Population-based validation of the recursive partitioning analysis based staging system for oropharyngeal cancer

Transcription

1 ORIGINAL ARTICLE Population-based validation of the recursive partitioning analysis based staging system for oropharyngeal cancer Florence K. Keane, MD, 1 Yu-Hui Chen, MS, MPH, 2 Roy B. Tishler, MD, PhD, 3 Jonathan D. Schoenfeld, MD, MPH, 3 Robert I. Haddad, MD, 4 Laura A. Goguen, MD, 5 Paul Catalano, ScD, 2 Bridget A. Neville, MPH, 6 Danielle N. Margalit, MD, MPH 3 * 1 Harvard Radiation Oncology Program, Harvard Medical School, Boston, Massachusetts, 2 Dana Farber/Harvard Cancer Center, Boston, Massachusetts, 3 Department of Radiation Oncology, Dana Farber Cancer Institute/Brigham and Women s Hospital, Boston, Massachusetts, 4 Department of Medical Oncology, Dana Farber Cancer Institute, Harvard Medical School and Department of Medicine, Brigham and Women s Hospital, Boston, Massachusetts, 5 Division of Otolaryngology, Department of Surgery, Brigham and Women s Hospital, Boston, Massachusetts, 6 Ariadne Labs, Boston, Massachusetts. Accepted 14 March 2016 Published online 15 April 2016 in Wiley Online Library (wileyonlinelibrary.com). DOI /hed ABSTRACT: Background. The American Joint Committee on Cancer (AJCC) staging system does not adequately distinguish prognostic groups in the era of human papillomavirus (HPV)-related oropharyngeal cancer. The purpose of this study was to validate a recursive partitioning analysis (RPA)-based stage grouping on a population-wide level. Methods. We identified 8427 patients in Surveillance, Epidemiology, and End Results (SEER) with nonmetastatic oropharyngeal cancer with unknown HPV-status diagnosed from 2004 to We estimated the overall survival (OS) and head and neck cancer-specific mortality by RPA stage and AJCC stage and compared the predictive power of the systems. Results. RPA stage was significantly associated with OS and head and neck cancer-specific mortality (p <.0001) with 5-year OS of 70% for RPA-I, 55.6% for RPA-II, and 44.3% for RPA-III. AJCC stage failed to divide patients into distinct subgroups. RPA stage had significantly improved predictive ability versus AJCC stage for OS (c-statistic: RPA vs AJCC) and head and neck cancer-specific mortality (c-statistic: RPA vs AJCC). Conclusion. The RPA-based stage grouping divided patients into prognostically distinct cohorts and provided superior prediction of OS and head and neck cancer-specific mortality compared to AJCC staging. VC 2016 Wiley Periodicals, Inc. Head Neck 38: , 2016 KEY WORDS: oropharyngeal squamous cell cancer, head and neck cancer staging, American Joint Committee on Cancer (AJCC) staging system, human papillomavirus (HPV)-related cancer, recursive partitioning analysis-based staging INTRODUCTION Prior studies demonstrated that the current American Joint Commission on Cancer (AJCC) staging classification system does not consistently distinguish prognostic subgroups for oropharyngeal squamous cell carcinoma (SCC). 1,2 This is largely thought to reflect the changing demographics of oropharyngeal SCC as the rate of human papillomavirus (HPV)-related cancers have increased rapidly relative to HPV-negative oropharyngeal SCC since the initial publication of the AJCC system in with an increase in the incidence of HPV-related oropharyngeal SCC from 16.3% in 1984 to 1989 to 71.7% in 2000 to Studies focusing exclusively on patients with HPV-related oropharyngeal SCC have also demonstrated the declining efficacy of the AJCC staging system. 7,8 *Corresponding author: D. N. Margalit, Department of Radiation Oncology, Harvard Medical School, Dana Farber Cancer Institute, 450 Brookline Avenue, Boston, MA, dmargalit@lroc.harvard.edu Additional Supporting Information may be found in the online version of this article. Recently, investigators from the Princess Margaret Cancer Centre (PMCC) proposed a modification of the current AJCC staging system derived from recursive partitioning analysis-based TNM classification groupings for an institutional cohort of patients with HPV-related oropharyngeal SCC treated with radiotherapy or chemoradiotherapy. 