Customize your JAMA Network experience by selecting one or more topics from the list below.
Hosni A, Dixon PR, Rishi A, et al. Radiotherapy Characteristics and Outcomes for Head and Neck Carcinoma of Unknown Primary vs T1 Base-of-Tongue Carcinoma. JAMA Otolaryngol Head Neck Surg. 2016;142(12):1208–1215. doi:10.1001/jamaoto.2016.3083
What is the potential therapeutic value of lingual tonsillectomy in head or neck carcinoma of unknown primary (CUP) site?
In this cohort study of 115 patients, intensity-modulated radiotherapy for head and neck CUP site and small base-of-tongue (BOT) carcinoma had similar clinical outcomes. However, radiotherapy volumes and dosimetric characteristics differed between the carcinomas.
Successful identification of hidden primary BOT carcinoma presenting as CUP site using transoral robotic surgery or transoral laser microsurgery would lead to (1) less extensive treatment (in terms of low-dose mucosal clinical target volumes), which could result in less morbidity; (2) more specific treatment (in terms of high-dose mucosal clinical target volumes), which could result in better local tumor control; and (3) frequent bilateral neck irradiation, which could decrease the concerns of regional failure in the contralateral neck.
Transoral robotic surgery– or transoral laser microsurgery–assisted lingual tonsillectomy may improve the identification rate of hidden base-of-tongue (BOT) carcinoma presenting as head or neck carcinoma of unknown primary (CUP) site.
To evaluate the potential impact of lingual tonsillectomy in CUP site by comparing differences in radiotherapy volumes, dosimetry, and clinical outcomes for CUP site and T1-category BOT carcinoma.
Design, Setting, and Participants
Retrospective study of 115 patients treated at a tertiary cancer center between January 1, 2005, and December 31, 2013, that included patients with BOT carcinoma (category T1N1-3M0) and CUP site (category T0N1-3M0) with known p16 status. Fifty-four patients with T1-category BOT carcinoma (50 [92.6%] p16-positive) were treated with definitive intensity-modulated radiotherapy (IMRT), including 34 (63%) who received concurrent chemotherapy. Sixty-one patients with CUP site (38 [62.3%] p16-positive) received definitive (42 [68.9%]) or postoperative (19 [31.1%]) IMRT, including 22 (36%) who received concurrent chemotherapy.
Definitive or postoperative IMRT, with or without concurrent chemotherapy.
Main Outcomes and Measures
Characteristics of mucosal clinical target volume (CTV-T), nodal CTV, and organ-at-risk dosimetry; local, regional, and distant control; cause-specific and overall survival; and Radiation Therapy Oncology Group grade 3 or higher late toxic effects.
Of 115 participants, 104 (90.4%) were male; mean (SD) age was 59 (10) years. High-dose CTV-T was prescribed in all 54 patients with BOT carcinoma and 23 (37.7%) with CUP site (effect size [Δ], 62%; 95% CI, 50%-74%). Low-dose CTV-T included mucosal pharyngeal sites outside the oropharynx in no patients with BOT carcinoma and 26 (42.6%) (95% CI, 30%-54%) with CUP site, with greater low-dose CTV-T volume in CUP site than BOT carcinoma (113 vs 84 cm3; Δ, 30 cm3; 95% CI, 10-49 cm3). Bilateral neck irradiation was used in 53 of 54 patients (98.1%) with BOT carcinoma and 46 of 61 (75.4%) with CUP site (Δ, 23%; 95% CI, 12% to 34%). Patients with BOT carcinoma received a higher maximum dose to the mandible (71 vs 67.2 Gy; Δ, 3.8 Gy; 95% CI, 1.6 to 6 Gy), with a nonsignificantly higher maximum dose (66.1 vs 62.8 Gy; 3.2 Gy; 95% CI, −0.1 to 6.5 Gy) and lower mean dose to the larynx (43.8 vs 47.1 Gy; 3.3 Gy; 95% CI, −0.3 to 6.9 Gy). There were no significant differences in local control, regional control, distant control, cause-specific survival, and overall survival between the BOT carcinoma and CUP site groups stratified by p16 status. Grade 3 Radiation Therapy Oncology Group late toxic effects occurred in 2 patients (3.3%) with CUP site (both neck fibrosis) and 5 (9.3%) with BOT carcinoma (2 neck fibrosis, 2 osteoradionecrosis, and 1 dysphagia).
