Quantitative Assessment of p16 Expression in FNA Specimens From Head and Neck Squamous Cell Carcinoma and Correlation With HPV Status

Abstract

BACKGROUND:

This study investigated p16 by immunohistochemistry (IHC) on cellblocks (CBs) and human papillomavirus (HPV) by polymerase chain reaction (PCR) in fine-needle aspiration (FNA) of head and neck squamous cell carcinoma (HNSCC).

METHODS:

Receiver operating characteristic (ROC) curve analysis was used to assess test performance in CBs compared with p16 IHC in 42 surgical specimens from patients with HNSCC and in correlation with HPV by PCR in cytology specimens. The study assessed HPV by PCR in FNA specimens as a substitute for p16 IHC in surgical specimens.

RESULTS:

Of 42 cases, 38 CBs showed malignant cells as cohesive clusters of viable cells with or without single tumor cells, whereas 4 specimens were composed exclusively of single tumor cells and degenerated cells. All p16-negative surgical specimens showed an absence of p16 staining in the corresponding CBs (n = 16). In the p16-positive surgical cases (n = 26), corresponding CBs with tumor clusters (n = 23) showed heterogeneous p16 expression ranging from 40% to 100%; however, scoring single cells was challenging and unreliable because of cellular degradation. ROC curve inspection showed the optimal threshold to be at least 40% p16 staining in tumor clusters with 100% sensitivity and specificity. In cases with inadequate CBs, HPV by PCR on needle rinse showed 88% sensitivity and 100% specificity for p16 expression in surgical specimens.

CONCLUSIONS:

A cutoff of at least 40% p16 expression in tumor clusters may be appropriate for p16 positivity in cytology CB specimens. A positive HPV finding by PCR on needle rinse can be used as a substitute for p16 expression in surgical specimens.

INTRODUCTION

Although the incidence of conventional, smoking, and alcohol-related head and neck squamous cell carcinoma (HNSCC) has seen a considerable decrease in the United States, the overall incidence of HNSCC is on the rise, primarily driven by an increase in human papillomavirus (HPV)–related HNSCC, particularly oropharyngeal squamous cell carcinoma (OPSCC).^1–3 HPV and non-HPV OPSCC are 2 distinct entities with regard to molecular alterations, histology, and clinical outcome.^3–6 HPV-related OPSCC tends to affect a younger, non-smoker male population with exposure to high-risk HPV and is more likely to present with small tumors and early lymph node metastasis.⁷ However, it is chemosensitive and radiosensitive and displays a more favorable response to treatment and a better prognosis compared with HPV-negative HNSCC.⁷ Because of the adverse effects of curative surgery or chemoradiation in this younger population, the confirmation of HPV-related disease indicates a more favorable clinical outcome justifying deintensified therapy to mitigate the morbidity associated with chemoradiation.^8,9 Consequently, determining that an OPSCC is an HPV-driven tumor has significant implications for patient management and prognosis and has been integrated into the recently up-dated AJCC staging manual.¹⁰ Furthermore, HPV status determines patient eligibility for clinical trials investigating new treatment strategies and modalities.^6,11

Currently, the HPV status in HNSCC can be detected by HPV DNA by polymerase chain reaction (PCR), HPV E6/E7 messenger RNA (mRNA) by quantitative reverse transcriptase-PCR (qRT-PCR), DNA or RNA in situ hybridization (ISH), and p16 protein by immunohistochemistry (IHC). HPV DNA by PCR is considered to be highly sensitive, but it may show a positive reaction in HPV-harboring rather than HPV-driven OPSCC. Detection of viral oncoproteins E6/E7 mRNA is regarded as the gold standard as these confirm the presence of transcriptionally active virus. The main disadvantages of E6/E7 mRNA by qRT-PCR are its high cost and limited efficacy in formalin-fixed, paraffin-embedded samples. ISH techniques are popular by means of specificity, but technically challenging and expensive.^12,13 Although testing for HPV E6/E7 transcripts by RNA ISH is considered the ideal test to confirm a transcriptionally active HPV,^14–17 it does not offer an added benefit compared with p16 IHC alone in OPSCC¹⁸ and is not widely used yet in many clinical laboratories, which still rely on p16 IHC in daily practice.

