ABSTRACT
Objective
To develop and validate a prognostic nomogram that integrates clinical variables and biomarker statuses for predicting progression‐free survival (PFS) and overall survival (OS) in patients with advanced epithelial ovarian cancer (OC) receiving first‐line poly (ADP‐ribose) polymerase inhibitor (PARPi) maintenance therapy.
Methods
Clinical data from 145 OC patients who received first‐line PARPi maintenance therapy from August 2018 to May 2024 were retrospectively analyzed. Univariate and multivariate regression analyses were performed to identify the predictive factors. A nomogram was constructed using a multivariate Cox regression model.
Results
Patients with OC who received first‐line PARPi maintenance treatment showed improved PFS (median PFS: 48.53 months) and OS. Cox regression analysis identified several independent prognostic factors for prolonged PFS, including BRCA mutations, R0 resection (no residual disease), FIGO stage III (vs. IV), and a higher CA‐125 elimination rate constant (KELIM) score (KELIM > 1). Additionally, BRCA mutations and younger age were significant predictors of better OS. Analysis of subsequent treatment in patients who experienced recurrence revealed that platinum‐based second‐line chemotherapy combined with bevacizumab improved outcomes, whereas a longer duration of PARPi therapy (> 12 months) appeared to be associated with poorer prognosis.
Conclusion
This study establishes and validates the nomogram that integrates clinical factors (FIGO Stage, Residual Disease, and Age) with biomarkers (BRCA and KELIM) to predict outcomes in patients with advanced OC receiving First‐Line PARPi maintenance therapy. Furthermore, our finding suggest that an extended duration of PARPi treatment may be a risk factor for subsequent treatment.
Keywords: nomogram, ovarian cancer, PARP inhibitor, prognostic factors
1. Introduction
Ovarian cancer (OC) is the most lethal gynecologic malignancy, accounting for over 207,000 annual deaths worldwide and ranking eighth in cancer‐related mortality among women [1]. Despite advancements in cytoreductive surgery and platinum‐based chemotherapy, the 5‐year survival rate for advanced‐stage disease has stagnated at 30%–40% [2], underscoring the critical need for novel therapeutic strategies.
The evolution of maintenance therapy has reshaped OC management [3]. While anti‐angiogenic agents like bevacizumab in the GOG‐0218 trial have demonstrated modest progression‐free survival (PFS) benefits (HR = 0.717, 95% CI: 0.625–0.824) without overall survival (OS) improvement (HR 0.96；95% CI: 0.85 ~ 1.09) [4, 5], the emergence of poly (ADP‐ribose) polymerase inhibitors (PARPis) represents a paradigm shift. Through synthetic lethality targeting homologous recombination deficiency (HRD), PARPis have shown differential efficacy across molecular subgroups [6]. Landmark trials including SOLO‐1 and PRIMA established the role of PARPis in first‐line maintenance, particularly in patients with BRCA mutations (SOLO‐1: median PFS 56.0 vs. 13.8 months, HR = 0.33) and HRD‐positive populations (PRIMA: median PFS 13.8 vs. 8.2 months, HR = 0.66) [7, 8, 9]. Long‐term follow‐up results of the SOLO‐1 study confirmed that, compared to placebo, olaparib maintenance therapy significantly prolonged the median OS (not reached vs. 75.2 months) (HR = 0.55, 95% CI 0.40–0.76) and increased the 7‐year survival rate by 20.5% [10]. For patients with HRD‐positive newly diagnosed advanced OC, the long‐term follow‐up results of the PAOLA‐1 study confirmed that compared with bevacizumab monotherapy, olaparib combined with bevacizumab maintenance therapy could significantly prolong the median OS of patients (75.2 months vs. 57.3 months) (HR = 0.62, 95% CI 0.45–0.85) [11]. These data suggest that patients with BRCA mutations or HRD‐positive status derive significant long‐term survival benefits from first‐line maintenance therapy with olaparib ± bevacizumab. Notably, the therapeutic landscape of OC has rapidly expanded beyond PARPi. Recent identification of diverse genetic alterations has led to the emergence of several novel molecular inhibitors, such as immune checkpoint inhibitors and BRAF or MEK inhibitors, which exhibit promising therapeutic potential [12, 13, 14].
