The prevalence of HPV‐associated oropharyngeal squamous cell carcinoma (OPSCC) has steadily risen over the past 30 years and now accounts for 33%–71% ¹ , ² of all OPSCCs worldwide. Clinical features suggestive of HPV‐associated OPSCC include younger age, location within the tonsil or base of the tongue, a small primary tumour with larger and often cystic nodal involvement, and non‐keratinizing morphology. ³ HPV subtype 16 accounts for approximately 90% of all HPV‐associated OPSCCs. ⁴ Accurately determining the HPV status in OPSCC is critical due to the higher rates of response to induction chemotherapy and higher sensitivity to radiation of HPV‐associated tumours. ⁵ , ⁶ Therefore, de‐escalated therapy may be indicated in HPV‐driven tumours, as opposed to the conventional, higher intensity chemoradiation employed for the treatment of HPV‐independent tumours. ⁷ Further, HPV‐associated OPSCC has better overall and progression‐free survival compared to HPV‐independent OPSCC. ⁸ The prognostic significance of HPV involvement in squamous cell carcinoma (SCC) of other head and neck (HN) sites is uncertain and represents an area of ongoing investigation.

The most common approach for identifying HPV involvement in HN SCC is by p16 immunohistochemistry (IHC). This is used as a surrogate marker of HPV infection in the oropharynx and lateral neck lymph nodes and is defined as ‘positive’ by the College of American Pathologists as nuclear and cytoplasmic expression with at least moderate to strong intensity in at least 70% of tumour cells. ⁹ Overall, p16 is a reliable surrogate marker, with sensitivity and specificity for HPV involvement being 94% and 83%, respectively, in a recent meta‐analysis. ¹⁰ However, p16‐HPV discordance exists and has recently been shown to have prognostic implications. ¹¹ , ¹² Consequently, direct testing methods such as HPV DNA PCR or HPV DNA or RNA in situ hybridization (ISH), which have greater sensitivities (98%, 85% and 97%, respectively) and specificities (84%, 88% and 93%, respectively) ¹⁰ , ¹³ compared to p16 IHC, are occasionally relied upon to ensure an accurate diagnosis. Underlying causes of this p16‐HPV discordance include, but are not limited to, p16 antibody variation, ¹⁴ application of different standard cut‐offs for p16 positivity (e.g. 70%, ⁹ 60% ¹⁵ and 10% ¹⁶ ), alternative oncogenic mechanisms such as HPV‐independent RB1 loss‐induced p16 overexpression, ¹⁷ as well as potential inter‐ and intra‐observer variability in p16 evaluation. In this series, we highlight four cases of p16−/HPV+ discordant HN SCC, which all demonstrated CDKN2A deletions that likely caused the observed discordance. To the best of our knowledge, this is the first report documenting the molecular basis for the discordance in p16‐negative HPV‐associated HN SCC.

Materials and Methods

Through a multi‐institutional collaboration between Mount Sinai Hospital, Toronto, ON, The Hospital for Sick Children, Toronto, ON, The Ottawa Hospital, Ottawa, ON, University of Wisconsin Hospital, Madison, WI and the University of British Columbia, Vancouver, BC, four patients with non‐keratinizing SCC with negative p16 staining by IHC and positivity for HPV RNA ISH were identified in a search of archival cases. For each patient, a review of the medical history, histological features and ancillary work‐up was compiled. This retrospective study was approved by the Mount Sinai Hospital Research Ethics (protocol #22‐0216‐E) and deemed exempt from the requirement for obtaining informed consent from potential subjects.

Histology and Immunohistochemistry

Histological review was performed for each specimen using standard haematoxylin and eosin staining. p16 immunohistochemistry was performed using multiple validated protocols. Staining with the E6H4™ clone (Ventana Medical Systems, Tucson, AZ) was conducted either manually (case 1) or on the Ventana Discovery XT automated platform (cases 3 and 4), following the manufacturer's protocol. The JC2 clone (Cell Marque) on the Dako Omnis platform with the EnVision FLEX+ detection system (Dako) was also used (case 2) with automated staining as per the manufacturer's instructions.

