Race and sex disparities in long-term survival of oral and oropharyngeal cancer in the United States

Abstract

Purpose

To investigate the effect of race and sex on long-term survival of oral and oropharyngeal cancer.

Methods

The Surveillance, Epidemiology and End Results database was queried for adult oral and oropharyngeal cancer patients with at least 25-year follow-up. Kaplan–Meier survival curves and cox proportional hazards model were used to identify differences.

Results

Of the 22,162 patients identified, 70.3 % were males. Only 8.9 % were alive at 25 years post-diagnosis. Black males show the poorest overall and disease-specific survival rates (p < 0.001). After controlling for covariates, Blacks had a 40 % higher hazard of mortality compared with Whites (HR 1.40, 95 % CI 1.35–1.46), while females had a 9 % reduction in mortality risk (HR 0.91, 95 % CI 0.88–0.94).

Conclusions

Overall and disease-specific survival is poor for oral and oropharyngeal cancer patients, and Black men fare worst. This illustrates the need for long-term cancer survival plans incorporating disparity effects in overall cancer outcomes.

Introduction

Oral cavity and oropharyngeal cancer (OCOPC) is found in the lip, tongue, mouth, and the oropharynx (Warnakulasuriya 2009), and it is one of the top 8 commonly diagnosed cancers among males in the USA (American Cancer Society 2015). OCOPC accounts for 3 % of all new cases of cancer in the USA, and it is estimated that in 2015 there will be 45,780 new cases resulting in 8650 deaths (American Cancer Society 2015). The major causal factors associated with OCOPC include tobacco, alcohol and human papillomavirus (HPV) (Neville and Day 2002; American Cancer Society 2015). While tobacco- and alcohol-associated OCOPC are decreasing, HPV-associated head and neck cancer, especially oropharyngeal cancer, has increased by approximately 225 % since the 1980s (D’Souza and Dempsey 2011; Malloy et al. 2013; Marur et al. 2010).

OCOPC has poorer 5-year relative survival rates (all stage) when compared with other commonly diagnosed cancers such as prostate, breast, colorectal, and cervical cancers (American Cancer Society 2015). Blacks fare worse than Whites in 5-year relative survival (45 vs. 67 %) (American Cancer Society 2015), and Black males have higher age-adjusted incidence rates than females for cancer of the oral cavity and pharynx (Goodwin et al. 2008).

Although OCOPC survival rates are relatively poor, survival has gradually increased in the last two decades (Barker et al. 2005; Jemal et al. 2010; Lin et al. 2014; Pulte and Brenner 2010). In 2013, there were at least 280,763 (185,242 males and 95,521 females) head and neck cancer survivors in the USA (de Moor et al. 2013). Some factors that have contributed to the changing oropharyngeal and oral cavity cancer survival landscape include the increasing incidence in HPV-associated head and neck cancer, improvement in treatment modalities, and the shift in age and profile of survivors from sixth or seventh generation to much younger individuals (Chaturvedi et al. 2011; Deschler et al. 2014; Lin et al. 2014; Patel et al. 2011; Pulte and Brenner 2010; Schantz and Yu 2002). However, the increased survival of oral cavity and oropharyngeal cancer is also associated with long-term sequelae, which include several toxicities associated with treatment modalities, structural, functional, and esthetic compromise, and other quality of life issues (Barker and Barker 2001; Epstein et al. 2012; Nicolatou-Galitis et al. 2011; Vickery et al. 2003). This is excluding significant costs associated with managing complications (de Moor et al. 2013; de Souza and Seiwert 2014).

The growing number of OCOPC survivors, as well as the unintended outcomes of their cancer treatments, necessitates longer-term surveillance than the commonly reported 5- to 10-year survival rates (Tiwana et al. 2014). Increasing survival rates creates the need for surveillance, survivorship plan, and management of late treatment complications and comorbidities, and longer-term survival rates may provide a better prognostic estimate of overall survival. There is dearth of information about longer-term disparities associated with survival of OCOPC, even though a major focus of public health service in the USA has been health disparities, exemplified by the Healthy People 2020 overarching objective of eliminating health disparities (Services; Tiwana et al. 2014).

The objective of this study is to describe disparities in long-term survival after diagnosis of OCOPC, stratified by race and sex. We hypothesize that early disease-specific survival racial and sex disparities observed in other studies will persist over long-term follow-up.