However, critical knowledge gaps remain. Only 20% of ovarian cancer patients harbor BRCA1/2 mutations, and less than half are HRD‐positive [15, 16]. Moreover, among those with BRCA mutations, approximately 20%–30% experience disease progression within 24 months of initiating first‐line PARPi maintenance therapy [17, 18]. Approximately 50% of patients with HRD develop PARPi resistance within 24 months, whereas HRD‐negative subgroups derive even more limited benefits [19]. This highlights that the current biomarker stratification relying solely on BRCA/HRD status has suboptimal predictive accuracy. Beyond BRCA/HRD, other biomarkers not directly related to the HRR pathway but are expected to predict the efficacy of PARPis, such as ADP‐ribosylation levels [20], methylation status of the HOXA9 promoter [21], HELQ [22], and Schlafen 11 [23], potentially overlook clinically relevant variables. Emerging evidence suggests that treatment timing and surgical outcomes significantly modulate PARPi efficacy; for instance, patients with no residual disease show a better prognosis [24]. Although some studies have incorporated clinical indicators, these markers predict the prognosis of PARPi in isolation. Given the complexity of OC, it is necessary to include as many clinical characteristics as possible and establish a prognostic model for PARPis.
We hypothesized that integrating multidimensional clinicopathological variables with molecular markers would enhance prognostic precision. This study aimed to develop a novel nomogram incorporating biomarker status, tumor stage, surgical outcomes, and the CA‐125 elimination rate constant (KELIM) score to address three critical challenges: (1) identification of PARPi non‐responders despite favorable molecular profiles, (2) stratification of HRD‐negative patients who may benefit from PARPi, and (3) optimization of maintenance therapy sequencing. By leveraging real‐world evidence from a retrospective cohort, we aimed to establish a clinically actionable decision‐support tool for first‐line PARPi.
2. Materials and Methods
2.1. Study Design and Participants
This retrospective cohort study included patients in the department of Gynecology at Xiangya Hospital, covering the period from August 2018 to May 2024. This study was approved by the Ethics Committee of Xiangya Hospital, Central South University (No. 202507081), and informed consent was obtained from all participants. The inclusion criteria were as follows: (1) histological confirmation of epithelial OC; (2) FIGO stage III‐IV; (3) age over 18 years; and (4) for patients receiving a PARPi as first‐line treatment, the duration of treatment must exceed 6 months. The exclusion criteria were incomplete clinical information, lack of surgery, second or subsequent lines of PARPi treatment, and loss to follow‐up.
2.2. Data Collection
The following clinical data were collected: Age, histological type, FIGO stage, BRCA status, details regarding neoadjuvant chemotherapy (NACT), residual disease status, primary CA‐125 levels, CA‐125 levels before PARPi treatment, hyperthermic intraperitoneal chemotherapy (HIPEC) data, and first‐line chemotherapy response. PFS was calculated from the time of the last chemotherapy session until disease progression, death from any cause, or the date of the last follow‐up in the comparison between the PARPi cohort and the no‐PARPi cohorts. In the comparison between subgroups in the PARPi cohort, PFS was calculated from the start of PARPi treatment until disease progression, death from any cause, or the date of the last follow‐up. OS was measured from the date of diagnosis to the date of death from any cause or the date of the last follow‐up. Time to second progression (TTSP) was defined as the duration from initiation of subsequent therapy post PARPi maintenance therapy to the second progression or death.
2.3. CA‐125 Elimination Rate Constant (KELIM) Score Calculation
The CA‐125 KELIM score was calculated online (https://www.biomarker‐kinetics.org/CA‐125) using at least three CA‐125 values measured within the first 100 days (or less) after the start of chemotherapy after tumor reductive surgery.