HPV Genotype Testing

Genotyping of HPV via PCR was performed for case 1 using the Canadian National Microbiology Laboratory's (Winnipeg, MB, Canada) Luminex assay, a multiplex PCR capable of detecting 45 different HPV genotypes of low‐, indeterminate‐, high‐ and unknown‐risk.

High‐Risk HPV RNA ISH and CDKN2A Fish Testing

High‐risk HPV RNA testing by in situ hybridization (ISH) was performed on all specimens using the ACD RNAscope™ high‐risk probe cocktail, which probes for 18 types of transcriptionally active high‐risk HPV types (16, 18, 25, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 68, 73 and 82) using either manual or automated protocols (Advanced Cell Diagnostics, Newark, CA). The specimens also received fluorescence in situ hybridization (FISH) testing to assess for CDKN2A deletions. Hybridization was performed according to the manufacturer's instructions using a dual colour locus‐specific probe for CDKN2A (9p21.3, red) and the control region D9Z3 (9q12, green) (Cytocell, Cambridge, UK).

Results

Patient sex, age, tumour site, tumour size, primary intervention, tumour stage, status of lymphovascular and perineural invasion, neoadjuvant and adjuvant therapy status, results of p16 IHC, HPV RNA ISH, CDKN2A FISH, as well as the status of disease recurrence are provided in Table 1.

Table 1.

Clinicopathologic features of p16‐HPV discordant head and neck squamous cell carcinoma

Case	1	2	3	4
Sex	M	M	M	M
Age	49	38	79	73
Tumour site	Tonsil	Larynx	Mandible alveolar ridge	Tonsil
Tumour size	N/A	1.1 cm	4.5 cm	1.5 cm
Primary procedure	Biopsy	CO2 laser excision	Composite resection	Tonsillectomy
Primary radiation	Yes	No	No	No
Stage	pT2N1	pT2N0	pT4aN0	pT1N0
Lymph node status	Positive by imaging	Negative by imaging	0/56	0/30
Perineural invasion	No	No	No	No
Lymphovascular invasion	Yes	No	No	No
Neoadjuvant/adjuvant therapy	No	Radiation	Radiation	No
p16 IHC	−	−	±	−
HPV ISH	+	+	+	+
HPV genotype	HPV16	N/A	N/A	N/A
NGS results	No CDKN2A mutations	N/A	N/A	N/A
CDKN2A FISH	Heterozygous deletion	Homozygous deletion	Homozygous deletion	Homozygous deletion
Disease recurrence	No	No	Yes; at 3 months	No
Last follow‐up interval	6 months	2 years	14 months	168 months

Case 1

A 49‐year‐old male with a history of alcohol consumption and remote, minimal smoking presented with difficulty swallowing and a right tonsillar mass. Imaging showed mild asymmetric enlargement of the right tonsil with an enlarged, partially cystic and necrotic right level IIA lymph node, findings suggestive of a metastatic OPSCC (Table 1). An incisional biopsy of the right tonsil was performed and microscopically demonstrated a squamous cell carcinoma with a predominantly non‐keratinizing morphology (Figure 1A). Focal keratinization was present but not beyond what can be seen in typical HPV‐associated tumours. IHC for p16 was performed as a surrogate marker for HPV involvement and was negative (0% staining) (Figure 1D). Given the predominantly non‐keratinizing morphology and clinical features (i.e., younger patient, oropharyngeal location, cystic nodal involvement), HPV RNA ISH was performed and was positive (Figure 1G). Follow‐up FISH testing revealed a heterozygous deletion of CDKN2A (probe for 9p21.3) in 63% (128/200) of nuclei examined (Figure 1J). No additional documented evidence of disease recurrence has been made 6 months following the initial diagnosis.

Figure 1.

Morphology, p16 expression, HPV Status and CDKN2A copy number profile in cases with only invasive carcinoma identified. Haematoxylin and eosin (A–C, medium power). Immunohistochemical expression of p16 (D–F medium power). Transcriptionally active high‐risk HPV RNA ISH expression (G–I, high power) CDKN2A fluorescence in situ hybridization copy number assessment showing heterozygous deletion (J, inset: single cell) and homozygous deletion (K, L, inset: single cell).

