Annual percentage change (APC) was calculated using log-linear regression. P < .05 denotes a statistically significant change in the incidence of OPSCC.
Annual percentage change (APC) was calculated using log-linear regression. P values are based on the APC of each age subgroup within a given cancer site.
P values represent the comparison of the linear regression line and a line with slope equal to zero for each age group.
eTable 1. Baseline characteristics of patients diagnosed with oropharyngeal cancer from 2000 to 2012 in 18 SEER registries
eTable 2. The mean age-adjusted incidence (AAI) and the annual percentage change (APC) for tobacco and alcohol-associated squamous cell carcinomas of the head and neck from 2000-2012, stratified by age at diagnosis
eTable 3. The mean age-adjusted incidence (AAI) and the annual percentage change (APC) for tobacco-associated lung cancers and anal cancer, an HPV-associated malignancy, are given for patients diagnosed between 2000 and 2012, stratified by age at diagnosis
eTable 4. Trends in Kaplan-Meier estimated 3 year overall survival (OS), cause-specific survival (CSS), relative survival (RS) compared to the age-matched United States population for patients diagnosed with oropharyngeal squamous cell carcinoma from 2000 to 2009, stratified by age at diagnosis
eFigure 1. Use of definitive local therapy for patients with oropharyngeal squamous cell carcinoma diagnosed from 2000 to 2012, stratified by age at diagnosis
eFigure 2. The age-adjusted incidence and annual percentage change (APC) of base of tongue and tonsil cancers, in comparison to cancers arising in other oropharynx sites, from 2000 to 2012, stratified by age at diagnosis
eFigure 3. The ratio of the incidence of oropharyngeal squamous cell carcinoma (OPSCC) in men in comparison to women over time, stratified by age
eFigure 4. The age-adjusted incidence of oropharyngeal squamous cell carcinoma from 2000 to 2012, stratified by age of diagnosis, sex and race
eFigure 5. The age-adjusted incidences of A) lung squamous cell carcinoma and B) small cell lung cancer from 2000 to 2012, stratified by age at diagnosis
eFigure 6. Estimated three year A) overall survival and B) relative survival in comparison to the expected survival of an age-matched population based on United States 2000 census data, for patients with oropharyngeal squamous cell carcinoma, based on year of diagnosis and stratified by age
eFigure 7. Estimated three year cause specific survival for a) laryngeal cancer, b) oral cavity cancer, and c) hypopharyngeal cancer, based on year of diagnosis and stratified by age
Customize your JAMA Network experience by selecting one or more topics from the list below.
Zumsteg ZS, Cook-Wiens G, Yoshida E, et al. Incidence of Oropharyngeal Cancer Among Elderly Patients in the United States. JAMA Oncol. 2016;2(12):1617–1623. doi:10.1001/jamaoncol.2016.1804
Is the increasing incidence of oropharyngeal cancer, initially observed primarily in middle-aged patients, also occurring in elderly patients in recent years?
This study of SEER data found that the incidence of oropharyngeal cancer increased between 2000 and 2012 in patients 65 years and older, whereas smoking-associated head and neck cancers decreased in this population. This was mostly driven by cancers in white men.
Given the rapidly increasing prevalence of oropharyngeal cancer in elderly patients, the poor representation of this group in clinical trials, and the challenges treating these patients with multimodality therapy, prospective clinical trials specifically focusing on this group should be performed.
An escalating incidence of human papillomavirus (HPV)-related oropharyngeal squamous cell carcinoma (OPSCC) has been reported predominantly among middle-aged adults. However, HPV infection is believed to occur many years before cancer develops, and tissue studies suggest that HPV DNA is found in the majority of OPSCC diagnosed in patients 65 years or older.
To update the trends in OPSCC incidence using US cancer registry data, with an emphasis on age-specific trends.
Design, Setting, and Participants
Data from the Surveillance, Epidemiology, and End Results (SEER) database (2000-2012) were queried to compare changes in incidence and survival trends in OPSCC with selected tobacco-related cancers (larynx, oral cavity, hypopharynx, lung) and an HPV-related cancer (anus). A total of 40 264 patients who received a diagnosis of OPSCC from 2000 to 2012 were included. Elderly patients were defined as those 65 years or older.