2.4. Statistical Analysis
Statistical analyses were performed using GraphPad Prism version 8.0.1 software (GraphPad Software Inc., La Jolla, CA, USA) and R software (version 4.2.1). For comparisons between two groups, continuous variables were analyzed using Student's t‐test for normally distributed data, or the Mann–Whitney U test for non‐normally distributed data. Categorical variables were compared using the chi‐square test or Fisher's exact test when expected frequencies were < 5. Data conforming to a normal distribution were presented as mean ± standard deviation (x̄ ± s) and median (min, max). Categorical variables were reported as counts and percentages (n, %). Survival curves were generated using the Kaplan–Meier method and compared using the log‐rank test. Univariate Cox proportional hazards regression analyses were first performed to identify prognostic factors. Variables with a p‐value < 0.05 in the univariate analysis were entered into a multivariate Cox regression model to identify independent prognostic factors. The proportional hazards assumption was tested using the survival package in R. Based on the independent factors identified by multivariate analysis, a prognostic nomogram was constructed using the rms package in R. The consistency index (C‐index), calibration curve, and area under the receiver operating characteristic curve (AAUC, ROC) were used to evaluate the clinical predictive efficacy of the nomogram.
3. Results
3.1. Clinicopathologic Characteristics
From August 2018 to May 2024, 347 patients with advanced epithelial OC who were treated at Xiangya Hospital were retrospectively reviewed. Among these, 145 patients received first‐line PARPi treatment, while 202 did not receive maintenance therapy. After applying our exclusion criteria, 145 patients who received first‐line PARPi treatment (PARPi cohort) and 168 who did not receive maintenance therapy (no‐PARPi cohort) were included in the statistical analysis (Figure 1). The PARPi cohort included 76 patients with BRCA1/2 mutations (BRCA‐mut) and 69 patients without BRCA1/2 mutations (BRCA‐wt). The baseline clinical characteristics of the two cohorts are shown in Table 1. Overall, significant differences between the cohorts were observed in the proportions of patients with BRCA mutations, those who underwent HIPEC, those with no residual disease (R0), and primary CA‐125 levels.
FIGURE 1.
Flowchart of patients' selection. PARPi, PARP inhibitors.
TABLE 1.
Characteristics of ovarian cancer patients with or without PARPi for first‐line maintenance treatment.
| no‐PARPi (n = 168) | PARPi (n = 145) | p | |
|---|---|---|---|
| Age | |||
| Mean (SD) | 55.5 (9.68) | 56.0 (8.93) | 0.66 |
| Median [Min, Max] | 56.0 [23.0, 77.0] | 55.4 [33.2, 79.8] | |
| Histological type | |||
| HGSOC | 147 (87.5%) | 137 (94.5%) | 0.05 |
| no‐HGSOC | 21 (12.5%) | 8 (5.5%) | |
| FIGO stage | |||
| III | 123 (73.2%) | 106 (73.1%) | 1 |
| IV | 45 (26.8%) | 39 (26.9%) | |
| BRCA status | |||
| wt | 8 (4.8%) | 69 (47.6%) | 0.2 |
| mut | 3 (1.8%) | 76 (52.4%) | |
| Missing | 157 (93.5%) | 0 (0%) | |
| NACT | |||
| No | 111 (66.1%) | 85 (58.6%) | 0.21 |
| Yes | 57 (33.9%) | 60 (41.4%) | |
| Residual disease | |||
| R0 | 78 (46.4%) | 90 (62.1%) | 0.01 |
| > R0 | 90 (53.6%) | 55 (37.9%) | |
| Primary CA125 | |||
| Mean (SD) | 1370 (1460) | 1230 (1270) | 0.38 |
| Median [Min, Max] | 793 [16.9, 5000] | 772 [10.9, 5000] | |
| CA125 after Chemotherapy | |||
| Mean (SD) | 97.9 (425) | 13.5 (10.7) | 0.01 |
| Median [Min, Max] | 14.7 [4.20, 3980] | 11.2 [4.46, 103] | |
| HIPEC | |||
| No | 107 (63.7%) | 109 (75.2%) | 0.04 |
| Yes | 61 (36.3%) | 36 (24.8%) | |
| Chemotherapy Response | |||
| CR | 0 (0%) | 126 (86.9%) | NA |
| PR | 0 (0%) | 19 (13.1%) | |
| Missing | 168 (100%) | 0 (0%) | |
Note: Student's t‐test, the chi‐square test, or Fisher's exact test was used in this table.
