Racial and Ethnic Disparities in Cervical Cancer Incidence, Survival, and Mortality by Histologic Subtype

PURPOSE

We conducted an integrated population-based analysis of histologic subtype–specific cervical cancer incidence, survival, and incidence-based mortality by race and ethnicity, with correction for hysterectomy prevalence.

METHODS

Using the SEER 21 and 18 registries, we selected primary cases of malignant cervical cancer diagnosed among women ≥ 15 years. We evaluated age-adjusted incidence rates among cases diagnosed between 2000 and 2018 (SEER21) and incidence-based mortality rates among deaths from 2005 to 2018 (SEER18), per 100,000 person-years. Rates were stratified by histologic subtype and race/ethnicity (incidence and mortality), and stage, age at diagnosis, and county-level measures of social determinants of health (incidence only). Incidence and mortality rates were corrected for hysterectomy using data from the Behavioral Risk Factor Surveillance System. We estimated 5-year relative survival by histologic subtype and stratified by stage at diagnosis.

RESULTS

Incidence rates of cervical squamous cell carcinoma were highest in Black and Hispanic women, while incidence rates of cervical adenocarcinoma (ADC) were highest among Hispanic and White women, particularly for localized ADC. County-level income and education variables were inversely associated with squamous cell carcinoma incidence rates in all racial and ethnic groups but had less influence on ADC incidence rates. Black women had the highest overall mortality rates and lowest 5-year relative survival, irrespective of subtype and stage. Disparities in survival were particularly pronounced for Black women with regional and distant ADC, compared with other racial/ethnic groups.

CONCLUSION

Although Black women are less likely to be diagnosed with ADC compared with all other racial/ethnic groups, they experience the highest mortality rates for this subtype, likely attributed to the poor survival observed for Black women with regional and distant ADC.

INTRODUCTION

Cervical cancer incidence rates have declined substantially in the United States following the introduction of screening; however, these trends differ by histologic subtype. Although overall declines have been driven largely by a reduction in incidence of cervical squamous cell carcinoma (SCC; approximately 68% of cases), incidence rates of cervical adenocarcinoma (ADC; approximately 21% of cases) have been increasing.^1,2 Few population-based studies have evaluated differences in survival and mortality between these two subtypes; some prior studies have suggested that ADC has worse prognosis compared with SCC,³ whereas others have reported no differences after accounting for tumor stage.⁴

CONTEXT

• Key Objective

• To assess subtype-specific racial and ethnic differences with respect to cervical cancer incidence, survival, and mortality.

• Knowledge Generated

• Despite Black women having the lowest incidence of cervical adenocarcinoma (ADC) compared with other racial and ethnic groups, they have the highest mortality rates and the lowest 5-year relative survival for this subtype, with particularly pronounced disparities for regional and distant ADC. Every other racial and ethnic group had worse survival for cervical squamous cell carcinoma compared with ADC, except for Black women who had comparatively worse survival for ADC.

• Relevance (G.F. Fleming)

• The high incidence rates of cervical cancer in Black and Hispanic women older than 65 years should prompt further research regarding the safety of screening cessation recommendations.*

  *Relevance section written by JCO Associate Editor Gini F. Fleming, MD.

Epidemiologic studies have shown higher incidence rates of SCC but lower incidence rates of ADC among Black individuals compared with White individuals.^1,3,5,6 Reasons for these differences may involve both biological factors (ie, differences in human papillomavirus [HPV] genotype prevalence)⁷ and nonbiological factors such as structural inequities that affect access to and quality of care for screening, follow-up, and/or treatment.⁸ Black individuals have worse survival and higher hysterectomy-corrected cervical cancer mortality rates than White individuals^9,10; however, few studies have evaluated survival differences by race/ethnicity and histologic subtype,^3,11,12 and none report data since 2007. Furthermore, there have been no population-based studies evaluating hysterectomy-corrected cervical cancer mortality rates for histologic subtypes by race/ethnicity.

We hypothesized that there are important racial and ethnic differences with respect to subtype-specific cervical cancer incidence and mortality. To investigate this hypothesis, we conducted an integrated, population-based analysis evaluating hysterectomy-corrected cervical cancer incidence, incidence-based mortality rates, and relative survival by histologic subtype, stage, and age at diagnosis using the SEER database.^13,14

METHODS

SEER Database and Case Selection

We selected microscopically confirmed primary cases of malignant cervical cancer among individuals coded as female (hereafter, women) older than 15 years, diagnosed between 2000 and 2018 from the SEER21 database. We excluded cases diagnosed by autopsy or death certificate only. This database represents approximately 37% of the US population and includes 21 population-based registries.¹³ For analyses of incidence-based mortality and 5-year relative survival, we used the same selection criteria to identify cases in the SEER18 database (28% of the US population), since these data are not yet available in all 21 registries.¹⁴ We categorized race and ethnicity according to the SEER race and origin recode as non-Hispanic White, non-Hispanic Black, non-Hispanic American Indian/Alaska Native (AI/AN), non-Hispanic Asian or Pacific Islander (API), and Hispanic, all races. Those with unknown race/ethnicity were excluded.