Methods

Data were obtained from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program’s database with a focus on sex and racial disparities for oral and oropharyngeal cancers. We selected oral and oropharyngeal cancer cases diagnosed between 1975 and 1986 to assure that each patient had the opportunity to be followed for at least 25 years from time of cancer diagnosis, since the most complete SEER database followed patients until 2011. Sites were identified based on the SEER site recode ICD-O-3/WHO 2008 variable which corresponds to ICD-O-3 site codes: C000–C029, C030–C069, C090–C109, C140, C142, and C148. Using the SEER categorization system tongue includes both oral tongue and tongue base. Cases were not selected based on histologic subtype. The sites we included are: lip, tongue, floor of mouth, gum and other mouth, tonsil, oropharynx, and other oral cavity and pharynx. We excluded nasopharynx and hypopharynx from our analysis. There are organizations such as the American Cancer Society (http://www.cancer.gov/types/head-and-neck/patient/oral-prevention-pdq#section/_7) and The Oral Cancer Foundation (http://oralcancerfoundation.org/cdc/cdc_chapter1.php) that group oral cavity and oropharynx together and do not include nasopharynx and hypopharynx. For this study, we chose to group oral cavity and oropharynx; while these are two separate subsites within the head and neck, they share some common features including risk factors and the opportunity for transoral screening. Although some organizations group oral cavity and oropharynx with these other pharyngeal sites (including nasopharynx and hypopharynx) due to anatomic proximity, we excluded them because nasopharyngeal and hypopharyngeal cancers typically have distinct clinical presentations and epidemiology.

Records required pathologically proven cancers to be included and records limited to data from death certificates only were excluded. We also excluded patients <18 years old.

From the SEER 18 database (1973–2012), we selected states with the SEER 9 records so that trends in survival could be followed for a longer follow-up in the same cohort. Race was defined as White, Black, and Other (American Indian/AK Native, Asian/Pacific Islander, and unknown) categories. The “Other” group was used rather than more specific groupings to allow for examination of long-term survival trends.

Data abstracted included age, cancer site, marital status, treatment type, and county-level demographics. A socioeconomic status index variable was created using a composite of three county-level variables available in the SEER database based on methods employed by Du et al. (2006) and others (Danzig et al. 2014; Robert et al. 2004): specifically, (1) poverty percentage, (2) median family income, and (3) percentage of high school graduates. The variables were separated into quintiles and coded so that the highest quintile represented the highest SES. The three variables were then combined into a SES Index variable representing lowest to highest SES. We also abstracted data on observed survival and disease-specific survival.

Statistical analyses were completed using SPSS version 22.0 (IBM SPSS Statistics for Windows 2013). Kaplan–Meier curves were produced to examine differences in survival (overall and disease specific) based on race and sex. The log-rank test was used to determine whether there were significant differences between groups. In an effort to evaluate possible mechanisms of sex and racial disparities, a Cox proportional hazards survival analysis was performed using the following variables: sex, race, age, marital status, cancer site, and treatment (radiation and/or surgery), stage (local, regional, distant), and the SES Index variable. Potential variables were evaluated in univariate analysis and then significant variables were included in the multivariate analysis. Statistical significance was set at α = 0.05.

Results

Participant characteristics

Our cohort included 22,162 cancer patients (Fig. 1); the majority were males (70.3 %) and minority were Blacks (8.9 %). Most of the cohort was married at the time of diagnosis (60.4 %). The most common cancer sites were tongue (24.0 %), lip (21.0 %), and gum/other mouth (20.7 %). The majority were treated with surgery (64.3 %), while just under half (47.1 %) had radiation as part of their treatment regimen (Table 1).

Fig. 1.

Patient selection flowchart. This figure describes how sample population was derived from the SEER cancer database, using the inclusion and exclusion criteria

Table 1.