The protein p16 is a tumor-suppressor protein encoded by the cyclin-dependent kinase inhibitor 2A (CDKN2A) gene. In OPSCC, expression of HPV E6 and E7 viral oncoproteins interferes with the tumor suppressors p53 and Rb pathways, and inactivation of Rb by E7 leads to upregulation of p16 expression.^6,19 Because of the very high pretest probability of oncogenic HPV infections in OPSCC, p16 IHC is widely used as a surrogate marker for HPV in OPSCC.^12,20,21 It has shown excellent performance as a stand-alone test, leading a panel of experts to recommend p16 IHC as the preferred testing method to detect transcriptionally active HPV infection in OPSCC tissue specimens.^12,18 A p16 IHC test is a relatively straight-forward, widely available, cost-effective, reliable, and sensitive method for detecting HPV-related squamous cell carcinoma (SCC) of the head and neck and the recent 8th edition of the AJCC Cancer Staging Manual divides OPSCC into HPV-positive and HPV-negative tumors based on p16 IHC expression.¹⁰

Patients with HPV-driven OPSCC often present with a small tumor at the primary site and prominent cervical lymphadenopathy.¹⁸ As a result, cervical lymph nodes are often sampled by fine-needle aspiration (FNA) to establish a diagnosis, and the cytologic material may be the only specimen available for p16 IHC assessment in anticipation of neoadjuvant chemoradiation. The College of American Pathologists (CAP) has published guidelines for HPV testing in surgical specimens from HNSCC, advocating for p16 IHC as a surrogate marker for HPV infection when there is at least 70% nuclear and cytoplasmic expression with moderate to strong intensity in OPSCC.¹⁸ However, there are no recommendations or established standard criteria to determine p16 positivity in cytology specimens. The objective of the current study was to assess p16 IHC on cytology cellblock (CB) material from metastatic HNSCC in correlation with p16 IHC on corresponding surgical specimens and to evaluate HPV by PCR in FNA samples.

MATERIALS AND METHODS

Case Selection

This study was approved by the institutional review board. The institutional pathology database was searched for HNSCC with corresponding neck mass FNAs performed at our institution between January 2013 and June 2018. Clinicopathologic data and demographics were recorded. The cytology cases included in the study had a histopathology-confirmed HNSCC (either biopsy or resection), had a CB with tumor in clusters or single cells, and p16 immunostain performed on cytology and surgical specimens. Additionally, we correlated HPV status performed by PCR on residual FNA material with p16 IHC and HPV DNA studies (PCR or ISH) on corresponding tissue specimens.

Specimen Preparation

For FNA specimens, both direct air-dried (Diff-Quik stain) and alcohol-fixed (Papanicolaou stain) smears were prepared. The aspiration needles were then rinsed in CytoRich Red solution (Thermo Scientific, Hanover Park, IL), which was used to prepare a ThinPrep slide (Hologic, Marlborough, MA), and a CB, either a Cellient CB on the Cellient automated cell block system for low cellularity specimens (Hologic, Marlborough, MA) or a traditional HistoGel-based formalin-fixed paraffin embedded CB for cases with a significant amount of cell pellet. An aliquot of FNA specimen was set aside in PreservCyt solution at room temperature for potential high-risk HPV testing.

p16 Immunohistochemistry

Five-μm-thick sections cut from paraffin-embedded CBs and tissue sections were deparaffinized. Antigen retrieval was performed using heat-induced epitope retrieval with citrate buffer. Staining of p16 was performed using the CINtec p16 histology assay (mouse monoclonal anti-body, clone E6H4; Ventana Medical Systems Inc, Tucson, AZ) on a BenchMark Ultra automated immunostainer (Ventana Medical Systems Inc) following the manufacturer’s protocol. Appropriate positive and negative controls were used.