Conclusions and Relevance
Intensity-modulated radiotherapy for CUP site or T1-category BOT carcinoma had similar clinical outcomes. Identifying hidden BOT primary carcinoma with novel approaches (eg, transoral robotic surgery and transoral laser microsurgery) may lead to changes in the radiotherapy target volume and dose prescription. Studies are needed to investigate the effect of these differences on quality of life and functional outcomes.
Head and neck carcinoma of unknown primary (CUP) site is relatively uncommon. Standard diagnostic workup includes comprehensive physical examination with flexible endoscopy, diagnostic imaging with computed tomography, magnetic resonance imaging and/or positron emission tomography with computed tomography, and a diagnostic surgical workup including panendoscopy with directed biopsies and palatine tonsillectomy. Despite this thorough diagnostic approach, the primary tumor site remains unknown in approximately 50% of patients.1-3
Although considerable treatment variability exists, the mainstay of treatment for patients with CUP site is radiotherapy (RT) with or without concurrent chemotherapy. Wide-volume irradiation of the entire pharyngeal axis, even with intensity-modulated radiotherapy (IMRT), is associated with significant toxic effects.4 Efforts should be made to identify hidden primary tumors and facilitate targeted IMRT to potentially avoid irradiation of the pharynx. Proposed selective irradiation of specific mucosal sites based on risk features (eg, oropharynx-targeted RT in patients with p16-positive CUP site) has become increasingly popular in the IMRT era.5 However, the concern of geographic miss exists with this approach.
The base of tongue (BOT) and palatine tonsil are by far the most common sites harboring an occult primary tumor.2,6 Palatine tonsillectomy has shown7 a higher identification rate of a hidden tonsillar primary tumor compared with deep tonsillar biopsy. Performing lingual tonsillectomies with transoral robotic surgery or transoral laser microsurgery may improve the identification rate of hidden primary BOT carcinoma presenting as CUP site.6,8,9 We evaluated the potential therapeutic value for this approach by comparing differences in IMRT characteristics and clinical outcomes for CUP site and small BOT carcinoma with known p16 immunohistochemistry status.
A retrospective study was undertaken for all adults (aged ≥18 years) with previously untreated, pathologically confirmed CUP site (category T0N1-3M0) and BOT carcinoma (category T1N1-3M0) with known p16 immunohistochemistry status (surrogate for human papillomavirus association) who were treated with curative intent at our institution between January 1, 2005, and December 31, 2013, with definitive or postoperative IMRT with or without concurrent chemotherapy. Patients with CUP site and isolated supraclavicular lymphadenopathy, concurrent parotid nodal metastases, nodal disease staining positive for Epstein-Barr virus–encoded RNA, or a history of carcinoma of the head and neck were excluded.
The study was approved by the research ethics board at University Health Network, Toronto, Ontario, Canada. A waiver of consent was provided by the board to perform retrospective analysis of clinical and pathologic variables, treatment characteristics, and clinical outcomes.
Tumor p16 status was determined using immunohistochemistry on tissue blocks prepared from core and open biopsies or neck dissections. If p16 staining had not been performed at the time of treatment, available tissue blocks were retrieved and staining was performed retrospectively, provided there were sufficient viable tumor cells remaining in the specimen. The IMRT characteristics with mucosal clinical target volume (CTV-T), nodal CTV, and organ-at-risk dosimetry were obtained for all patients. Clinical and outcome data were prospectively collected at the point of care and retrieved from the Head and Neck Anthology of Outcomes database.10
Standard staging included comprehensive physical examination and computed tomography of the head, neck, and chest with or without magnetic resonance imaging (typically in BOT carcinoma). Panendoscopy with directed biopsies of potential primary sites along with diagnostic palatine tonsillectomy (either unilateral or bilateral at the discretion of the surgeon) was used in patients with CUP site. Positron emission tomography with computed tomography was not routinely used during the study period.
In our institution, treatment for patients with CUP site with N1 disease was managed by RT alone. Patients with N2-3 disease were generally treated with definitive chemoradiotherapy, with altered-fractionation RT alone reserved for those unsuitable for chemotherapy (aged >70 years, poor performance status, and comorbidity) or when the patient refused chemotherapy.
