Determination of high-risk human papillomavirus (HPV) status of head and neck squamous cell carcinoma using the Roche Cobas HPV test on cytologic specimens and acellular supernatant

Abstract

Background:

High-risk human papillomavirus-positive (hrHPV+) oropharyngeal squamous cell carcinomas comprise a subset of head and neck squamous cell carcinomas (HNSCC) with distinct biology and prognosis. Commonly, the diagnosis of HNSCC is rendered on fine needle aspiration (FNA). As cell blocks may be insufficient for determining HPV status using microscopy-based techniques, the ability of liquid-based assays was pursued.

Methods:

The performance of the Roche Cobas 4800 platform was evaluated on the FNA material from the cell pellet and corresponding cell-free supernatants of primary and metastatic HNSCCs. These results were compared to the p16 immunostain result from the histologic material of the same patient. Discrepant cases were adjudicated by hrHPV ribonucleic acid (RNA) in situ hybridization (ISH).

Results:

Forty-one samples (23 primary and 18 nodal metastases) were acquired from 34 patients with HNSCC. Primaries included: oropharynx (n=20), oral cavity (n=13), larynx (n=3), and skin (n=3). In two cases, a primary could not be identified. Twenty-three (56%) were p16+ by immunohistochemistry. 22 were positive on Cobas hrHPV testing from both cell pellet and cell-free supernatant. Two cell-free supernatants yielded indeterminate Cobas results. Upon additional hrHPV RNA ISH analysis, one Cobas-positive cell pellet was deemed false positive. The sensitivity of the Cobas assay was 100% for pellet material and cell-free supernatant, with respective specificities of 94.7% and 100%.

Conclusions:

Cobas hrHPV testing of HNSCC showed high concordance with p16 immunohistochemistry on the corresponding cell block/tissue specimen. Utilizing the cell-free supernatant in this platform could provide accurate HPV results, while conserving material for cytomorphologic analyses.

Precis for use in the Table of Contents:

The Roche Cobas HPV test is highly sensitive and specific for determining high-risk HPV status of head and neck squamous cell carcinomas on both cellular cytologic specimens and their corresponding cell-free supernatants.

Background

High-risk human papillomavirus-positive (hrHPV+) oropharyngeal squamous cell carcinomas (OPSCC) are recognized as a distinct biologic type of head and neck SCC (HNSCC). In the United States and parts of Europe, up to 70 to 80% of OPSCC are hrHPV+. In contrast to the declining rates of conventional HNSCC, the incidence of hrHPV+ OPSCC has risen dramatically over the past few decades.^1,2 Their associated patient demographics and clinical behavior further distinguish them from non-HPV-related HNSCC. HrHPV+ OPSCC typically present in Caucasian men of higher socioeconomic status, are associated with sexual risk factors more than with smoking or alcohol, are more sensitive to treatment, and generally have more favorable prognosis.^{1, 2, 3, 4} The most recent (8th) edition of the American Joint Committee Cancer Staging Manual recognizes this distinction with separate staging systems, which are, in turn, reflected in the most recent College of American Pathologists (CAP) protocol for pharyngeal cancers.^{5, 6, 7, 8} In addition to ensuring appropriate staging, accurate identification of hrHPV+ OPSCC is important because of the evolving approach to treating these tumors. Their generally good prognosis and chemoradiosensitivity has led to explorations of reduced-intensity treatment in an effort to decrease toxicity.^{9, 10} Novel biomarkers and the potential roles of targeted therapies and immunotherapies are also under investigation.^{11, 12}