Population characteristics

Patient characteristic	Entire cohort, n = 22,162	White, n = 19,628 (88.6 %)	Black, n = 1964 (8.9 %)	Other, n = 570 (2.6)
Age, mean (SD) (min–max 19–99)	62.50 (12.33)	63.15 (12.22)	56.34 (10.82)	61.27 (14.68)
Sex [n (%)]
Male	15,580 (70.3)	13,798 (70.3)	1415 (72.0)	367 (64.4)
Female	6582 (29.7)	5830 (29.7)	549 (28.0)	203 (35.6)
Marital status
Single, never married	2226 (10.9)	1741 (9.6)	419 (23.3)	66 (12.7)
Married	12,362 (60.4)	11,404 (62.8)	639 (35.6)	319 (61.6)
Divorced or separated	2641 (12.9)	2187 (12.0)	415 (23.1)	39 (7.5)
Widowed	3236 (15.8)	2820 (15.5)	322 (17.9)	94 (18.1)
Cancer site
Tongue	5322 (24.0)	4581 (23.3)	544 (27.7)	197 (34.6)
Oropharynx	723 (3.3)	633 (3.2)	76 (3.9)	14 (2.5)
Floor of mouth	3299 (14.9)	2871 (14.6)	338 (17.2)	90 (15.8)
Gum and other mouth	4588 (20.7)	3967 (20.2)	462 (23.5)	159 (27.9)
Lip	4662 (21.0)	4592 (23.4)	25 (1.3)	45 (7.9)
Tonsil	2699 (12.2)	2243 (11.4)	403 (20.5)	53 (9.3)
Other	869 (3.9)	741 (3.8)	116 (5.9)	12 (2.1)
Surgery, yes	14,242 (64.3)	12,920 (65.8)	973 (49.5)	349 (61.2)
Radiation, yes	10,210 (47.1)	8725 (45.3)	1200 (65.1)	285 (50.3)

Overall survival and racial disparities

The 10-year overall survival rate was 30.8 % and continued to decline throughout the period of follow-up, with only 8.9 % still alive at 25 years post-diagnosis and treatment. Figures 2 and 3 show the overall and disease-specific survival, respectively, and racial survival disparities were observed in patients diagnosed between 1975 and 1986. Overall observed survival curves show lower survival rates among Black males compared with all other racial groups, log rank χ ² = 542.43, p < 0.001 (Fig. 2). There was also a significant difference in disease-specific survival, log rank χ ² = 661.35, p < 0.001 (Fig. 3). Specific survival rates by race can be seen in Table 2, survival rates by race and sex are presented in Table 3.

Fig. 2.

Overall survival by race and sex. This figure is a survival curve showing overall survival of patients. Survival curve for Black females eventually converges with White males and females around the 20-year mark. For males, the curve fails to converge even at the 25-year mark

Fig. 3.

Disease-specific survival. This figure is a survival curve showing oral cavity and oropharyngeal cancer disease-specific survival. Curves indicate the disparities that exist between Blacks, both male and females, versus Whites. Black males fare worse

Table 2.

Survival rates (% surviving) by race and sex

Race	2 years (%)	5 years (%)	10 years (%)	15 years (%)	20 years (%)	25 years (%)
White	65.2	48.1	32.0	21.2	13.9	9.1
White male	65.6	48.0	31.7	21.0	13.6	8.9
White female	64.2	48.6	32.6	21.8	14.6	9.6
Black	45.4	28.2	16.7	9.4	6.3	4.0
Black male	41.7	23.7	12.9	6.9	4.0	2.4
Black female	55.2	39.9	26.4	15.7	12.2	8.3
Other	62.8	53.2	40.2	31.9	24.0	19.6
Other male	59.9	49.2	35.2	26.6	19.2	15.1
Other female	67.9	60.5	49.4	41.1	33.0	27.9

Table 3.