In patients with CUP site, initial neck dissection (ND) was performed for those who underwent previous nondiagnostic fine-needle aspiration or core biopsies, for presumptive diagnosis of branchial cyst, for patients in whom the intent was single-modality treatment, or for those with advanced nodal disease when it was believed that definitive chemoradiotherapy was insufficient for disease control. In some instances, patients presented with disease with skin involvement and it was believed that these patients may have persistent nonhealing skin wounds after RT; as such, these patients were also offered initial ND with reconstruction. Following ND, postoperative RT was delivered to patients with multiple positive nodes, and postoperative chemoradiotherapy was delivered to patients with pathologic evidence of extracapsular extension.
Patients with T1N1 BOT carcinoma received either definitive chemoradiotherapy or altered-fractionation RT alone. Patients with T1N2-3 BOT carcinoma were generally treated with definitive chemoradiotherapy, with altered-fractionation RT alone reserved for those who were unsuitable for or refused chemotherapy.
Intensity-modulated RT has been the standard-of-care RT technique in our institution since 2005. Concurrent chemotherapy generally consisted of high-dose cisplatin (100 mg/m2) on days 1, 22, and 43.
Posttreatment imaging evaluation was routinely performed 10 to 12 weeks after completion of RT and then as clinically indicated. Patients were typically seen in the multidisciplinary head and neck clinic with physical examination, including fiberoptic endoscopy, performed at each follow-up every 3 months for the first 2 years, every 4 months in the third year, every 6 months in the fourth and fifth years, and annually thereafter until the tenth year.
Descriptive statistics were used to describe patient and treatment characteristics. The t test was used for comparison of continuous variables, and the χ2 test (Fisher exact test as appropriate) was used for comparison of categorical variables. The outcomes were further stratified by p16 status, and the estimated incidence rates of local, regional, and distant control as well as cause-specific survival were analyzed by the competing risk method. Disease-free survival and overall survival were calculated by the Kaplan-Meier method. Multivariable analysis using Cox proportional hazards regression models was applied to identify predictors for disease-free survival and overall survival. All tests were 2-sided. Effect size (Δ) and 95% CIs around the effect size were calculated. Statistical analyses were applied using SAS, version 9.4 (SAS Institute Inc) and R, version 3.1.2 (R Foundation) software.
A total of 54 patients with BOT carcinoma and 61 patients with CUP site were identified; 104 participants (90.4%) were male and the mean (SD) age was 59 (10) years. The proportion of p16-positive tumors was higher in the BOT carcinoma group than in the CUP site group (53 of 54 [98.1%] vs 38 of 61 [62.3%]; Δ, 31%; 95% CI, 17% to 45%). There was no significant difference in the distribution of N category in patients with BOT carcinoma vs CUP site. Respective N classifications included N1 (8 [14.8%] vs 5 [8.2%]), N2a (9 [16.7%] vs 19 [31.1%]), N2b (15 [27.8%] vs 22 [36.1%]), N2c (13 [24.1%] vs 5 [8.2%]), and N3 (9 [16.7%] vs 10 [16.4%]). However, patients with p16-positive tumors had fewer cases of advanced (N2c-N3) nodal category than did those with p16-negative tumors (24 of 88 [27.3%] vs 13 of 27 [48.1%]; Δ, 21%; 95% CI, −0.1% to 42%). In the p16-positive cohort, patients with CUP site had fewer cases of advanced (N2c-N3) nodal category than did patients with BOT carcinoma (4 of 38 [10.5%] vs 20 of 50 [40%]; Δ, 29%; 95% CI, 12% to 46%). The distribution of N category and other clinical characteristics stratified by p16 status are summarized in Table 1.
Among patients with CUP site (n = 61), 42 patients (68.9%) were treated with definitive IMRT. The other 19 patients (31.1%) had initial ND followed by postoperative RT. For 15 of these 19 patients (78.9%), the initial intent was single-modality treatment (ie, ND alone) with adjuvant therapy indicated based on pathologic findings (pathologic evidence of extracapsular extension [n = 4], pathologic upstaging [n = 3], or both [n = 8]); in the remaining 4 patients, initial ND was planned as part of multimodality management of advanced N3 category disease. Concurrent chemotherapy was used in 36% of patients with CUP site (22 of 61: 18 in the definitive setting and 4 with postoperative RT). All 54 patients with BOT carcinoma were treated with definitive IMRT, of whom 34 (63%) received concurrent chemotherapy. Chemotherapy consisted of high-dose cisplatin in all patients except in 4 individuals with p16-positive BOT carcinoma who received epithelial growth factor receptor inhibitors in a clinical trial setting. High-dose CTV-T was prescribed in 54 patients (100%) with BOT carcinoma and 23 (37.7%) of those with CUP site (Δ, 62%; 95% CI, 50%-74%). Low dose CTV-T included mucosal sites outside the oropharynx (ie, nasopharynx, hypopharynx, and/or larynx) in 0% of patients with BOT carcinoma and 26 (42.6%) of those with CUP site (Δ, 42%; 95% CI, 30%-54%), with a greater volume of low-dose CTV-T in those with CUP site than in those with BOT carcinoma (mean [SD], 113  vs 84  cm3; Δ, 30 cm3; 95% CI, 10-49 cm3). Bilateral neck irradiation was used in 53 of 54 patients (98.1%) with BOT carcinoma and 46 of 61 (75.4%) of those with CUP site (Δ, 23%; 95% CI, 12%-34%). Dosimetric criteria of organ at risk demonstrated that patients with BOT carcinoma received a larger maximum dose to the mandible, with a nonsignificantly higher maximum dose and lower mean dose to the larynx. The details of IMRT characteristics and organ-at-risk dosimetry are summarized in Table 2.
