Risks of Second Primary Gynecologic Cancers following Ovarian Cancer Treatment in Asian Ethnic Subgroups in the United States, 2000–2016

An estimated 22,530 U.S. women were diagnosed with ovarian cancer in 2019, the fifth leading cause of cancer deaths in women (1). Overall, 5-year survival of ovarian cancer is less than 50% (2), and women with ovarian cancer living beyond 5 years have a 31% increased risk of developing second malignancy compared with the general population (3). Given this elevated risk of second cancers, and that 11% of second solid tumors are gynecologic (4), a nuanced understanding of racial and interethnic differences in incidence is imperative for targeted clinical and public health efforts.

Disparities according to race and ethnicity in ovarian cancer outcomes exist. From 2000 to 2013, the incidence of epithelial ovarian cancer (EOC) in the United States decreased, but the magnitude of change differed across racial groups (5). A significant decrease was identified among White, Black, and Hispanic women; among Asian women, the decrease was smaller and not statistically significant. In addition, increases in incidence of clear cell ovarian cancer, an epithelial subtype with limited treatment options, were observed, and this significant elevation occurred primarily among Asian women. Racial differences in survival following ovarian cancer also exist, with non-Hispanic Black women experiencing poor survival compared with non-Hispanic White, Hispanic, and Asian/Pacific Islander (API) women (6, 7). Relatively less, however, is known about differences in the risk of second malignancy in women with primary ovarian cancer by race and between Asian ethnic subgroups, in particular.

In a recent study of women diagnosed with ovarian cancer in the Surveillance, Epidemiology, and End Results (SEER) registries between 1992 and 2012, Kanninen and colleagues reported that incidence rates of second primary gynecologic cancers (SPGC) differed with regard to race (4). Marked disparities were observed among API women, in which an overall risk of SPGC was increased, particularly uterine and vaginal cancers. However, grouping all API patients as a single entity is problematic (8). Asian Americans are not a monolithic group; there is substantial ethnic diversity and heterogeneity in health behaviors, cancer outcomes, biomarkers, and sociodemographic characteristics (9–13). These differences are evident in the differences in incidence rates and average age at diagnosis for specific histologic subtypes (14). Compared with White women, Chinese, Filipina, and Japanese women have an elevated risk of clear cell ovarian cancer. Differences in mean age at diagnosis also exist, with 37% of Vietnamese women diagnosed before the age of 50 compared with 19% of Korean women. In addition, studies have reported significant differences in BRCA mutations between Chinese and non-Chinese Asian ethnicities (15). Given that about 20%–27% of the risk of EOC can be explained by germline mutations in BRCA1/2 gene (16), risk cannot be generalized to all Asian ethnic groups. For these reasons and multiple other sociodemographic health determinants that vary by Asian ethnic subgroups, we hypothesize that the burden of second primary tumor (SPT) risk is unequal across populations of Asian women living in the United States.

Our objective was to describe disparities in the incidence of SPTs, specifically second gynecologic cancers among women with primary ovarian malignancy that received definitive treatment with surgery and chemotherapy. We aimed to elucidate ethnic-specific disparities in risk of SPGCs following ovarian cancer among API women.

We performed a retrospective cohort study of women diagnosed with malignant epithelial ovarian carcinoma using data from the SEER Program registries (17). The SEER Program aggregates cancer incidence data from 18 population-based registries and includes patient demographics, tumor site, morphology, stage, first treatment course, and vital status. The registries in this study included the following: Alaska Natives, Atlanta, Connecticut, Detroit, Hawaii, Iowa, Los Angeles, New Mexico, San Francisco–Oakland, San Jose–Monterey, Seattle–Puget Sound, Rural Georgia, Utah, Greater California, Greater Georgia, Kentucky, Louisiana, and New Jersey. County-level demographic information was collected based on linkage to the 2000 U.S. Decennial Census long form survey and American Community Survey 5-year estimates from 2012 to 2016, provided through the NCI SEER Program (18). This analysis utilized deidentified secondary data obtained from NCI SEER repository under agreement, and was conducted in accordance with the U.S. Common Rule. This study was reviewed and determined to be exempt from human subjects research requiring written informed consent by the institutional review board of the University of Illinois at Chicago (UIC, Chicago, lL; IRB# 2019-1413).

In this study, index cancer cases included women ages 20 years and older who were diagnosed with grades I–IV ovarian carcinoma between January 2000 and December 2016 that received definitive treatment with both surgery and chemotherapy (with or without radiotherapy). Women were excluded from the analysis if they were identified only by death certificate or autopsy or diagnosed with any SPT within 2 months after diagnosis of the index ovarian cancer.