Abbreviations: CR, complete remission; FIGO, International Federation of Gynecology and Obstetrics; HGSOC, high‐grade serous ovarian carcinoma; HIPEC, hyperthermic intraperitoneal chemotherapy; mut, mutation; N, number; NACT, neoadjuvant chemotherapy treatment; PR, partial remission; SD, standard deviation; wt, wild type.
3.2. Survival Outcomes: PARPi Vs. No‐PARPi Maintenance
As anticipated, in the PARPi cohort, the median PFS was significantly longer than that of patients in the no‐PARPi cohort (48.53 months vs. 17.80 months; HR = 0.4473; 95% CI: 0.3167–0.6318; p < 0.001) (Figure 2A). Consistent with the PFS results, patients receiving PARPi treatment exhibited superior OS compared with those in the no‐PARPi cohort (p < 0.0001; HR = 0.2339; 95% CI: 0.1433–0.3819) (Figure 2B).
FIGURE 2.
Kaplan–Meier curves for PFS and OS. (A, B) Kaplan–Meier curves show PFS (A) and OS (B) of patients in the PARPi cohort and the no‐PARPi cohort.
3.3. Prognostic Factors for PFS and OS in the PARPi Cohort
Univariate and multivariate Cox regression analyses were performed to elucidate the clinical factors influencing prognosis. In the univariate Cox regression analysis, BRCA status, residual disease, FIGO stage, and CA125 KELIM score were identified as potential factors affecting PFS (Figure 3A). Upon including these factors in the multivariate Cox regression analysis, all the factors mentioned above were confirmed as independent risk factors for PFS (Figure 3B). We performed a parallel analysis for OS. Based on the univariate Cox regression analysis, patient age and BRCA status were included as potential predictors in the multivariate regression model (Figure 3C), both of which remained independent in the multivariate analysis (Figure 3D).
FIGURE 3.
Univariate and multivariate Cox regression analysis in the PARPi cohort. (A, B) Univariate (A) and multivariate (B) Cox regression analysis of factors associated with PFS. (C, D) Univariate (C) and multivariate (D) Cox regression analysis of factors associated with OS. HGSOC, High‐grade serous ovarian cancer; FIGO, the International Federation of Gynecology and Obstetrics; NACT, neoadjuvant chemotherapy; last CT, last chemotherapy; CR, complete response; PR, partial response.
Survival analysis was performed based on the results of univariate and multivariate Cox regression analyses. Patients with BRCA mutations, R0 resection, FIGO stage III, or higher KELIM score (> 1) had significantly better PFS (Supplementary Figure 1A–D). Patients with BRCA mutations or younger patients (age < 55 years) had a markedly longer OS (Supplementary Figure 1E,F).
3.4. Construction and Evaluation of the Prognostic Nomogram
Based on the independent prognostic factors identified through Cox regression analysis, a nomogram was constructed to predict the probability of 1‐year, 2‐year, and 3‐year PFS (Figure 4A). By calculating the total score corresponding to each variable, the probabilities of 1‐, 2‐, and 3‐year PFS could be estimated. The model demonstrated good discriminatory ability, with a C‐index of 0.793 (95% CI, 0.759–0.827). In addition, the calibration curves demonstrated a high degree of consistency between the predicted and observed survival rates for both models (Figure 4B). Time‐dependent receiver operating characteristic (ROC) curves were generated using these models. In the PFS nomogram model, the area under the curve values at 1, 2, and 3 years were 0.829, 0.818, and 0.752, respectively (Figure 4C). Decision curve analysis (DCA) further confirmed the clinical utility of the nomogram, showing significant net benefits across a wide range of threshold probabilities for 1‐, 2‐, and 3‐year PFS predictions (Figure 4D). These results indicated that the models exhibited high predictive accuracy and clinical applicability.
FIGURE 4.
Nomogram for predicting PFS in ovarian cancer patients with PARPi for first‐line maintenance therapy. (A) Nomogram for predicting PFS. (B) Calibration plots for 1‐(left), 2‐(middle), and 3‐(right) year PFS in PARPi cohort. (C, D) Receiver operating curves (C) and decision curve analysis (D) for 1‐(left), 2‐(middle), and 3‐(right) year PFS in PARPi cohort. AUC, area under the curve. FIGO, the International Federation of Gynecology and Obstetrics.