p16 Interpretation in Cytology and Surgical Specimens

Cellularity and cytologic integrity were determined by reviewing the hematoxylin-eosin–stained CB slide. Only CBs with at least 10 viable tumor cells, in clusters or individual cells, were included in the study. In cases with tumor clusters, the presence of at least 1 cluster was deemed adequate. CB quality (tumor clusters versus single cells) was recorded. Number of clusters, largest cluster size, and number of tumor cells were assessed quantitatively. Both nuclear and cytoplasmic staining in viable tumor cells were scored for intensity and extent: The percentage of p16-positive tumor cells was recorded in 5% increments and intensity of staining reported as follows: negative (absence of staining), weak (faint staining), moderate (darker staining), and strong (dense staining).^22,23 Scoring was assessed independently by 2 cytopathologists (R.A-R. and S.G.) in a blinded manner without knowledge of the p16 status in the corresponding histopathologic sections. Discrepancies were resolved over a double-headed microscope. Surgical specimens were reviewed for extent and intensity of p16 IHC staining by 1 of the head and neck pathologists. p16 was considered positive when moderate-to-strong, diffuse nuclear and cytoplasmic block-like staining was observed in ≥70% of tumor cells and used as the gold standard to compare p16 staining on cytology CB material.

High-Risk HPV Detection

Following review of cytology for each case, the aliquot from the residual fluid specimen was referred for high-risk HPV testing if the cytologic material was diagnostic or suggestive of SCC. The specimen was prepared and analyzed using the Cobas HPV assay (Cobas 4800; Roche Molecular Systems, Pleasanton, CA), which detects high-risk HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68 by cellular DNA extraction and real-time PCR amplification according to the manufacturer’s protocol as previously described.²⁴

Statistical Analysis

To assess interrater reliability of assessing the p16 percentage from FNAs, we analyzed the intraclass correlation coefficient (ICC) using the “irr” package in R.²⁵ For the analysis of the p16 percentage as a function of cellularity, we performed a logistic regression analysis using R’s glm function. A P value equal or less than .05 (≤.05) was considered statistically significant. Receiver operating characteristic (ROC) curves were plotted to determine the test’s performance using p16 positivity (≥70%) in surgical specimens as a reference test.²⁶

RESULTS

Clinicopathologic Characteristics

A total of 172 HNSCC patients with cervical lymph node aspirate and corresponding surgical resection or biopsy specimens were identified. Eighty-four cases had a CB available, 42 of which had sufficient cellularity and integrity and therefore were included in the study. The clinical and pathologic findings are summarized in Table 1.

TABLE 1.

Clinicopathologic Characteristics of Patients and Tumors

	p16 Positive (n = 26)	p16 Negative (n = 16)
Age, median (range), y	59 (35–72)	63 (40–80)
Sex, No. (%)
Male	24 (92)	12 (75)
Female	2 (8)	4 (25)
Tumor site, No. (%)
Oropharynx	25 (96)	2 (12)
Larynx	0 (0)	5 (32)
Hypopharynx	0 (0)	3 (19)
Oral cavity	0 (0)	3 (19)
Tongue	0 (0)	2 (12)
Unknown	1 (4)	1 (6)
Cellblock quality, No. (%)
Clusters	7 (27)	6 (38)
Single cells	3 (11)	1 (6)
Clusters and single cells	16 (62)	9 (56)