Main Outcomes and Measures
The annual percentage change in OPSCC incidence from 2000 to 2012, stratified according to age group.
Among the 40 264 patients who received a diagnosis of OPSCC from 2000 to 2012, 13 313 (33.1%) were aged 65 years or older and 80.3% were male. Significant increases in the age-adjusted incidence of OPSCC were observed during the study period for both younger adults aged 45 to 64 years (annual percentage change [APC], 2.31; 95% CI, 1.76-2.86; P < .001) and patients 65 years or older (APC, 2.92; 95% CI, 2.32-3.51; P < .001). These changes were driven predominantly by base-of-tongue and tonsil cancers in men. Concomitantly, the incidence of tobacco-associated head and neck cancers decreased for elderly patients (larynx: APC, −1.54; 95% CI, −2.00 to −1.08; P < .001; oral cavity: APC, −1.23; 95% CI, −1.84 to −0.62; P = .001; hypopharynx: APC, −2.44; 95% CI, −3.01 to −1.86; P < .001), whereas the incidence of anal cancer significantly increased (APC, 4.42; 95% CI, 3.28 to 5.57; P < .001). Furthermore, improved overall and cause-specific survival over time were observed for both younger and elderly patients with OPSCC. Nevertheless, absolute cause-specific survival remained worse for elderly patients (3-year CSS, 60.8%; 95% CI, 59.6%-61.9%) in comparison with those aged 45 to 64 years (75.7%; 95% CI, 75.1%-76.4%; P < .001).
Conclusions and Relevance
The incidence of OPSCC is increasing among elderly patients in the United States, likely driven by HPV-associated cancers. Given the unique challenges related to treating elderly patients with OPSCC, their limited enrollment in clinical trials, and the aging US population, clinical studies investigating improved therapeutic strategies for elderly patients with HPV-positive OPSCC should be performed.
Over the past several decades, the epidemiologic and clinical landscape of head and neck cancer has radically transformed in the United States and other economically developed countries.1-7 This transformation has been driven by substantial shifts in the prevalence of the 2 major risk factors underlying head and neck carcinogenesis, with markedly declining tobacco use8 and increasing prevalence of human papillomavirus (HPV) in oropharyngeal squamous cell carcinomas (OPSCCs) from these regions.9 These altered risk profiles have led to significant decreases in incidence of cancers arising in tobacco-associated head and neck cancer sites, such as the oral cavity and larynx, but increased incidence of HPV-associated OPSCCs.1-3 In fact, a recent meta-analysis found that the prevalence of HPV detected in OPSCCs in the United States increased from approximately 20% prior to 1990 to 65% in 2000.10 As a result of these trends, OPSCC is projected to represent more than half of all head and neck cancers in the United States by 2030.2
Studies to date have suggested that HPV-associated OPSCCs are more commonly diagnosed in younger patients1,3,11-14 and that the increasing incidence of OPSCC is found almost exclusively among middle-aged individuals.1,3 These reports have led to the widespread belief that HPV-associated OPSCC affects predominantly younger, otherwise healthy patients.15 However, these studies describe incidence trends that were established more than a decade ago and do not account for more recent changes in patterns of HPV infection and associated cancers in the United States. In fact, an analysis of 2009 through 2010 National Health and Nutrition Examination Survey (NHANES) data shows the highest prevalence of oral HPV infection in 55- to 60-year-olds,16 and a recent study observed high-risk HPV DNA in 65% of OPSCC specimens from patients aged 70 years and older.17 Thus, HPV-associated cancers in elderly patients may be more common in contemporary clinical practice than suggested by earlier reports.
The elderly represent a unique subset of patients with head and neck cancer given that they do not generally benefit from intensification strategies such as concomitant chemoradiation therapy,18 concomitant cetuximab-based bioradiation therapy,19 or altered fractionation20 in comparison with radiotherapy alone. Furthermore, elderly patients have historically been underrepresented in head and neck cancer clinical trials.18 Thus, demographic shifts in HPV-associated head and neck cancer would have important implications for future clinical investigations. In this analysis, we explore trends in OPSCC incidence and survival in the United States from 2000 to 2012 using the Surveillance, Epidemiology, and End results (SEER) database, with a specific focus on the epidemiologic patterns among patients 65 years and older.
