Article Text
PDF
Abstract
Objective To describe sociodemographic and racial disparities in receipt of poly ADP-ribose polymerase inhibitors (PARPi) and bevacizumab among insured patients with ovarian cancer.
Methods This retrospective study used the Surveillance, Epidemiology, and End Results (SEER)–Medicare database to identify patients with advanced stage, high grade serous ovarian cancer diagnosed between 2010 and 2019. The primary outcome of interest was receipt of PARPi or bevacizumab at any time after diagnosis. χ2 tests were used to compare categorical variables. Factors independently associated with the receipt of PARPi and/or bevacizumab were identified using a multivariable logistic regression.
Results The cohort included 6242 patients; 276 (4.4%) received PARPi, 2142 (34.3%) received bevacizumab, and 389 (6.2%) received both. Receipt of either targeted treatment increased over the study period. On univariate analysis, patients who received either targeted therapy were younger (63% vs 48% aged <75 years; p<0.001), had a lower comorbidity index (86% vs 80% Charlson Comorbidity Index 0–1; p<0.001), and higher socioeconomic status (74% vs 71% high socioeconomic status; p=0.047) compared with those who did not receive targeted therapy. In the multivariable model, non-Hispanic black patients were less likely than non-Hispanic white patients to receive either targeted therapy (odds ratio 0.77; 95% confidence interval 0.61 to 0.98; p=0.032). Older patients (aged >74 years) were also less likely to receive PARPi or bevacizumab compared with those aged 65–69 years (all p<0.001).
Conclusion Sociodemographic and racial disparities exist in receipt of PARPi and bevacizumab among patients with advanced ovarian cancer insured by Medicare. As targeted therapies become more commonly used, a widening disparity gap is likely.
- Ovarian Cancer
- Retrospective Study
- Carcinoma, Ovarian Epithelial
Data availability statement
Data are available upon reasonable request. Data are available from the corresponding author upon reasonable request.
Statistics from Altmetric.com
What is already known on this topic
Racial, ethnic, and socioeconomic disparities have been demonstrated in access to novel cancer therapeutics in multiple tumor types, with a detrimental impact on oncologic outcomes.
What this study adds
This study demonstrated that non-Hispanic black patients with advanced ovarian cancer were >20% less likely than their white counterparts to received targeted therapy in the form of a poly ADP-ribose polymerase inhibitor or bevacizumab.
How this study might affect research, practice or policy
As targeted therapies become more widely used in cancer treatment, a widening disparity gap is likely.
Action must be taken to broaden access to tumor biomarker and somatic/germline genetic testing, which is a key determinant of the use of targeted therapy in cancer care.
Introduction
The 5 year survival rate for advanced ovarian cancer is <50%.1 However, for the first time in decades, new targeted therapies are beginning to move the needle in ovarian cancer survival. Poly ADP-ribose polymerase inhibitors (PARPi) and bevacizumab, a monoclonal antibody that inhibits vascular endothelial growth factor, were approved for the treatment of advanced ovarian cancer in 2014 based on improved oncologic outcomes.2 While initially approved in the recurrent setting only,3 4 in 2018 bevacizumab was approved by the US Food and Drug Administration (FDA) for upfront and maintenance therapy based on data from two phase III trials that demonstrated improved progression free survival in all patients and overall survival in a poor prognostic subgroup.5 6 Similarly, PARPi, which include olaparib, niraparib, and rucaparib, were also approved in the upfront maintenance setting in 2018 and have largely been used in patients with germline BRCA pathogenic variants or homologous recombination deficient tumors.7–10 Incorporation of these two maintenance therapy options in the past 10 years represents the first major change in frontline ovarian cancer treatment for decades, expanding the therapeutic options for a disease with high recurrence rates and overall poor survival rates.
Although the advent of targeted therapies in cancer management has led to significant advances in the field, there are several barriers that may result in inequitable access to these novel therapeutic options. A viewpoint published in JAMA Oncology in 2021 by Ashrafzadeh and colleagues demonstrated some of these potential barriers, which include high out-of-pocket costs, limited access to subspecialty care, and sociodemographic factors, all of which may result in lower rates of guideline concordant care.11 The authors concluded with a call to action for critical examination of disparities in differential receipt of targeted therapies so as not to widen the already existent disparity gap in cancer care.12 Racial, ethnic, and socioeconomic disparities have already been demonstrated in the use of immune checkpoint inhibitors in upfront treatment of multiple other cancer types, including advanced stage melanoma, hepatocellular carcinoma, and non-small cell lung cancer.13–15 Moreover, racial disparities exist in receipt of trastuzumab based therapy for HER2 positive breast cancer, with black patients significantly less likely to receive this highly effective therapeutic option than their white counterparts.16 Overall, however, there is a paucity of published data on disparities in receipt of targeted therapies in cancer care, including in ovarian cancer specifically.