8 The following 3 groupings were able to better differentiate subgroups with significantly different survival than the AJCC staging system: recursive partitioning analysis (RPA)-I (T1-3N0-N2b); RPA-II (T1-3N2c); and RPA-III (T4 and/or N3). An additional group, RPA-IV, consisting of patients with distant metastases (M1), was proposed but not included in the PMCC analysis. They also assessed prognostic groupings that included age and smoking status, but found that the RPA-based grouping consisting of only TNM classifications to be superior. We sought to validate the RPA-based TNM classification grouping for oropharyngeal SCC on a populationwide level using Surveillance, Epidemiology, and End Results (SEER) program research data. We hypothesized that, unlike the AJCC staging system, the RPA stage would divide patients with oropharyngeal SCC in the HPV-era into prognostically distinct subgroups and would 1530 HEAD & NECK DOI /HED OCTOBER 2016

2 SEER VALIDATION OF THE RECURSIVE PARTITIONING ANALYSIS BASED STAGING SYSTEM provide superior prediction of overall survival (OS) and cancer-specific survival. Additionally, we aimed to show that the RPA-based system was superior to AJCC staging even among a group of patients with unknown HPV status, an important consideration when selecting a staging system for worldwide applicability where p16-staining or direct HPV assessment may not be available. MATERIALS AND METHODS Study cohort We used SEER Program public-use data 9 to identify patients age 18 and older diagnosed with nonmetastatic oropharyngeal SCC between January 1, 2004, and December 31, The SEER Program, which is sponsored by the National Cancer Institute, collects cancer incidence and survival data from population-based registries that cover approximately 28% of the U.S. population, and captures 97% of incident cancer diagnoses. 9 Of note, SEER does not record HPV status; however, we selected 2004 as the start point of the study to capture a modern cohort of patients that are expected to have a majority of HPV-related cancers based on prior studies estimating the prevalence of HPV-related oropharyngeal SCC in the United States. 5,6 We included patients with International Classification of Diseases for Oncology, second and third editions (ICD-O-2 and ICD-O-3) primary site codes, including base of tongue (C01.9, C02.4), tonsil (C09.0 C09.1, C09.8 C09.9), and oropharynx not otherwise specified (NOS; C10.0 C10.4, C ). Histologies included were SCC and variants (ICD-O-3 morphology codes , 8054, , 8083, and 8094). We excluded patients for the following reasons: presence of distant metastases; missing pathologic confirmation of disease; incomplete information on T classification or N classification; diagnosis only obtained through death certificate or autopsy; or prior cancer diagnosis (except nonmelanoma skin cancer). Overall stage, T classification, and N classification were based on the AJCC sixth edition staging system 10 and were provided in SEER, in addition to codes that described the extent of disease, including tumor size, extension to surrounding organs, and lymph node involvement. Of note, SEER did not consistently report whether staging was based on clinical or pathologic information. Available treatment data included radiotherapy, surgery, or a combination of radiotherapy and surgery; SEER does not include chemotherapy information. Although SEER does capture some surgical data, it was not possible to determine whether patients received definitive-intent surgery or other diagnostic surgical procedures, such as tonsillectomies. We attempted to characterize the general percentage of patients that may have had radical surgery; patients were scored as having radical surgery if they had surgery to the primary site, excluding incisional or excisional biopsies. Details regarding radiation dose, technique, and treatment duration are not available in SEER. SEER also does not capture data on comorbidities or tobacco use. Outcome measures The primary outcomes were OS and head and neck cancer-specific mortality, defined as death attributed to head and neck cancer, as specified in SEER. SEER collects information on vital status (alive vs dead) and determines cause-specific death (cancer-related vs noncancerrelated) through methods including state health department records, the National Death Index, and information from treating physicians. OS was defined as the time from date of diagnosis to date of death, and head and neck cancer-specific mortality was defined as the time from date of diagnosis to date of head and neck cancerrelated death. Statistical analysis The Kaplan Meier method 11 was used to estimate OS and head