All data regarding demographic characteristics and cancer incidence were obtained from the SEER-18 database spanning 2000 through 2012, derived from 18 cancer registries across the United States, covering approximately 28% of incident cases in the United States (http://seer.cancer.gov). Race information was extracted from the SEER-18 database and analyzed given that previous publications have observed an increasing incidence of OPSCC only in white patients.1 This study was deemed to be exempt from full institutional review board review by Cedars-Sinai Medical Center due to the retrospective nature of the study.
Cases of OPSCC were identified on the basis of their International Classification of Diseases for Oncology, Third Edition (ICD-O-3), primary site codes as arising in the tonsil (C090-C091, C098-C099), base of tongue (C019, C024), Waldeyer ring (C142), vallecula (C100), soft palate (C051), uvula (C052), oropharyngeal wall (C102-103), or other oropharynx site (C108-C109). Anatomic codes considered ambiguous for oropharynx origin, including C028, C029, C101, and C140, were excluded. Given that tobacco and alcohol are the other main etiologic risk factors for OPSCC other than oral HPV infection, we also analyzed changes in the incidence of other head and neck sites, such as the larynx (C320-329), hypopharynx (C129-139), and oral cavity, that are driven primarily by these factors. Oral cavity cancers were tumors arising in the lip (C000-009), oral tongue (C020-023), gums (C030-031, C039), floor of mouth (C040-049), hard palate (C050), cheek (C060), vestibule of mouth (C061), or retromolar area (C062). Lung (C340-349) and anal (C210-219) cancers were also analyzed to determine changes in incidence trends in cancers driven primary by tobacco use and HPV, respectively, arising outside the head and neck.
Only patients with invasive squamous cell histologic subtype based on ICD-O-3 were included for all head and neck cancer sites (ICD-O-3 histology codes 8051-8052, 8070-8078, 8083, or 8084). For lung cancer, both squamous and small-cell histologic subtypes (8041-8045) were included, given that these histologic subtypes are most strongly associated with tobacco use.21
All analyses were performed using SEER*Stat, version 8.2.1, Joinpoint Regression Program, version 18.104.22.168, or R, version 3.2.2. Elderly patients were defined as those 65 years and older at the time of diagnosis. Younger patients were dichotomized into ages 45 to 64 and younger than 45 years, given the low incidence of OPSCC in the latter group. For more granular analysis, patients were also classified by 5-year age groups, except for subgroups containing patients younger than 40 years and 80 years or older, respectively.
Baseline characteristics were compared between elderly and nonelderly patients using the χ2 or Fisher exact test. Age-adjusted incidence rates, expressed per 100 000 persons, were calculated as the weighted average of crude cancer rates standardized to the US Census 2000 population. Annual percentage change (APC) was calculated with log-linear regression, and comparison of the parallelism of incidence trends was performed with the Joinpoint Regression Program.22 Given the relatively short study period, the maximum number of joinpoints was set to zero to avoid overfitting the data. Trends in male to female incidence ratios were analyzed with linear regression. Graphs were created in GraphPad Prism and Tableau.
Estimated overall survival and cause-specific survival (CSS) was calculated according to the Kaplan-Meier method. For relative survival, the ratio of the observed survival to the expected survival, calculated according to the Ederer II method, is reported.23 Time trends in 3-year survival, based on year of diagnosis, were assessed with linear regression. Treatment of the primary site (surgery, radiation, or none) was determined on the basis of the initial local treatment delivered. Two-sided P values <.05 were considered significant.
Among the 40 264 patients who received a diagnosis of OPSCC from 2000 to 2012, 13 313 (33.1%) were aged 65 years or older and 80.3% were male (eTable 1 in the Supplement). Patients 65 years or older at diagnosis were less likely to be treated with surgery and more likely to receive no treatment to the primary site compared with younger patients (eFigure 1 in the Supplement).