We collected demographic and clinical information on ovarian histologic subtypes, laterality, surgery, treatment with radiotherapy, and tumor grade. We characterized ovarian cancers according to International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3) histology in our analyses as serous (ICD-O-3 8441, 8460-8463, 9014), mucinous (ICD-O-3 8470-8490, 9015), endometrioid (ICD-O-3 8380-8383, 8560, 8570), and clear cell (ICD-O-3 8310-8313, 9110). County-level estimates of median household income, percentage with less than a high school education, percentage living below 150% of the poverty line, percentage language isolation, and percentage foreign born were assigned to patients according to their state-county recode (18).

The incidence of second primary cancers at any site, any solid SPTs, and SPGC (female genital system, cervix uteri, corpus uteri, ovary, vagina, vulva, and other female genital organs) in our cohort was compared with expected incidence rates in the general population using estimated standardized incidence ratios (SIR) and 95% confidence intervals (CI). The SIR was calculated by dividing the number of observed second primary cancer cases obtained from the SEER database by the number of expected cancers in the U.S. population with standardization based on age, race, and geographic location derived from the most recently available SEER data. Therefore, estimates were standardized for age, race (overall), and registry region. We determined absolute excess risks (ER) per 10,000 person-years in an effort to fully characterize the additional cancer burden. This value was calculated by dividing the difference between observed and expected number of cases by the number of person-years per risk and multiplied by 10,000. Analyses were conducted overall with further stratification by race (non-Hispanic White, non-Hispanic Black, Hispanic/Latina, API) and histology of the index ovarian cancer. Primary ovarian tumor grades were grouped as low (grades I) and high (grades II, III, and IV). Finally, analyses were performed in subgroups of Asian ethnicity in Chinese, Filipina, Indian, Japanese, and Vietnamese women. SEER*Stat version 8.3.5 software was utilized for all statistical analyses (19). All P values were two-sided, and the level of statistical significance was set a priori at 0.05.

Descriptive characteristics of the four most frequent Asian ethnic subgroups of women (Chinese, Filipina, Vietnamese, and Japanese) and an aggregate of all other API ethnic groups diagnosed with primary malignant EOC between 2000 and 2016 are reported in Table 1. The aggregated group was primarily composed of cases classified as other Asian (29%), Asian Indian or Pakistani (29%), and Korean (18%), among others. Among Asian women, Japanese women had a higher median age at ovarian cancer diagnosis [58 years, interquartile range (IQR) 51–71] relative to Chinese (53 years, IQR 47–62), Filipina (56 years, IQR 47–64), Vietnamese (53 years, IQR 47–61), and other Asian women (54 years, IQR 46–62). Clear cell tumors were more common in Vietnamese (21%) and Japanese (19%) women; endometroid tumors were more common in Filipina (22%) and Chinese women (20%). Across Asian ethnic groups, women were similar with respect to other clinical characteristics including tumor laterality, grade, and surgery.

More Chinese women (46%) were living in counties in the upper quartile of median household income compared with Japanese (15%), Filipina (27%), Vietnamese (31%), and other Asian (26%) women. A higher proportion of Chinese (27%), Filipina (33%), Vietnamese (22%), and other Asian (24%) women were living in counties in the upper quartile of low attained education (less than high school) compared with Japanese women (16%). More Filipina women (31%) were living in counties with the largest proportion of individuals whose income is less than 150% of the poverty level compared with Chinese (27%), other Asian (25%), Vietnamese (21%), and Japanese (15%) women. Higher proportions of Chinese women were living in areas with the highest quartile of foreign birth (48%) compared with Vietnamese (40%), Filipina (34%), other Asian (23%), and Japanese women (20%). A higher proportion of Filipina (29%) and Chinese (38%) women lived in areas with the greatest language isolation compared with Vietnamese (22%), other Asian (20%), and Japanese (18%) women.