We performed a subgroup analysis of patients with p16-positive disease (BOT carcinoma [n = 50] vs CUP site [n = 38]). The patients with p16-positive BOT carcinoma were treated more frequently with high-dose CTV-T (50 [100%] vs 15 [39.5%]; ∆, 60.5%; 95% CI, 49%-73%) and with bilateral neck irradiation (49 [98%] vs 31 [81.6%]; ∆, 16.4%; 95% CI, 7%-25%). Compared with the patients with p16-positive BOT carcinoma, those with p16-positive CUP site were treated more frequently with low-dose CTV-T that included mucosal sites outside the oropharynx (CUP site, 6 [15.8%] vs BOT carcinoma, 0; Δ, 15.8%; 95% CI, 7%-25%) and received a greater volume of low-dose CTV-T (mean [SD], 108  vs 78  cm3; ∆, 30 cm3; 95% CI, 12-48 cm3). These findings were all consistent with the analysis of the entire cohort.
Only 2 patients (1.7%) in the entire cohort developed local failure. One patient had p16-positive CUP site with initial presentation of category N2b disease. This patient eventually developed mucosal emergence at the BOT at 36 months, which was treated with salvage surgery; however, the patient subsequently developed regional failure. The second local failure was observed in a patient with p16-negative N1 BOT carcinoma who had a significant smoking history and developed late local failure at 82 months and was treated with salvage surgery.
Regional failure was reported in 5 patients (8.2%) with CUP site at a median of 29 (range, 3-66) months, and 2 patients (3.7%) with BOT carcinoma (at 2 and 16 months); all the patients were initially treated with bilateral neck irradiation, and only 2 tumors (1 CUP site and 1 BOT carcinoma) were p16-positive. Patients with regional failure initially presented with disease categories N2b (1 CUP site), N2c (1 BOT carcinoma), and N3 (1 CUP site and 1 BOT carcinoma). Treatment of patients with regional failure consisted of surgery (3 patients), chemotherapy (1), and supportive therapy (3).
A total of 9 patients developed distant metastasis: 6 patients (9.8%) with CUP site (3 p16-positive and 3 p16-negative) at a median time of 14 months (range, 4-20 months) and 3 (5.6%) with BOT carcinoma (all p16-positive) at 5, 17, and 23 months. Site of distant metastasis in patients with BOT carcinoma were all in the lung; in patients with CUP site, distant metastasis was in the lungs (2 patients), bones (3), and at multiple sites (1). All distant metastases were isolated first recurrences, except in 2 patients with synchronous regional failure (1 CUP site and 1 BOT carcinoma).
Overall, 10 patients (16.4%) with CUP site and 9 (16.7%) with BOT carcinoma had died by the last follow-up. There were no significant differences in local, regional, and distant control; cancer-specific survival; and overall survival between the BOT carcinoma and CUP site groups stratified by p16 status (Table 3). Subgroup analysis of all p16-positive (CUP site and BOT carcinoma) patients (n = 88) showed that concurrent chemotherapy was associated with better disease-free survival (hazard ratio [HR], 8.85; 95% CI, 1.70-46.14) and overall survival (HR, 8.75; 95% CI, 1.68-45.67) on multivariable analysis (Table 4). No grade 4 or 5 late toxic effects were reported; however, grade 3 Radiation Therapy Oncology Group late toxic effects were recorded in 2 patients (3.3%) with CUP site (both neck fibrosis) and 5 (9.3%) with BOT carcinoma (2 neck fibrosis, 2 osteoradionecrosis, and 1 dysphagia).