From 2000 to 2012, the age-adjusted incidence of OPSCC significantly increased for patients aged 65 years or older (APC, 2.92; 95% CI, 2.32-3.51; P < .001) at a similar rate (parallelism P = .24) to what was observed in those aged 45 to 64 years (APC, 2.31; 95% CI, 1.76-2.86; P < .001) (Figure 1A). In both subgroups, these changes were driven by significant increases in tonsil and base-of-tongue primaries, whereas other oropharynx subsites were stable or decreased (eFigure 2 in the Supplement). When stratifying patients into subgroups by 5-year increments, the incidence of OPSCC significantly increased for all subgroups between the ages of 50 and 79 years (Table). Although there was no overall increase in total OPSCC for patients aged 80 years or older (APC, 1.47; 95% CI, −0.24 to 3.21; P = .09), there was a significant increase in the incidence of tonsil and base-of-tongue primaries in these patients (APC, 2.01; 95% CI, 0.31-3.74; P = .02).
Stratifying by sex, similar to the overall cohort, men aged 65 years and older experienced an increased incidence of OPSCC from 2000 to 2012 (APC, 3.64; 95% CI, 2.94-4.34; P < .001) (Figure 1B and Table). Notably, this included a significant increase in the incidence of OPSCC among men 80 years or older (APC, 2.03; 95% CI, 0.11-3.99; P = .04). Additionally, the APC for elderly men was significantly higher (parallelism P = .01) than that observed in men aged 45 to 64 years (APC, 2.50; 95% CI, 1.86-3.15; P < .001). By contrast, although there was a modest increase in the incidence of OPSCC among women aged 45 to 64 years (APC, 1.25; 95% CI, 0.61-1.88; P = .001), no increase in OPSCCs was observed for any other age group, including women 65 years and older. As a result of these trends, increases in the male to female incidence ratios for OPSCC were observed in both elderly and younger patients (45 to 64 years) over time (eFigure 3 in the Supplement). When further stratifying based on race, both within the entire cohort and among elderly patients, increases in the incidence of OPSCC were observed primarily in white men (eFigure 4 in the Supplement).
In contrast to trends observed for OPSCC, there were significant decreases in the incidence of tobacco- and alcohol-related head and neck cancers from 2000 to 2012 among almost all subgroups of patients, including elderly patients (Figure 2 and eTable 2 in the Supplement). Similarly, significant decreases in the incidence of tobacco-related lung cancer histologic subtypes, such as squamous cell carcinoma and small-cell carcinoma, were nearly universally observed, except for squamous cell carcinoma in elderly patients (eFigure 5 and eTable 3 in the Supplement). By contrast, the incidence of anal cancer significantly increased for all age groups 45 years and older.
We found that for both younger and elderly patients with OPSCC, the probability of 3-year overall survival and CSS significantly increased from 2000 to 2009 (Figure 3 and eFigure 6A and eTable 4 in the Supplement). Similar trends were observed when analyzing survival in patients with OPSCC relative to the expected survival of an aged-matched population (eFigure 6B in the Supplement). Despite relative increases in survival across age groups, CSS remained significantly worse for patients 65 years and older (3-year CSS, 60.8%; 95% CI, 59.6%-61.9%) in comparison with those aged 45 to 64 years (75.7%; 95% CI, 75.1%-76.4%; P < .001) (Figure 4). Of note, overall, cause-specific, and relative survival were stable during this period for elderly patients with head and neck cancers arising in tobacco- and alcohol-related primary sites (eFigure 7 in the Supplement).