Case 2

A 38‐year‐old male with a history of ‘verrucous laryngeal cancer’ and marijuana smoking presented with a laryngeal mass, clinically concerning for recurrence, involving the vocal folds and subglottic regions bilaterally (Table 1). A CO₂ laser excision of the mass demonstrated an exophytic non‐keratinizing papillary squamous cell carcinoma (Figure 1B). The non‐keratinizing papillary morphology was noted to be atypical for conventional laryngeal SCC, so the possibility of involvement by high‐risk HPV was investigated. IHC for p16 was performed and was negative (Figure 1E). IHC for p53 was also performed and demonstrated basal sparing, consistent with an HPV‐associated staining pattern. ¹⁸ As TP53 is mutated in up to 84% of HPV‐independent HN SCC, ¹⁹ the wild‐type staining observed, in conjunction with the non‐keratinizing papillary features, was suggestive of involvement by high‐risk HPV, in spite of the p16 negativity. High‐risk HPV RNA ISH was performed and was positive (Figure 1H). Subsequently, FISH with a probe for the CDKN2A (p16) locus (9p21.3) was performed and demonstrated a deletion of both CDKN2A loci in 140/200 (70%) of nuclei examined (Figure 1K), consistent with a homozygous deletion, explaining our discordant findings. No additional documented evidence of disease recurrence has been made 2 years following the initial diagnosis.

Case 3

A 79‐year‐old male with a history of a 4.5 cm mass involving the mandibular alveolar ridge had a composite resection followed by radiation (Table 1). The resection demonstrated a warty squamous cell carcinoma (Figure 2A) with extensive in situ disease, with the appearance of HPV‐associated dysplasia (Figure 2B). IHC for p16 was performed and demonstrated minimal patchy positive staining in the invasive component, well below the 70% threshold for being called ‘positive’ according to the College of American Pathologists guidelines, ⁹ but was block‐positive in the in situ component (Figure 2C,D). Due to the observed peculiar p16 staining pattern and HPV‐associated morphological features, high‐risk HPV RNA ISH was performed. The HPV RNA ISH was positive in both the invasive and in situ components (Figure 2E,F). This was followed by CDKN2A FISH, which revealed a homozygous deletion exclusively within the invasive component (Figure 2G). A normal diploid signal was observed in the in situ component (Figure 2H). The patient developed a local recurrence 3 months following the primary resection and received adjuvant chemotherapy plus a salvage resection. No additional documented evidence of disease recurrence has been made 14 months following the initial recurrence.

Figure 2.

Morphology, p16 expression, HPV status, and CDKN2A copy number profile in cases with invasive and in situ carcinoma. Haematoxylin and eosin (A, B, Medium power). Immunohistochemical expression of p16 (C, D, medium power). Transcriptionally active high‐risk HPV RNA ISH expression (E, F, high power). CDKN2A fluorescence in situ hybridization copy number assessment showing heterozygous deletion (G, inset: single cell) and normal diploid profile (H, inset: single cell).

Case 4

A 73‐year‐old male with a history of a 1.5 cm mass involving the tonsils had a tonsillectomy performed without adjuvant or neoadjuvant therapy (Table 1). The resection demonstrated a non‐keratinizing squamous cell carcinoma (Figure 1C). IHC for p16 was performed and was negative (Figure 1F). High‐risk HPV RNA ISH was performed and was positive (Figure 1I). This was followed by CDKN2A FISH, which revealed a homozygous deletion (Figure 1L). Previous data published from this patient demonstrated no hypermethylation of CDKN2A, and RNA sequencing demonstrated a lack of CDKN2A expression, consistent with a homozygous deletion. ²⁰ The patient has had no documented evidence of disease recurrence 168 months following the initial diagnosis.