The recent CAP guidelines strongly recommend high-risk HPV testing at the time of diagnosis of OPSCC, regardless of the histologic subtype. This may be performed on the primary tumor or on a regional lymph node metastasis when the clinical or imaging findings are consistent with an oropharyngeal primary. Testing is also recommended for patients with unknown primary tumor site who present with metastasis in a cervical level II or III lymph node or cervical lymph node of unknown level. In tissue specimens, hrHPV status can be predicted by p16 immunohistochemistry, which is a prognostic marker and surrogate marker of high-risk HPV. Depending on the site of disease, tumor morphology, degree of staining, and clinician preference, immunohistochemical analysis may be followed by HPV-specific testing. The algorithm also endorses assessment of high-risk HPV on fine needle aspirations (FNAs) of the head and neck from patients with known OPSCC not previously tested, from patients with suspected OPSCC, or from those with metastatic SCC of unknown primary. These guidelines do not, however, recommend one specific testing modality for cytology specimens.⁵ Several studies have evaluated the reliability of p16 staining on FNA cell blocks. Although there is generally good concordance between p16 expression on cell blocks and biopsy/resection specimens, the often scant and degenerated nature of the cell block material can lead to false negative rates as high as 21% and sensitivities below 70% if the stringent 70% nuclear and cytoplasmic staining cutoff used for tissue is applied to cell block p16 staining.^{13, 14} Some studies suggest that any nuclear and cytoplasmic p16 positivity in cell blocks should be considered a positive result.¹⁵

Another option for detecting HPV in cytologic samples from patients with HNSCC is in situ hybridization (ISH). While it is a highly specific test that affords direct visualization of the hybrid signal in the nuclei of tumor cells, it also requires construction of a formalin-fixed paraffin embedded cell block which again may be suboptimal due to low cellularity and cell degradation. In the experience of Miller et al, HPV DNA ISH staining in small biopsies and cell blocks is frequently very focal and weak, making interpretation difficult.¹⁶ Extensive validation of HPV DNA ISH on cell block material is imperative in order to ensure accurate interpretation of results. RNA ISH is emerging as an excellent test on tissue sections and cell blocks due to signal amplification methods able to detect fragmented RNA in the specimens. Cocktail RNA ISH for most high-risk HPV types works well, but is expensive and not widely available in clinical practice.

Recently, non-tissue targeted approaches for HPV testing using the liquid-based assays developed for gynecological specimens.¹⁷ have been pursued. The Qiagen Hybrid Capture II [Qiagen, Gaithersburg, MD], Cervista HR HPV and Cervista HPV 16/18 [Hologic, Inc.], PCR-based Roche Cobas HPV Test [Roche Molecular Systems, Pleasanton, CA], and Aptima HPV Assay [Gen-Probe, Inc., San Diego, CA] have all been shown to have over 70–100% concordance when compared to HPV ISH and/or p16 as assessed on the corresponding histologic specimen.^{18, 19, 20, 21, 22, 23, 24, 25} These platforms are attractive alternatives for HNSCC fine needle aspirates because they are already in widespread use; many labs that process FNA samples commonly perform HR-HPV testing on cervical cytology specimens. Furthermore, these tests are commercially available, cost-effective, automated, quickly and easily performed with minimal specimen preparation, allow for specimen storage for modest periods of time, and provide dichotomous results requiring minimal interpretation.^{19, 26, 27} Nevertheless, while the basic analytic characteristics of these platforms are well-established (a strong advantage), none of them are currently approved by the Food and Drug Administration for use on non-cervical specimens. As a result, a diagnostic laboratory seeking to implement hrHPV testing for HNSCC specimens must locally evaluate/validate a commercial technology for this specimen type, in order to meet federal regulatory requirements.

As FNA samples often have limited cellularity, additional passes may be necessary to obtain adequately cellular material for hrHPV testing. Even with extra needle passes, cell block preparations often have scant or no cellularity. However, specimen processing essentially always produces supernatant, which, though typically discarded, represents a potentially valuable resource for cellular DNA and RNA. This is due to cell shearing during cytocentrifugation, which results in the release of DNA, RNA, and protein. In this study, we evaluated the performance of the Roche Cobas HPV assay on the cell pellet material obtained from needle passes of HNSCC specimens, as well as on the corresponding cell-free supernatants of these samples, and compared these results to the p16 IHC from corresponding cell block or tissue specimens. In cases where the cell pellet, supernatant, and p16 results were discordant, HPV status was adjudicated by cocktail high risk HPV RNA ISH.

Materials and methods

Specimen Collection

Following approval from The Vanderbilt University Medical Center Human Research Protection Program, cytologic material was prospectively collected via fine needle aspiration of either a palpable lymph node suspicious for metastatic HNSCC (n=4) or from surgical resections of patients with a history of a either a primary HNSCC or cutaneous SCC from the head and neck (n=37). The latter were identified from the VUMC operating room schedule and their clinical histories confirmed via review of the electronic medical record. FNAs were conducted by one of two experienced cytopathologists (KE and VW). Using a 25-gauge needle, up to 3 passes were made and the entire needle contents transferred into 20 mls of PreservCyte solution (Hologic, Marlborough, Mass).