Histology Classification and Stage at Diagnosis

SEER's broad histology categories were reviewed by an expert gynecologic pathologist (R.Z.) and reclassified when applicable (see the Data Supplement, online only, for more details) to minimize the potential for misclassification. We focused on the following histologic subtypes with adequate case numbers for statistical analyses: SCC, ADC, and cervical adenosquamous carcinoma (ADSC). Remaining cervical cancer cases were classified as other and those identified as noncervical by expert pathology review were excluded (Data Supplement). Stage at diagnosis was classified according to the SEER Combined Summary Stage (2004+) as localized (including 2018 International Federation of Obstetrics and Gynecology [FIGO] stages IA and IB), regional (including 2018 FIGO stages IIA, IIB, IIIA, and IIIB), distant (including 2018 FIGO stages IVA and IVB), or unknown/unstaged.¹⁵

County-Level Attributes

County attributes were estimated from the 2000 Census Bureau SF3 data to describe the demographics and socioeconomic status of a county in SEER.¹⁶ We selected median household income, educational attainment, and family poverty as three important social determinants of health that have been shown to be related to cervical cancer screening, survival, and incidence and mortality rates.^6,17-20 On the basis of the distributions of the data, we categorized median household income into tertiles, and dichotomized the percent of families below poverty and educational attainment (percent of women with at least a bachelor's degree) as either high (above the median) or low (at or below the median).

Incidence-Based Mortality Rates

We conducted an incidence-based mortality analysis linking cervical cancer deaths to incident cervical cancer cases in the SEER18 registry (2000-2018). Since many of the deaths occurred among cases diagnosed before 2000 (when incidence data and tumor characteristics were not available from SEER18), we restricted the analysis to cervical cancer deaths occurring between 2005 and 2018 to allow for adequate follow-up time (described in detail in the Data Supplement).

Statistical Analyses

We used SEER*Stat (version 8.3.9.2) to calculate age-adjusted incidence rates and incidence-based mortality rates per 100,000 person-years with 95% CIs, uncorrected for hysterectomy, by histologic subtype and race/ethnicity. Incidence rates were further stratified by age group (15-34, 35-44, 45-54, 55-64, 65-74, and 75+ years), stage, and county-level attributes, as described above. Hysterectomy correction is described in detail in the Data Supplement. We calculated incidence rate ratios with 95% CIs for both the corrected and uncorrected incidence rates for comparisons of histologic subtypes by racial and ethnic groups with non-Hispanic Whites as the reference, and of stage at diagnosis within racial and ethnic groups by histologic subtype using localized stage as the reference group. Age-specific incidence curves for SCC and ADC by race/ethnicity were plotted using a semilogarithmic scale.²¹ We also calculated median age at diagnosis for each racial and ethnic group.

Using the actuarial method, we estimated 5-year relative survival as the ratio of observed to expected survival rates among patients matched in the survival cohort to the socioeconomic, geographic, and race annual life tables by age, sex, race/ethnicity, calendar year, and county of residence at the time of diagnosis.²² Expected survival was estimated using the Ederer II method.²³ We generated Kaplan-Meier curves of the probability of survival by race/ethnicity and histologic subtype in STATA version 17.0. For all tests of significance, a P value < .05 was considered statistically significant.

RESULTS

Cervical Cancer Incidence Rates by Histologic Subtype and Race and Ethnicity

A total of 80,580 incident cervical cancer cases were diagnosed between 2000 and 2018. We excluded 5,158 (6.4%) cases identified as noncervical primaries following expert review. Of the remaining 75,422 cases, most were classified as SCC histology (54,023; 71.6%), followed by ADC (16,447; 21.8%), ADSC (2,963; 3.9%), and other (1,989; 2.6%; Data Supplement).

