Real-world Utilization of Second-line Poly (ADP-Ribose) Polymerase Inhibitor Maintenance Therapy Versus Active Surveillance in Community Oncology Practice Patients With Advanced Ovarian Cancer

J Clin Pathways. 2022;8(4):38-46. doi:10.25270/jcp.2022.05.3
Received October 30, 2021; accepted March 4, 2022.

Abstract

Purpose: This study analyzed second-line poly(ADP-ribose) polymerase inhibitor (PARPi) maintenance therapy in a real-world setting to understand how it is being utilized and the resultant impact on patients’ disease course. Methods: This was a retrospective analysis of adult females diagnosed with advanced ovarian cancer (OC) who received a second-line platinum-containing regimen for advanced OC and had ≥2 visits in an electronic health records database between January 1, 2016, and July 31, 2020. Results: Of 1051 patients who received second-line platinum-based chemotherapy, 513 (48.8%) subsequently received maintenance therapy (27.9% of 1,051 received bevacizumab, 25.8% PARPi and 3.0% received a nonplatinum chemotherapy). There were 639 patients who had a documented BRCA test result (14.4% BRCAmut, 85.6% BRCAwt). Significantly more BRCAmut patients received second-line PARPi maintenance (53.3%) than BRCAwt patients (30.5%; P<.0001). Overall survival at 24 months was significantly higher with PARPi maintenance versus active surveillance: 61.2% (95% CI, 52.4%–68.8%) versus  53.0% (95% CI, 47.1%–58.7%; log-rank P=.005). Conclusion: A significant proportion of recurrent OC patients are not receiving second-line PARPi maintenance therapy despite treatment guideline recommendations and real-world survival benefits associated with its use.

Ovarian cancer (OC) was the 10th most common cancer among women in the United States in 2020.¹ Although the age-adjusted incidence of OC is lower than other cancers among women, such as breast cancer (11.4 vs 126.8 per 100,000 women in the United States), the mortality rate of OC is much higher.¹ The 5-year survival rate of OC in the United States is only 49.1% vs 90.2% for breast cancer.² Given the high fatality rate, ensuring availability and access to life-prolonging therapies for the OC population is essential.

In phase 3 clinical trials, poly(ADP-ribose) polymerase inhibitor (PARPi) treatments have been shown to extend the time between second-line chemotherapy and relapse (ie, progression-free survival) in the metastatic OC population.^3–5 On the basis of these studies, PARPi treatments have been recommended since 2017 as second-line maintenance treatment by the National Comprehensive Cancer Network (NCCN) for patients with OC with or without BRCA1 or BRCA2 (BRCA) alterations.⁶

This study analyzed second-line maintenance therapy, with a particular focus on PARPi therapy, in a real-world setting to understand how it is being utilized and the resultant impact on the disease course of patients. Since PARPi treatments are a relatively new class of agent, it was of interest to study their use and outcomes in real-world setting. To that end, we present recent estimates of second-line PARPi maintenance utilization and a landmark analysis to 24 months comparing patients receiving PARPi with those who did not receive second-line maintenance therapy for OC.

Materials & Methods

This study was performed primarily using the iKnowMed electronic health record (EHR) database of the US Oncology Network (McKesson Corporation, Las Colinas, TX), which includes 1400 affiliated physicians operating in more than 470 sites of care in the United States and provides care for more than 1 million patients annually.⁷ The iKnowMed database is geographically diverse and representative of the US community oncology population.

iKnowMed captures outpatient practice encounter histories for patients under community-based care, including but not limited to patient demographics, such as age and sex; clinical information, such as disease diagnosis, diagnosis stages, performance status, and laboratory testing results; and treatment information, such as line of therapy and treatment administration within the US Oncology Network.

The Limited Access Death Master File of the Social Security Administration also was used as an additional source of vital status (death), in addition to death dates recorded in the EHR database. Common patient identifiers (patient ID number, name, and birth date) were used to link patients from the iKnowMed data warehouse.

The study protocol was granted an exception and waiver of informed consent by the US Oncology Network Institutional Review Board because of the deidentified data set being Health Insurance Portability and Accountability Act compliant and the study posing minimal risk to patients.