p16 Immunostaining

The CB sections showed both tumor clusters and single tumor cells, tumor clusters only, and single tumor cells only in 25 (59%), 13 (31%), and 4 (10%) cases, respectively (Table 1). CB cellularity ranged from 10 to >2000 cells in clusters and from 20 to >2000 cells in single cells. The number of clusters ranged from 1 to >50 clusters and the largest cluster size ranged from 10 to >1000 cells. All SCCs that were characterized as p16 negative on surgical specimens, were also negative for p16 on cytology CBs (complete absence of staining, n = 16). In contrast, corresponding cytology CBs of surgical specimens with p16-positive SCC showed moderate to strong p16 expression ranging from 40% to 100% in tumor clusters. Overall scoring of p16 in tumor cells clusters (n = 23) was not significantly different between the 2 observers (ICC, 0.982; 95% CI, 0.967–0.991) and differences were resolved by joint review. In contrast, p16 staining in single cells (n = 19) was more heterogeneous and scoring was challenging; although cases with single cells were selected on the basis of their viability, we encountered difficulty in scoring in the midst of cellular degeneration and poor preservation (Fig. 1). In cases with both clusters and single tumor cells (n = 16), only 1 case showed the same extent and intensity of staining in clusters and single cells, though staining was more heterogeneous, relatively lower, and weaker in single cells compared with tumor clusters in the remaining 15 cases (Fig. 2). Estimated staining of single tumor cells ranged from approximately 1% to 5% to around 70% (moderate intensity), but accurate scoring of p16 expression in single tumor cells had significant interobserver variability (ICC, 0.467; 95% CI, 0.132–0.707), which was difficult to resolve by joint review. The analysis was therefore performed on cases with tumor clusters (n = 38). The extent of p16 staining increased with the total number of tumor cells in clusters (predicted logit of p16 percentage = −0.7882 + 0.00133 * (number of cells); P = .0465) and largest cluster size (predicted logit of p16 percentage = −1.0123 + 0.00474 * (largest cluster size); P = .0255), whereas the extent of p16 staining and the number of clusters on a slide were not associated (predicted logit of p16 percentage = −0.117566 + 0.02122 * (number of clusters); P = .34).

Figure 1.

(A,B) Fine-needle aspiration from a right neck mass in a patient with oropharyngeal squamous cell carcinoma (cellblock). (A) Single tumor cells with inflammatory cells, degenerated cells, and cellular debris. No tumor clusters present (H & E, original magnification ×400). (B) Corresponding p16 immunohistochemistry with faint scattered staining (×400). (C,D) Follow-up neck biopsy shows (C) nonkeratinizing squamous cell carcinoma (H & E, original magnification ×200) and (D) corresponding p16 immunohistochemistry (×200).

Figure 2.

(A-D) Fine-needle aspiration from a cervical lymph node in a patient with metastatic squamous cell carcinoma (cellblock). (A) Single cells and cellular debris (H & E, original magnification ×400). (B) Corresponding p16 immunohistochemistry (×400). (C) Preserved tumor cluster (H & E, original magnification ×400). (D) Corresponding p16 immunohistochemistry (×400). (E,F) Follow-up tonsillar biopsy shows (E) moderately differentiated squamous cell carcinoma (H & E, original magnification ×200) and (F) corresponding p16 immunohistochemistry (×200).

In cases with tumor clusters, 20 out of 38 cases (53%) showed moderate to strong nuclear and cytoplasmic p16 staining in at least 70% of tumor cells in clusters. Applying this threshold for p16 positivity in cytology specimens, the concordance rate with surgical specimens was 92% (35 of 38), with a sensitivity at 87% and specificity at 100%. An ROC curve analysis was performed, yielding an ROC–area under the curve of 1. Inspection of the ROC curve showed the optimal threshold to be at least 40% p16 staining (moderate staining), with a concordance rate of 100% with p16 on tissue specimens and a sensitivity, specificity, positive predictive value, and negative predictive value at 100% (Table 2). At 40% threshold, there was a significant association with an oropharyngeal origin; p16 was positive in 22 out of 24 OPSCCs (92%) and negative in all 12 non-OPSCCs.

TABLE 2.

Performance of p16 Immunohistochemistry in Cytology Cellblock in Comparison to p16 Immunohistochemistry in Surgical Specimens

Thresholda (%)	Concordance Rate (%)	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
70	92	87	100	100	83
60	95	91	100	100	88
50	97	96	100	100	94
40	100	100	100	100	100

Abbreviations: NPV, negative predictive value; PPV, positive predictive value.

^aPercentage of tumor cells positive for p16 by immunohistochemistry in cytology cellblock specimens.

High-Risk HPV Status

Of 38 cases with tumor clusters, HPV testing results were available in 30 cytology cases. Nineteen (63%) FNA samples were positive and 11 (37%) were negative for HPV. All p16 negative on FNA samples were negative for HPV (n = 10). Of cases with p16 IHC ≥40% (n = 20), 19 (95%) were HPV positive and 1 (5%) was HPV negative (Table 3). A repeat HPV test on the corresponding surgical specimen in the latter case was negative, suggesting a p16-positive, HPV-unrelated SCC (unknown primary). There was a 97% overall agreement between p16 at a threshold of 40% on cytology CB material and HPV by PCR on needle rinse and a 100% concordance rate when the analysis was performed only on patients with OPSCC.