Cobas HPV Testing

For each FNA, 10 mL of the above PreservCyte suspension were set aside for molecular HPV testing. FNA samples underwent centrifugation, wherein the resultant supernatant was separated from the cell pellet. The latter was resuspended in 10 mL of fresh PreservCyte. Both the resuspended cell pellet and cell-free supernatant were analyzed by the Cobas HPV test on the Roche robotic platform using the standard protocol for cervical specimens (Figure 1). The methodologic and analytic performance of this assay have been well-described. In brief, the platform utilizes multiplex real-time PCR to independently detect (without quantification) HPV type 16, HPV type 18, and combined other high-risk HPV types (31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68). In parallel, human beta-globin is detected as an internal control (IC), both to assess the presence of patient-derived material within the specimen and to ensure appropriate amplification conditions. Study specimens that initially demonstrated indeterminate results (indicating IC failure) were subjected to one attempted repeat analysis.

Figure 1.

Workflow for Cobas Testing

Immunohistochemistry

Immunohistochemistry for p16 was evaluated on formalin-fixed, paraffin embedded tissues. Briefly, 3 to 5 micron sections were cut, deparaffinized, and subjected to antigen retrieval consisting of 20 minutes in the Leica Epitope Retrieval Solution. Expression was assessed using the p16 mouse monoclonal antibody (E6H4 clone, Ventana Medical Systems, Tucson, Arizona) on a Leica Bond autostainer (Leica Biosystems, Inc). A positive interpretation was rendered when strong, diffuse nuclear and cytoplasmic staining was present in greater than 70% of tumor cells on both histologic samples and cell blocks. P16 immunohistochemistry was performed on cell blocks (n=2) or from the corresponding tissue sections (n=21), if it had not been determined on prior material (n=17). P16 IHC for those cases not previously stained was interpreted by one pathologist (KE) and one pathology trainee (MFW). In cases where the Cobas and p16 result were discordant, HR HPV RNA ISH was performed.

In situ hybridization

In situ hybridization for high risk HPV E6/E7 mRNA was performed on discrepant cases using the RNAscope HPV kit (Advanced Cell Diagnostics, Inc., Hayward, CA), according to the manufacturer’s instructions. Probes covered HPV types 16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 68, 73, and 82. Each case included appropriately-staining control probes for the bacterial gene DapB (negative control) and for the housekeeping gene ubiquitin C (positive control evidence of adequate RNA). The slides were classified as either positive or negative, with the former exhibiting brown granular cytoplasmic and/or nuclear staining of greater intensity than that in the DapB negative control slide. HPV RNA ISH results were interpreted by two pathologists (KE and VW) and one pathology trainee (MFW).

Statistical analysis

Sensitivity and specificity were calculated using p16 IHC performed on the prior or subsequent corresponding surgical resection or, in the absence of a surgical resection specimen, the cell block as the gold standard. Confidence intervals (CI) were calculated using the Wilson method. For cases in which Cobas HPV testing and p16 were discrepant, ISH was performed for adjudication. Patients were considered hrHPV positive when both p16 IHC and Cobas results (on both cell pellet and supernatant) were positive, or, in discrepant cases, when both Cobas results (on both cell pellet and supernatant) and HPV RNA ISH were positive. Specimens that repeatedly yielded indeterminate results due to IC failure were excluded from the respective analyses.

Results

FNA samples were acquired from 41 sites of 34 different patients over a period of 11 months. There were 23 men and 11 women, with an average age of 63.6 years (range: 38–82). Thirty-seven of the specimens were obtained from surgical resections, and four were obtained from clinical fine needle aspirates. Primary sites consisted of the oropharynx (n=20), including base of tongue (n=11) and tonsil (n=9), facial skin (n=3), oral cavity (n=13), and larynx (n=3). There were two patients who presented with a lymph node metastasis of unknown primary. Aspiration material was obtained from the primary tumor (n=23) or a lymph node metastasis (n=18) (Table 1).

Table 1.