Study Design  

This was a retrospective, observational, real-world data study of US patients with advanced OC who received a second-line platinum-containing regimen for advanced OC and who had ≥2 visits within the database between January 1, 2016, and July 31, 2020 (ie, patient identification period) in the US Oncology Network. Eligible patients who received either a PARPi or underwent “active surveillance” were followed longitudinally to assess survival to 24 months until October 10, 2020, the last patient record, or date of death, whichever occurred earliest (Figure 1).

Study Population

The study population consisted of patients with advanced OC who initiated a qualifying regimen within the US Oncology Network. Inclusion criteria were as follows: (1) female patients with stage III or IV epithelial OC; (2) patients aged ≥18 years at first diagnosis of advanced OC; (3) patients with ≥2 visits within the database during the study observation period; and (4) patients who had received second-line treatment with platinum-containing regimen (cisplatin or carboplatin) for advanced OC during the study identification period. The study excluded patients who received treatment for other documented primary cancer diagnoses during the study observation period.

Study Time Periods

Overall response rate (ORR) was 94.5%, including The patient identification period was between January 1, 2016, and July 31, 2020. Among patients who received PARPi maintenance treactment, patients’ first visit during the study identification period served as their Index date for follow-up. Among patients who underwent active surveillance, the patients’ last date of second-line chemotherapy served as their index date since they did not have a second-line maintenance treatment. The study observation period was between January 1, 2016, and October 31, 2020. Patient follow-up for overall survival (OS) occurred within the study observation period. The minimum follow-up period was 3 months; the maximum follow-up period was 58 months.

Study Variables and Outcomes

Baseline demographics (eg, date of birth, sex, race) and baseline clinical characteristics (eg, menopausal status, other cancer diagnoses, date of initial OC diagnosis, date of advanced or metastatic OC diagnosis, stage at initial diagnosis, TNM staging, tumor grade, genetic testing, Eastern Cooperative Oncology Group [ECOG] performance status, labs, and comorbidities) were identified by using the closest visit date in the database prior to the index date and within 30 days. Medication utilization (eg, second-line index treatment regimen, second-line index treatment dates [start/end], duration of second-line index therapy, other lines of therapy/treatments) was ascertained throughout the study observation period. Data collected for clinical outcomes (eg, last visit date, death date, vital status) marked the end of the study observation period for each patient. If a patient had no event (eg, last visit date, death date, change in vital status), they were censored at the end of the overall study follow-up period on October 31, 2020.

Statistical Methods

Descriptive analyses were conducted to evaluate the demographic, clinical, and treatment characteristics. Categorical variables were reported as frequency and percentage. Continuous variables, such as age, were reported as mean and standard deviation. χ2 testing was used to assess associations between categorical variables or Fisher’s exact test (when distribution could not be assumed to be χ2). Depending on normality, analysis of variance/t tests or Kruskal-Wallis tests were used for continuous variables. An alpha level of .05 was used as the primary criterion for statistical significance of this study.

Rates of any second-line maintenance therapy and PARPi second-line maintenance therapy following discontinuation of second-line platinum-based chemotherapy were determined for the following groups: (1) all patients, (2) patients with a BRCA gene alteration (BRCAmut), and (3) patients without a BRCA gene alteration (BRCAwt). These rates were reported as frequency and percentage. Rates of any second-line maintenance utilization and second-line PARPi maintenance utilization were compared between BRCAmut and BRCAwt populations using χ2 tests.

The OS time-to-event outcome to 24 months was assessed between the second-line PARPi maintenance and watchful waiting groups using the Kaplan-Meier method with 95% CIs, and summary tables of the number of events and censored patients at each month interval were included. Statistical testing for the difference in OS was compared between PARPi and active surveillance groups using the log-rank test.

Results

A total of 11,494 patients were identified with a diagnosis of advanced OC. Of those, 1051 met the study inclusion criteria (Table 1).

Of 1051 patients who received second-line platinum-based chemotherapy, 513 (48.8%) subsequently received any second-line maintenance therapy (27.9% of 1051 received bevacizumab, 25.8% received PARPi and 3.0% received a nonplatinum chemotherapy), and 538 (51.2%) did not receive maintenance therapy (ie, underwent “active surveillance”)
Figure 2). This study analyzed the comparison between second-line PARPi maintenance vs active surveillance.