3.5. Stratified Analysis of Prognostic Factors by BRCA Status in PARPi‐Treated Patients
To investigate the potential predictors of differential PARPi efficacy, we conducted BRCA‐stratified analyses in 145 PARPi‐treated patients. The cohort was divided into BRCA‐mut (n = 76) and BRCA‐wt (n = 69) subgroups. No significant differences in baseline clinical variables were observed between the two BRCA subgroups (Table S1). In the BRCA‐mut subgroup, univariate Cox regression identified FIGO stage IV (p = 0.006), residual disease > R0 (p = 0.033), elevated post‐chemotherapy CA125 (p = 0.037), and KELIM score ≤ 1 (p = 0.028) as adverse factors for PFS (Supplementary Figure 2A). Multivariate analysis confirmed FIGO stage IV (HR = 4.566, 95% CI: 1.196–17.429; p = 0.026) as independent adverse prognostic factors and KELIM > 1 (HR = 0.11, 95% CI: 0.013–0.95; p = 0.045) as an independent favorable prognostic factor (Supplementary Figure 2B). In contrast, no clinical variables were significantly associated with PFS in the BRCA‐wild‐type subgroup (Supplementary Figure 2C).
3.6. Prognostic Outcomes of Recurrent Patients After First‐Line PARPi Treatment
In the PARPi cohort, 42 patients experienced disease progression after the initiation of first‐line maintenance treatment with PARPis. To further analyze the factors influencing the prognosis of patients with recurrence, we included BRCA status, type of second‐line chemotherapy (whether platinum was included), whether bevacizumab was used, and duration of PARPi use in the multivariate Cox regression analysis. The results showed that the use of platinum‐based chemotherapy and bevacizumab were key factors in prolonging the TTSP of patients, while the use of PARPi over 12 months may also have been a hazardous factor (Figure 5A). To further analyze the impact of PARPi on the efficacy of subsequent chemotherapy, we conducted a stratified analysis of the objective remission rate (ORR) and disease control rate (DCR) of patients after chemotherapy. No significant difference was observed between the two groups with PARPi duration < 12 months and > 12 months, regardless of whether platinum‐based chemotherapy was included (Figure 5B,C).
FIGURE 5.
Prognostic outcomes of recurrent patients after first‐line PARPi treatment. (A) Univariate analysis for PFS in recurrent patients with first‐line PARPi treatment. (B, C) The chi‐square test is used to examine the impact of the duration of PARPi on the response to second‐line chemotherapy. (B) objective response rate; (C) disease control rate; (D) Univariate analysis for PFS in recurrent BRCA‐wt patients with first‐line PARPi treatment. BEVA, bevacizumab; 12 m, 12 months; CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease; ORR, objective response rate; DCR, disease control rate; ns, no significance.
Because it has been widely established that the BRCA‐mutated population can derive OS benefits from first‐line PARPi maintenance therapy, we further analyzed subsequent therapy in the BRCA‐wild‐type relapsed population. By analyzing the clinical data of 30 patients with BRCA wild‐type recurrence, multivariate Cox regression analysis showed that platinum‐based chemotherapy remained the key factor for prolonging TTSP (Figure 5D), whereas a PARPi duration longer than 12 months again showed a trend towards a shorter TTSP duration.
4. Discussion
The approval of olaparib for patients with BRCA‐mutated advanced OC has catalyzed the rapid development of PARPis and the expansion of their indications. To date, PARPis have been approved for all patients with advanced epithelial ovarian cancer who achieve a complete response (CR) or partial response (PR) following first‐line chemotherapy. Despite this progress, no other biomarkers beyond BRCA and HRD testing are routinely used to predict prognosis. Therefore, there is an urgent need for large‐scale real‐world studies to supplement the prognostic research on PARPis. In this context, our study retrospectively analyzed data from a large single center in China over a 5‐year period, focusing specifically on the first‐line maintenance treatment with PARPi in patients with advanced OC. Key clinical factors were identified and integrated into a nomogram, providing a novel perspective for clinical decision‐making.