Patient demographics and tumor sites

Age	63.6 (38–82)
Sex
Male	23 (67.6%)
Female	11 (32.4%)
Primary tumor site
Oral Cavity	13 (31.7%)
Oropharynx	20 (48.8%)
Larynx	3 (7.3%)
Skin	3 (7.3%)
Unknown	2 (4.9%)
FNA site
Primary tumor	23 (56.1%)
Lymph node	18 (43.9%)

Twenty-three of 41 specimens (19/20 primary or metastatic OPSCCs, 3/19 primary or metastatic non-OPSCCs, and 1/2 metastases from unknown primary site) were p16+ by IHC, and 22 (95.7%) of these specimens had correspondingly positive Cobas hrHPV results on both cell pellet material and cell-free supernatant. Of 23 hrHPV+ cytologic specimens, 19 were positive for HPV16 and four were positive for an hrHPV strain other than HPV16 or HPV18. None of the samples was positive for HPV18. One cutaneous SCC metastasis that was p16+ by IHC was hrHPV negative on both cell pellets and supernatant; follow-up HPV RNA ISH was negative. This result represents a false positive for p16 IHC and a true negative by Cobas on both cell pellet material and supernatant.

The remaining 18 specimens were p16-negative by IHC, and 15 of these specimens had correspondingly negative cobas hrHPV results on both cell pellets and cell-free supernatant. One oral cavity SCC was hrHPV+ (HPV16) on cell pellet but negative on supernatant; follow-up HPV RNA ISH was negative, classifying it as a false positive for cell pellet material and a true negative for supernatant and p16 by IHC. Two specimens, a primary laryngeal SCC and an intraparotid lymph node non-OPSCC metastasis, had negative hrHPV results on cell pellets, but testing on the supernatants yielded indeterminate results due to IC failure (Table 2). In theory, these results could be due to either the presence of PCR inhibitors within those supernatant-specimens or the insufficient presence of patient DNA (assessed via the beta-globin target). Given the valid, negative result of each corresponding pellet fraction, however, the latter interpretation is more likely.

Table 2.

Classification by origin, FNA site, and HPV status

Patient	1° tumor site	FNA site	Cobas result on cell pellet	Cobas result on cell-free supernatant	p16 IHC	HPV RNA ISH	Result (TP, TN, FP, FN)
1	tonsil	tonsil	HPV16	HPV16	Pos		TP
2	oral cavity	oral cavity	HPV16	HPV16	Pos		TP
3	tonsil	tonsil	HPV16	HPV16	Pos		TP
4	tonsil	lymph node (II-IV)	HPV16	HPV16	Pos		TP
5	base of tongue	lymph node (II-IV)	HPV16	HPV16	Pos		TP
6	base of tongue	base of tongue	HPV16	HPV16	Pos		TP
7	tonsil	tonsil	HPV16	HPV16	Pos		TP
8	unknown	lymph node (neck)	Neg	Neg	Neg		TN
9	base of tongue	base of tongue	HPV16	HPV16	Pos		TP
10	base of tongue	lymph node (neck)	HPV16	HPV16	Pos		TP
11	oral cavity	lymph node (neck)	Neg	Neg	Neg		TN
12	oral cavity	lymph node (neck)	Neg	Neg	Neg		TN
13	base of tongue	lymph node (neck)	HPVoth	HPVoth	Pos		TP
14	oral cavity	oral cavity	Neg	Neg	Neg		TN
15	skin	lymph node (parotid)	Neg	Neg	Neg		TN
16	tonsil	pharynx	HPV16	HPV16	Pos		TP
17	tonsil	pharynx	HPV16	HPV16	Pos		TP
18	tonsil	nasopharynx	HPV16	HPV16	Pos		TP
19	tonsil	nasopharynx	HPV16	HPV16	Pos		TP
20	oral cavity	skin	Neg	Neg	Neg		TN
21	oral cavity	lymph node (II-IV)	Neg	Neg	Neg		TN
22	oral cavity	oral cavity	Neg	Neg	Neg		TN
23	oral cavity	lymph node (II-IV)	Neg	Neg	Neg		TN
24	base of tongue	lymph node (II-IV)	HPVoth	HPVoth	Pos		TP
25	base of tongue	base of tongue	HPVoth	HPVoth	Pos		TP
26	unknown	lymph node (II-IV)	HPV16	HPV16	Pos		TP
27	oral cavity	oral cavity	HPV16	Neg	Neg	Neg	FP (cells), TN (supernatant)
28	tonsil	lymph node (II-IV)	HPVoth	HPVoth	Pos		TP
29	oral cavity	lymph node (perifacial)	Neg	Neg	Neg		TN
30	base of tongue	lymph node (neck)	HPV16	HPV16	Pos		TP
31	base of tongue	base of tongue	HPV16	HPV16	Pos		TP
32	larynx	larynx	Neg	IC failure	Neg		TN (cells)
33	oral cavity	oral cavity	Neg	Neg	Neg		TN
34	base of tongue	base of tongue	Neg	Neg	Neg		TN
35	skin	lymph node (parotid)	Neg	IC failure	Neg		TN (cells)
36	oral cavity	oral cavity	Neg	Neg	Neg		TN
37	oral cavity	lymph node (unspecified)	Neg	Neg	Neg		TN
38	larynx	larynx	HPV16	HPV16	Pos		TP
39	larynx	larynx	Neg	Neg	Neg		TN
40	skin	lymph node (IA)	Neg	Neg	Pos	Neg	TN
41	base of tongue	base of tongue	HPV16	HPV16	Pos		TP