Black women had the highest hysterectomy-corrected cervical cancer incidence rate (16.8 per 100,000). Both uncorrected and corrected incidence rates of SCC were highest in Black women (9.1 and 13.8, respectively), followed by Hispanic (8.9 and 11.8, respectively), AI/AN (6.1 and 8.6, respectively), API (5.6 and 6.9, respectively), and White women (5.2 and 6.8, respectively). Compared with White women, corrected SCC incidence rates were two-fold higher in Black women. By contrast, both uncorrected and corrected incidence rates of ADC were highest in Hispanic women (2.4 and 3.0, respectively), followed by White (2.0 and 2.5, respectively), AI/AN (1.8 and 2.5, respectively), API (1.8 and 2.1, respectively), and Black women (1.2 and 2.0, respectively). Hispanic women had the highest incidence rates of ADSC (0.7 [corrected]), while API and White women had the lowest (0.4 [corrected]; Table 1 and Data Supplement).

TABLE 1.

Age-Adjusted Incidence Rates and Rate Ratios Expressed per 100,000 Person-Years of Cervical Cancer by Histologic Subtype and Race and Ethnicity Corrected for Hysterectomy Prevalence Among Women Age 15-75+ Years in the United States in SEER21 (2000-2018)

Stage-Specific Incidence Rates by Histologic Subtype and Race and Ethnicity

Among White women, corrected incidence rates of localized SCC were highest, whereas Black, API, AI/AN, and Hispanic women had higher corrected incidence rates of regional SCC compared with localized (Table 2).

TABLE 2.

Age-Adjusted Incidence Rates and Rate Ratios of Cervical Cancer Expressed per 100,000 Person-Years by Histologic Subtype, Race and Ethnicity, and Stage Corrected for Hysterectomy Prevalence Among Women Age 15-75+ Years in the United States in SEER21 (2000-2018)

Incidence rates of localized ADC were higher compared with regional and distant disease among all racial/ethnic groups. Differences among Black compared with Hispanic and White women were most apparent for localized disease (Table 2).

Age-Specific Incidence Rates by Histologic Subtype and Race and Ethnicity

Among White women, incidence rates of SCC peaked at age 35-44 years and remained relatively stable thereafter, whereas incidence rates continued to increase with age among Black and Hispanic women, peaking at age 65-74 years (Fig 1).

FIG 1.

Age-specific incidence rates expressed per 100,000 person-years by race and ethnicity for (A) cervical SCC and (B) ADC subtypes, corrected for hysterectomy prevalence among women in the United States age 15 to 75+ years according to SEER21 (2000-2018). ADC, cervical adenocarcinoma; AI/AN, American Indian and Alaska Native; API, Asian or Pacific Islander; SCC, squamous cell carcinoma.

ADC incidence rates in White women were highest at age 35-44 years, decreasing thereafter. Similar age patterns were observed among other racial/ethnic groups except for Black women, who experienced increasing incidence rates of ADC with increasing age, peaking at 65-74 years (Fig 1). Age-specific incidence rates by stage at diagnosis for each subtype are shown by race/ethnicity in the Data Supplement.

Overall and by subtype, Black women were diagnosed with cervical cancer at an older median age group (50-54 years) than White (45-49 years), Hispanic (45-49 years), and AI/AN women (40-44 years).

Subtype-Specific Incidence Rates by County-Level Attributes and Race and Ethnicity

SCC incidence rates were higher among counties with a lower median household income, higher percentage of families in poverty, and lower percentage of women with a bachelor's degree among White, Black, Hispanic and API women. By contrast, incidence rates of ADC did not vary substantially by these factors across most racial/ethnic groups (Data Supplement).

Five-Year Relative Survival by Histologic Subtype and Race and Ethnicity

Overall, 5-year relative survival was higher for ADC compared with SCC among all racial and ethnic groups except Black women, who had lower survival for ADC compared with SCC. For both histologic subtypes at every stage, Black women had the lowest 5-year relative survival, whereas Hispanic women generally had the highest survival irrespective of subtype and stage (Fig 2). Compared with White and Hispanic women, Black women diagnosed with regional and distant ADC had dramatically lower 5-year relative survival (regional: 37.6 v 61.5 and 65.1, respectively, and distant: 9.2 v 21.1 and 24.9, respectively) and much shorter survival times, whereas differences in corresponding survival estimates for SCC were much less pronounced. API women also had particularly lower survival for regional and distant ADC when compared with White and Hispanic women (54.5 and 12.9, respectively; Fig 2 and Data Supplement).

FIG 2.

Five-year relative survival by race and ethnicity and stage at diagnosis for cervical SCC and ADC among women in the United States age 15 to 75+ years according to SEER18 (2000-2018). Relative survival was estimated as the ratio of observed to expected survival rates among patients matched in the survival cohort using the actuarial method. Expected survival was estimated using the Ederer II method. ADC, cervical adenocarcinoma; AI/AN, American Indian and Alaska Native; API, Asian or Pacific Islander; SCC, squamous cell carcinoma.