TABLE 3.

p16 Immunohistochemistry in Relation to HPV Status in Cytology Specimens by Tumor Site

	p16 ≥ 40% (n = 20)			p16 < 40% (n = 10)
	OPSCC	Non-OPSCC	Unknown	OPSCC	Non-OPSCC	Unknown
HPV positive (n = 19)	19	0	0	0	0	0
HPV negative (n = 11)	0	0	1	2	8	0

Abbreviations: HPV, human papillomavirus; OPSCC, oropharyngeal squamous cell carcinoma.

To determine whether HPV by PCR could be used as a substitute for p16 IHC, we correlated HPV status in 107 FNA-needle rinse fluids with unavailable/inadequate CB material and available HPV results with p16 IHC on corresponding surgical specimens. Three samples had an indeterminant HPV result and were excluded. Of the remaining 104 samples, 39 (38%) were p16 negative and 65 (62%) were p16 positive. All surgical samples with negative p16 IHC were negative for HPV on FNA samples (n = 39). Of 65 surgical specimens with positive p16 IHC, 57 (88%) were HPV positive (52 OPSCCs) and 8 (12%) were HPV negative (6 OPSCCs) on FNA samples. A repeat HPV test on surgical specimens in the latter cases was positive in 5 cases, suggesting a false-negative HPV result in the FNA specimen and negative in 3 cases, suggesting a p16-positive/HPV-unrelated SCC. The sensitivity and specificity of HPV PCR in determining p16 IHC status in surgical specimens were 88% and 100%, respectively. In 37 cases with available HPV in cytology and surgical specimens, there were 5 false negatives and no false positives, yielding a concordance rate of 86%.

DISCUSSION

p16 overexpression is a powerful marker of favorable outcome in patients with OPSCC and predicts response to treatment.^12,27 p16 IHC is therefore often requested as part of the initial workup, often on cytology specimens collected from neck metastasis. Furthermore, up to one-third of HPV-related metastatic HNSCCs elude detection of the primary site despite a thorough clinical, radiologic, and pathologic evaluation,²⁸ making cytology specimens from metastatic HNSCC the only sample available for p16 evaluation. In the quest to establish criteria for p16 positivity in cytology specimens, several studies have used p16 staining in tissue specimens as the gold standard for comparison^22,23,29,30 and/or HPV ISH in some studies.^23,31–34 Most have indicated a high false-negative rate when applying the CAP recommended 70% threshold in surgical specimens and have suggested a lower threshold to improve sensitivity.^22,23,32 These studies have proposed any staining in tumor cells to a cutoff of 50%.^{22,23,29,31,32,34,35}Depending on where the authors have set their cutoff point for test positivity, the sensitivity of the test has ranged between 74% and 100%, and the specificity between 52% and 100% (Table 4).^{22,23,29–34} In our study, applying the threshold of 70% positive staining resulted in a high number of false-negative results, providing a sensitivity of 87% and specificity of 100%. These results are similar to other studies and provide further evidence that a lower threshold needs to be defined in cytology specimens. In a series of HPV-related HNSCC, p16 in cytology specimens demonstrated heterogeneous nuclear and cytoplasmic staining with a threshold as low as 1% leading to a sensitivity of 93%.³² Similarly, in a study by Holmes et al in which CBs comprised a significant proportion of the samples,³⁵ there was a 98% correlation between p16 IHC and HPV ISH on preoperative lymph node samples and subsequent surgical specimens, considering any cytoplasmic and nuclear staining observed, even in the presence of cellular degradation.

TABLE 4.

