Racial/Ethnic Disparities in Ovarian Cancer Treatment and Survival

Ovarian cancer is the second most common gynecologic cancer and the leading cause of death from gynecologic malignancies in the United States (1). Among patients with ovarian cancer, there are well-known racial disparities, with African American (AA) women being less likely to receive adequate treatment and more likely to experience worse survival compared with white women (2, 3). These differences have been attributed to unequal access to care and receipt of treatment (3). AA women are also more likely to be obese and to have related comorbidities (4, 5), which are known to affect chemotherapy dosing, and dose reduction has been shown to reduce ovarian cancer survival (6). Previous studies have not taken these factors into account, and possible disparities among other racial/ethnic groups have received little attention. We evaluated race/ethnicity differences in chemotherapy dosing and survival in a cohort study of epithelial-invasive ovarian cancer cases diagnosed and treated as members of Kaiser Permanente Northern California (KPNC) and who thus had equal access to health care.

The Kaiser Permanente-Research on Ovarian Cancer Survival (KP-ROCS) Cohort Study has been described in detail elsewhere (6). In brief, cases of invasive epithelial ovarian cancer 21 years or older and diagnosed from 2000 to 2013 were identified through the KPNC Cancer Registry. Information on patient demographic and clinical characteristics including dosing was obtained from KPNC electronic medical records, including its Virtual Data Warehouse (7, 8). Information on race/ethnicity was obtained from the VDW's Tumor File, which reflects data from the KPNC Cancer Registry. There were 1,307 whites, 106 AA, 167 Hispanics, 253 Asians, and 13 of “other” race. We excluded these 13 women as the group was too small for meaningful analyses.

Included in the chemotherapy subcohort were patients who received intravenous adjuvant first-line therapy of carboplatin and paclitaxel with curative intent, with complete dosing data, resulting in 793 patients. Demographics and clinical characteristics were similar among women included in the chemotherapy subcohort and the full cohort, with the exception that those receiving chemotherapy were less likely to be over 70 years of age or to have advanced disease, which was expected (6).

Relative dose (RD: actual dose/expected dose for first cycle) and Relative dose intensity (RDI: actual dose administered per week/expected dose per week) were computed for paclitaxel and carboplatin separately, and for the combination regimen by computing the average RD (ARD) and average RDI (ARDI), as calculated by others (9–11) and described in detail elsewhere (6). Expected doses were based on National Comprehensive Cancer Network (NCCN) Guidelines (www.nccn.org). Chemotherapy dose reduction was defined as an RDI for the full regimen of less than 85% (12–14). Early discontinuation was defined as not completing the full six scheduled treatments (15), or alternatively, receiving less than four cycles. Treatment delay was defined as a delay in receiving scheduled chemotherapy treatment of more than 7 days.

Outcomes included overall mortality and ovarian cancer–specific mortality identified through the KPNC Mortality Linkage System through December 2014, representing a median follow-up of 38 months (50 months for the chemotherapy subcohort). The study was approved by the Institutional Review Boards at KPNC and Rutgers Biomedical and Health Sciences.

Distributions for demographic and clinical characteristics were compared across racial/ethnic groups using χ² or Fisher exact test, as appropriate. In stratified analyses by race/ethnicity we compared the proportions of women with dose reduction (RDI < 85%), early discontinuation, and treatment delay using χ² tests, as well as mean values of RD, RDI, ARD and ARDI, actual dose of paclitaxel and carboplatin, number of cycles, and treatment duration using ANOVA.

Survival for AAs, Hispanics, and Asians was compared with that of white women by conducting proportional hazards regression to calculate HRs and 95% confidence intervals (CIs) for overall and ovarian cancer–specific survival, after adjusting for relevant covariates. Potential covariates included age at diagnosis, stage, grade, histologic subtype, body mass index (BMI) at diagnosis, chemotherapy-related toxicities, use of G-CSF for prophylaxis or treatment of neutropenia, relevant comorbidities, post-treatment CA-125 as a marker of residual disease (16, 17), ARDI, and type of surgery. Chemotherapy-related toxicities considered as potential covariates included severe myelosuppression (severe neutropenia or thrombocytopenia) and neuropathy, grade III/IV, according to the National Cancer Institute's Common Terminology Criteria for Adverse Events (NCI CTCE), version 3.0. Comorbidities included those likely to affect dosing or survival (diabetes, hypertension, cardiovascular disease, and acute kidney disease and chronic renal insufficiency). SAS version 9.2 (SAS Institute) was used for analyses.