Given the significant impact of PARPi and bevacizumab on ovarian cancer care in the past decade, and the known disparities that already exist in receipt of targeted therapies among other cancer types, we sought to assess sociodemographic and racial disparities in receipt of targeted therapies among an insured population of patients with advanced stage, high-grade serous ovarian cancer.
Methods
Data Source and Cohort Definition
We used the Surveillance, Epidemiology, and End Results (SEER)–Medicare database. Linkage of these two datasets provides clinical, demographic, and cause of death information for representative individuals with cancer paired with Medicare claims.17 Medicare is a US federal health insurance program primarily for individuals aged ≥65 years; it also provides coverage for certain younger individuals with disabilities and end stage renal disease patients. Given that the SEER database includes cancer incidence and survival data from population based registries covering almost half of the US population,18 results generated from SEER–Medicare data are considered fairly representative of elderly patients with cancer in the U.S.
We identified patients aged >65 years with advanced stage (stages III–IV), high grade serous ovarian, fallopian tube, or primary peritoneal cancer diagnosed between 2010 and 2019. The starting year of 2010 for diagnosis was selected given the use of these medications in the recurrent ovarian cancer setting at the beginning of the time interval. Therefore, although PARPi and bevacizumab were not approved for ovarian cancer before 2014, patients diagnosed before this may still have received them at the time of recurrence. We also performed a sensitivity analysis that included only patients diagnosed from 2014 to 2019, so as only to include those diagnosed at a time when PARPi and bevacizumab were approved for use by the FDA. Patients without at least 12 months of continuous Medicare part A and B before and after diagnosis, those who did not receive platinum based chemotherapy, and those with health maintenance organization insurance plans were excluded (Figure 1).
Flowchart of cohort selection. HMO, health maintenance organization.
Study Outcomes and Covariates
The primary outcome of interest was receipt of PARPi (niraparib, olaparib, or rucaparib) or bevacizumab, defined by at least one claim for any of the targeted agents, at any time after ovarian, fallopian tube, or primary peritoneal cancer diagnosis. Covariates of interest included race and ethnicity (non-Hispanic white, non-Hispanic black, Asian, Hispanic, other, and unknown), marital status at diagnosis (single, married, separated or divorced, widowed, unmarried or domestic partner, and unknown), age group at diagnosis (65–69, 70–74, 75–79, 80–84, and ≥85 years), year of diagnosis (2010–19), cancer stage (III or IV), geographical region (Midwest, Northeast, South, and West), rurality (metropolitan, urban, and rural), Charlson Comorbidity Index (0, 1, and ≥2), and YOST index, a census tract level socioeconomic status variable that is categorized into quintiles. Rurality was categorized into metropolitan, urban, and rural. The socioeconomic status variables included in the YOST index include median household income, median house value, median rent, per cent <150% of poverty line, education index, per cent working class, and per cent unemployed.19 20 In this study, the YOST index was dichotomized into low socioeconomic status (quintiles 1–2) and high socioeconomic status (quintiles 3–5).
Statistical Analysis
Summary statistics such as means, SD, ranges, frequencies, and percentages were used to describe the study population, and comparisons were performed using the χ2 test or Fisher exact test, depending on the underlying distribution of the data with respect to the primary outcome. Logistic regression was used to estimate the odds ratio (OR) with a 95% confidence interval (CI) of treatment utilization based on patient characteristics. The comparison group in the model consisted of patients who did not receive any targeted agents. The backward selection method was applied to select covariates that were significantly associated with the outcome. Variables with a p value <0.1 in the full model were retained in the final model to estimates the ORs. The final model controlled for race and ethnicity, age group, year of diagnosis, region, and Charlson Comorbidity Index. A sensitivity analysis was performed that included only those diagnosed in the years 2014–19. A p value <0.05 was considered significant. All statistical analysis was performed using SAS Enterprise Guide V.7.1.