Comparison of Studies That Investigated the Use of p16 Immunohistochemistry in Cytology Specimens

	No. of Cases	Cytologic Preparation	Tumor Cellularity	Clusters/Single Cells Scored	Positivity Criteria	Reference Test	P16 Performance
							Sensitivity (%)	Specificity (%)
Begum 200731	77	CB	Any	Both	Any staining	HPV16-ISH in cytology specimens	92	94
Jannapureddy 201034	40	CB	Adequate material, NOS	Both	Any nuclear and cytoplasmic staining	HPV-ISH in cytology specimens	100	77
Jakscha 201330	25	Smears	Adequate material, NOS	Both	Strong cytoplasmic staining in ≥5% of tumor cells	P16 IHC in surgical specimens	88	100
Jalaly 201529	48	CB	Any	Clusters only	Strong staining in ≥15% of tumor cells	P16 IHC in surgical specimens	97	100
						HPV-ISH in cytology specimens	90	100
Xu 201623	60	CB, smear, ThinPrep slide	Any	Both	Moderate-strong staining in ≥1 % of tumor cells	P16 IHC in surgical specimens	95	100
					Moderate-strong staining in ≥10% of tumor cells	HPV-ISH in cytology specimens	94	75
Hou 201633	73	CB	At least 100 cells	Both	Strong nuclear and cytoplasmic staining in ≥70% of tumor cells	HPV-ISH in cytology specimens	98	52
Wong 201932	97	CB	Any	Both	Any staining	HPV-ISH/PCR in cytology CBs	93	92
Yang 201922	50	Smear	Not reported	Both	Weak nuclear and cytoplasmic staining ≥50% of tumor cells	P16 IHC in surgical specimens	74	100
Current study	42	CB	At least 10 cells	Clusters	Moderate nuclear and cytoplasmic staining in 40% of tumor cells (at least)	P16 IHC in surgical specimens	100	100

Abbreviations: CB, cellblock; HPV, human papillomavirus; IHC, immunohistochemistry; ISH, in situ hybridization.

p16 intratumoral heterogeneity in SCC is well-described and illustrated by multiple studies, suggesting that p16 staining in <70% of tumor cells in surgical specimens was also associated with HPV.^7,36,37 However, multiple guidelines have recommended that a positive p16 interpretation be based on moderate to strong p16 staining in 70% or more of tumor cells in surgical specimens; <50% staining should be considered negative. In equivocal cases (<70% but >50% moderate to strong staining or diffuse weak staining), HPV-specific testing is recommended.¹⁸ In tumors classified as negative per those recommendations, however, p16 staining is not always totally absent (ie, moderate to strong staining <50%); areas with staining that do not meet the criteria for positivity, both in extent and intensity, are sometimes present. This finding has significant implications in cytology specimens where the needle randomly samples the lesion without being targeted to regions with the highest p16 staining, suggesting caution in classifying any p16 staining on cytology specimens as a positive result. A threshold that is too low would increase sensitivity, though at the expense of increasing the false-positive rate.²³ Two studies have proposed a threshold in the range of 10% to 15% with excellent concordance rate with surgical specimens.^23,29 The study by Xu et al included cases with single cells and assessed p16 IHC in cytology specimens against p16 in surgical specimens and HPV ISH in cytology. Interestingly, though this study showed the best sensitivity, specificity, and concordance rate with p16 IHC on surgical specimens when a 1% cutoff was applied, increasing the cutoff to 10% achieved a better correlation with HPV ISH, a threshold recommended to avoid a potential false positive as well. In contrast, Jalaly et al scored only clusters of tumor cells and found a concordance rate of 98% with surgical specimens when at least 15% of tumor cells in CBs were positive for p16 IHC.²⁹ The inclusion of single and even degenerated cells for p16 scoring in some studies,^23,32,35 probably accounts for the lower threshold reported. In our study, we have at-tempted to score single tumor cells. Akin to those studies, when p16 IHC staining was evaluated separately in single cells and tumor clusters, a lower threshold, estimated at 1% to 5% (moderate intensity) p16 staining in single cells, showed a 100% concordance rate with surgical specimens. However, we found accurate scoring of single tumor cells to be challenging with very poor interobserver agreement; distinguishing basaloid-type SCC single cells that have metastasized to cervical lymph nodes from lymphocytes, degenerated cells, and cystic background on a p16 immunostain slide, proved challenging in many cases, precluding accurate assessment (Fig. 1). Scoring tumor clusters showed excellent interobserver agreement, with a 100% concordance rate and 100% sensitivity and specificity for p16 in tissue specimens when considering a threshold of at least 40% as a positive staining for p16 in CBs. This percentage increased with cellularity and size of tumor clusters. Furthermore, there was 97% agreement with HPV PCR in HNSCC overall and 100% correlation in OPSCC, confirming p16 IHC with at least 40% staining as a reliable surrogate marker for the presence of HPV.