and neck cancer-specific survival and associated 95% confidence intervals (CIs) by overall AJCC stage and RPA stage. Cox proportional hazards regression 12 was used to assess the effect of patient characteristics and disease factors on OS and head and neck cancer-specific mortality. The adjusted models included age at diagnosis, sex, race, geographic category, marital status, location of the primary tumor, radiotherapy, and surgery to the primary site (excluding incisional, needle, or aspiration biopsy), and we confirmed that the proportional hazards assumption was satisfied for these variables. We compared the adequacy of risk prediction of the 2 staging systems using the c-statistic, 13 which is similar to the area under the receiver operating characteristic curve. The difference in concordance probability between the AJCC staging system and the RPA-based stage grouping was also estimated. We also assessed the hazard discrimination, or heterogeneity between stages, and the hazard consistency, or homogeneity of survival within each stage, for both staging systems. All computed p values were 2-sided. Statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC) and R software (The R Foundation, Vienna, Austria). The study was reviewed and considered exempt by the institutional review board. RESULTS Population characteristics We identified 8427 patients age 18 years and older diagnosed with oropharyngeal SCC between January 1, 2004, and December 31, The baseline patient characteristics are summarized in Table 1. The age, race, and sex of the cohort were largely consistent with the typical demographic of HPV-related oropharyngeal SCC. 4,8,14 The median age was 58 years. The majority of patients were men, white, and married. Overall survival With a median follow-up of 50 months, there were 2946 deaths, including 2203 (74.8%) deaths from head and neck cancer. Both AJCC stage and RPA stage were significantly associated with OS in unadjusted (Supplementary Table 1) and adjusted (Table 2) models. RPA stage also effectively divided patients with oropharyngeal HEAD & NECK DOI /HED OCTOBER

3 KEANE ET AL. TABLE 1. Patient characteristics. No. of patients % Age at diagnosis, y > Median (range) 58 (21 96) Sex Female Male Race White Black American Indian/Alaska Native Asian or Pacific Islander Unknown 32 Spanish/Hispanic origin No Yes Geographical category for registry Northeast South Midwest West Marital status at diagnosis Married Single Separated Divorced Widowed Unknown 246 Primary site Base of tongue Tonsil Oropharynx, NOS T classification N classification a b c Overall AJCC stage I II III IVA IVB RPA stage I II III Radiotherapy No Yes Radical surgery No Yes Unknown Total 8427 Abbreviations: NOS, not otherwise specified; AJCC, American Joint Committee on Cancer; RPA, recursive partitioning analysis. SCC into prognostically distinct cohorts, with good hazard discrimination between stage groups, as the 95% CIs of the hazard ratios did not overlap (Table 3, Figure 1A). Although RPA stage had a strong monotonic relationship with OS (p <.0001), the relationship between AJCC stage and OS was not linear (Table 3, Figure 1B), and the survival curves overlapped over time, demonstrating the nonproportionality of AJCC stage in the model. For example, the patients with AJCC stage III grouping had better rates of survival than patients with AJCC stage I and II cancers. The magnitude of difference in overall mortality risk between AJCC stages, such as stage I to stage IV (adjusted hazard ratio [AHR] ; 95% CI ) was also modest compared to the magnitude of the difference between RPA-I and RPA-III (AHR ; 95% CI ; Table 2). Head and neck cancer specific survival Both RPA stage and AJCC stage were significantly associated with head and neck cancer-specific mortality in unadjusted (Supplementary Table 2) and adjusted (Table 4) models. However, although there was a linear relationship between RPA stage and head and neck cancer-specific mortality (Table 5, Figure 2A), the relationship between AJCC stage and head and neck cancer-specific mortality was not consistent. AJCC stage III was associated with improved 3-year and 5-year head and neck cancer-specific survival rates compared to AJCC stage II (Table 5, Figure 2B). Furthermore, unlike RPA stage, there was little distinction between head and neck cancer-specific survival rates, with overlapping confidence intervals