Multivariate cox regression analysis

Patient characteristic	Overall mortality		Disease-specific mortality
	HR (95 % CI)a	p value	HR (95 % CI)a	p value
Race
White (ref)	–	–	–	–
Black	1.43 (1.39–1.48)	<0.001	1.40 (1.35–1.46)	<0.001
Other	0.82 (0.78–0.87)	<0.001	0.97 (0.90–1.03)	0.34
Sex, female	0.80 (0.78–0.81)	<0.001	0.91 (0.88–0.94)	<0.001
Age	1.04 (1.04–1.04)	<0.001	1.02 (1.02–1.02)	<0.001
Marital status
Married (ref)	–	–	–	–
Single, never married	1.34 (1.31–1.39)	<0.001	1.36 (1.31–1.41)	<0.001
Divorced or separated	1.41 (1.37–1.45)	<0.001	1.38 (1.33–1.43)	<0.001
Widowed	1.33 (1.29–1.37)	<0.001	1.39 (1.33–1.45)	<0.001
Cancer site
Tongue (ref)	–	–	–	–
Oropharynx	1.24 (1.18–1.31)	<0.001	1.17 (1.10–1.24)	<0.001
Floor of mouth	1.21 (1.17–1.24)	<0.001	1.10 (1.05–1.15)	<0.001
Gum and other mouth	0.96 (0.93–0.99)	0.007	0.95 (0.91–0.98)	0.005
Lip	0.67 (0.65–0.70)	<0.001	0.20 (0.18–0.22)	<0.001
Tonsil	0.85 (0.82–0.88)	<0.001	0.77 (0.74–0.80)	<0.001
Other	1.48 (1.41–1.56)	<0.001	1.51 (1.43–1.61)	<0.001
Stage
Localized	–	–	–	–
Regional	1.46 (1.43–1.50)	<0.001	1.98 (1.87–2.10)	<0.001
Distant	2.56 (2.47–2.66)	<0.001	3.82 (3.56–4.11)	<0.001
Unstaged	1.36 (1.30–1.41)	<0.001	1.72 (1.58–1.86)	<0.001
SES index
Quintile 1	1.15 (1.12–1.19)	<0.001	0.94 (0.88–1.00)	0.07
Quintile 2	1.05 (1.02–1.08)	0.002	0.96 (0.91–1.02)	0.24
Quintile 3	1.07 (1.04–1.11)	<0.001	0.77 (0.71–0.85)	<0.001
Quintile 4	0.98 (0.95–1.02)	0.38	0.95 (0.89–1.02)	0.13
Quintile 5 (ref)	–	–	–	–
Surgery, yes	0.61 (0.60–0.63)	<0.001	0.56 (0.54–0.58)	<0.001
Radiation, yes	1.01 (0.98–1.04)	0.27	1.04 (1.01–1.08)	0.02

^a HR hazard ratio, 95 % CI 95 % confidence interval

The largest absolute racial disparity in disease-specific survival was 24.9 % and occured 11–13 years after diagnosis. After this period the disease-specific survival curves for all groups began to plateau; however, survival for Black males was consistently lower than that for all other groups. Meanwhile, females of other racial groups showed the best survival throughout long-term follow-up. Black females and White overall survival eventually converged beyond 20 years, whereas disease-specific survival showed persistent disparities (see Fig. 2). With regard to disease-specific survival, opposite trends were also observed by sex between racial groups with Black males showing poorer survival than Black females while White males had better survival than White females (Fig. 3).

Multivariate cox proportional hazards analysis

Results from the final adjusted Cox proportional hazards model are presented in Table 3. After adjusting for age, marital status, cancer site, stage, treatment type, and socioeconomic status, there was still a significant effect of race and sex on overall survival. Blacks had a 43 % higher hazard of mortality compared with Whites (HR 1.43, 95 % CI 1.39–1.48), while those of other racial backgrounds saw a 28 % lower hazard of mortality (HR 0.82, 95 % CI 0.78–0.87) after an oral or oropharyngeal cancer diagnosis. Females also had a 20 % lower hazard of mortality after the cancer diagnosis compared with males (HR 0.80, 95 % CI 0.78–0.81). This relationship holds for disease-specific survival, with a 40 % higher hazard of mortality due to the cancer diagnosis for Blacks compared with Whites (HR 1.40, 95 % CI 1.35–1.46). Females showed a 9 % lower hazard of cancer-related mortality compared with males (HR 0.91, 95 % CI 0.88–0.94). Socioeconomic status had a modest, but statistically significant association with survival; lower socioeconomic status was an indicator of higher hazard of mortality when compared with those with the highest socioeconomic status (HR 1.15, 95 % CI 1.12–1.19). Marital status was also a significant predictor of overall and specific survival. For OCOPC-specific survival, patients who were single, divorced, or widowed had 36–39 % higher hazard of mortality, compared with those that were married.

