Introduction
Prospective randomized trials of poly(ADP-ribose) polymerase (PARP) inhibitors as maintenance therapy for advanced ovarian cancer, particularly high-grade serous carcinoma, have yielded inconsistent results regarding overall survival; although some studies reported a survival benefit, others did not report such benefits (1-3). Several studies have suggested that overall survival outcomes may differ depending on the homologous recombination deficiency status (2,3). In addition, survival advantages have not been consistent for recurrent ovarian cancer.
In Japan, following the Study 19 trial, PARP inhibitors were approved as maintenance therapy for patients with platinum-sensitive recurrent ovarian cancer, irrespective of the homologous recombination deficiency status (4-13). In the recurrent setting, olaparib has also been evaluated as an oral targeted treatment option, rather than cyclic intravenous cytotoxic chemotherapy, in comparison with conventional non-platinum chemotherapy for patients with platinum-sensitive relapsed ovarian cancer and a germline BRCA1/2 mutation (5). In routine practice, PARP inhibitor maintenance therapy may impose a different type of treatment burden from repeated intravenous cytotoxic chemotherapy and may reduce patients' cumulative exposure to conventional cytotoxic agents. Moreover, the SOLO2 trial demonstrated improved progression-free survival and preserved health-related quality of life among patients treated with olaparib maintenance therapy (6).
The total burden of cytotoxic chemotherapy is clinically relevant because patients with advanced ovarian cancer often receive multiple lines of treatment over the course of their disease. A reduction in the cumulative number of cytotoxic chemotherapy cycles may represent a measurable change in treatment burden, although it does not directly capture quality of life, toxicity, or patient-reported outcomes. Therefore, real-world analyses describing the number of chemotherapy regimens and cumulative cytotoxic chemotherapy exposure may provide complementary information to survival-focused clinical trials.
Several randomized trials allowed subsequent PARP inhibitor administration after progression in patients who were initially assigned to placebo or comparator arms, which could complicate the interpretation of overall survival differences (14,15). A historical cohort may describe treatment patterns before and after the clinical introduction of PARP inhibitors, but such a design is inherently non-randomized and is susceptible to treatment-selection bias, era effects, and baseline imbalances. In the present study, we aimed to describe real-world treatment patterns, the number of chemotherapy regimens received, cumulative exposure to cytotoxic chemotherapy, and exploratory survival outcomes in patients with stage III-IV high-grade serous ovarian, fallopian tube, or primary peritoneal carcinoma treated before and after the introduction of PARP inhibitors at a single institution. This study was not designed to establish a causal survival benefit of PARP inhibitors.
Discussion
In this descriptive single-center historical cohort restricted to patients with stage III-IV high-grade serous ovarian, fallopian tube, or primary peritoneal carcinoma, the clinical introduction of PARP inhibitors was associated with lower cumulative exposure to cytotoxic chemotherapy. The cumulative number of cytotoxic chemotherapy cycles was significantly lower in the PARP inhibitor-treated group than in the PARP inhibitor-untreated group, whereas the number of chemotherapy regimens did not differ significantly between the groups. Overall survival appeared longer in the PARP inhibitor-treated group; however, this finding should be interpreted as exploratory because of the retrospective historical design, treatment-selection bias, era effects, and baseline imbalances between the groups.
This endpoint was selected as an objective measure of cytotoxic treatment exposure that could be consistently retrieved from medical records. Patients with advanced high-grade serous ovarian carcinoma often receive multiple lines of systemic treatment over the course of their disease, and repeated exposure to cytotoxic agents may contribute to treatment burden. PARP inhibitor maintenance therapy has been shown to prolong progression-free survival in pivotal trials, and health-related quality of life analyses have suggested that maintenance therapy can delay disease progression without a major deterioration in patient-reported outcomes (1-3,6). In this context, the lower cumulative number of cytotoxic chemotherapy cycles observed in the PARP inhibitor-treated group may represent a measurable change in real-world treatment patterns after the introduction of PARP inhibitor maintenance therapy. However, this finding should not be interpreted as direct evidence of improved quality of life or reduced toxicity, because patient-reported outcomes and treatment-related adverse events were not systematically evaluated in this retrospective analysis. Instead, cumulative cytotoxic chemotherapy cycles should be regarded as a surrogate measure of treatment exposure rather than a validated measure of patient burden. The number of chemotherapy regimens did not differ significantly between the two groups, suggesting that PARP inhibitor introduction may alter cumulative cytotoxic exposure without necessarily reducing the number of treatment lines.
Pivotal trials have demonstrated progression-free survival benefits of PARP inhibitor maintenance therapy, whereas overall survival benefits have been inconsistent and appear to vary according to BRCA mutation status, homologous recombination deficiency status, treatment line, and subsequent therapies (1-3). In the recurrent setting, the ENGOT-OV16/NOVA, ARIEL3, ARIEL4, SOLO3, QUADRA, and OReO trials highlighted the complexity of evaluating survival outcomes and subsequent treatment strategies after PARP inhibitor exposure (7-10,14,15). Unlike randomized trials, however, the present historical cohort study could not control for treatment indication, supportive care, surgical strategy, or era-related factors.