Consistent with prior evidence, PARPi‐treated patients demonstrated significantly prolonged PFS and OS compared with those without maintenance therapy (median PFS: 48.53 vs. 17.80 months). Subgroup analyses further revealed superior outcomes in BRCA‐mutated cohorts vs. BRCA‐wild‐type subgroups (median PFS: 48.00 vs. 24.17 months). Notably, our study showed a substantially improved PFS in the PARPi cohort compared with the PRIME trial involving Chinese populations (median 48.53 vs. 24.8 months) [25]. Even in the no‐PARPi cohort, our institutional outcomes surpassed PRIME results (median PFS: 17.80 vs. 8.3 months), suggesting enhanced multidisciplinary management of advanced ovarian cancer at our center. However, similar to PRIME, median OS remained immature in both cohorts owing to limited follow‐up. Subgroup comparisons highlighted differential PARPi efficacy: BRCA‐wild‐type patients in our cohort showed modest improvement over PRIME (24.17 vs. 19.3 months), underscoring the limited clinical benefit of PARPi in this population and the imperative for refined predictive biomarkers. Regrettably, HRD testing was unavailable for the majority of BRCA‐wild‐type PARPi recipients owing to financial constraints, precluding HRD‐based stratification [26, 27]. This gap emphasizes the urgent need for standardized HRD testing in China [28].
Cox proportional hazards regression analysis identified four independent prognostic factors for PFS in advanced ovarian cancer patients receiving first‐line PARPi maintenance therapy: BRCA mutations, residual disease, FIGO stage, and KELIM score. While prior studies predominantly emphasized BRCA status and post‐chemotherapy response (CR/PR) as predictors, our findings diverged by highlighting FIGO stage and KELIM score as novel clinical determinants. Intriguingly, although HRD status [29], NACT, and age were previously implicated in PFS outcomes [30, 31], NACT and age were not statistically significant in our multivariate model, and HRD status was not included in this study.
Based on the identified independent prognostic factors, we constructed a nomogram for predicting PFS in advanced epithelial ovarian cancer patients receiving first‐line PARPi maintenance therapy. The models demonstrated robust discrimination and calibration. While a multicenter retrospective study of niraparib similarly developed a PFS nomogram incorporating age, BRCA status, residual disease, and treatment interval [30], our model uniquely integrated the FIGO stage and KELIM score, achieving superior predictive accuracy. The key limitations of this study are as follows. Despite the international guidelines endorsing PARPi maintenance for advanced ovarian cancer, financial constraints limited our cohort size, precluding external validation through training/validation set partitioning. Although there was no validation set for verification, we employed various methods to test the performance of our model, including calibration, ROC, and DCA curves, all of which demonstrated excellent performance. These findings underscore the need for multicenter validation (ideally prospective) to confirm generalizability.
Our BRCA‐stratified analysis suggested distinct prognostic patterns between the subgroups. In BRCA‐mutant patients, FIGO stage IV and KELIM ≤ 1 appeared to be associated with poorer PFS outcomes, while no significant clinical predictors were identified in the BRCA‐wild‐type subgroup. These observations suggest that: (1) in BRCA‐mutant disease, PARPi efficacy could be influenced by both tumor burden (FIGO stage) and chemotherapy sensitivity (KELIM score), whereas (2) in BRCA‐wild‐type cases, the PARPi response might depend on factors beyond conventional clinical parameters, possibly including unexplored molecular features. However, these interpretations should be considered preliminary given the single‐center nature of our study. Larger multicenter studies with comprehensive molecular profiling are required to validate these patterns and explore their biological underpinnings.
Cross‐resistance between PARPi and platinum has been mechanistically established [32]; however, its clinical impact on post‐PARPi chemotherapy remains debatable. In this study, we analyzed the subsequent treatment and prognosis of 42 patients with recurrence. Platinum‐based chemotherapy, bevacizumab, and shorter maintenance periods (< 12 months) seemed to indicate a longer TTSP. However, in terms of the response to second‐line chemotherapy, the duration of PARPi did not affect the ORR or DCR. The results of the PRIMA study also showed that niraparib had no effect on second‐line chemotherapy [8]. At the same time, there are also many real‐world studies that support this view [33, 34]. However, the results of the SOLO‐2 study, as well as a real‐world study from South Korea, presented an opposing view, suggesting that the use of PARPi would weaken the efficacy of subsequent chemotherapy [35, 36]. Thus, whether PARPi maintenance therapy affects the efficacy of subsequent chemotherapy requires high‐level evidence from large population randomized double‐blind studies. Clinically, while focusing on chemotherapy efficacy after the progression of PARPis, more attention should be paid to the patients' long‐term survival benefits.