for AJCC stages I, II, and III (Table 5). Comparison of the staging systems The RPA-based TNM stage grouping had significantly improved discriminatory power compared to AJCC staging between different AHRs for OS and head and neck cancerspecific mortality, as well as improved consistency with distinct and nonoverlapping CIs between RPA stages (Table 3, Table 5). Although neither staging system explained all the variation in the model for OS or head and neck cancer-specific mortality, the c-statistic of RPA stage was significantly higher than that of AJCC stage for both OS and head and neck cancer-specific mortality prediction. There was also a significant difference in the concordance probability between the 2 staging systems. The complete distribution of the c-statistics for both OS and head and neck cancer-specific mortality is shown in Table 6. DISCUSSION In this population-based validation study of patients with oropharyngeal SCC, we confirmed that the recently proposed RPA-based TNM stage grouping designed for HPV-related oropharyngeal SCC 8 provides superior OS and head and neck cancer-specific mortality prediction compared with the current AJCC staging system. Additionally, we demonstrated that the RPA-based TNM stage grouping provides better stratification of distinct risk groups with statistically significant differences in hazard 1532 HEAD & NECK DOI /HED OCTOBER 2016

4 SEER VALIDATION OF THE RECURSIVE PARTITIONING ANALYSIS BASED STAGING SYSTEM TABLE 2. Multivariable analysis for overall survival. AJCC stage RPA stage Variables AHR 95% CI p value AHR 95% CI p value Disease stage <.0001 <.0001 I II III IV Age at diagnosis, y <.0001 < > Sex Female Male Race <.0001 <.0001 White Black Other Spanish/Hispanic origin No Yes Geographical category for registry <.0001 <.0001 West Northeast South Midwest Marital status at diagnosis <.0001 <.0001 Married Other Primary site <.0001 <.0001 Tonsil Base of tongue Oropharynx, NOS Radiotherapy <.0001 <.0001 No Yes Radical surgery <.0001 <.0001 No/unknown Yes Abbreviations: AJCC, American Joint Committee on Cancer; RPA, recursive partitioning analysis; AHR, adjusted hazard ratio; CI, confidence interval; NOS, not otherwise specified. ratios per stage compared with that of the AJCC staging system, and is generalizable to include a population with unknown HPV status. It has been proposed that the declining prognostic significance of the AJCC staging system on OS and head and neck cancer-specific mortality reflects the increasing TABLE 3. Overall survival rates at 3 and 5 years. OS rate (95% CI) No. of patients No. of deaths 3 y 5 y Overall stage % ( ) 62.1% ( ) AJCC I % ( ) 68.2% ( ) II % ( ) 64.0% ( ) III % ( ) 68.3% ( ) IV % ( ) 59.0% ( ) RPA I % ( ) 70.0% ( ) II % ( ) 55.6% ( ) III % ( ) 44.3% ( ) Abbreviations: OS, overall survival; CI, confidence interval; AJCC, American Joint Committee on Cancer; RPA, recursive partitioning analysis. HEAD & NECK DOI /HED OCTOBER

5 KEANE ET AL. FIGURE 1. (A) Overall survival (OS) by recursive partitioning analysis (RPA) stage. (B) OS by American Joint Committee on Cancer (AJCC) stage. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] prevalence of HPV-related oropharyngeal SCC. The AJCC staging system, 10,15 which was first published in 1977, 3 was developed based on outcomes from patients with unknown HPV status, in an era when there was a lower prevalence of HPV, and primarily smokingrelated oropharyngeal SCC. This staging system relies in part on the assumption of a linear relationship between increasing N classification and worse prognosis. 3 However, in HPV-related oropharyngeal SCC, the excellent outcomes, 5,14,16 despite the presence of advanced nodal disease, limit the predictive ability of the AJCC staging system. 17 Several surgical series have reported the declining prognostic impact of AJCC N classification on outcomes in patients with HPV-related oropharyngeal SCC. 7,18,19 In addition, prior studies have demonstrated a nonlinear relationship between N classification and head and neck cancer-specific mortality in HPV-related oropharyngeal SCC, with variable outcomes specifically within the N2 subgroup. 14,20,21 Huang et al 8 found that OS was increased in patients with N2a disease compared to N0 or N1 disease, a finding that was also demonstrated in a population-level analysis of patients with oropharyngeal SCC in the HPV-era. 2 Unlike the AJCC staging system, the RPAbased TNM stage grouping accounts for the nonlinear relationship of N classification with OS and head and neck cancer-specific mortality by grouping all patients with non-t4 primary tumors and unilateral nodal disease (excluding N3) into 1 category. Patients in the RPA-III group (those with T4 and/or N3 disease) still have poor outcomes, which may be related in part to patients who also have significant tobacco histories 7,22 24 or significant comorbidities. 25 Unlike the AJCC stage IVA group, which includes some T4 tumors 1534 HEAD & NECK DOI /HED OCTOBER 2016