Incidence-Based Mortality Rates by Histologic Subtype and Race and Ethnicity

There were 10,488 cervical cancer deaths from 2005 to 2018 recorded in SEER18 corresponding to a hysterectomy-corrected incidence-based mortality rate of 3.0 per 100,000 person-years. Black women had the highest uncorrected and corrected overall mortality rates (3.2 and 5.0, respectively), followed by AI/AN (2.6 and 4.0, respectively), Hispanic (2.5 and 3.5, respectively), White (1.8 and 2.6, respectively), and API women (1.6 and 2.1, respectively).

A total of 7,876 deaths were among cases with SCC (75.1%) and 1,675 among those with ADC (16.0%; Table 3 and Data Supplement). Black women had the highest SCC mortality rates (corrected: 4.0), followed by AI/AN (2.9), Hispanic (2.7), White (1.9), and API (1.4) women. Despite having the lowest incidence of ADC, Black women also had the highest mortality rates of ADC (corrected: 0.6) compared with Hispanic (0.5), API (0.4), and White women (0.4; Table 3).

TABLE 3.

Age-Adjusted Incidence-Based Mortality Rates and Mortality Rate Ratios of Cervical Cancer Expressed per 100,000 Person-Years by Histologic Subtype and Race and Ethnicity Corrected for Hysterectomy Prevalence Among Women Age 15-75+ Years in the United States in SEER18 (2005-2018)

DISCUSSION

Racial disparities in cervical cancer incidence and mortality have persisted for decades in the United States, representing one of the largest Black-White mortality gaps for any cancer.^24,25 To date, few studies have evaluated these disparities by histologic subtype, particularly with respect to mortality. In line with our current findings, studies have suggested that although cervical SCC incidence rates are highest among Black women, ADC incidence rates are lowest compared with all racial/ethnic groups.^1,3,5,6 A limitation of some of these previous studies is that they only looked at individual metrics, like incidence, which may result in misleading conclusions. By performing an integrated population-based analysis of subtype-specific incidence, survival, and incidence-based mortality, we have, to our knowledge, for the first time, demonstrated that although incidence rates of ADC may be lowest in Black women, there are profound racial disparities in ADC survival, leading to Black women having the highest ADC mortality rates compared with all other groups. Notably, although Black women also had the highest SCC incidence and mortality rates, survival disparities were much less pronounced for this subtype, suggesting subtype-specific differences that may be related to both systemic inequities that influence receipt of high-quality care as well as potential biological factors distinctive of ADC.

Some,^3,26 but not all, studies⁴ have previously suggested that ADC has worse prognosis than SCC. Our findings suggest that at the population level, ADC has better stage-specific survival compared with SCC among all groups except Black women. Most understanding of cervical cancer, including optimal treatment, has resulted from studies in predominantly White patients with SCC.^27-29 Currently, there are no explicit differences in treatment recommendations for cervical cancer by histologic subtype, despite important histologic differences with respect to clinical behavior,³⁰ recurrence risk,²⁸ response to therapy,^31,32 and prognosis.^32-36 For example, studies have shown a greater benefit of posthysterectomy adjuvant radiation among patients diagnosed with ADC compared with those with SCC.³¹ Others have shown that addition of cisplatin-based chemotherapy to radiation versus radiation alone confers a greater benefit among patients with locally advanced ADC compared with SCC.³² Furthermore, in this study, chemoradiation mitigated survival disparities observed among Black patients treated with radiation alone who had worse overall survival compared with White women. Subtype-specific differences in response to therapy may point to the contribution of treatment-related factors, such as receipt of guideline-concordant care and/or differential access to skilled gynecologic oncologists, to the subtype-specific racial survival disparities observed in our study. Another possible explanation could be that some cases diagnosed as cervical ADC are of endometrial origin. Although immunohistochemical markers (eg, p16 and hormone receptors) can differentiate endocervical adenocarcinomas from low-grade endometrioid adenocarcinomas, these markers are less helpful for distinguishing cervical ADC from endometrial serous adenocarcinomas, which are more common in Black women.^37,38 This distinction is critical for optimizing management and could lead to disparities if misclassification occurs more often in Black women.