To evaluate the robustness of our results, we used the E value to assess for unmeasured confounding. The E value was introduced by VanderWeele and Ding in 2017.21 22 Unlike other methods for conducting sensitivity analyses, the E value does not require input of subjective investigator assumptions about the data, but uses the estimates derived from the study itself to quantify how strong an unmeasured confounder would have to be to explain away an observed exposure–outcome association. In the context of this study, we calculated the magnitude of the association that an unmeasured confounder must have with both the exposure (eg, race) and the outcome (receipt of a targeted therapy) to drive the study derived estimate to the null (see the Online Supplemental Appendix for more details).
Supplemental material
In accordance with the journal’s guidelines, we will provide our data for independent analysis by a selected team by the editorial team for the purposes of additional data analysis or for the reproducibility of this study in other centers if such is requested.
Results
Patient Characteristics
Of the 6242 patients who met inclusion criteria and were included in the final cohort (Figure 1), 276 (4.4%) received PARPi, 2142 (34.3%) received bevacizumab, and 389 (6.2%) received both. Patient characteristics stratified by receipt of any targeted therapy are shown in Table 1. Receipt of targeted treatment increased over the study period (Figure 2), with the majority of patients (55.3%) receiving treatment diagnosed in 2015 or later. On univariate analysis, patients who received either targeted therapy were more likely to be married compared with those who did not receive targeted therapy (32.7% vs 28.2%). Those who received PARPi and/or bevacizumab were also more likely to be younger (63% vs 48% aged <75 years; p<0.001), have a lower Charlson Comorbidity Index (86% vs 80% Charlson Comorbidity Index 0–1; p<0.001), and to be of higher socioeconomic status (74% vs 71% high socioeconomic status; p=0.047) compared with those who did not receive targeted therapy. Race and ethnicity, cancer stage, and rurality were not significantly different between the two groups (all p>0.05). Characteristics in the two groups were similar in the sensitivity analysis that included only patients diagnosed in 2014–19 (n=3629; Table 2).
Medication utilization over time. Percentages represent yearly medication utilizations (based on year of ovarian cancer diagnosis) over the number of patients diagnosed each year. PARPi, poly ADP-ribose polymerase inhibitor.
Patient characteristics by receipt of targeted therapy (yes/no) (2010–19)
Patient characteristics by receipt of targeted therapy (yes/no) (2014–19 sensitivity analysis)
Logistic Regression Model
Results from the multivariable logistic regression are demonstrated in Table 3. In the multivariable model, non-Hispanic black patients were 23% less likely than non-Hispanic white patients to receive either targeted therapy (OR 0.77; 95% CI 0.61 to 0.98; p=0.032). Hispanic patients also had decreased odds of receiving targeted therapy compared with non-Hispanic white patients (OR 0.88), but this was not statistically significant in the multivariable model (p=0.23). Older patients (aged >74 years) were less likely to receive PARPi or bevacizumab compared with those aged 65–69 years (all p<0.001); those aged ≥85 years had the lowest odds (OR 0.34) of receiving targeted therapy (95% CI 0.27 to 0.42) compared with the youngest cohort (those aged 65–69 years). Similarly, a Charlson Comorbidity Index of ≥2 was associated with decreased odds of receipt of targeted therapy (OR 0.68, 95% CI 0.59 to 0.78; p<0.001) compared with a Charlson Comorbidity Index of 0. In terms of geographical region, patients residing in the Northeast, South, and West were significantly more likely to receive targeted therapy compared with those in the Midwest region (OR 1.40, 95% CI 1.47 to 1.49, respectively; all p<0.01). A sensitivity analysis that included patients diagnosed in 2014–19 demonstrated similar results (Table 4); in this logistic regression model, non-Hispanic black patients were 18% less likely than non-Hispanic white patients to receive either targeted therapy (OR 0.82; 95% CI 0.60 to 1.12; p=0.21). Similar to the initial analysis (2010–19), older patients (>74 years) were significantly less likely to receive PARPi or bevacizumab, as were those with a Charlson Comorbidity Index of ≥2. In the sensitivity analysis, the odds of receiving targeted therapy was significantly higher in the years 2017 and 2018 compared with 2014 (both OR 1.46).