6 SEER VALIDATION OF THE RECURSIVE PARTITIONING ANALYSIS BASED STAGING SYSTEM TABLE 4. Multivariable analysis for head and neck cancer specific mortality. AJCC stage RPA stage Variable AHR 95% CI p value AHR 95% CI p value Disease stage <.0001 <.0001 I II III IV Age at diagnosis, y <.0001 < > Sex Female Male Race <.0001 <.0001 White Black Other Spanish/Hispanic origin No Yes Geographical category for registry West Northeast South Midwest Marital status at diagnosis <.0001 <.0001 Married Other Primary site <.0001 <.0001 Tonsil Base of tongue Oropharynx, NOS Radiotherapy <.0001 <.0001 No Yes Radical surgery <.0001 <.0001 No/unknown Yes Abbreviations: AJCC, American Joint Committee on Cancer; RPA, recursive partitioning analysis; AHR, adjusted hazard ratio; CI, confidence interval; NOS, not otherwise specified. with more favorable T1 to T2 tumors, the RPA-based stage grouping creates a separate category for patients with particularly advanced disease. Although the RPA-based stage grouping had improved predictive ability compared with the AJCC staging classification system (c-statistic vs 0.55 for head and TABLE 5. Head and neck cancer specific survival rates at 3 and 5 years. No. of patients CSS rate (95% CI) No. of head and neck related deaths 3 y 5 y Overall stage % ( ) 70.7% ( ) AJCC I % ( ) 81.3% ( ) II % ( ) 75.4% ( ) III % ( ) 76.7% ( ) IV % ( ) 66.8% ( ) RPA I % ( ) 79.1% ( ) II % ( ) 61.8% ( ) III % ( ) 52.1% ( ) Abbreviations: CSS, cancer-specific survival; CI, confidence interval; AJCC, American Joint Committee on Cancer; RPA, recursive partitioning analysis. HEAD & NECK DOI /HED OCTOBER

7 KEANE ET AL. FIGURE 2. (A) Head and neck cancer specific survival (HNCSS) by recursive partitioning analysis (RPA) stage. B: HNCSS by American Joint Committee on Cancer (AJCC) stage. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary. com.] neck cancer-specific mortality), it did not explain all variations in the model for OS or head and neck cancerspecific mortality. Incorporation of smoking status, comorbidity, and other biologic and demographic factors may enhance the predictive ability of the staging systems. Accurate prediction of patient outcomes is essential, given the ongoing focus on deintensification of treatment in HPV-related oropharyngeal SCC. Eastern Cooperative Oncology Group (ECOG) 1308, 26 a phase II study, demonstrated in preliminary results that patients with T1-2N0-2b HPV-related oropharyngeal SCC with a complete clinical response to induction chemotherapy had excellent OS and progression-free survival despite receiving dosereduced radiotherapy. Currently accruing trials include TABLE 6. C-statistics for head and neck cancer specific mortality and overall survival. AJCC stage RPA stage Endpoint Est SE 95% CI Est SE 95% CI OS Head and neck cancer specific mortality Abbreviations: AJCC, American Joint Committee on Cancer; RPA, recursive partitioning analysis; Est, point estimates; CI, confidence interval; OS, overall survival HEAD & NECK DOI /HED OCTOBER 2016

8 SEER VALIDATION OF THE RECURSIVE PARTITIONING ANALYSIS BASED STAGING SYSTEM Radiation Therapy Oncology Group (RTOG) 1016, 27 ECOG 3311, 28 and NRG-HN RTOG is a phase III trial in which patients will receive intensitymodulated radiotherapy with either cisplatin or cetuximab in an effort to deintensify systemic treatment. ECOG uses transoral resection to risk-stratify and randomize patients with HPV-related stage III and IVA oropharyngeal SCC to standard versus lower-dose radiotherapy. NRG-HN is a phase III trial of intensity-modulated radiotherapy with or without cisplatin. Both RTOG 1016 and NRG-HN002 are testing the potential for decreasing systemic treatment in HPVrelated oropharyngeal SCC. The present study had several strengths. SEER is a nationwide registry that collects much