It is not clear why Black women have the lowest incidence of ADC despite having the highest incidence of SCC. We hypothesized that racial differences by subtype could be partially explained by social determinants of health associated with access to care and screening. However, whereas SCC incidence rates decreased with increasing county-level socioeconomic status (SES) measures, ADC incidence rates were less affected. Notably, disparities in SCC incidence rates persisted in all strata (ie, Black women from high SES counties still had much higher incidence rates than White women). It is possible that differences in the distribution of HPV genotypes and variants³⁹ contribute to subtype-specific differences in incidence. For example, Black women have a lower prevalence of HPV16 compared with White women, and HPV16 is more likely to cause ADC compared with other HPV types.^7,40-42 It is unlikely that proposed risk factors for ADC such as nonsmoking⁴³ and obesity⁴⁴ contribute to racial differences in incidence, as obesity is more prevalent and smoking is less prevalent in Black women compared with White women.^45,46

Interestingly, although Hispanic women had high incidence of SCC and the highest incidence of ADC and ADSC, they also had the highest 5-year relative survival compared with all other racial and ethnic groups. A recent population-based study also found Hispanic women have higher incidence of cervical cancer overall, but similar 5-year relative survival compared with non-Hispanic Whites⁴⁷; however, differences in survival appear to vary by factors such as nativity.^48,49 Although incidence and mortality rates were generally lowest among API women in our study, we noted relatively worse survival for those diagnosed with regional and distant ADC compared with their White and Hispanic counterparts. More research is needed to better understand these differences, particularly by country of origin and ancestry.

We observed important racial and ethnic differences with respect to age-specific incidence rates. For example, in White women, incidence rates of SCC and ADC peaked at age 35-44 years and remained stable thereafter, whereas incidence rates continued to increase with age among Black women, peaking at age 65-74 years for both subtypes. Incidence rates of SCC also peaked at age 65-74 years among Hispanic women. High incidence rates of cervical cancer in Black and Hispanic women older than 65 years are concerning and may reflect inadequate screening in the preceding decade. Our findings underscore the need for more research in clinical populations age 65+ years to ensure the safety of screening cessation recommendations for all individuals.⁵⁰

Our study has many strengths as the first and largest nationally representative population-based study, to our knowledge, of hysterectomy-corrected cervical cancer incidence and mortality rates by histologic subtype and race/ethnicity. Furthermore, we conducted expert pathology review of all histologic diagnosis categories coded by SEER to reduce misclassification, although we were unable to evaluate individual cases for accurate designation of histologic cell type and HPV status. Other limitations are also worth noting. Although we evaluated differences in incidence by county-level SES and accounted for SES when matching individuals in the survival cohort analysis, we were unable to evaluate individual-level SES; furthermore, numbers were limited to assess mortality differences by county-level SES measures. Misclassification of race/ethnicity is possible despite SEER registries using standardized codes and procedures.⁵¹ Additionally, SEER categorizes sex as male or female⁵²; therefore, transgender individuals with a cervix could have been unknowingly excluded from our analysis. Furthermore, SEER stage categories do not directly correspond to clinical FIGO staging, and we lacked access to imaging data that are increasingly important for tumor staging; however, most cases in our cohort were diagnosed before 2018, before FIGO staging guidelines were updated to emphasize the importance of imaging.⁵³ Finally, we do not have reliable data on treatment variables in SEER. There are many nonbiological and biological factors that could explain the racial/ethnic differences in survival that we observed in our study, and these factors need to be comprehensively studied in diverse clinical populations with high-quality individual-level data.

In conclusion, despite Black women having the lowest incidence of ADC, we show dramatic and profound disparities in cervical cancer survival affecting Black women with regional and distant ADC, resulting in higher ADC mortality rates among Black women compared with all other racial and ethnic groups. Although racial and ethnic differences in incidence rates of SCC are likely due to disparities in access to care and screening, the reasons underlying the observed racial and ethnic differences in incidence rates of ADC are less clear. More research is needed to better understand the influence of factors such as differences in detection (screening; including later ages at diagnosis), treatment, and management, as well as potential biological differences, on these disparities.

AUTHOR CONTRIBUTIONS

Conception and design: Camryn M. Cohen, Nicolas Wentzensen, Philip E. Castle, Mark Schiffman, Megan A. Clarke

Financial support: Nicolas Wentzensen, Megan A. Clarke

Collection and assembly of data: Camryn M. Cohen, Rosemary Zuna, Megan A. Clarke

Data analysis and interpretation: Camryn M. Cohen, Nicolas Wentzensen, Rebecca C. Arend, Megan A. Clarke

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Racial and Ethnic Disparities in Cervical Cancer Incidence, Survival, and Mortality by Histologic Subtype

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).