Changes in Unknown Primary Squamous Cell Carcinoma of the Head and Neck at Initial Presentation in the Era of Human Papillomavirus

Importance  The presence of human papillomavirus (HPV) in unknown primary squamous cell carcinoma (UPSCC) of the head and neck at initial presentation focuses the investigation for the primary tumor on the oropharynx. The trends, frequency, and detection rate of UPSCCs have not been evaluated in the context of HPV tumor status.

Objectives  To determine the frequency of UPSCC over time and to evaluate the proportion of HPV-positive UPSCCs.

Design, Setting, and Participants  Retrospective, single-institutional case series of patients diagnosed with UPSCC and evaluated at the Johns Hopkins Hospital from January 1, 2005, to June 1, 2014. Human papillomavirus tumor status was determined by p16 immunohistochemical analysis and/or high-risk HPV DNA by in situ hybridization as clinically available.

Main Outcomes and Measures  Number and clinical characteristics of UPSCC cases over time.

Results  Eighty-four UPSCC cases were eligible for analysis. The mean age of the patients was 57.3 years (range 29-80 years), and 88.1% (n = 74) were male. The frequency of UPSCC increased significantly over time (P for trend = .01) and was significantly higher during later calendar periods (14 cases during 2005-2008 vs 39 cases during 2012-2014, P = .03). A total of 69 cases (90.7%) with available HPV tumor status were HPV-positive. The patients with HPV-positive UPSCC were significantly more likely to be male (91% vs 42.9%, P = .005) and younger (56.1 vs 67.7 years, P = .002) than the HPV-negative patients with UPSCC. The overall primary tumor site detection rate was 59.3% (n = 48). There was a nonsignificant increase in the detection rate from calendar periods 2005-2008 to 2012-2014 (50.0% vs 64.9%, P = .38). Since transoral robotic surgery was adopted in the diagnostic evaluation of UPSCC in 2011, a nonsignificant increase in the detection of primary tumors was observed (53.8% vs 64.3%, P = .34).

Conclusions and Relevance  The frequency of UPSCC has increased significantly in recent calendar periods, and most cases are HPV-positive. As expected, patients with HPV-positive UPSCC tend to be male and younger.

Unknown primary squamous cell carcinoma (UPSCC) of the head and neck is a rare entity characterized by clinical nodal metastases without an identifiable primary tumor site. Approximately 4% of head and neck SCCs present as UPSCC.¹ Diagnostic evaluation of the primary site historically included visualization and directed biopsies of the nasopharynx, larynx, hypopharynx, and oropharynx.^2,3 However, contemporary diagnostic evaluation has evolved to include physical examination, endoscopic and radiographic imaging, palatine and lingual tonsillectomy, and human papillomavirus (HPV) and Epstein-Barr virus detection.¹ Human papillomavirus–positive tumor status of UPSCC has emerged as an important biomarker for primary disease arising from the oropharynx.^4-6 Therefore, the presence of an HPV-positive tumor status in UPSCC permits a focused diagnostic evaluation of the oropharynx.^6-8

A disproportionate number of HPV-positive oropharyngeal SCCs compared with HPV-negative oropharyngeal SCCs initially present as UPSCC.⁹ When these HPV-positive oropharyngeal primary tumors are ultimately discovered, they are commonly found in the cryptic lymphoepithelium of the palatine and lingual tonsils.¹⁰ Although UPSCCs are strongly associated with HPV-positive tumor status,^4-9 the proportion that are HPV-positive is unknown. Several retrospective series of UPSCC arising from all head and neck sites with sample sizes ranging from 25 to 63 have estimated that 22% to 74% are related to HPV.^10-16 One series of 26 cases limited to UPSCCs that were ultimately found to arise exclusively from the oropharynx reported the proportion of HPV-positive tumor status to be as high as 81%.⁹ With the incidence of HPV-related oropharyngeal cancer increasing in the United States and other countries,^17,18 we hypothesized that both the frequency of UPSCCs and the proportion that are HPV-positive are increasing.

Identification of the primary tumor when a patient presents with a UPSCC has important clinical implications. Localization allows for more targeted therapy and potentially decreased morbidity and improved survival.^19-21 Considering that treatment outcomes depend on whether the primary tumor is detected, the diagnostic investigation remains paramount for UPSCCs. Estimates of the primary tumor detection rate for UPSCC in the literature are broad and do not account for HPV tumor status.^7,20,22 Transoral robotic surgery (TORS) and transoral laser microsurgery are surgical methods touted to increase the detection of primary tumors.²³ Whether UPSCC detection rates have improved in the context of these novel surgical tools is unknown. The aims of this study were to determine the changes in frequency of UPSCC evaluated over time, estimate the proportion of UPSCCs that are HPV-positive, and evaluate changes in the detection rate of primary tumors in UPSCC over time.