Descriptive characteristics by race for women diagnosed with primary malignant ovarian cancer between 2000 and 2016 included in our study are reported in Table 2. Among 27,602 women, 20,641 (75%) were White, 1,628 (6%) were Black, 2,279 (8%) were API, and 3,054 (11%) were Hispanic/Latina. The median (IQR) age at diagnosis for White women (60 years, IQR 52–69) was higher than API (55 years, IQR 47–63), Black (58 years, IQR 49–67), and Hispanic/Latina (55 years, IQR 46–64) women. Across all racial/ethnic groups, the most common histologic subtype was serous ovarian tumors. However, among API women, higher proportions of clear cell (16%) and endometroid tumors (18%) were observed compared with other groups. API women were more often diagnosed with unilateral ovarian tumors (57%) versus other racial/ethnic groups (White 47%; Black 49%; Hispanic/Latina 47%). Distribution of ovarian tumor grades between racial/ethnic groups was similar. More White women (51%) received debulking surgery compared with the API women (44%), whereas slightly more API women received oophorectomy plus omentectomy (37%) compared with White women (33%).

More API women (51%) were living in counties with the highest quartile of median household income compared with White (25%), Black (14%), and Hispanic/Latina (22%) women. A greater proportion of API women were living in counties with the highest quartile of foreign-born residents (51%) and language isolation (50%) compared with White (20% and 20%) and Black (19% and 22%) women.

Subgroup analyses reporting risks of female genital, uterine, and vaginal cancer by Asian ethnicity are presented in Fig. 1. Among Asian women, only Filipina women had a significantly elevated risk of SPGC (SIR 2.92; 95% CI, 1.26–5.76; ER 16.45) and uterine cancer specifically (SIR 3.94; 95% CI, 1.45–8.58; ER 14.0). Elevated risks for SPGC were also identified for Chinese (SIR 2.06; 95% CI, 0.56–5.27; ER 8.64), Vietnamese (SIR 2.5; 95% CI, 0.3–9.04; ER 12.1), and Indian women (SIR 1.88; 95% CI, 0.05–10.45; ER 6.6), but CIs were wide and included 1.0 for non-Filipina ethnicities.

We report SIR estimates with 95% CI and ER per 10,000 women for second primary cancers at any site, any second primary solid tumor (SPST), and SPGC stratified by race in Table 3. Overall, White women (SIR 0.92; 95% CI, 0.86–0.98; ER −8.42) did not have a greater than expected risk of SPST. However, increased risk of any SPST was observed among API women (SIR 1.39; 95% CI, 1.11–1.71; ER 22.03). Hispanic/Latina women had a nonsignificantly increased rate of SPST (SIR 1.18; 95% CI, 0.98-1.40; ER 14.48), as did Black women (SIR 1.27; 95% CI, 0.98–1.60; ER 23.18). Estimates for incidence of second primary cancers at any site were similar with no increased incidence among White women (SIR 0.96; 95% CI, 0.90–1.02; ER −4.91), higher incidence among API women (SIR 1.37; 95% CI, 1.11–1.68; ER 23.24), and a modestly increased rate among Hispanic/Latina women (SIR 1.19; 95% CI, 1.00–1.40; ER 17.12) and Black women (SIR 1.26; 95% CI, 1.00–1.58; ER 25.79). A significantly lower than expected incidence of SPGC was observed among White women (SIR 0.73; 95% CI, 0.59–0.89; ER −3.64); a higher than expected incidence of SPGC was observed among Black (SIR 1.80; 95% CI, 0.96–3.08; ER 9.29) and Hispanic/Latina (SIR 1.26; 95% CI, 0.76–1.97; ER 2.98), but CIs included 1.0; and a significantly elevated incidence was observed among API women (SIR 1.83; 95% CI, 1.07–2.93; ER 7.02). Overall, all women had a higher than expected incidence of vaginal cancers (White SIR 6.14; 95% CI, 3.17–10.72; ER 1.02; Black SIR 13.13; 95% CI, 1.59–47.44; ER 2.96; Hispanic/Latina SIR 9.78; 95% CI, 1.18–35.33; ER 1.36); the highest incidence of vaginal cancers was observed among API women (SIR 26.76; 95% CI, 5.52–78.2; ER 2.62). API women also had increased risk of second primary uterine cancers (SIR 2.53; 95% CI, 1.35–4.33; ER 7.14), an association not observed among White, Black, or Hispanic/Latina women.