Multivariable logistic regression model estimating the odds ratio of targeted therapy utilization based on patient characteristics (2010–19)
Multivariable logistic regression model estimating the odds ratio of targeted therapy utilization based on patient characteristics (2014–19 sensitivity analysis)
Sensitivity Analysis Using E Values
Assessment of the sensitivity of our findings using E values demonstrated that after adjusting for measured covariates, an unmeasured confounder would have to be associated with both non-Hispanic black race and receipt of a targeted therapy by an OR of 1.92 to drive the estimate to the null. For a Charlson Comorbidity Index of ≥2, the magnitude of association for an unmeasured confounder with both high Charlson Comorbidity Index and receipt of targeted therapy would have to be at least 1.72. Additional E values for covariates with significant ORs in the multivariable logistic regression model are provided in the Online Supplemental Table S1 and Figure S1, respectively.
Discussion
Summary of Main Results
The results of this study indicated that among patients with advanced ovarian cancer enrolled in Medicare, sociodemographic and racial disparities existed in the receipt of PARPi and bevacizumab. Non-Hispanic back patients, those with lower socioeconomic status, and patients who resided in the Midwest region of the US were less likely to receive these targeted therapies. Although this study focused on a specific subgroup of insured patients with ovarian cancer, the results highlight a deviation from standard of care and potential access problems for a substantial portion of cancer patients. As targeted therapies become more used in routine cancer care, this disparity gap is likely to widen if action is not taken to ensure equitable access.
Results in the Context of Published Literature
Disparities in receipt of novel cancer therapeutics, including immunotherapy and targeted therapies, have been demonstrated in multiple cancer types.13 14 16 Increasing age and comorbidity index have both been associated with a lower odds of receiving immunotherapy in metastatic melanoma13; these factors are more likely to be associated with a side effect profile of therapies being considered and evaluation of the risk–benefit ratio for older, more medically complicated patients. In contrast with these more straightforward clinical variables, disparities based on race and socioeconomic status are more complex and more likely related to systemic barriers to quality, guideline concordant care, and systematic racism. It is considered standard of care, for example, for advanced ovarian cancer patients with a germline or somatic BRCA1/2 deleterious mutation or homologous recombination deficient tumor to receive upfront maintenance PARPi.2 However, genetic testing rates among newly diagnosed ovarian cancer patients are as low as 53% in some studies, indicating varying practice patterns among gynecologic oncologists.23 24
Additionally, evidence is mixed on whether or not racial differences exist in the incidence of these genetic signatures.25 Given the unclear association with race, and the fact that black breast and ovarian cancer patients are less likely than their white counterparts to receive germline genetic testing,23 it is difficult to ascertain to what extent this may influence differential receipt of targeted therapy. Rather, systematic inequities seem to exist in receipt of standard of care cancer therapy for ovarian cancer, with non-Hispanic black patients less likely than non-Hispanic white patients to receive guideline concordant treatment in a recently published SEER–Medicare study.26
Importantly, given that receipt of these targeted therapies in the frontline maintenance setting has been demonstrated to improve progression free survival, as well as overall survival for particular subsets of patients (BRCA mutated for PARPi and stage IV disease for bevacizumab), it is reasonable to assume that limited access to these therapies will detrimentally impact oncologic outcomes disproportionately.5 9 27 28 This has been previously demonstrated in advanced hepatocellular carcinoma, with Hispanic and black patients less likely to receive immunotherapy (adjusted OR 0.63 and 0.71, respectively) compared with white patients, and receipt of immunotherapy associated with improved overall survival.14
Strengths and Weaknesses
This study had a number of important strengths and limitations. A key strength was the inclusion of a large, registry based cohort that is designed to be representative of cancer patients across multiple regions in the US. Limitations included lack of granularity in the available data given the use of a claims based database, with limited data on important covariates, such as BRCA/homologous recombination deficient status, provider practice patterns, reasons for differential receipt of these therapies, and relevant clinical outcomes, such as progression free survival; lack of available secondary outcomes limited our ability to translate the disparities in receipt of targeted therapy to oncologic outcomes. Additionally, all patients had continuous enrollment in Medicare insurance, so the results do not capture patients who are uninsured or underinsured and likely face even more significant barriers.
Finally, the primary outcome was broad and included receipt of targeted therapy at any time point in the disease course. Given the years included in the study and timing of FDA approvals, many patients likely received PARPi and/or bevacizumab in the recurrent setting, as opposed to the upfront maintenance setting, where these therapies are currently more commonly utilized in clinical practice today. Similarly, inclusion of patients diagnosed before FDA approval of these medications may have introduced selection bias. However, we attempted to account for this by performing a sensitivity analysis that only included patients diagnosed in the year 2014 or later.