Other studies have reported higher thresholds as well. Such studies included p16 IHC on smears, a less well-described method to assess HPV or its surrogate marker in cytology, which was explored by Yang et al,²² who found the best p16 threshold to be 50% yielding 74% sensitivity and 100% specificity. Using a similar approach and a threshold at only 5%, Jakscha et al reported a concordance rate of 92%; interestingly, the 2 false-positive smears had staining in 10% of tumor cells only and no false negatives. Thus, increasing the threshold to >10%, the authors report a 100% concordance rate.³⁰

The difference in p16 cutoff in CB and intensity of staining across these studies is not entirely clear. The selection of cases and definition of adequacy are neither well-defined nor uniform throughout and have definitely played a role. As mentioned previously, scoring of single tumor cells has almost certainly contributed to a lower threshold in the corresponding studies.^23,31,32 Moreover, the fixation technique used in cytologic collections has probably conferred a significant impact on p16 expression. The results reported by Buonocore et al³⁸ and several other studies^22,23,29,31 suggest that alcohol-fixed cytology specimens (CBs and smears) displayed a weaker and in-consistent p16 expression^22,23,29 compared with strong and diffuse staining in formalin-fixed material.^31,38 It is difficult to confirm this finding with certainty because all CBs in our study were initially fixed in alcohol.

Although p16 threshold in cytology material remains unsettled, the one area of broad agreement across studies is the false-negative rate, attributed mostly to suboptimal CBs caused by scant cellularity.^23,31,32 Additionally, in cases lacking tumor clusters, accurate quantitative analysis can be challenging when dealing with scattered preserved cells, particularly in HPV-related metastasis where cystic degeneration and low cellularity are common findings.³¹ Those facts have created few drawbacks in this study because CBs were often unavailable or paucicellular with rare viable tumor cells with decreased staining. In those cases, liquid-phase PCR assays might be helpful in determining HPV status in residual FNA rinse, sidestepping the prerequisite adequate cellularity of CBs. Although DNA PCR has been advocated for its high sensitivity, particularly in OPSCC, concerns have been raised that it may detect a transcriptionally silent virus instead of one that is acting as a driver of malignant transformation. However, the quantitative real-time PCR approach, which can currently measure viral load, allows this distinction; studies have shown that tumors with a high viral load are more likely to express E6/E7 mRNA and be driven by HPV.^13,40 Additionally, it has shown excellent correlation with p16 and HPV ISH in tissue from HNSCC, even when cellular integrity is not preserved and only necrotic debris are present.^24,39,40,41 Our study is in line with those findings. HPV PCR has demonstrated excellent specificity and good sensitivity for p16 IHC and has shown 86% concordance rate with HPV status on surgical specimens, suggesting that a positive HPV result by PCR is an excellent marker for positive p16 IHC and HPV status in tissue specimens. Because of the retrospective nature of our study, however, HPV data were not available for all cases. Another limitation in our series was the lowest threshold of 40% p16 positivity we encountered in tumor clusters. Although p16 positivity threshold in at least 40% of tumor correlates with p16 in tissue specimens, there were no cases with p16 staining between 1% and 40% in tumor clusters. It is unknown whether a lower threshold would have yielded the same results.

In conclusion, standardization of p16 IHC in cytology specimens is still in the works. Studies with a larger number of patients are needed before p16 IHC is integrated in cytology practice. Meanwhile, our study supports suggestions that the cutoff for positive p16 staining in cytology specimens is lower compared with tissue specimens. We found p16 staining in at least 40% of tumor clusters to yield a sensitivity and specificity at 100% and to confidently predict p16 positivity in surgical specimens. Accurate scoring of single tumor cells proved more challenging. In those cases, HPV PCR on needle rinse might offer an alternative for p16 IHC. However, in negative HPV results, a repeat p16 IHC should be performed on tissue specimens when available to avoid false-negative results.