Patients with a clinical diagnosis of UPSCC as determined by a head and neck surgical oncologist at the Johns Hopkins Hospital from January 1, 2005, to June 1, 2014, were eligible for this study. This was a single-institution retrospective review approved by the Johns Hopkins Hospital Institutional Review Board. Patients with recurrent disease or evidence of a primary tumor site determined by previous biopsies or clinical workup were excluded. All records were reviewed to confirm that a Johns Hopkins Hospital head and neck surgeon had provided a clinical diagnosis of UPSCC and documented that there was no clinical evidence or suspicion of a primary lesion on clinical, radiographic, or flexible endoscopic evaluation.

Medical records of eligible patients were abstracted. Clinical variables of interest included age, sex, race, smoking and alcohol exposure, tumor site, HPV tumor status, TNM classification, and overall stage based on surgical pathologic findings as defined by the American Joint Committee on Cancer.²⁴ Human papillomavirus tumor status was determined by p16 immunohistochemical analysis and/or high-risk HPV DNA in situ hybridization, as clinically available. Patients were considered to be HPV-positive if p16 immunohistochemical analysis and/or in situ hybridization results for high-risk HPV were positive. In addition, receipt of surgical diagnostic interventions, including direct laryngoscopy and palatine and lingual tonsillectomy, was of interest. Of note, TORS was offered for UPSCC detection after 2011 for lingual tonsillectomy in the event that imaging, endoscopy, directed surgical biopsies, and/or palatine tonsillectomy did not identify a primary tumor consistent with previous literature.²⁰

Positron emission tomography (PET) results were reviewed. Patients with equivocal PET scans were eligible for analysis. If a primary tumor was suspected on a PET scan but no primary site was identified after surgical evaluation, these patients were considered to have UPSCC and were therefore eligible for analysis. However, patients with a suspicious lesion on a PET scan, which was pathologically confirmed as the site of the primary tumor, were ineligible. Furthermore, if direct laryngoscopy revealed a visually suspicious lesion, which subsequently tested positive for carcinoma, these patients were also ineligible.

Descriptive statistics were used to summarize frequencies, proportions, means, and ranges. Associations between HPV tumor status and clinical variables of interest were evaluated using Fisher exact tests for categorical variables and 2-sample t tests for continuous variables. Nonparametric tests for trend were performed across periods for the number of UPSCCs evaluated per year and the proportion of cases with primary site detection. Prevalence ratios (PRs) for primary site detection by period were estimated using log-binomial regression. The median number of surgical procedures before and after the potential use of TORS in diagnostic evaluation of UPSCC was compared using a nonparametric Wilcoxon rank sum test. A 2-sided P < .05 was considered statistically significant. Data analysis was performed using STATA statistical software, version 11.2 (StataCorp).

The characteristics of the study population are summarized in Table 1. In total, 84 cases of UPSCCs were eligible for analysis. The mean age of the patients was 57.3 years (range, 29-80 years). Most of the study population was male (74 of 84 [88.1%]) and had a history of ever smoking (50 of 78 [64.1%]). Human papillomavirus tumor status of metastatic cervical lymph nodes was available for 76 of 84 patients (90.5%). Of the 75 patients with HPV tumor status available, 68 (90.8%) were HPV-positive. The characteristics of patients with and without HPV tumor status were similar (P = .12).

The characteristics of HPV-positive and HPV-negative patients with UPSCCs were compared. A significantly greater proportion of HPV-positive patients with UPSCCs were male than the HPV-negative patients with UPSCCs (91.3% vs 42.9%, P = .005). The HPV-positive patients with UPSCCs were on average significantly younger than the HPV-negative patients with UPSCCs (56.1 vs 67.7 years, P = .002). The HPV-positive patients with UPSCCs were more likely to present with advanced nodal stage compared with the HPV-negative patients with UPSCCs (P = .05). A higher proportion of HPV-positive patients with UPSCCs were never smokers relative to the HPV-negative patients with UPSCCs, although this finding was not statistically significant (37.5% vs 0%, P = .09).