Considering p16 IHC, or, in discrepant cases, HPV RNA ISH, as the gold standard, the Roche Cobas assay had a sensitivity (SN) of 100%, a specificity (SP) of 94.7%, a positive predictive value (PPV) of 95.7%, and a negative predictive value (NPV) of 100% when performed on cell pellets. On cell-free supernatant, results showed SN 100%, SP 100%, PPV 100%, and NPV 100%.

In the majority of cases, Cobas on the material from the cell pellet (95.1%, n=41) and supernatant (97.4%, n=39) and p16 IHC correlated in determining hrHPV status (Table 3).

Table 3.

Correlation between Cobas hrHPV and p16 IHC or HPV RNA ISH on cell pellet and cell-free supernatant

Cell pellet
	P16 IHC+ or HPV ISH+	P16 IHC− or HPV ISH−	Total
Cobas hrHPV+	22	1	23
Cobas hrHPV−	0	18	18
Total	22	19	41
Sensitivity = 100% (95% CI 85.1–100%) Specificity = 94.7% (95% CI 75.4–99.1%) Positive predictive value = 95.7% (95% CI 79.0–99.2%) Negative predictive value = 100% (95% CI 82.4–100%)
Cell-free supernatant
	P16 IHC+ or HPV ISH+	P16 IHC− or HPV ISH−	Total
Cobas hrHPV+	22	0	22
Cobas hrHPV−	0	17	17
Total	22	17	39
Two Cobas results had IC failure and were excluded from this table. Sensitivity = 100% (95% CI 85.1–100%) Specificity = 100% (95% CI 81.6–100%) Positive predictive value = 100% (95% CI 85.1–100%) Negative predictive value = 100% (95% CI 81.6–100%)

Discussion

Aspirated cells from a cervical lymph node metastasis are frequently the first material available for the determination of HPV status in OPSCC patients. In this context, tumor HPV status has significant prognostic and treatment implications, and, as positive tumors can be effectively treated non-surgically, the FNA biopsy may be the sole sample on which this testing can be performed. CAP guidelines recommend determination of tumor HPV status only in patients with oropharyngeal primary tumors, because the significance of p16 positivity in non-oropharyngeal HNSCC is unclear. However, given that the primary tumor is not always identifiable and that non-oropharyngeal HNSCC may metastasize to cervical lymph nodes, we included cases of non-oropharyngeal HNSCC and cutaneous SCC in this study. Of note, CAP guidelines also indicate that for patients with metastatic p16+ SCC of unknown primary involving a cervical lymph node of unknown level or metastatic p16+ keratinizing SCC of unknown primary involving a level II or III lymph node, follow-up HPV-specific testing should be performed for final determination of tumor HPV status.⁸