Clinical characteristics by race/ethnicity are shown in Table 1. Hispanic and Asian women were younger at diagnosis. AA women were more likely to be diagnosed with advanced disease (26.4%), to have hypertension (75.5%), cardiovascular disease (63.21%), or renal disease (58.5%), not to have surgery (12.3%), and to have elevated post-treatment CA-125 (35.8%), a marker of residual disease, than any other racial/ethnic group. Both AA and Hispanic women had high prevalence of obesity, with Hispanic women having the highest prevalence of diabetes (32%). Compared with white women, Asian women were less likely to have advanced disease or be obese and more likely to have endometrioid and clear cell tumors. Severe neutropenia was more commonly diagnosed among whites and they were more likely to receive G-CSF than any other racial/ethnic group.

Compared with whites, AAs were more likely to have dose reduction (ARDI < 85%), treatment delay, and early discontinuation, while Hispanics were also more likely to have dose reduction, but less likely to have early discontinuation or treatment delay (Table 2). However, only early discontinuation was statistically significant.

Mean actual dose (mg/kg of body weight) of paclitaxel and carboplatin was considerably lower for AA women than any other group (P = 0.006 and <0.0001, respectively), shown in Table 2. Notably, compared with whites, AA women tended to have higher RD in the first cycle but fewer cycles, while Hispanics had lower RD in the first cycle but longer treatment duration (more cycles), resulting in similar mean total ARDI for AA and Hispanics, with both receiving slightly lower ARDI than white women. In contrast, mean ARDI was highest among Asian women (Table 2).

For the full cohort, although mortality rates were highest for AA women, we did not observe significant differences in survival by race/ethnicity after adjusting for major prognostic factors, including treatment. However, a different picture emerged when analyses were restricted to women receiving the carboplatin and paclitaxel, the most common chemotherapy regimen in the treatment of ovarian cancer (18). Compared with white women with ovarian cancer, AA patients had the worst survival among race/ethnicity groups after taking into account potential clinical and treatment differences (Table 3). After controlling for age at diagnosis, stage, grade, histology, BMI at diagnosis, comorbidities potentially affecting dosing (diabetes, hypertension, cardiovascular diseases, renal disease), surgery type, and post-treatment CA 125, adjusted HRs (95% CI) for overall mortality were 1.56 (1.01–2.40) for AAs; 0.95 (0.66–1.39) for Asians; and 1.41 (0.97–2.03) for Hispanics. Further adjusting for ARDI of carboplatin–paclitaxel received, chemotherapy-related toxic effects, and G-CSF use did not impact risk estimates. Findings for ovarian cancer–specific mortality were similar (data not shown).

Disparities in ovarian cancer survival have been reported using Surveillance, Epidemiology, and End Results (SEER) data, SEER-Medicare data, and a National Cancer Database with most reports focusing on differences in AA versus white women (3). Our study confirms some of the known disparities in ovarian cancer disease presentation, treatment, and survival in AAs compared with whites. We found that AAs were more likely to be diagnosed with advanced disease, not to have surgery and have post-treatment residual disease, to have the serous histologic subtype, and to be obese and have related comorbidities, and to experience worse survival compared with whites, similar to what has been reported in the literature (2). We also found that they were more likely to have treatment delay and early discontinuation, and to receive the lowest mean ARDI for the regimen compared with other racial/ethnic groups, which to our knowledge was not previously evaluated.