In this study, we show that contrary to the conventional belief that the epidemic of HPV-driven OPSCC affects predominantly middle-aged individuals, the incidence of OPSCC in the United States has increased dramatically across a broad spectrum of patient age groups since 2000, including patients aged 65 years or older. In fact, the APC of OPSCC for elderly patients closely mirrored that of younger patients aged 45 to 64 years in the SEER data set. Although the SEER database does not collect data on p16 immunohistochemical analysis or other HPV-specific testing, multiple lines of evidence suggest that the increased incidence of OPSCCs among elderly patients is a direct consequence of increased rates of oral HPV infections and consequent HPV-driven cancers in this population. For example, the increase in OPSCC in the elderly was primarily driven by cancers in white men and tumors arising from the tonsil and base of tongue, consistent with what would be expected from HPV-associated cancers. Moreover, during the study period, near-universal decreases in the incidence of tobacco- and alcohol-related head and neck cancers such as laryngeal cancer, oral cavity cancer, and hypopharyngeal cancer were observed across virtually all age subgroups with increasing OPSCC incidence, including patients aged 65 years or older. Similar negative trends were observed in tobacco-associated lung cancer histologic subtypes, such as squamous cell and small-cell carcinoma. By contrast, the incidence of anal cancer, a disease associated with HPV in more than 90% of cases,24 markedly increased across most age groups from 2000 to 2012, including in the elderly population. Furthermore, given that HPV-related OPSCC has been associated with improved survival in comparison with HPV-negative OPSCCs,12-14,25,26 we performed survival analysis among elderly patients with OPSCC and observed significantly improved overall, cause-specific, and relative survival during the interval of this study at a similar rate to what was observed for younger patients. By contrast, survival outcomes were stable for elderly patients with head and neck cancers arising in tobacco- and alcohol-associated primary sites, suggesting that these survival improvements for OPSCC in the elderly may be due to changes in the etiology of this disease rather than improvements in treatment of head and neck cancers. Taken together, these data provide strong circumstantial support for HPV as the primary driver of the changes in incidence and outcomes for OPSCC among elderly patients.
Several other data sets support the high prevalence of HPV-associated OPSCC among the elderly in the modern era. For example, in a recent analysis of the 2009 to 2010 NHANES data set comprising more than 9000 patients aged 14 to 69 years, the highest incidence of active oral HPV infection, as detected by polymerase chain reaction, was observed in the subgroup aged 55 to 60 years.16 Given that the typical latency of development of an HPV-driven head and neck cancer is estimated to be 10 to 30 years, almost all cancers due to these infections would be expected to occur in patients aged 65 years or older. Additionally, a recent US Centers for Disease Control and Prevention study of 557 OPSCCs diagnosed between 1995 and 2005 reported that oncogenic HPV DNA was detected in 65% of OPSCCs from patients 70 years or older.17 This was identical to the proportion of HPV-positive tumors found in patients aged 60 to 69 years, and only approximately 10% lower than the rate detected in patients 59 years or younger. In combination with our results, these data provide an emerging picture of the relatively high, and increasing, burden of HPV oral infection and HPV-associated OPSCC among elderly patients since 2000.
Our data diverge significantly from what has been reported previously. For example, in one of the seminal reports in the epidemiology of OPSCC, also using the SEER database, there was no significant increase in the incidence of OPSCC among the population 60 years and older in the United States from 1973 to 2004.1 Similarly, an international study using the Cancer Incidence in Five Continents database found that in 7 of 9 countries with an increasing incidence of OPSCC, significantly stronger trends were observed in patients younger than 60 years in comparison with older patients from 1983 to 2002.3 These differences in age-specific trends in OPSCC in data sets that are little more than a decade older than that used for this analysis highlight the rapidly changing epidemiology of OPSCC in the United States.
The ramifications of these changes in the incidence of OPSCC among elderly patients will be amplified by concomitant changes in the demographic characteristics of the US population over the next several decades. Specifically, the number of adults 65 years and older is projected to increase from 35 million in 2000 to 72 million by 2030.27 A multiplicative effect on the absolute number of elderly patients with OPSCC would thus be expected to result from the combination of a growing number of elderly Americans and an increased proportion harboring oral HPV infections. Moreover, older men seem to have lower clearance rates of oral HPV than younger men,28,29 which may result in a higher rate of malignant progression among elderly patients. Collectively, these data support that HPV-associated OPSCC in the elderly will become an increasingly common medical condition facing patients and clinicians in the United States.