While there are several different modalities for detecting hrHPV in HNSCC, a universally-accepted technique has yet to be defined for cytologic preparations. Of the conventional methods for assessing hrHPV, p16 IHC, ISH for hrHPV DNA, and ISH for the E6 an E7 messenger RNA (mRNA) transcripts typically require creation of a cell block, which may yield unreliable results if paucicellular, or for which there may be insufficient material. Implementation of the available liquid phase methods already validated for cervical neoplasia screening to detect hrHPV in HNSCCs would circumvent the need for a cell block. Such testing has been conducted on a variety of specimen types including dedicated passes, needle rinses, cells ‘lifted’ from alcohol fixed smears, brushings, and scrapings from surgical specimens.^{18, 19, 20, 21, 23, 24, 25}

Guo et al applied the Cervista (HPV HR and HPV 16/18) assays to cells removed via scalpel blade from alcohol fixed Papanicolaou (Pap) stained smears of HNSCCs. The results from these assays were compared to the results from HPV ISH and p16 immunostaining in the corresponding carcinoma specimens. With their labor-intensive cell lifting technique, there was overall agreement with p16/hrHPV ISH in 92% of cases. When compared to p16 alone, the Cervista assays were 91.3% sensitive and 75% specific.²¹ The other study correlating the Cervista assay results (specifically the HPV HR assay) to p16 immunohistochemistry used needle rinses from lymph nodes with metastatic HNSCC. That study showed an overall concordance of 74% compared to p16 IHC as performed on cell blocks, with a sensitivity of 95.5% and specificity of 38.5%.²²

Another signal amplification platform for hrHPV detection is Hybrid Capture 2. Similar to the Cervista HPV HR assay, it does not subtype into HPV16 and HPV 18. Three investigators have reported successful use of HC2 to evaluate the presence of hrHPV in HNSCC. Bishop et al brushed and aspirated the primary tumor and metastatic implants, respectively, of excised HNSCCs. For samples obtained by FNA, second passes were performed to assess adequacy. Extra passes (although the exact number was not indicated) were made if the specimen was of low tumor cellularity. Through these techniques they correctly classified all cytologic preparations when benchmarked against p16 IHC on the corresponding tissue sections (sensitivity and specificity of 100%).²³ Smith et al obtained dedicated passes from patients with cervical lymphadenopathy for HC2 analysis after an initial aspirate was performed to assess overall specimen cellularity. HC2 was concordant with p16 as performed on the resected cancers in 84% of cases (sensitivity of 66.7% and specificity of 100%). The sensitivity was improved to 100% for cases in which the presence of tumor cells was confirmed in the cytologic sample.²⁴ Finally, Chute et al widened the application of the HC2 test to the supernatant of the residual material from the FNA needle rinse vial. HC2 was 72% sensitive and 100% specific for hrHPV when compared to p16 and chromogenic in situ hybridization (CISH) as conducted on histologic material from the primary tumor.²⁵ A limitation of the HC2 assay, particularly relevant in the context of Chute’s study design, is the lack of an internal control to verify specimen adequacy. Therefore, it is not possible to separate a true negative result from a false negative due to an insufficient amount of DNA.²⁸

Han et al used the RNA amplification-based Aptima HR-HPV platform to perform hrHPV testing on cell harvests from DiffQuik-stained aspirate smeared slides obtained from FNA samples of HNSCC cervical lymph node metastases and compared the results to p16 immunostaining of the corresponding resection specimens. For each of the aspirates, areas containing approximately 200 cells were identified, scraped off, and resuspended before testing. The number 200 was chosen because a pilot study demonstrated positivity at a threshold of 100 cells. They obtained a sensitivity of 87.5% and a specificity of 100%.¹⁸ The potentially low cellularity of these FNA specimens is one of the primary reasons for identifying alternatives to p16 IHC for HPV testing, so a platform requiring more than 100 cells may be of somewhat limited value for many labs. Additionally, Aptima relies on RNA amplification, which, similarly to HC2, makes it impossible to distinguish true negative from false negative results.