Among the 1051 patients who met inclusion criteria, 1020 had nonmissing baseline characteristics (Table 2).

In univariate analyses, patients who underwent active surveillance were significantly older at baseline than patients who received PARPi (mean, SD) 65.4, 11.7 vs 63.2, 10.4; P=.006. Race groups had significantly different proportions between groups (P=.0008), but most patients in both groups (approximately 80%) were White, and the PARPi group had more patients in the “other” category. The groups had statistically different distributions in ECOG performance status (P=.007), with more patients who received PARPi being assigned an ECOG performance status of 0. There were no statistically significant differences between patients who received PARPi and patients who underwent active surveillance with respect to practice region, stage at diagnosis, or number of metastases (Table 2).

Utilization of Second-line PARPi Maintenance Versus Active Surveillance Over Time

The proportion of eligible patients utilizing active surveillance (n=538) increased over the study period, from 27.1% in 2017 to 48.7% in 2020. Of the 271 patients who received PARPi second-line maintenance therapy, the utilization increased from 17.2% in 2017 to 34.8% in 2019 but decreased to 22.3% in 2020 (Figure 3).

Utilization of Second-line PARPi Maintenance by BRCA Status

Among the 1020 patients who met eligibility criteria and had nonmissing baseline data, 639 (63%) patients had a documented BRCA test result, and 381 (37.4%) did not. Of the 639 patients with a documented BRCA test result, 92 (14.4%) were BRCAmut and 547 (85.6%) were BRCAwt (Table 2).

BRCAmut patients received second-line PARPi maintenance treatment at a significantly higher rate (49/92, 53.3%) than BRCAwt patients (167/547, 30.5%; P<.0001; Figure 4).

OS Analysis

Overall median follow-up time from initiation of second-line maintenance to end of observation was 24.9 months (range, 0–57.9 months). Comparison of survival was performed among patients with valid death/vital information between those who received a PARPi (n=244) and patients with no second-line maintenance (n=414). Their baseline characteristics are compared in Table 3. Survival from the beginning of the second-line maintenance period to 24 months was significantly higher for PARPi maintenance versus active surveillance: 61.2% (95% CI, 52.4%–68.8%) versus 53.0% (95% CI, 47.1%–58.7%; log-rank P=.0045; Figure 5).

Discussion

This study provides insight into the real-world utilization of second-line maintenance therapy in advanced OC, and the outcomes associated with treatment choices.

Garofalo et al. performed a real-world, retrospective study with EHR data that included 3629 patients with OC with at least 2 doctor office visits in between July 2016 and April 2018.⁸ A total of 398 (11%) patients had a complete/partial response to second-line platinum-based chemotherapy between January 2017 and July 2018. Of these 398 patients, only 49% ended up receiving any maintenance therapy; 39 were BRCAmut, and 359 were BRCAwt. Use of second-line maintenance therapy was higher in patients with BRCAmut disease (56%) when compared with those with BRCAwt disease (49%). Among patients who received second-line maintenance therapy, 47% received PARPi, 27% bevacizumab, and 26% nonplatinum chemotherapy.

A similar real-world, retrospective utilization study of electronic charts was published by Moss et al that examined patients diagnosed with OC between January 2011 and July 2019. ⁹ A total of 222 patients were included who completed second- or third-line platinum-based chemotherapy on or after March 1, 2017, and who had ≥2 months of active surveillance or received maintenance therapy with PARPi or bevacizumab. In this study, 147 (66%) patients received PARPi or bevacizumab second-line or third-line maintenance, while 78 (35%) underwent active surveillance. Forty-six (21%) were BRCAmut, 132 (59%) were BRCAwt and 47 (21%) had an unknown BRCA status. Of BRCAmut patients, 63% received a PARPi, while 40% of patients who were BRCAwt received a PARPi. The study found that PARPi use increased on average by 1.3% every 3 months (P=.02). However, it should be noted that this study included both second- and third-line maintenance therapies, so the estimates differ from the current study and from Garofalo et al.