Risks of SPGC stratified by grade and histology are reported in Supplementary Table S1. Second primary cancers occurred more often in women diagnosed with high-grade first primary ovarian tumors. Among women with first primary high-grade ovarian cancer, a lower incidence of second primary cancer at all sites was identified among White (SIR 0.92; 95% CI, 0.86–0.98; ER −9.51) women, whereas Black (SIR 1.26; 95% CI, 0.98–1.60; ER 26.20), Hispanic/Latina (SIR 1.25; 95% CI, 1.04–1.48; ER 23.14), and API women (SIR 1.22; 95% CI, 0.95–1.54; ER 13.89) experienced increased rates of second primary cancers. Rates of second vaginal cancer after high-grade ovarian cancer were elevated for all racial groups but highest for API women (SIR 30.81; 95% CI, 6.35–90.05; ER 3.12). Among API women, increased incidences of SPGC were observed with endometroid (SIR 5.10; 95% CI, 2.45–9.39; ER 33.95) and clear cell ovarian tumors (SIR 3.25; 95% CI, 1.73–5.56; ER 18.77).

In this large, retrospective population-based study of women with malignant ovarian cancers treated with definitive surgery and chemotherapy, API women had the greatest elevation in the incidence of SPGCs with Filipina women having the highest observed rates compared with other Asian women. API women in our sample differed with respect to sociodemographic characteristics; more Filipina women lived in areas with greater rates of poverty and lower educational attainment, whereas Chinese women were more likely to live in areas with higher proportions of foreign-born residents and household language isolation.

Our findings are consistent with other studies describing the incidence of second cancers after ovarian malignancies in the United States (4). In the study by Kanninen and colleagues, between 1992 and 2012 incidences of uterine and vaginal cancers were also observed to be higher among API and Black women, while an inverse association with SPGC was observed among White women. Our study expanded on this work, extending follow-up to 2016. We focused primarily on SPGCs and restricted our analysis to women who received treatment with both surgery and chemotherapy, suggesting a curative goal. We also investigated more specifically the population of API women by ethnic origin instead of considering API women to be a single homogenous group.

Disparities in ovarian cancer care between non-Hispanic White and Black women are well documented (20–23), while fewer studies focus on Asian-American women (6, 7). Compared with White women, Asian women are 20% less likely to receive definitive therapy for high-grade ovarian cancer (6). In our study, API women were less likely than White women to receive debulking surgery, which is consistent with other reports (7). As racial disparities in treatment and outcomes have been identified with ovarian malignancies, our inclusion criteria requiring definitive therapy with both surgery plus chemotherapy suggests that these differences persist even with receipt of the standard of care (6, 24, 25).

Biologic mechanisms for the observed differences in ovarian cancer outcomes could explain our findings. Variations in the BRCA gene, which are linked to breast and ovarian cancer, exist among different racial groups (26) and Asian ethnicities (15, 27, 28). Currently, BRCA testing is recommended by the American College of Obstetrics and Gynecology (29) and the National Comprehensive Cancer Network (30) for women with a personal history of EOC. Ensuring adherence to screening guidelines in Asian-American women is essential, as two predictive models used to assess BRCA1/2 carrier risk underpredicted the prevalence of mutations in Asian Americans by two-fold (31). A spectrum of homologous recombination deficiency is another biologic mechanism that could impact differential rates of ovarian cancer between races (32) and Asian ethnicities (33, 34).

The impact of socioeconomic status and other demographic factors including health behaviors and disease prevalence on ovarian cancer outcomes has not been well studied among Asian ethnicities. Intra-Asian differences in the incidence of human papillomavirus (HPV), which is associated with elevated rates of cervical, vulvular, and vaginal cancer (35) has not been well described. While differences in prevalence of HPV among Asian Americans is not well documented, there is a higher adjusted prevalence of HPV in Southeast Asia compared with Central Asia (36), and a higher incidence of cervical cancer in the Philippines than China, Japan, and Vietnam (37). Sociodemographic differences among Asian women may also contribute to differences in the incidence of HPV. Residence in census tracts with lower median incomes and educational attainment is associated with increased rates of HPV-associated cancers including cervical and vaginal cancer (35). In our study, sociodemographic differences were observed among Asian Americans, with Filipina women residing in areas with greater rates of poverty and lower educational attainment. Variation in health behaviors, including cervical cancer screenings between Asian ethnicities (38, 39) and HPV vaccination willingness between women who spoke English versus non-English languages at home have also been identified (40). These differences in health behaviors may extend to cancer surveillance following a first primary ovarian cancer. Directed public health efforts toward Asian communities in the United States are less prominent compared with other racial/ethnic minorities (41, 42), which may further compound disparities in health outcomes.