Importantly, these results also have implications for future clinical investigations in elderly patients, a group that is historically underrepresented in head and neck cancer clinical trials.18 It is widely recognized that the treatment of elderly patients with head and neck cancer is challenging.30 Intensification strategies, including the addition of concomitant chemotherapy, concomitant cetuximab, and altered fractionation, have demonstrated improvements in both locoregional control and overall survival in patients with head and neck cancer but have failed to show a survival benefit among the elderly subgroup compared with radiotherapy alone.18-20 This is likely due, at least in part, to the poor tolerance of these intense treatment regimens in a relatively frail population. Nevertheless, the elderly population is heterogeneous, and elderly patients with HPV-positive OPSCCs may tolerate multimodality therapy better than the lifelong tobacco and alcohol users who accounted for the majority of patients in historical head and neck cancer trials.
Given the excellent locoregional outcomes observed for HPV-positive OPSCCs,12,13,26 deintensification strategies are currently being investigated in this disease and may help decrease toxicity while maintaining acceptable oncologic outcomes.31 Although these approaches appear particularly appealing for elderly patients, they should be implemented with caution in this population given that our study found CSS to be worse for elderly patients in comparison with younger patients with OPSCC. Although these worse outcomes could be due to treatment-related morbidity, hesitance to administer aggressive multimodality therapy, differences in utilization of definitive treatments, or persistent absolute differences in the proportion of HPV-positive OPSCCs in elderly vs younger patients, they underscore the need for more robust therapeutic paradigms in this population. Thus, trials that optimize management specifically for this unique population are imperative.
Aside from the absence of information on p16 immunohistochemistry or other markers that distinguish between HPV-positive and -negative OPSCCs, a further caveat of our analysis is that we cannot entirely rule out the possibility that changes in OPSCC survival could be influenced by treatment-related improvements involving radiation therapy techniques, systemic therapy, or supportive care. However, none of these strategies has demonstrated the ability to improve survival in elderly patients in clinical trials. Furthermore, no change in CSS was seen among elderly patients for other tobacco- and alcohol-related head and neck treatment sites during the period studied.
We report that the incidence of OPSCC from 2000 to 2012 increased over a broad population of patient age groups, including the elderly, with no evidence of plateau in this subgroup. Based on multiple lines of circumstantial evidence, the increased incidence of OPSCC in elderly patients is likely attributable to increased HPV-associated carcinogenesis. Thus, HPV-associated OPSCC should no longer be considered a disease primarily afflicting younger patients. In combination with projected increases in the size of the US elderly population in the coming decades, HPV-associated OPSCCs in older patients will likely represent an exponentially growing problem in the coming decades. Given the unique challenges of treating elderly patients with head and neck cancer and their historical underrepresentation in clinical trials, future clinical investigations are needed to identify optimal therapeutic strategies for this population.
Accepted for Publication: April 19, 2016.
Corresponding Author: Zachary S. Zumsteg, MD, Department of Radiation Oncology, Cedars-Sinai Medical Center, 8700 Beverly Blvd, Los Angeles, CA 90048 (firstname.lastname@example.org).
Published Online: July 14, 2016. doi:10.1001/jamaoncol.2016.1804
Author Contributions: Dr Zumsteg had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Zumsteg, Lee, Jeon, Goodman, Ho.
Acquisition, analysis, or interpretation of data: Zumsteg, Cook-Wiens, Yoshida, Shiao, Lee, Mita, Jeon, Goodman.
Drafting of the manuscript: Zumsteg, Cook-Wiens, Yoshida, Goodman, Ho.
Critical revision of the manuscript for important intellectual content: Zumsteg, Shiao, Lee, Mita, Jeon, Goodman, Ho.
Statistical analysis: Zumsteg, Cook-Wiens, Shiao, Jeon, Goodman, Ho.
Administrative, technical, or material support: Yoshida, Shiao, Ho.
Study supervision: Zumsteg, Lee, Ho.
Conflict of Interest Disclosures: Dr Zumsteg is on the external advisory board for the Scripps Proton Therapy Center. Dr Lee is on the external advisory board for Merck, Vertex Pharmaceuticals, and Pfizer. Dr Goodman is a consultant for Johnson and Johnson. No other disclosures are reported.