Implications for Practice and Future Research
Leaders in oncology must consider and implement effective initiatives to improve equitable access to targeted therapies or it is likely that the existent disparities will worsen with the more routine use of these therapeutics in clinical practice. Academic societies within the field of gynecologic oncology specifically should emphasize in practice guidelines the importance of universal germline and somatic genetic testing at the time of ovarian cancer diagnosis to foster broader adoption of these practices. Importantly, a key determinant of the use of targeted therapy in cancer care is access to tumor biomarker and somatic and germline genetic testing. Despite the fact that all women diagnosed with advanced stage epithelial ovarian cancer should be offered BRCA genetic testing (germline first and, if negative, somatic tumor testing), racial disparities in both referral and completion of genetic testing are known to exist among patients with gynecologic malignancies.29–31 Key barriers for racial and ethnic minority populations include lack of knowledge of the purpose of genetic testing, time required to attend genetics appointments, and other socioeconomic factors.29 Additional provider and healthcare system level barriers undoubtedly exist, including a shortage of genetic counselors and lack of knowledge regarding reimbursement policies.32
One potential solution to improve uptake of genetic testing could be to incorporate genetic counseling, either in person or virtually, and testing into routine oncology visits in a more systematic fashion, and therefore eliminate the need to attend separate appointments. However, a key barrier to this approach exists, as Medicare does not currently reimburse for genetic counseling, unless the counselor is under direct supervision of a physician. Policy changes are needed to expand reimbursement for this critical portion of cancer care for Medicare beneficiaries. Another possible intervention is to incorporate a short educational video into routine care, which is currently being investigated in a randomized controlled trial for black women at high risk of hereditary breast and ovarian cancer.33
Conclusions
This study highlighted the existing racial, socioeconomic, and regional disparities in receipt of targeted maintenance therapies for patients with advanced ovarian cancer. Future efforts must focus on increasing genetic testing and access to guideline concordant care for these at risk populations, especially as the use of targeted and immune therapies continues to increase across cancer care.
Data availability statement
Data are available upon reasonable request. Data are available from the corresponding author upon reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
This study was approved by the institutional review board (IRB) at the University of Texas MD Anderson Cancer Center (IRB approved protocol PA14-0949 Patterns, costs, and outcomes of cancer care).
Acknowledgments
The collection of cancer incidence data used in this study were supported by the California Department of Public Health pursuant to California Health and Safety Code Section 103885; Centers for Disease Control and Prevention’s National Program of Cancer Registries, under cooperative agreement 1NU58DP007156; the National Cancer Institute’s Surveillance, Epidemiology and End Results Program under contract HHSN2612018000321 awarded to the University of California, San Francisco, contract HHSN2612018000151 awarded to the University of Southern California, and contract HHSN2612018000091 awarded to the Public Health Institute. The ideas and opinions expressed herein are those of the authors and do not necessarily reflect the opinions of the State of California, Department of Public Health, the National Cancer Institute, and the Centers for Disease Control and Prevention or their Contractions and Subcontractors.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
X @AnneKniselyMD
Contributors Conceptualization: AK and RNW. Methodology: AK, C-FW, JAR-H, and RNW. Formal analysis: CW. Resources: SG and JAR. Data curation: AK, C-FW, and RNW. Writing—original draft: AK. Writing—review and editing: AK, C-FW, AK, NA, KAJ, AM, SHG, JAR-H, and RNW. Visualization: AK and C-FW. Supervision: SHG, JAR-H, and RNW. Project administration: SHG, JAR-H, and RNW. Funding acquisition: SHG and JAR-H. RNW is the guarantor for this study and therefore accepts full responsibility for the finished work and/or the conduct of the study, had access to the data, and controlled the decision to publish.
Funding This research was supported in part by the National Institutes of Health through MD Anderson’s Cancer Center support grant (P30CA016672) and the MD Anderson T32 training grant for gynecologic oncologists (AK, KAJ, RNW; T32CA101642). NA is supported by the Fundación Alfonso Martin Escudero. AM is supported by the Department of Defense Ovarian Cancer Research Program (OC210024). SHG is supported by CPRIT RP210140, Komen SAC 150061, and NIH R01 AR078484. This work was also supported by grants from the National Institutes of Health/National Cancer Institute (JAR-H; K08CA234333, R01MD017999). The funding sources were not involved in the development of the research hypothesis, study design, data analysis, or manuscript writing.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.