The widely recognized stereotype of API groups as “model minorities” in the United States has adverse health consequences (43), including the substantial lack of data equity for some Asian ethnic subgroups and minorities (44). Data equity with respect to cancer disparities is important in understanding differential risks in API communities. Filipina breast cancer survivors have an over 10% lower 5-year overall survival compared with other non-Filipina API women (45). Disparities also exist by nativity. Health disparities in cardiovascular disease (46, 47) and cancer (48–50) have been observed among Asian immigrant communities. The impact of factors that disproportionately affect immigrant groups, such as language isolation and foreign-born status, on cancer epidemiology and outcomes have largely been studied among Asian Americans as a composite measure—ethnic enclave residence (9, 12, 51). Differences in breast cancer incidence and mortality have also been identified between foreign-born and American-born Asian women (52, 53). Foreign-born Asian women are more than twice as likely to be diagnosed with breast cancer (48), and tend to have more advanced disease at diagnosis with subsequent lower 5-year survival than their American-born counterparts (13). Relatively less is known about the independent impact of language isolation in Asian Americans (54). This study includes multiple ecologic variables for non-Hispanic White, non-Hispanic Black, Hispanic/Latina, API, and large Asian ethnic subgroups. Disaggregating these sociodemographic factors allows for more granular between-group comparisons. Our study aimed to increase knowledge in this important area of disparities research into ovarian cancer. The diversity described here between different Asian ethnic groups in the risk of SPGC affirms that determining cancer risks and outcomes of women in these heterogeneous API communities solely as one group is insufficient.

This study is the largest analysis evaluating the incidence of SPGC following ovarian cancer focused on differences among Asian women by ethnic subgroup in the United States. Together, the SEER registries allowed us to sample women from a large population-based cohort that covers approximately 35% of the U.S. population overall and 56% of Asian and 69% of Hawaiian/Pacific Islander communities residing in the United States in particular. Our study also had limitations. The SEER classification of Asian ethnicity has a limited sensitivity and positive predictive value for API women. While this study focused on differences among Asian ethnicities, we did not have a sufficient number of SPGC cases in all Asian ethnic groups to provide a full accounting of second primary cancer risks, including when stratifying by important clinical factors such as tumor grade. There also exists a possibility of misclassification of recurrences as SPTs. However, in 2007, SEER implemented a new rule to promote consistent and standard definitions of multiple primary cancers versus recurrence. We used characteristics linked from the American Community Survey and the U.S. Decennial Census survey to describe socioeconomic status aggregated at the geographic level and not at the individual level. In addition, although we described differences in sociodemographic characteristics, our standardized incidence estimates were not adjusted for these variables in the primary analysis. Still, community-level characteristics such as language isolation are a meaningful way to elucidate how cancer disparities may persist in minority communities and immigrant enclaves. The disparity in the incidence of SPGCs between Asian ethnic subgroups of women suggests that cancer prevention and control efforts specific to structural barriers and health behaviors in communities of Filipina women would be well motivated to close this gap.

In conclusion, this study provides new insights into differences that exist among API survivors of ovarian cancer who experience an SPGC. Disaggregation of sociodemographic data creates a more nuanced picture of the communities in which these women live, but further study of individual levels factors is warranted. Asian American, and specifically Filipina women, may benefit from policies and public health interventions that increase cancer surveillance. Recognizing that cancer incidence differs among Asian ethnicities underscores the importance of detailed racial/ethnic reporting and studying interethnic cancer disparities.

N.H. Mukand reports Pfizer fellowship (March–November 2019; $25,000 was awarded by Pfizer to the University of Illinois Cancer Center to support a pharmacy student of the Cancer Center's choice who was engaged in cancer research). The cost of publication will be covered by the fellowship. Pfizer had no involvement in any part of this study, including but not limited to the design, data collection, interpretation, or writing this manuscript. N.H. Mukand had no communication with Pfizer. N.Y. Ko reports grants and personal fees from Pfizer outside the submitted work. G.S. Calip reports grants from Pfizer outside the submitted work, as well as current employment at the time of acceptance with Flatiron Health, Inc., which is an independent subsidiary of the Roche group. No potential conflicts of interest were disclosed by the other authors.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the funders.

N.H. Mukand: Data curation, formal analysis, writing–original draft, writing–review and editing. A. Zolekar: Data curation, formal analysis, writing–review and editing. N.Y. Ko: Writing–review and editing. G.S. Calip: Conceptualization, software, formal analysis, funding acquisition, investigation, methodology, writing–original draft, project administration, writing–review and editing.

N.H. Mukand was supported by the University of Illinois Cancer Center Pfizer Fellowship. G.S. Calip was supported by the NIH, National Center for Advancing Translational Sciences through grant number KL2TR002002.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.