The frequency of patients presenting with a clinical diagnosis of UPSCC has significantly increased over time (P for trend = .01; Figure 1). When considering the mean number of UPSCCs per year, an increase was observed by calendar period (Figure 2). The mean number of UPSCCs per year increased significantly from 3.5 cases in the earliest calendar period (2005-2008) to 10.3 cases in 2009-2011 and 15.6 cases in the most recent calendar period (2012-2014) (P = .03). The proportion of UPSCCs that were HPV-positive across calendar periods was similar (P = .64) although most UPSCCs in recent calendar periods were HPV-positive (Figure 3). In the earliest calendar period, there were a substantial number of cases with unknown HPV tumor status (n = 8 [9.5%]; Figure 3).

Most UPSCCs underwent diagnostic evaluation in an attempt to localize the primary site (81 of 84 [96.4%]). Therefore, analysis of detection was limited to these 81 patients. Overall, primary tumors were ultimately identified in 48 of 81 UPSCCs (59.3%). Most patients with identified primary tumors were HPV-positive (44 of 46 [95.7%]). All identified primary tumors were found in the oropharynx. Twenty-seven (56.3%) were determined to be primary tumors of the base of tongue, 20 (41.7%) of the palatine tonsils, and 1 (2.1%) that was overlapping the base of tongue and palatine tonsil.

When evaluating changes in detection over time, the proportion of identified primary tumors increased, although nonsignificantly, from 50.0% in 2005-2008 to 64.9% in 2012-2014 (P for trend = .38; PR = 1.3; Table 2). To determine whether the availability of TORS lingual tonsillectomy in 2011 affected the detection rate of primary tumors, the proportion of primary tumors identified before and after TORS was compared. A lower proportion of UPSCCs was identified in the pre-TORS period compared with the post-TORS period, although this finding was nonsignificant (53.8% vs 64.3%; PR = 1.2; P = .34; Table 2). Among HPV-positive UPSCCs, the detection rate over time was similar (P for trend = .14; eTable in the Supplement). Finally, the mean number of surgical procedures per patient with UPSCC before and after TORS was compared. There was no increase in the mean number of surgical procedures performed per patient after the introduction of TORS into the UPSCC diagnostic workup (1.85 vs 1.88; range, 1-4; P = .95).

Among the 48 patients with primary tumors detected, 26 (54.2%) underwent bilateral palatine tonsillectomy or bilateral lingual tonsillectomy. Of those patients, 2 (7.7%) had synchronous primary lesions. In both these cases, the tumor was HPV-positive.

This analysis describes an increasing frequency of UPSCCs in recent calendar periods, most of which are HPV-positive. Although the increasing frequency of UPSCCs has not previously been described, it is not unexpected given the evolving epidemiologic features of head and neck cancer.

The increasing incidence of oropharyngeal squamous cell cancers in the United States and abroad is attributable to HPV infection.^17,18 Human papillomavirus–positive oropharyngeal cancers are more commonly small primary tumors with advanced nodal disease; consequently, UPSCCs are disproportionately HPV-positive.^7,25 In light of these trends, the high proportion of HPV-positive UPSCCs in this series is consistent with the observed epidemiologic features of head and neck cancer. In this large, albeit single-institution series, 91% of contemporary UPSCCs are HPV-positive. Prior estimates varied from 22% to 81%, with most series reporting that less than 50% of cases are HPV-positive by either p16 immunohistochemical analysis or in situ hybridization.^10-15 Similar to the epidemiology of oropharyngeal cancer, the variation may be a reflection of geographic heterogeneity^26,27 or HPV detection methods.⁸ The observation that most UPSCCs are HPV-positive suggests that HPV is indeed driving the increased frequency of UPSCCs.

Although the question of greater significance is whether the incidence of UPSCCs is increasing, it is challenging to determine this at a population level. The UPSCCs in the US-based population registry (Surveillance, Epidemiology, and End Results Program of the National Cancer Institute) is not specific to the head and neck but rather captures unknown primary tumors of all anatomical sites, including the abdomen and chest. Given this limitation, we investigated this question at a tertiary medical center and evaluated frequency over time but highlight the inability to provide a rate.

