Treatment Duration, Adherence, Interruptions, and Discontinuations Associated With Maintenance Regimens for Recurrent Ovarian Cancer

J Clin Pathways. 2023;9(3):9-15. doi:10.25270/jcp.2023.05.01

Ovarian cancer was the 11th most common cancer among women in the US in 2021.¹ Although the age-adjusted incidence of ovarian cancer is lower than that for other female cancers, such as breast cancer, ovarian cancer’s mortality rate is much higher. The 5-year survival rate of ovarian cancer in the US is only 49% vs 90% for breast cancer.² Most women are diagnosed with advanced-stage disease, and no screening tools currently exist for early detection of this cancer.¹ Given the high fatality rate, ensuring access to life-prolonging therapies for the ovarian cancer population is essential.

Although most patients will receive platinum-based che­motherapy for advanced ovarian cancer and have a clinical re­sponse to treatment, the majority will ultimately experience disease recurrence, requiring additional systemic treatment.³ Remission duration between treatment and relapse is generally shorter with each subsequent treatment.⁴ Based on this natural history, maintenance therapy has been used with the goal to extend time to progression and time to subsequent therapies.^5-8

Clinical trials have shown that maintenance therapy after response to platinum-based chemotherapy prolongs progres­sion-free survival (PFS) in patients with recurrent ovarian can­cer. In several phase 3 trials, poly (ADP-ribose) polymerase (PARP) inhibitors (rucaparib, olaparib, and niraparib) have been shown to extend the time between second-line chemo­therapy and relapse (ie, PFS) in patients with platinum-sensi­tive ovarian cancer.^5-7 Similarly, bevacizumab has also shown improved PFS in this setting.^8,9

Despite these data and the demonstrated benefit in improv­ing PFS, real-world use data for these relatively new mainte­nance agents are lacking. This analysis examines current re­al-world practice patterns for patients receiving maintenance therapy for recurrent ovarian cancer and assesses duration of therapy, treatment adherence, dose reductions, and treatment interruptions among patients receiving maintenance therapy after second-line chemotherapy.

Some of these results were previously published as an abstract.^₁₀

Time Periods

This retrospective, observational, real-world cohort study ana­lyzed data on adult patients in the US with recurrent ovarian cancer from a large administrative claims database. The study periods and timelines are summarized in Figure E1 in the on­line supplement. Patients were assessed for inclusion in the study during the identification period between July 1, 2010, and December 1, 2019. The index date for follow-up was defined as first date for a claim including a second-line mainte­nance treatment for recurrent ovarian cancer. The baseline period was defined as 6 months before the index date. The follow-up period was until death, end of study (December 1, 2019), or health plan disenrollment.

The cancer date was defined as the first date of a nondiagnos­tic claim with a diagnosis for ovarian cancer in any position, be­tween July 1, 2010, and December 1, 2019. The medication index date was the first systemic chemotherapy date on or after the can­cer date, as late as December 1, 2019. The baseline start date (fixed) was defined as 180 days prior to the cancer date.

The 6 months prior to the cancer date was used to assess preindex (baseline) characteristics. A variable follow-up pe­riod consisting of a minimum of 30 days on and following the medication index date and ending with the earliest of disen­rollment, death, or study end was used to assess lines of thera­py treatment patterns.

Patients eligible for second-line maintenance therapy were identified, and the maintenance index date was defined as the first date of maintenance therapy, with January 1, 2017, to December 1, 2019, as the maintenance identification period. The maintenance identification period start date was defined as January 2017, as the US Food and Drug Administration (FDA) approved PARP inhibitors for maintenance therapy in 2017.

Study Population

Individuals from both commercial and Medicare Advantage insurance plans were included in this analysis of the Optum Research Database (Supplemental Methods in the online supplement). Identification of patients in the study population occurred with two steps.

The first step identified a population of patients with ovar­ian cancer (Table 1) and quantified their lines of therapy. First-line therapy was defined as the first administration of a systemic anticancer therapy in the database. First-line therapy regimens consisted of all drugs received within the 30-day period on or after the first ovarian cancer diagnosis (index date). The end of first-line therapy was defined as the earli­est date among the following: discontinuation of treatment, 60-day gap in receipt of first-line therapy drug regimen, start of a new therapy, death, disenrollment from the plan or end of study period, or discontinuation of any one agent in the line (used to capture bevacizumab maintenance monotherapy). The second line of therapy started with the date of the subse­quent systematic cancer therapy observed after first-line thera­py. Definitions of the end of second-line therapy were identical to those for the end of first-line therapy.