Beyond the established role of BRCA and HRD in predicting PARPi benefit, it is increasingly being recognized that the prognostic and predictive value of a biomarker is intrinsically linked to the histological and molecular subtypes of ovarian cancer. For instance, while our nomogram aids in stratifying outcomes within the high‐grade serous ovarian carcinoma PARPi maintenance setting, other subtypes like low‐grade serous ovarian carcinoma are driven by distinct alterations, such as BRAF or KRAS mutations. These alterations not only confer a different prognosis but also predict the response to a separate class of targeted therapies (e.g., MEK inhibitors) [12, 14, 37]. This underscores the need for molecular subtyping in therapeutic decision‐making. The clinical integration of diverse biomarkers, including spanning homologous recombination, MAPK signaling, and other pathways, into future prognostic tools is crucial for advancing personalized sequence‐based treatment strategies for all ovarian cancer patients. Although our nomogram provides a focused tool for prognosis within the PARPi maintenance setting, optimizing personalized therapy requires a broader view of the molecular landscape of the tumor. Beyond BRCA, alterations in other homologous recombination repair genes (e.g., RAD51C and PALB2) can influence platinum and PARPi sensitivity [19, 38]. Furthermore, mismatch repair deficiency (dMMR/MSI‐H) identifies a distinct subgroup with a differential prognosis and marked response to immunotherapy [39]. This highlights that the predictive value of a biomarker is inherently linked to a specific therapeutic context. In recent years, beyond conventional nomograms constructed using traditional clinical characteristics, several studies have explored the use of radiomics, pathomics, and related approaches for predicting the prognosis of patients with ovarian cancer. Tumor nuclear features, TME features, and DEGs are closely associated with patient OS and chemotherapy effects, suggesting the potential role of the TME in benefiting from bevacizumab treatment [40]. In addition, it has been reported that radiomics features derived from CT images of OC have the potential to predict HRD score and status, and the developed nomograms can enrich the range of applicable population of PARPi, prolong PFS, and provide personalized treatment for OC patients [41]. Future clinical decision‐making will likely depend on integrating multiple biomarkers to guide a rational sequence of therapies, determining not only who benefits from PARPi but also how to effectively utilize subsequent options, such as platinum rechallenge or immunotherapy, based on the individual molecular profile.
In conclusion, we developed and validated the first nomogram that integrates clinical factors (FIGO stage, residual disease, and age) with biomarker status (BRCA and KELIM) to predict PFS and OS in patients receiving first‐line PARPi maintenance therapy for advanced ovarian cancer. Our analysis of patients with recurrence suggests that platinum‐based chemotherapy combined with bevacizumab may improve outcomes, whereas a PARPi maintenance duration exceeding 12 months may be a risk factor. This study provides a comprehensive, clinical perspective on first‐line PARPi maintenance treatment and may guide personalized PARPi utilization while highlighting socioeconomic barriers to optimal care delivery in resource‐limited settings.
Author Contributions
Yu Zhang: conceptualization, funding acquisition, writing – review and editing, supervision, resources. Shuran Tan: investigation, writing – original draft, methodology, validation, formal analysis, data curation. Xuerui Duan: investigation, data curation. Siyu Yang: investigation, methodology, validation, software, formal analysis, data curation.
Funding
This work was supported by the National Natural Science Foundation of China (82073323).
Ethics Statement
Ethics approval and consent to participate: This study received approval from the ethics committee of Xiangya Hospital, Central South University (Ethics No: 202507081). Written informed consent was obtained from the legally authorized representatives of participants.