Consistent with the clinicodemographic distinctions between HPV-positive and HPV-negative oropharyngeal cancers, analogous clinicodemographic differences were observed between HPV-positive and HPV-negative UPSCCs.^28,29 Human papillomavirus–positive UPSCCs are more commonly found in males, younger patients, and nonsmokers compared with HPV-negative UPSCCs. This finding reinforces the notion that HPV confers a distinct clinicodemographic profile on patients with head and neck malignant neoplasms.^28,30 This finding may reflect demographic differences in exposure to oral HPV infection.^31-35

Identification of the primary tumor for patients with UPSCC is of prognostic significance and has long been a priority. Bilateral tonsillectomies after negative direct laryngoscopy and directed biopsies have historically been recommended.^2,3 However, lingual tonsillectomy has been advocated as a means of further increasing detection rates of unknown primary tumors.²⁰ Small case series with variable selection criteria have reported identification in 72% to 90% of cases.^19,20 A large series restricted to HPV-positive UPSCCs reported a detection rate of 89% using a TORS algorithm.²³ In this study, there was a nonsignificant increase in the detection rate of primary tumors before vs after TORS. Whether this can be attributed to TORS or the knowledge that HPV-positive UPSCCs are more likely to arise from the oropharynx and therefore increased diagnostic scrutiny of this site remains unknown. Given the potential morbidity of this surgical approach, including bleeding, dysphagia, and strictures, there was not a significant increase in the number of procedures per patient. New strategies, such as transcervical ultrasonography, may help guide primary tumor identification and reduce the breadth of the TORS operative field and therefore potential morbidities.^36-39

Of the cases in this study that underwent bilateral palatine or lingual tonsillectomy with the primary lesions ultimately identified, 2 of 26 (7.7%) had synchronous primary lesions. Both these synchronous primary lesions were HPV-positive. Human papillomavirus–positive synchronous primary tumors have been reported, but the pathophysiologic features remain poorly understood.^40-43 Groups exploring the molecular causes of synchronous HPV-positive primary tumors hypothesize migration of HPV-infected cells throughout the Waldeyer ring or independent inoculation events by the same virus leading to field cancerization.⁴⁰ Other studies^19,23 investigating the utility of TORS for detecting primary tumors in UPSCCs have also reported synchronous primary tumors; thus, it is important to consider this possibility when evaluating patients with HPV-positive UPSCC. Whether synchronous primary tumors are more or less common in HPV-positive than HPV-negative oropharyngeal cancer is unknown. Determining this and improved estimates of the frequency of synchronous primary tumors will inform the extent of diagnostic interventions at the time of UPSCC evaluation.

Although, to our knowledge, this is the largest single-institution series of UPSCCs in the context of HPV tumor status and describes trends over time, there are important limitations that warrant attention. As previously discussed, changes in frequency, not incidence, are described. Therefore, the frequency changes could be a reflection of institutional referral patterns and marketing, which may affect the total number of patients with head and neck cancer, rather than driven by increasing incidence of this entity. In addition, this is a retrospective single-institution review and therefore has the inherent biases and shortcomings of such a study, including geographic and institutional factors that may influence HPV detection. Of note, HPV detection results are not available for more than half of the earlier calendar period UPSCCs. Human papillomavirus detection results in this calendar period may be biased by physicians ordering HPV tumor detection for UPSCCs with clinical characteristics of patients with oropharyngeal cancer, which could explain why 100% of patients tested in the earliest time point were HPV-positive. Therefore, had all the UPSCCs in the earlier calendar period been appropriately tested, it would be expected that the proportion that were HPV-positive would be commensurate with later calendar periods. Finally, the distribution of calendar years per calendar periods is not equal; therefore, mean number of cases was used to evaluate questions of interest. This was dictated by the number of cases over time and calendar periods.

The findings of this study indicate an increasing frequency of UPSCCs and that most of these cases are HPV-positive. It is likely that the increase in HPV-related tumors is driving the increase in UPSCC. Human papillomavirus–positive UPSCCs differ from HPV-negative UPSCCs by clinicodemographic characteristics.

Submitted for Publication: August 20, 2015; final revision received October 13, 2015; accepted November 15, 2015.

Corresponding Author: Carole Fakhry, MD, MPH, Department of Otolaryngology–Head and Neck Surgery, Johns Hopkins School of Medicine, 601 N Caroline St, Johns Hopkins Outpatient Center, Sixth Floor, Baltimore, MD 21287 (cfakhry@jhmi.edu).

Published Online: January 14, 2016. doi:10.1001/jamaoto.2015.3228.

Author Contributions: Dr Motz and Mr Qualliotine had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Motz, Fakhry.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Motz, Qualliotine, Rettig, Richmon, Fakhry.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Motz, Qualliotine, Rettig, Fakhry.

Obtained funding: Fakhry.

Administrative, technical, or material support: Eisele, Fakhry.

Study supervision: Rettig, Richmon, Eisele, Fakhry.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported by grant P50DE019032 from the National Institute of Dental and Craniofacial Research and the Oral Cancer Foundation.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.

Previous Presentation: This study was presented as a poster at the American Head and Neck Society 2015 Translational Research Meeting: April 21-22, 2015; Boston, Massachusetts.