A subset (19 [31.1%]) of patients with CUP site was treated with initial ND. To evaluate whether management modality affected IMRT characteristics or clinical outcomes, we performed a subgroup analysis. Group 1 included patients with BOT carcinoma treated with definitive IMRT (ie, without ND), group 2 included patients with CUP site treated with definitive IMRT (ie, without ND), and group 3 included patients with CUP site treated with ND followed by postoperative IMRT (Table 5). Compared with group 1 patients, group 2 patients were treated more frequently with a high-dose CTV-T (54 [100%] vs 20 [47.6%]; ∆, 53%; 95% CI, 38%-68%) and a greater volume of low-dose CTV-T (117 vs 84 cm3; ∆, 33; 95% CI, 11.5-54 cm3) and received a higher mean dose to both the larynx and inferior constrictor muscle, with a lower maximum dose to the mandible. There was no difference in the proportion of patients whose treatment was managed with bilateral neck irradiation in group 1 vs group 2. No significant differences in clinical outcomes were detected between the groups.
The present study compared treatment characteristics and clinical outcomes for CUP site of the head and neck and small BOT carcinomas. We uncovered differences in RT volumes and doses to organs at risk that provide important lessons for the potential impact of identifying hidden BOT primary tumors with novel approaches, such as lingual tonsillectomy. Similar patients’ demographics (eg, age, sex, and smoking history) and the high proportion of p16-positive tumors in patients with CUP site and BOT carcinoma (38 [62.3%] and 50 [92.6%], respectively) corroborate the possibility that many CUP sites may, in fact, represent occult oropharyngeal primary tumors. Patients treated with IMRT for CUP site and small BOT carcinoma tumors had similar clinical outcomes; however, patients with BOT carcinoma had more frequent bilateral neck irradiation and use of high-dose CTV-T targeting the small identified tumor, with a lesser volume of low-dose CTV-T resulting from more frequent sparing of mucosal sites outside the oropharynx. Although many patients with CUP site had unilateral neck irradiation in this series, this practice was predominantly seen among patients who underwent initial ND as part of multimodality management.
Localization of the unknown primary tumor remains a challenge. Panendoscopy under general anesthesia with tonsillectomy is the mainstay diagnostic approach for patients with CUP site, and almost 50% of patients will have the primary site detected following directed biopsies of clinically and/or radiologically suspicious areas.2 Standard imaging with computed tomography and/or magnetic resonance imaging identify suspicious areas to direct biopsies and could potentially help identify the occult primary tumor in up to 20% of pateints.1,11,12 The likelihood for identification of a primary tumor may be further increased by the addition of positron emission tomography or positron emission tomography with computed tomography, which may help to identify as many as 24% to 37% of primary tumors in systematic reviews.13-15 The rate of hidden palatine tonsil tumor identification after ipsilateral and contralateral palatine tonsillectomy ranges from 18% to 45%2,16-20 and 10% to 23%,21,22 respectively. Furthermore, the yield of detecting an occult palatine tonsil primary tumor was 30% for palatine tonsillectomy vs 3% for deep tonsil biopsies,7 raising the possibility that lingual tonsillectomy may increase the likelihood of finding a hidden lingual tonsil tumor than deep biopsies of BOT.
The most common site for a CUP is the oropharynx, with the palatine and lingual tonsils accounting for 90% of all identified CUP sites.2,6 Recently, transoral robotic surgery– or transoral laser microsurgery–assisted lingual tonsillectomy for patients with CUP site has been evolving with a diagnostic success rate of CUP site origin approaching 90% and an ability to identify tumors as small as 2 mm.6,8,9 Furthermore, this approach could be an alternative therapeutic strategy by combining the removal of an identified small primary BOT tumor with a therapeutic ND.23,24 Patients without high-risk features (ie, involved margins and/or extracapsular extension) and limited nodal disease (pN1) would not require adjuvant therapy, and those with more advanced nodal disease (pN2-3) may be offered adjuvant RT without the need for higher-dose RT (66-70 Gy/33-35 fractions) or concurrent chemotherapy.25 Consideration must also be given to possible morbidity associated with this approach, such as swallowing dysfunction, bleeding, and the need for tracheostomy.23,24
The toxic effects of comprehensive mucosal irradiation for CUP site is significantly more pronounced than targeting a specific subsite, even with the use of IMRT.4 Wallace et al26 recommended exclusion of the larynx and hypopharynx in patients with CUP site with no level III nodal involvement after they reported a 5-year local control of 100% for 28 patients with CUP site treated with mucosal portals limited to the nasopharynx and oropharynx. Mourad et al27 reported the outcome of oropharynx-targeted RT in 68 patients with CUP site and the 3-year local control was 98.5%. In our institution, the philosophy of selecting elective CTV-T is based on proper definition of clinical, radiologic, pathologic, and molecular risk features. For example, human papillomavirus association indicates a likely oropharyngeal primary site, and Epstein-Barr virus–encoded RNA positivity indicates a likely nasopharyngeal primary site. Successful identification of the primary hidden BOT tumor through lingual tonsillectomy obviates the need for entire mucosal irradiation and decreases the risk of geographic miss, which may happen with a selective target volume approach of elective CTV-T. This study was not designed to evaluate whether patients could be spared treatment of the pharyngeal mucosa with IMRT following successful identification and resection of a small BOT primary site.