When looking across the 3 studies in the literature, including the current one, the estimates of the utilization of second-line PARPi maintenance treatment ranged between 23% and 45% of eligible patients (Table 4). Patients on active surveillance regimens ranged between 35% and 51% of eligible patients. The current study found the exact same overall percentage of utilization of any second-line maintenance treatment (49%) as Garofalo et al.8 and had very similar estimates of PARPi utilization (26% in the current study versus 23% from Garofalo et al.). All studies found that there was a higher rate of PARPi use in the BRCAmut population than in the BRCAwt population (Table 4).^8,9

Most importantly, this study found that survival at 24 months was significantly higher in patients treated with second-line PARPi maintenance therapy versus active surveillance. We and other published studies have found that second-line PARPi maintenance therapy is significantly underutilized in this advanced OC population, ^8,9 despite (1) several US Food and Drug Administration–approved options for therapy,^10–12 (2) clinical trial data affirming its safety and efficacy,^3–5 (3) NCCN Treatment Guideline recommendations for its use since 2017,⁶ and (4) a survival benefit to the patients in this analysis.

Limitations

The results of this study should be considered within the context of the strengths and weaknesses of its design. The use of real-world data is advantageous because it reflects community oncology practice trends, as opposed to tightly controlled clinical trials. The study utilized clinical data from the EHR system of the US Oncology Network, which includes 1400 affiliated physicians operating in more than 450 sites of care in the United States. This group of physicians treats approximately 1 million cancer patients annually.⁷ Leveraging these data is expected to provide key insights to examine patient profiles, treatment, and genetic patterns and clinical outcomes among patients with OC treated in the community oncology setting.

Data in the iKnowMed database are not collected for research purposes; however, these data can be applied to gather insights about routine clinical practice in a community medical oncology setting. This may impede the standardization of the data collection methods and instruments and the reporting practices of the physician. As with all administrative databases, iKnowMed data are subject to coding errors of omission and commission. Problems with inadequate or inaccurate codes in the databases may introduce some level of misclassification bias of certain diagnoses, events, or procedures of interest in the study. Likewise, some variables of interest may not be as complete across the entire population. The iKnowMed EHR contains information on patients only when they are seen by US Oncology Network physicians. Services and procedures provided outside of the US Oncology Network are not captured by the database, as well as drugs received by patients from pharmacies not affiliated with US Oncology Network practices.

A patient’s treatment history prior to his/her first encounter at a US Oncology Network practice may be only available in physician progress notes and is not well captured in the iKnowMed EHR. We cannot rule out the possibility that some patients coded as being treatment naïve for advanced disease in the iKnowMed EHR actually had previous chemotherapy for advanced disease in health care facilities outside the US Oncology Network.

The iKnowMed structured data do not differentiate between deleterious alterations in BRCA (pathogenic alterations/mutations) versus those that are variants of unknown significance. This is possible through a chart review, but we did not undertake a full chart review for this study.

Additionally, the iKnowMed system is used for clinical practice reasons, not solely for research purposes; as such, associations, but not causality, can be detected, and thus, bias may be introduced by confounding factors. For example, data may be collected with an intent-to-treat approach, meaning based on when treatment is assigned rather than received. In particular, it will not be possible to determine if oral therapies were dispensed or taken; therefore, we will use an intent-to-treat approach for analysis, with the assumption that all prescribed oral therapies were taken. Likewise, patients who do and do not receive the treatment at one point in time may be fundamentally different than those who received treatment during the observation period. These confounding factors are most likely to affect the outcomes being considered for this study. Because of the nature of the study design, there is potential for bias to be introduced into the calculations of clinical outcomes. Specifically, patients who initiated treatment during the patient identification period may be meaningfully different from other patients who initiated therapy prior to or after the study identification period.

Lastly, not all community oncology practices are included in the iKnowMed data set. Furthermore, not all the US Oncology Network practices utilize the full capabilities of the iKnowMed EHR. The US Oncology Network encourages use of evidence-based treatment guidelines. Therefore, practices that participate in the US Oncology Network may be different from other community oncology practices in the patient population that is seen or the prescribing practices of the physicians.

Conclusion

Our data suggest a significant proportion of eligible patients are not receiving second-line PARPi maintenance therapy despite treatment availability, guideline recommendations, and apparent survival benefits associated with its use.