Although the analysis was restricted to high-grade serous carcinoma to reduce histologic heterogeneity, important baseline imbalances remained. The FIGO stage, use of neoadjuvant chemotherapy, and cytoreduction status differed between the groups. Suboptimal cytoreduction was more frequent in the PARP inhibitor-untreated group, whereas neoadjuvant chemotherapy was more frequent in the PARP inhibitor-treated group. These differences likely reflect changes in treatment strategy over time and differences in patient selection. Therefore, the observed survival association in the stratified multivariable Cox model should not be interpreted as causal evidence that PARP inhibitors improve overall survival in this cohort. It should be regarded as hypothesis-generating rather than confirmatory. Although multivariable Cox regression analysis was performed using clinically relevant prognostic factors, statistical adjustment could not fully account for residual confounding, treatment-selection bias, or unmeasured factors in this historical cohort.
The association between neoadjuvant chemotherapy and poorer overall survival should be interpreted as confounding by indication, because neoadjuvant chemotherapy is generally selected for patients with greater tumor burden, more extensive disease, poorer resectability, or reduced suitability for primary debulking surgery. Similarly, bevacizumab use was more frequent in the PARP inhibitor-treated group, likely influenced by disease extent, reimbursement status, and physician judgment.
In the exploratory analysis of patients who experienced recurrence or progression after PARP inhibitor therapy, platinum sensitivity after progression was associated with longer post-progression overall survival. However, because this subgroup included only 26 patients, the analysis was underpowered and vulnerable to statistical instability. Therefore, this finding should be interpreted as descriptive and hypothesis-generating. Previous studies have reported that prior PARP inhibitor exposure may influence response to subsequent chemotherapy, supporting the clinical importance of treatment sequencing after PARP inhibitor therapy (11,12). As new agents such as antibody-drug conjugates are being incorporated into the treatment landscape for platinum-resistant ovarian cancer, understanding post-PARP treatment sensitivity, sequencing, and biomarker-guided treatment selection remains clinically important (16).
This study has some limitations. First, this was a retrospective single-center historical cohort study with substantial treatment-selection bias and era effects; therefore, causal inference regarding the effect of PARP inhibitors on overall survival could not be made. Second, although the cohort was restricted to high-grade serous carcinoma, baseline imbalances remained, particularly in FIGO stage, neoadjuvant chemotherapy, and cytoreduction status. In addition, immortal time bias may have affected the survival analysis, as patients in the PARP inhibitor-treated group had to survive and achieve disease control after platinum-based chemotherapy before receiving PARP inhibitor maintenance therapy. Although we added the response to first-line platinum-based treatment and the response immediately before PARP inhibitor initiation, this bias could not be eliminated in the retrospective design. Third, comprehensive BRCA mutation status and homologous recombination deficiency data were unavailable for the full cohort. Many patients had been treated before routine BRCA/HRD testing became available, and some had died by the time of this retrospective analysis; therefore, comprehensive post hoc biomarker testing was not feasible. Because BRCA and HRD status are closely associated with PARP inhibitor benefit, residual confounding by unmeasured biological factors cannot be excluded (1-3). Fourth, approximately half of the patients in the PARP inhibitor-treated group received PARP inhibitors as later-line maintenance therapy, which complicated the interpretation of survival outcomes and treatment exposure. Fifth, quality of life, adverse events, dose reductions, treatment interruptions, and hospitalization were not systematically evaluated in this retrospective analysis. These variables were not evaluated because patient-reported outcomes were not routinely collected, and treatment-related adverse events were not consistently recorded using standardized grading criteria throughout the long study period. In addition, the reasons for dose reductions and treatment interruptions were heterogeneous and could not be reliably classified retrospectively. Hospitalizations were also difficult to interpret as a surrogate for treatment burden or toxicity, because planned admissions for chemotherapy administration, admissions based on patient preference or social factors, disease-related admissions, and adverse-event-related admissions could not be consistently distinguished from medical records. Including these incompletely and inconsistently documented variables could have introduced substantial missing data and misclassification bias. Therefore, the lower cumulative number of cytotoxic chemotherapy cycles should not be interpreted as direct evidence of reduced toxicity or improved quality of life. Sixth, the subgroup analysis of post-progression overall survival after PARP inhibitor therapy included only 26 patients. To reduce the risk of overfitting, the main Cox regression analysis in this subgroup was limited to univariable analysis, and the multivariable model was presented only as an exploratory supplementary analysis. Therefore, the association between platinum sensitivity after progression and post-progression overall survival should be interpreted as descriptive and hypothesis-generating. Finally, differences in follow-up duration may have influenced cumulative cytotoxic chemotherapy exposure and exploratory survival outcomes, and may partly reflect differences in survival because follow-up duration was defined as the time from first-line treatment initiation to death or last follow-up.
In conclusion, this descriptive single-center cohort study restricted to advanced high-grade serous carcinoma suggests that the clinical introduction of PARP inhibitors was associated with lower cumulative exposure to cytotoxic chemotherapy. However, the number of chemotherapy regimens did not differ significantly between the groups, and no causal inference regarding a survival benefit could be made. Platinum sensitivity after recurrence or progression following PARP inhibitor therapy remained associated with post-progression survival in the exploratory analysis. Larger multicenter studies with biomarker data, toxicity data, and patient-reported outcomes are needed to clarify how PARP inhibitors influence long-term treatment burden and subsequent treatment sequencing in real-world practice.