The second step used information from the lines of therapy to identify patients who were eligible for maintenance thera­py, who received second-line maintenance treatment, or who were eligible for maintenance (active surveillance).

Study Variables and Outcomes

Baseline characteristics

Baseline characteristics were evaluated on or before the diag­nosis index date. Demographic baseline variables were age, in­surance type, US region, and maintenance therapy index year. Patient clinical baseline characteristics were represented by the Charlson Comorbidity Index score, a severity measurement that incorporates both physician and inpatient claims data and is validated in ovarian cancer.¹¹ Scores range from 0 to 29, with higher scores indicating a greater number of comorbid condi­tions and reduced survival.

Utilization and time on maintenance treatment variables

We calculated the prevalence of second-line maintenance therapy use and active surveillance. Use of second-line main­tenance treatment, treatment duration, adherence with thera­py, dose reductions, dose interruptions, and drug discontinu­ations were assessed during the follow-up period.

Maintenance therapy prevalence was assessed using the National Drug Codes in each member’s pharmacy prescrip­tion claims data in the Optum Research Database. Total dura­tion of time on maintenance treatment was assessed based on dates and refills of the maintenance therapy. Time on treat­ment was calculated individually for each patient, with start dates from 2017 to 2019 based on the availability date for PARP inhibitors.

Dose adjustments were assessed by increases or reductions in dosages in subsequent refills of the same maintenance therapy. Interruptions were defined as a gap in drug supply (fill) for at least 30 days, followed by a restart. Treatment discontinuations were assessed by flagging the last date of the last prescription filled by the patient for a PARP inhibitor, or the last infusion of bevacizumab during the maintenance therapy period.

Adherence to therapy for PARP inhibitors and bevaci­zumab was calculated using the proportion of days covered method, which calculates the percentage of covered days by dividing the number of days of medication supply dispensed in a period (numerator) by the number of days in the period (denominator) and multiplying by 100.^12,13

Time to treatment discontinuation was calculated as the length of time between second-line maintenance index thera­py start date and the maintenance index treatment discontinu­ation date. Time to treatment interruption was calculated by subtracting the second-line maintenance index therapy start date from the last day of therapy before a ≥30-day gap in sec­ond-line maintenance treatment.

Data Analysis

A description of statistical methods used for data analysis is presented in the Supplemental Methods in the online supplement.

Data may be obtained from a third party and are not pub­licly available.

Baseline Characteristics

A total of 1,092 patients were identified as meeting the study inclusion criteria (Table 1). Baseline demographics are pre­sented in Table 2. Among all patients included in the study population, 737 (68%) were over 65 years old and 774 (71%) patients participated in a commercial insurance plan. The South (48%) and the Midwest (27%) were the two largest US regions represented. Overall, 973 (89%) patients had a Charl­son Comorbidity Index score of ≥ 3, and 908 (83%) had a score of ≥ 5, suggesting this clinical population was complex with a variety of underlying conditions.

The analyses in this paper focus on the groups of patients that received second-line maintenance therapies (PARP in­hibitor or bevacizumab), but these groups were similar to the active surveillance group with respect to demographics and clinical characteristics (Table 2).

Utilization

Among the 1,092 patients in the study eligible for second-line maintenance, 482 (44.1%) received a second-line main­tenance therapy and 610 (55.9%) underwent active surveil­lance with no receipt of maintenance treatment. Of the patients who received a second-line maintenance therapy, 262 (54.3%) received a PARP inhibitor (38 [14.5%] ruca­parib, 114 [43.5%] olaparib, and 110 [42.0%] niraparib), 184 (38.2%) received bevacizumab, and 36 (7.5%) patients re­ceived “other” heterogeneous maintenance regimens (Figure E2 in the online supplement).

Time on Treatment

Average durations of maintenance treatment in days were as follows: olaparib, 187.6 (SD, 178.7); bevacizumab, 185.2 (SD, 149.2); rucaparib, 147.2 (SD, 155.4); and niraparib, 124.2 (SD, 122.9). Treatment adherence was similar across all thera­pies (86%-88%) regardless of once- (niraparib) or twice-daily (olaparib and rucaparib) dosing (Table 3).

Dose adjustments and dose reductions were numerically more common during this period in the niraparib group (29.1% and 28.1%, respectively) than in the rucaparib (21.1% for both) or olaparib (23.7% and 20.2%, respectively) groups, although the difference in frequency did not reach statistical significance (P = .39 and P = .16, respectively).