The elective nodal volume requiring irradiation in CUP site remains controversial. Several single-institution retrospective studies comparing bilateral and ipsilateral neck irradiation have not shown any advantage to more extensive RT,28-33 but other studies have observed improvements in regional control34 and overall survival5 with bilateral neck irradiation. Consequently, no firm conclusions on this issue can be made. When a hidden BOT tumor is identified, bilateral neck irradiation is warranted. In our study, bilateral neck irradiation was used in 98% of patients with BOT carcinoma and 75% of those with CUP site.
Despite the retrospective design of the study, clinical and outcome data were prospectively collected. Our results indicate that patients with successful identification of hidden primary BOT carcinoma presenting as CUP site could benefit from (1) less-extensive treatment (in terms of low-dose CTV-T), which could result in less morbidity and avoidance of unnecessary inclusion of pharyngeal subsites in target volume; (2) more-specific treatment (in terms of high-dose CTV-T), which could result in a higher probability of local tumor control; and (3) bilateral neck irradiation almost always, which could decrease the legitimate concerns of regional failure in the contralateral neck in these patients. However, doses to surrounding organs at risk could be affected by the use of high-dose CTV-T and bilateral neck irradiation. We believe that our data add to the growing body of literature providing a rationale of performing lingual tonsillectomy in the initial diagnostic workup of CUP site; however, further prospective research is required to assess whether successful identification of hidden BOT carcinoma tumors will have an effect on oncologic, quality-of-life, and functional outcomes. The potential value of a negative lingual tonsillectomy for guiding ipsilateral elective nodal irradiation in patients with CUP site should also be investigated.
Patients treated with IMRT for CUP site or small BOT carcinoma had similar clinical outcomes. Identifying small-volume hidden BOT primary carcinomas may lead to changes in RT target volume and dose prescription. Future studies considering the role of transoral robotic surgery or transoral laser microsurgery in the workup of CUP site are required to investigate their success rate in identification of hidden primary tumors and their potential effect on oncologic, quality-of-life, and functional outcomes.
Corresponding Author: John R. de Almeida, MD, MSc, Room 3-955, Department of Otolaryngology–Head & Neck Surgery, Surgical Oncology, Princess Margaret Cancer Centre, 610 University Ave, Toronto, ON M5G 2M9, Canada (email@example.com).
Accepted for Publication: August 13, 2016.
Published Online: November 3, 2016. doi:10.1001/jamaoto.2016.3083
Author Contributions: Drs Bratman and de Almeida have 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: Hosni, Xu, Goldstein, Huang, O’Sullivan, Waldron, de Almeida.
Acquisition, analysis, or interpretation of data: Hosni, Dixon, Rishi, Au, Xu, Song, Chepeha, Huang, Kim, O’Sullivan, Bratman, de Almeida.
Drafting of the manuscript: Hosni, Xu, Song, Bratman, de Almeida.
Critical revision of the manuscript for important intellectual content: Hosni, Dixon, Rishi, Au, Xu, Chepeha, Goldstein, Huang, Kim, O’Sullivan, Waldron, Bratman, de Almeida.
Statistical analysis: Hosni, Dixon, Xu, Song, Huang, de Almeida.
Administrative, technical, or material support: Hosni, Dixon, Au, Chepeha, Huang, O’Sullivan.
Study supervision: Xu, Waldron, Bratman.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Previous Presentation: This study was presented at the American Head & Neck Society Ninth International Conference on Head and Neck Cancer; July 19, 2016; Seattle, Washington.