Hispanics represent a rapidly growing group in the United States, with approximately 16.3% of the total U.S. population in the 2010 Census, representing a 43% increase in the previous decade (19). However, little is known about ovarian cancer disparities in treatment and survival in this population. Our study agrees with an earlier report using SEER data by Ibeanu and Diaz-Montes, which found that Hispanics tended to have an earlier age at diagnosis, compared with white and AA women (20). However, in contrast to our study, the report by Ibeanu and Diaz-Montes, which included nonepithelial tumors, found that Hispanic women tended to be diagnosed at an earlier stage and experience better survival than other groups; analyses adjusted for prognostic factors were not presented. Our study is the first to evaluate the differences in chemotherapy dosing and to present analyses accounting for detailed clinical information. We observed differences in treatment and poorer survival for this population compared with whites, after adjusting for detailed treatment information including chemotherapy dosing, but risk estimates did not reach statistical significance.

Our results are in general agreement with the few reports evaluating ovarian cancer survival disparities in Asians. As with others (21), we found that Asian women tended to be diagnosed at an earlier age and with more localized disease, were more likely to have endometrioid and clear cell tumors, and to experience better survival, compared with whites. They tend to have lower body mass index, which may facilitate earlier detection. Our study is the first to evaluate chemotherapy dosing, and we found that Asians were also less likely to have residual disease and more likely to receive higher RDI than any other racial/ethnic group, which may in part explain their survival advantage.

A limitation of the study is the classification of race/ethnicity in broad groups. For example, Asians include Japanese, Chinese, Korean, Vietnamese, Filipino, and South Asian women of Indian or Pakistani descent, and these subgroups have been shown to have different 5-year ovarian cancer–specific survival experience using SEER data, ranging from 62.1% for Vietnamese to 48.2% for Asian Indian/Pakistani (21). It also includes immigrant versus U.S.-born Asians, who also were found to have somewhat different 5-year disease-specific survival (55% vs. 52%, respectively), but both better than U.S. whites (48%) (21). The overall observation of better survival for Asians agrees with our findings after controlling for major prognostic factors. Similarly, for Hispanics, we had no information on country of origin and, to our knowledge, no studies have evaluated differences in survival according to subethnicity/indigenous ancestry.

Our study is the first to evaluate disparities using detailed clinical information, including variables that allowed the calculation of RDI for the most common regimen. It is also one of the first to evaluate disparities in an observational study under conditions of potentially equal access to health care. We were also able to take into account differences in severe chemotherapy-related toxicities and G-CSF use, which may have affected chemotherapy dosing. We did not find that severe neutropenia occurred more frequently in AA women, despite their known tendency to have lower white blood cell and absolute neutrophil counts than white women (22). Nevertheless, even after adjusting for these clinical variables, disparities in survival persisted, with AA women experiencing poorer survival. However, there are other factors at the personal level, such as education and cultural beliefs, that may have affected both personal and clinician decisions on treatment. For example, we demonstrated previously that obese women are more likely to have dose reduction, and those with dose reduction have poorer survival (6). There may also be biological differences in the disease, by which, for example, AAs may be diagnosed with more aggressive disease and have pre-existing conditions that complicate treatment. Similar observations are well known in breast cancer (5, 23).

In conclusion, survival disparities persisted in a population covered by and who received care from a single large integrated health care provider, after adjusting for detailed treatment and prognostic characteristics. Future studies should explore biological, personal, and social factors to further understand and address these inequities in treatment and survival.

No potential conflicts of interest were disclosed.

Conception and design: E.V. Bandera, L.H. Kushi

Development of methodology: E.V. Bandera, L.H. Kushi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): V.S. Lee, B. Powell, L.H. Kushi

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.V. Bandera, V.S. Lee, B. Powell, L. Rodriguez-Rodriguez, L.H. Kushi

Writing, review, and/or revision of the manuscript: E.V. Bandera, V.S. Lee, L. Rodriguez-Rodriguez, B. Powell, L.H. Kushi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.V. Bandera, V.S. Lee, L.H. Kushi

Study supervision: E.V. Bandera, L.H. Kushi

This work was supported by grants from the National Cancer Institute (K22 CA138563, UC2 CA148185, U24 CA171524), and the Kaiser Permanente Center for Safety and Effectiveness and Safety Research.

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.