Dose interruptions and discontinuations were more common in the niraparib group (14.6% and 21.8%, respective­ly) than in the rucaparib (7.9% for both) or olaparib (12.3% and 14.0%) groups. Mean time (days) to treatment discontinuation was shortest in the niraparib (65.4 [±59.9]) group vs the ruca­parib (74.3 [±45.0]) or olaparib (102.9 [±94.9]) groups. Mean time (days) to dose interruption was longest in the rucaparib group (44.7 [±10.4]) vs the olaparib (37.5 [±8.5]) or niraparib (39.4 [±7.3]) groups.

Multivariable Analysis

In a Cox proportional hazards model adjusting for base­line covariates, niraparib was associated with a significantly shorter time to dose interruption or discontinuation com­pared with rucaparib, which was the reference group in the model (hazard ratio [HR]: 3.7; 95% confidence interval [CI]: 1.1%-12.6%; P = .03) (Table 4 and Figure 1). Olaparib and bevacizumab also had higher HRs in this model for dose interruptions or discontinuations compared with ru­caparib, but these estimates were not statistically significant (HR: 1.5; 95% CI: 0.4%-5.4%; P = .50; and HR: 1.4; 95% CI: 0.4%-4.7%; P = .58, respectively) (Table 4 and Figure 1).

Summary of Main Results

This study examined real-world practice patterns for main­tenance treatment based on claims data in patients with re­current ovarian cancer after second-line chemotherapy. In this patient population, less than half received a second-line maintenance therapy, while the remainder underwent active surveillance. Of those who received second-line main­tenance, the most common drug class received was a PARP inhibitor (50%), followed by bevacizumab (40%). Treatment adherence was similar across all second-line maintenance treatments (86%-88%) regardless of once- or twice-daily dosing. Less than one-third of patients in any treatment group required a dose reduction, and treatment discontinua­tions were less than 25% in all groups. Treatment continuity for niraparib was poorer when compared with olaparib and rucaparib, with shorter time on treatment and numerically higher rates of all three dose-adjustment metrics as well as treatment discontinuations.

Results in the Context of Published Literature

The treatment landscape for patients with ovarian can­cer, including maintenance therapy, has been rapidly evolv­ing with the FDA approval of multiple agents for mainte­nance treatment in frontline and recurrent disease. As patients are typically diagnosed with advanced-stage disease, ovarian cancer tends to become a chronic disease process with a high response rate to systemic chemotherapy but also high rates of disease recurrence.14 The use of maintenance therapy is controversial, with some questioning the benefit of prolonging PFS without overall survival benefits and others pointing to data that patients value prolonging the time to recurrence. Our real-world data reflect this debate, with only half of eligible patients receiving maintenance therapy and the other half undergoing active surveillance. Three PARP inhibitors have been shown to prolong PFS after response to platinum in those with recurrent ovarian cancer. These seminal trials, NOVA, ARIEL3, and SOLO2, were published in 2016 and 2017, with the FDA indications for these drugs following in 2017 and 2018.5-7 Bevacizu-mab has been shown to prolong PFS among patients with platinum-sensitive recurrent ovarian cancer and, based on the results of three trials, was given an FDA indication for this use in 2016.8,15,16 Despite bevacizumab having an earlier indication for this population, we found more patients treated with PARP inhibitors, suggesting that a shift away from bevacizumab in this setting and toward PARP inhibitors may be occurring in this real-world population.

Maintenance treatments aim to extend PFS intervals by maximizing patient well-being and minimizing patient ex­posure to cytotoxic chemotherapy. Although these data do not directly address toxicity, treatment adherence was high in all maintenance therapies, ranging from 86% to 88% for the 4 drugs in patients on treatment for 4-6 months. Dose adjustments and reductions occurred in approximately 20%- 30% of patients on PARP inhibitor maintenance, consistent with published data.17 Perhaps unexpectedly, treatment discon­tinuation was also similar to that for clinical trial populations for both rucaparib (8% vs 13% in ARIEL3) and olaparib (12% vs 11% in SOLO2). For niraparib, we found a 22% discontinu­ation rate, which was higher than the 14.7% seen in the NOVA trial. In general, as clinical trial populations discontinue at low­er rates compared with real-world use, the similarity of these real-world rates speaks to these agents’ tolerability.