additional data, with results representing generalizable trends across a large population. The large patient and event numbers in SEER allowed us to compare the predictive ability of both staging systems, in addition to controlling for known factors that impact cancer-specific and OS, such as age, race, and marital status The primary limitation of this study was that SEER does not capture HPV status and, therefore, it was not possible to analyze the RPA-based TNM stage grouping exclusively in HPV-related oropharyngeal SCC. However, we focused our analysis on the recent time period of 2004 to 2008 in which HPV-related oropharyngeal SCC has previously been shown to be more common than HPV-negative oropharyngeal SCC, with an estimated prevalence of HPV-related oropharyngeal SCC of 65.4% in 2000 to 2013 versus 51.4% in 1990 to Despite the inclusion of patients with HPV-negative oropharyngeal SCC, the RPA-based TNM stage grouping was still able to divide patients into distinct prognostic subgroups by RPA stage, and was, in fact, superior to the AJCC staging system. An additional limitation was the smaller relative patient numbers in the RPA-II cohort, which, in turn, limited the power to estimate OS and head and neck cancer-specific survival. Of note, the rates of OS and head and neck cancer-specific survival in RPA-II were also lower in our cohort as compared to the PMCC cohort, which may reflect the heterogeneity of HPV status of the patients in our study. Although we were able to control for multiple factors that impact outcomes, including age, sex, race, and marital status, SEER does not include other patientspecific variables, such as tobacco and alcohol use, 14,33,34 which are known to impact cancer-specific survival and OS. As such, we were unable to assess a secondary prognostic system proposed by PMCC that incorporated tobacco use. There is also no information on comorbidity or performance status in SEER, which would impact both receipt of treatment, cancer-specific survival, and OS. 25 SEER does not include data on chemotherapy, and there is limited information on the extent of surgery and details of radiotherapy. Additional information on receipt of treatment, particularly chemotherapy, may have impacted OS and head and neck cancer-specific survival estimates, particularly for patients with N2c disease. 8 There is also no metric in SEER assessing the quality of various treatments. Despite these limitations, the proposed RPA-based TNM stage grouping is validated in the general U.S. population of patients with oropharyngeal SCC with unknown HPV status and provides superior survival prediction and hazard discrimination compared with the AJCC staging classification system for oropharyngeal cancer. This study supports current efforts to update the AJCC staging system for oropharyngeal cancer and to refine the staging system using the proposed RPA-based staging assignment. REFERENCES 1. Dahlstrom KR, Calzada G, Hanby JD, et al. An evolution in demographics, treatment, and outcomes of oropharyngeal cancer at a major cancer center: a staging system in need of repair. Cancer 2013;119: Keane FK, Chen YH, Neville BA, et al. Changing prognostic significance of tumor stage and nodal stage in patients with squamous cell carcinoma of the oropharynx in the human papillomavirus era. Cancer 2015;121: Beahrs OH, Carr DT, Rubin P. AJCC Manual of Staging for Cancer. 1st ed. Chicago, IL: American Joint Committee; Chaturvedi AK, Engels EA, Anderson WF, Gillison ML. Incidence trends for human papillomavirus-related and -unrelated oral squamous cell carcinomas in the United States. J Clin Oncol 2008;26: Chaturvedi AK, Engels EA, Pfeiffer RM, et al. Human papillomavirus and rising oropharyngeal cancer incidence in the United States. J Clin Oncol 2011;29: Stein AP, Saha S, Yu M, Kimple RJ, Lambert PF. Prevalence of human papillomavirus in oropharyngeal squamous cell carcinoma in the United States across time. Chem Res Toxicol 2014;27: Spector ME, Gallagher KK, Bellile E, et al. Patterns of nodal metastasis and prognosis in human papillomavirus-positive oropharyngeal squamous cell carcinoma. Head Neck 2014;36: Huang SH, Xu W, Waldron J, et al. Refining American Joint Committee on Cancer/Union for International Cancer Control TNM stage and prognostic groups for human papillomavirus-related oropharyngeal carcinomas. J Clin Oncol 2015;33: National Cancer Institute. Surveillance, Epidemiology, and End Results Program. Cancer statistics. Available at: Seer.Cancer.Gov/Data. Accessed July 1, Edge SB, Byrd DR, Compton CC, Greene FL, Fritz AG, Greene FL, Trotti A III, eds. AJCC Cancer Staging Manual. 