For all cases, two theoretical explanations for the p16‐negative HPV‐positive discordance were considered. The first, less likely, possibility was involvement of a rare HPV genotype that would not impact the cell cycle in the same way as the typical high‐risk HPV genotypes and would not lead to compensatory p16 upregulation, a phenomenon thus far undescribed in the scientific literature. This unlikely theory was investigated in case 1 by HPV PCR for the specific genotype. The typical high‐risk HPV genotype 16 was detected (Table 1). This subtype, along with other high‐risk subtypes, is well known to induce p16 overexpression ²¹ , ²² and thus did not explain the p16‐HPV discordance we observed. The second, more likely possibility was CDKN2A mutation, deletion, or hypermethylation that would silence the gene and impact production of the p16 protein. To investigate this theory, the tumour in case 1 underwent next‐generation sequencing using the TruSight RNA Pan‐Cancer panel (Illumina), which evaluates 1385 different genes. No evidence of mutations in CDKN2A or related genes was detected (Table 1), and expression of CDKN2A transcript was present, thus strongly suggesting the absence of a homozygous deletion or epigenetic silencing. Although coverage was good and a homozygous CDKN2A deletion was unlikely, definitive assessment of partial gene deletions was investigated with FISH, which ultimately demonstrated a heterozygous deletion of CDKN2A (Figure 1J), explaining our discordant findings. Similarly, p16‐HPV discordance in cases 2–4 was investigated with CDKN2A FISH testing, and all revealed a homozygous deletion (Figures 1K,L and 2G), also explaining the discordant findings.

Discussion

HPV‐positivity in the absence of p16 staining is an atypical finding, but not one that is commonly pursued. Once p16 is negative in an HN SCC, the diagnostic workup typically stops and the tumour is presumed to be HPV‐independent. The patient then proceeds to receive conventional chemoradiation, as opposed to consideration for de‐escalated trials. In this case series, due to predominantly non‐keratinizing morphology, other HPV‐associated morphologic features and/or clinical presentations suggestive of HPV‐driven tumours, direct HPV testing via HPV RNA ISH was pursued and was positive in all cases, despite the p16 IHC being negative for each case. In fact, the complete absence of p16 staining in cases 1, 2 and 4, was suspicious, as in conventional keratinizing squamous cell carcinoma there is not infrequently some staining for p16, not reaching the cut‐off for interpreting the stain as positive (<70%, mosaic). ²³ Although HPV involvement outside of the oropharynx is currently of unknown significance, investigation was nevertheless pursued by the case pathologists in an attempt to account for the unique morphologic and immunohistochemical features in each case. Additionally, gathering institutional experience with these tumours will help elucidate the significance of HPV involvement in extra‐oropharyngeal sites.

Ultimately, the cause for our observed p16‐HPV discordance in our cases was attributable to a heterozygous or homozygous deletion in the gene that produces p16, CDKN2A, a rare finding in HPV‐positive OPSCC. Notably, of the 36 HPV‐positive HN SCC tumours analysed in TCGA data, only two (5.6%) had a heterozygous CDKN2A deletion, and none were found to have a homozygous deletion, altered methylation status, point mutation, or splice mutation in the gene. ¹⁹ The reason for the complete absence of p16 staining in the case with heterozygous CDKN2A was not clear, and may be due to antibody sensitivity or potentially other forms of epigenetic silencing. Interestingly, and in contrast, CDKN2A aberrations are one of the most common genetic findings in HPV‐independent HN SCC, with 54%–57% exhibiting CDKN2A mutations or loss of the gene. ¹⁹ , ²⁴ It is believed that this may in part explain the wide variability and typical p16‐negativity in HPV‐negative HNSCC. Further, CDKN2A copy number loss predicts poorer survival among HPV‐negative HN SCCs, and it has been proposed that this may be related to the relative lack of p16 and its role as a tumour suppressor. ²⁵ Similarly, in select central nervous system tumours, homozygous CDKN2A deletion alone portends a poor prognosis and is relied upon for increasing tumour grade. ²⁶