Finally, two groups have assessed the PCR-based Roche Cobas HPV test on aspirate samples from HNSCC. Kerr et al (sensitivity of 100% and specificity of 86%) used material obtained from both needle rinses and cells scraped from surgical specimens at the time of intraoperative frozen section consultation, while Baldassarri et al (100% sensitivity and specificity) tested the residual cellular FNA material remaining after direct smear preparation. Both reported good overall concordance when compared to standard methods. Corresponding testing with ISH was concordant in 91% and 100% of cases, respectively.^{19, 20}

Our study employed the Roche Cobas platform because it is already in use in our laboratory for gynecologic Pap samples. Furthermore, it includes beta-globin to ensure the presence of DNA content within the specimen and affords identification of HPV16 and HPV18 as well as twelve other high-risk HPV amplicons. The Cobas results from material derived from both the cell pellets and the cell-free supernatants strongly correlated with the gold standard, showing sensitivity, specificity, PPV, and NPV over 94%. To the best of our knowledge, we are the first to compare the results of the Cobas method as performed on cell-free supernatant of FNAs from HNSCCs with the HPV status as determined by p16 IHC, and, in discrepant cases, HPV RNA ISH, on the corresponding surgical specimens. These data support the earlier work of Kerr et al and Baldassarri et al and additionally indicate that the cell-free supernatant, which is typically decanted and discarded, is a valuable source of material for the determination of tumor HPV status by the Cobas test.^{19, 20} In practice, this method of testing could conserve the entirety of the cell pellet for microscopic analysis, potentially sparing patients additional FNA passes and enabling faster, more definitive diagnosis for scant FNA specimens.

Despite strong correlation between IHC and Cobas assay results, there were two cases in which the p16 and Cobas assay results were discrepant and were subsequently adjudicated by HPV RNA ISH. One was a p16-negative oral cavity SCC, which was positive for HPV16 on Cobas testing of the cell pellet but negative on the supernatant and by HPV RNA ISH. This represents a false positive result for the cell pellet material, which is a known risk with the Cobas assay. Such highly-sensitive PCR-based platforms can detect contaminating viral DNA from another specimen as well as from HPV infections that are clinically insignificant (i.e., so-called “passenger” HPV).²⁶ However, it is worth noting that oral cavity lesions would not be aspirated in practice. The remaining discordant case was of a cutaneous SCC metastasis which was p16+ by IHC but hrHPV-negative on cell pellet, supernatant and HPV RNA ISH. Approximately 30–40% of primary cutaneous SCCs are immunohistochemically positive for p16 but will lack hrHPV upon further testing.^{25, 29} This is important, because similarly to OPSCC, cutaneous squamous cell carcinomas of the head and neck are known to be capable of metastasizing to lymph node levels II-IV.³⁰ Though a cutaneous primary will likely be recognized, in the event of an occult cutaneous primary tumor with cervical nodal metastases, p16 IHC may be misleading, and Cobas testing could yield a more accurate result.

Two specimens yielded negative results on Cobas testing of the cell pellet material but “invalid” results on testing of supernatant alone. The internal control for the Cobas assay is human beta-globin, which is intrinsically present if there is adequate DNA in the sample. An “invalid” read occurs when no HPV target is detected, but neither is the beta-globin signal. This can occur in the setting of relatively low DNA content, perhaps from a low-cellularity FNA needle stick, and/or in the presence of a PCR inhibitor such as gross blood. This Cobas assay design requiring a beta-globin signal reduces the risk of a false negative, conferring an important advantage with respect to HC2, the other platform that has been tested on cell-free supernatant. However, it does not eliminate the risk of a specimen failing to obtain a result, potentially from sampling error or from interfering substances. Although beta-globin ensures the presence of cellular DNA content within the sample, there is no inherent control for the presence of tumor cells in the original FNA specimen, so whether a negative result is due to sampling error may not be readily apparent.²⁷ Lastly, an important limitation of this study is that the majority of our FNA samples were collected on surgical resection specimens rather than in vivo, which likely improved operator accuracy, increased FNA cellularity (and thus DNA content), and decreased the chances of sampling error. As a result, the Cobas success rate on both cell pellets and cell-free supernatants in this study may not reflect how it would perform in a clinical setting.

Conclusion

In summary, this study shows that the Roche Cobas, a liquid-based assay developed for hrHPV testing of gynecological cytology specimens, has excellent correlation with more established methods for determining hrHPV status in HNSCC and can be performed on cell-free supernatant, which is available from every FNA specimen. This method enables rapid diagnosis and addresses some of the biggest challenges posed by limited FNA specimens, which may preclude the performance of p16 IHC or HPV ISH or yield unreliable results.