6 ed. Chicago, IL: Springer; Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53: Cox DR. Regression models and life tables. J R Stat Soc Series B Stat Methodol 1972;34: Uno H, Cai T, Pencina MJ, D Agostino RB, Wei LJ. On the C-statistics for evaluating overall adequacy of risk prediction procedures with censored survival data. Stat Med 2011;30: Ang KK, Harris J, Wheeler R, et al. Human papillomavirus and survival of patients with oropharyngeal cancer. N Engl J Med 2010;363: Edge S, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, eds. AJCC Cancer Staging Handbook. 7 ed. Chicago, IL: Springer; Fakhry C, Westra WH, Li S, et al. Improved survival of patients with human papillomavirus-positive head and neck squamous cell carcinoma in a prospective clinical trial. J Natl Cancer Inst 2008;100: Goldenberg D, Begum S, Westra WH, et al. Cystic lymph node metastasis in patients with head and neck cancer: an HPV-associated phenomenon. Head Neck 2008;30: Hong AM, Martin A, Armstrong BK, et al. Human papillomavirus modifies the prognostic significance of T stage and possibly N stage in tonsillar cancer. Ann Oncol 2013;24: Haughey BH, Sinha P. Prognostic factors and survival unique to surgically treated p161 oropharyngeal cancer. Laryngoscope 2012;122 Suppl 2:S13 S Haughey BH, Hinni ML, Salassa JR, et al. Transoral laser microsurgery as primary treatment for advanced-stage oropharyngeal cancer: a United States multicenter study. Head Neck 2011;33: O Sullivan B, Huang SH, Perez Ordonez B, et al. Outcomes of HPVrelated oropharyngeal cancer patients treated by radiotherapy alone using altered fractionation. Radiother Oncol 2012;103: Rich JT, Milov S, Lewis JS Jr, Thorstad WL, Adkins DR, Haughey BH. Transoral laser microsurgery (TLM) 1/- adjuvant therapy for advanced stage oropharyngeal cancer: outcomes and prognostic factors. Laryngoscope 2009;119: Maxwell JH, Kumar B, Feng FY, et al. Tobacco use in human papillomavirus-positive advanced oropharynx cancer patients related to increased risk of distant metastases and tumor recurrence. Clin Cancer Res 2010;16: HEAD & NECK DOI /HED OCTOBER

9 KEANE ET AL. 24. Rades D, Seibold ND, Gebhard MP, Noack F, Schild SE, Thorns C. Prognostic factors (including HPV status) for irradiation of locally advanced squamous cell carcinoma of the head and neck (SCCHN). Strahlenther Onkol 2011;187: Rose BS, Jeong JH, Nath SK, Lu SM, Mell LK. Population-based study of competing mortality in head and neck cancer. J Clin Oncol 2011;29: Cmelak A, Li S, Marur S, et al. E1308: Reduced-dose IMRT in human papilloma virus (HPV)-associated resectable oropharyngeal squamous carcinomas (OPSCC) after clinical complete response (ccr) to induction chemotherapy (IC). J Clin Oncol 2014;32(5 suppl); abstract LBA ClinicalTrials.gov. Radiation therapy with cisplatin or cetuximab in treating patients with oropharyngeal cancer. Available at: gov identifier NCT Accessed July 1, ClinicalTrials.gov. Transoral surgery followed by low-dose or standarddose radiation therapy with or without chemotherapy in treating patients with HPV positive stage III-IVA oropharyngeal cancer. Available at: www. clinicaltrials.gov identifier NCT Accessed July 1, ClinicalTrials.gov. Treatment de-intensification for locally advanced HPVassociated oropharyngeal cancer. Available at: identifier NCT Accessed July 1, Aizer AA, Chen MH, McCarthy EP, et al. Marital status and survival in patients with cancer. J Clin Oncol 2013;31: Liu GF, Ranck MC, Solanki AA, et al. Racial parities in outcomes after radiotherapy for head and neck cancer. Cancer 2014;120: Inverso G, Mahal BA, Aizer AA, Donoff RB, Chau NG, Haddad RI. Marital status and head and neck cancer outcomes. Cancer 2015;121: Gillison ML, Zhang Q, Jordan R, et al. Tobacco smoking and increased risk of death and progression for patients with p16-positive and p16- negative oropharyngeal cancer. J Clin Oncol 2012;30: Mayne ST, Cartmel B, Kirsh V, Goodwin WJ Jr. Alcohol and tobacco use prediagnosis and postdiagnosis, and survival in a cohort of patients with early stage cancers of the oral cavity, pharynx, and larynx. Cancer Epidemiol Biomarkers Prev 2009;18: HEAD & NECK DOI /HED OCTOBER 2016