A recent multicentre international review of 7654 patients with oropharyngeal cancer found an overall p16‐HPV discordance rate of 9.2% and rates of 3.8% (289/7654 cases) and 5.4% (415/7654 cases) for p16−/HPV+ and p16+/HPV−, respectively. ¹¹ Similar rates of discordance in OPSCC have been found by others, including an 18‐year retrospective single‐institutional study of 5.2% (37/709) and 5.5% (39/709) for p16‐/HPV+ and p16+/HPV−, respectively, ²⁷ and a large American OPSCC cohort of 467 patients using HPV RNA ISH, of 4.9% (3.4% p16−/HPV+ and 1.5% p16+/HPV−). ¹² Interestingly, these studies also found that overall and disease‐free survival rates varied depending on whether a tumour was HPV‐associated and p16‐HPV discordant, with the largest differences being between p16‐HPV concurrent double positive and concurrent double negative carcinomas. ¹¹ , ¹² While confirmation of HPV status and p16 stain interpretation vary based on antibody type and pathologist practice, and some of these values may be overestimates, true discordance between HPV and p16 is a well‐recognized phenomenon. Combining these survival data with the known variation in treatment sensitivity between HPV‐associated and HPV‐independent OPSCC, correct diagnosis is crucial to appropriately allocate patients to de‐escalated or conventional chemoradiation in a trial setting. Further, considering that p16/HPV discordant patients have worse survival outcomes than concordant p16+/HPV+ patients, use of de‐escalated or conventional therapy ought to be carefully considered. This worse relative survival may potentially be due to the well‐described oncogenic effects of homozygous CDKN2A loss. Once again, it is important to note that the significance of HPV involvement outside of the oropharynx is an area in need of further study.

This study raises conceptual considerations regarding the utility of p16 immunohistochemistry as a surrogate marker for HPV involvement. While the primary reason for the general high concordance of p16 staining with HPV status is due to the rarity of CDKN2A loss in HPV‐driven tumours, the relatively common occurrence of CDKN2A deletions in conventional HN SCC surely also contributes to this effect. Although HPV involvement and CDKN2A deletion generally do not co‐occur, we have demonstrated that these processes are not mutually exclusive. In clinical practice, the presence of CDKN2A deletion, detected by FISH, next‐generation sequencing, copy number analysis or complete absence of p16 immunohistochemical staining, is an indication that the assessment for HPV involvement was not effective, and more direct methods should be pursued if there is clinical or morphologic suspicion (which likely exists if p16 immunohistochemistry was performed in the first place). Additionally, one must wonder if these cases represent an unfavourable subgroup of HPV‐driven HN SCC due to the additional molecular ‘hit’ of CDKN2A silencing.

Conclusion

In conclusion, we present four cases of p16‐negative, HPV‐associated head and neck squamous cell carcinoma with predominantly homozygous CDKN2A deletions. In the oropharynx and levels II–IV of the lateral neck, p16 is used as a surrogate marker for HPV involvement. Outside of the oropharynx, p16 staining is often used as a ‘rule‐in’ test, and when positive, is an indication to perform direct molecular assessment for the presence of HPV. ⁹ , ²⁸ While in the majority of cases this approach leads to the correct diagnosis, these cases underscore and explain a limitation of this practice. Through our workup, we identified the molecular basis for the misleading IHC features, while also highlighting the strengths and weaknesses of different molecular assays. Future cases with clinical and morphological features of HPV involvement, with p16 negativity, especially complete absence of p16 staining, should require molecular testing (either by ISH or PCR) to definitively rule out HPV involvement. Additionally, it is important to be aware that oropharyngeal, cervical and anal squamous cell carcinoma cases undergoing next‐generation sequencing with detection of a CDKN2A aberration may have negative p16 staining, even if they are HPV‐associated. As the understanding of HPV‐related cancers improves, both within the oropharynx and beyond, refining diagnostic protocols will be pivotal in providing patients with the most tailored and effective treatment strategies.

Author contributions

EH: conception and design. TT, RH, AN, MS, YK, CR, RS, BP, EH: acquisition of data, data analysis and/or interpretation. TT and EH wrote the article. TT, RH, AN, MS, YK, CR, RS, BP and EH reviewed and approved the final article.

Funding information

This study was not supported by any funding.

Conflict of interests

The authors have no relevant financial or non‐financial interests to disclose.

Data availability statement

All data supporting the findings of this study are available from the corresponding author (EH) upon reasonable request.