Strengths and Weaknesses

Although second-line maintenance treatment has been approved for recurrent ovarian cancer since 2016, this study is one of the first to compare real-world use of PARP inhibitors and bevacizumab as maintenance therapy. Comparing PARP inhibitors to each other in real-world databases has been difficult because of low patient numbers. Real-world data analyses such as this one may help provide insight into the use of maintenance therapy for patients with recurrent ovarian cancer.

This study’s results should be interpreted within the limita­tions inherent to claims-based studies. An algorithm was de­veloped based on International Classification of Diseases-9 and -10 diagnostic and procedure codes and pharmacy codes on claims to identify patients eligible for inclusion in the anal­ysis. Precise disease grading and the TNM Classification of Malignant Tumors staging were not available; therefore, dis­ease diagnosis date and treatment date and type were used. Individuals with conflicting data were excluded, and missing data were not imputed.

Additionally, this population does not include patients on Medicaid, in the Veterans Administration system, or who are uninsured, nor does it include any detailed demographic in­formation, including race and ethnicity. Thus, this analysis may not be generalizable to other populations.

Lastly, data from this study come from a managed care population, and results may not be universal to all patients with ovarian cancer. The database is large and comprehen­sive of the commercial managed care and Medicare popu­lations in the US. In 2017, approximately 8% of the US commercially enrolled population, plus 21% of the Medicare Part C (Medicare Advantage) population, were represented in the Optum Research Database (Optum data on file).

Implications for Practice and Future Research

As more patients each year are treated with PARP inhibitor maintenance, these estimates will be refreshed with larger pa­tient groups. Additionally, comparing data from patients re­ceiving recurrent, second-line maintenance to data from those using PARP inhibitor maintenance in frontline maintenance, as these drugs are now available in the frontline setting as well, will bring additional insight. Further research should also ex­amine PARP inhibitor therapy treatment patterns in first-line maintenance settings.

In this cohort of patients with recurrent ovarian cancer, less than half subsequently received a maintenance therapy with a PARP inhibitor or bevacizumab after a second-line system­ic chemotherapy. For those patients receiving a second-line maintenance therapy, while adherence was high in all PARP inhibitor regimens regardless of once- or twice-daily dosing, levels of dose reductions or discontinuation varied depending on the treatment received. Niraparib was associated with the shortest time to dose interruption or discontinuation among the agents evaluated.

This article has supplementary material, which can be accessed here.

Authors: Emma L. Barber, MD^1-3; Allison M. Saiz, MD¹; Nicole M. Engel-Nitz, PhD⁴; Scott H. Bunner, MPH⁴; Katrine L. Wallace, PhD^5,6*

Affiliations: ¹Northwestern University, Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, Chicago, IL; ²Robert H. Lurie Comprehensive Cancer Center, Northwestern University, Chicago, IL; ³Surgical Outcomes and Quality Improvement Center, Institute for Public Health in Medicine, Chicago, IL; ⁴Health Economics and Outcomes Research, Optum, Eden Prairie, MN; ⁵Health Economics and Outcomes Research, Clovis Oncology, Inc., Boulder, CO; ⁶University of Illinois at Chicago, School of Public Health, Chicago, IL

*K.L.W. was employed at Clovis Oncology at the time the study was completed.

Acknowledgements: Editorial support was funded by Clovis Oncology, Inc, and provided by Kathleen Blake, PhD, of Ashfield Medical Communications, an Inizio company.

Contributions: E.L.B.: conception, critical analysis, drafting/final editing, guarantor. A.M.S.: critical analysis, final editing, guarantor. N.M.E.-N.: conception, data acquisition, data management, statistical analysis, critical analysis, guarantor. S.H.B.: concep­tion, data acquisition, data management, statistical analysis, critical analysis, guarantor. K.L.W.: conception, study design, data management, critical analysis, drafting/final editing, guar­antor. All authors contributed significantly to this work.

Address correspondence to: 

Katrine L. Wallace, PhD
University of Illinois Chicago
School of Public Health
Division of Epidemiology and Biostatistics
1601-3 W Taylor St, MC 923
Chicago, IL 60612
Email: kwalla2@uic.edu
ORCID: 0000-0002-5003-0617

Disclosures: E.L.B. has received grant funding to her institution from Eli Lilly, GOG Foundation, and NIA. A.M.S. disclosed no financial or other conflicts of interest. N.M.E.-N. is an employee of Optum and owns stock in UnitedHealth Group. S.H.B. owns stock in UnitedHealth Group. K.L.W. was an employee of Clovis Oncology, Inc, at the time the study was done, and may own stock or have owned stock options in that company.