Discussion

The objective of this study was to assess whether racial and sex disparities existed in long-term survival of oral and oropharyngeal cancer in the USA. We found significant disparities between racial groups, both in overall and in disease-specific survival. These results are consistent with our hypothesis that racial disparities will persist over long-term follow-up. Our findings are novel as the numerous studies that have examined racial disparities in survival of several head and neck cancer subsites have only reported 5- to 10-year survival rates (Amit et al. 2013; McMahon et al. 2011; Ruggeri et al. 2005; Tomita et al. 2010) and none have reported racial disparities after 25 years of follow-up. The longest term follow-up reported to date is 25 years, from a cancer database in British Columbia, Canada (Tiwana et al. 2014); however, this study did not include race in its analysis. Several explanations have been given why racial disparities in cancer outcomes occur, both biologic, genetic and environmental factors (Arbes et al. 1999; Freeman 1989; Henschke et al. 1973; Tomar et al. 2004; Whitworth 2006; Yoo et al. 2004). Among these factors include stage at diagnosis, lack of access to quality care, lower socioeconomic status, lack of health insurance, and possibly comorbidities (Arbes et al. 1999; Goodwin et al. 2008; Morse and Kerr 2006; Piccirillo 2000; Tomar et al. 2004). A recent paper from our institution (Osazuwa-Peters 2015) indicates that head and neck cancer patients without health insurance are more than 10 times more likely to present with advanced stage disease than those with health insurance. The role of HPV positivity of head and neck tumors in survival continues to be explored; some studies have shown that Whites have a higher proportion of HPV-positive tumors, which is associated with a more favorable prognosis (D’Souza et al. 2007; Lin et al. 2013; Panwar et al. 2014; Settle et al. 2009). However, we were not able to include HPV status in our study as the SEER database did not have this information available. Most previously published studies that analyzed race show a disparity in survival for Blacks, although this disparity is attenuated after controlling for other covariables, such as age, marital status, tumor stage, socioeconomic status, treatment type, and anatomic site (Arbes et al. 1999; Franco et al. 1993; Moore et al. 2001). The same trend was shown in our results; hazard ratio was lower though still significant after controlling for covariables such as age, sex, marital status, cancer site, stage, SES index, and treatment type. The knowledge that racial disparities in survival persists after two decades of survival of OCOPC illustrates the need for interventions aimed at mitigating the underlying mechanisms associated with disparities and health inequality.

We also found there was a sex disparity in long-term survival of OCOPC patients, with males in general having a higher hazard of death compared with females and Black males faring worse than all other races and sex, both in overall and in disease-specific survival. This trend was observed from 2-year survival right up to 25-year survival. This survival advantage in females has been observed in a previously published study (Franco et al. 1993; Funk et al. 2002; Moore et al. 2001; Tiwana et al. 2014). An explanation offered for this sex disparity included compliance to treatment; females are more likely than males to seek care and comply to treatment (Thies and Travers 2006). It could also be that males are more likely to engage in behaviors linked with tobacco-associated OCOPC, while females are more likely than males to present with HPV-positive oropharyngeal cancer, which has better survival rates than the tobacco-associated OCOPC.

This study is significant as it describes two independent predictors of long-term survival of OCOPC in the United States. Understanding factors associated with survivorship in cancer generally and OCOPC specifically has become increasing important as overall survivorship continues to improve. The National Institute of Medicine had previously stated that there is a need to improve the quality of follow-up care for each cancer survivor in the US healthcare system (Hewitt et al 2006), as toxicities and chronic side effects of treatment persist long after cancer therapy and cancer patients are often “lost in transition” (Hewitt et al 2006; Institute of Medicine 2013). This has become even more important in head and neck cancer care as patients present at a younger age and survive longer.

There are some limitations in the present study. The study is limited to data available through the SEER database. While this is a high-quality database, it suffers from limitations inherent in retrospective cohort studies. Many covariates were included, but HPV status, which has become very important in outcomes of head and neck cancer, was not examined due to reasons previously stated. We also did not examine the role of comorbidities in survival. The authors hope to explore the potential effect of comorbidities in long-term survivorship of OCOPC in the future. Another limitation was that we could only present county-level measures and compute a valid proxy measure for SES (Geronimus and Bound 1998; Krieger 1992; Kwok and Yankaskas 2001). Despite these limitations, the study benefits from a large sample size, multiple sites, quality controls, and population-based long-term follow-up.

We conclude that the overall survival for patients with OCOPC is poor, and Black males experience the worst outcomes. We also conclude from our results that racial and sex disparities exist in survival of those with OCOPC, and these disparities have not shown in appreciable improvement since 1975. Access to care is pivotal to reducing health disparities, which is the overarching goal of the Healthy People 2020 initiative. Realizing this goal will require a multidimensional approach to the myriad of causal and risk factors associated with survival disparities elucidated in this study.

Compliance with ethical standards

Conflict of interest

None of the authors have any conflict of interest to declare.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. No ethics approval was sought for this study by authors as data used were the National Cancer Institute’s Surveillance, Epidemiology, and End Result (SEER) database, which is a publicly available de-identified population-based database.