Consent
Informed consent was obtained from all participants.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Figure S1: Kaplan–Meier curves for PFS in the PARPi cohort. (A) Kaplan–Meier curves show PFS of patients in BRCA subgroups. (B) Kaplan–Meier curves show PFS of patients in residual disease subgroups. (C) Kaplan–Meier curves show PFS of patients in FIGO subgroups. (D) Kaplan–Meier curves show the PFS of patients in KELIM subgroups. (E) Kaplan–Meier curves show OS of patients in BRCA subgroups. (F) Kaplan–Meier curves show OS of patients in age subgroups.
Figure S2: Prognostic factor analysis stratified by BRCA status in PARPi‐treated ovarian cancer patients. (A) Forest plot of univariate Cox regression for PFS in the BRCA‐mutant subgroup. (B) Multivariate analysis of significant predictors in BRCA‐mutant patients. (C) Univariate analysis for PFS in the BRCA‐wild‐type subgroup. HGSOC, High‐grade serous ovarian cancer; FIGO, the International Federation of Gynecology and Obstetrics; NACT, neoadjuvant chemotherapy; last CT, last chemotherapy; CR, complete response; PR, partial response.
Table S1: Characteristics of PARPi‐treated ovarian cancer patients stratified by BRCA status.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
References
- 1. Bray F., Laversanne M., Sung H., et al., “Global Cancer Statistics 2022: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries,” CA: a Cancer Journal for Clinicians 74, no. 3 (2024): 229–263, 10.3322/caac.21834. [DOI] [PubMed] [Google Scholar]
- 2. Webb P. M. and Jordan S. J., “Global Epidemiology of Epithelial Ovarian Cancer,” Nature Reviews. Clinical Oncology 21, no. 5 (2024): 389–400, 10.1038/s41571-024-00881-3. [DOI] [PubMed] [Google Scholar]
- 3. Kuroki L. and Guntupalli S. R., “Treatment of Epithelial Ovarian Cancer,” BMJ 371 (2020): m3773, 10.1136/bmj.m3773. [DOI] [PubMed] [Google Scholar]
- 4. Tewari K. S., Burger R. A., Enserro D., et al., “Final Overall Survival of a Randomized Trial of Bevacizumab for Primary Treatment of Ovarian Cancer,” Journal of Clinical Oncology 37, no. 26 (2019): 2317–2328, 10.1200/jco.19.01009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. You B., Purdy C., Copeland L. J., et al., “Identification of Patients With Ovarian Cancer Experiencing the Highest Benefit From Bevacizumab in the First‐Line Setting on the Basis of Their Tumor‐Intrinsic Chemosensitivity (KELIM): The GOG‐0218 Validation Study,” Journal of Clinical Oncology 40, no. 34 (2022): 3965–3974, 10.1200/jco.22.01207. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Lord C. J. and Ashworth A., “PARP Inhibitors: Synthetic Lethality in the Clinic,” Science 355, no. 6330 (2017): 1152–1158, 10.1126/science.aam7344. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Banerjee S., Moore K. N., Colombo N., et al., “Maintenance Olaparib for Patients With Newly Diagnosed Advanced Ovarian Cancer and a BRCA Mutation (SOLO1/GOG 3004): 5‐Year Follow‐Up of a Randomised, Double‐Blind, Placebo‐Controlled, Phase 3 Trial,” Lancet Oncology 22, no. 12 (2021): 1721–1731, 10.1016/s1470-2045(21)00531-3. [DOI] [PubMed] [Google Scholar]
- 8. González‐Martín A., Pothuri B., Vergote I., et al., “Niraparib in Patients With Newly Diagnosed Advanced Ovarian Cancer,” New England Journal of Medicine 381, no. 25 (2019): 2391–2402, 10.1056/NEJMoa1910962. [DOI] [PubMed] [Google Scholar]
- 9. Monk B. J., Barretina‐Ginesta M. P., Pothuri B., et al., “Niraparib First‐Line Maintenance Therapy in Patients With Newly Diagnosed Advanced Ovarian Cancer: Final Overall Survival Results From the PRIMA/ENGOT‐OV26/GOG‐3012 Trial,” Annals of Oncology 35, no. 11 (2024): 981–992, 10.1016/j.annonc.2024.08.2241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. DiSilvestro P., Banerjee S., Colombo N., et al., “Overall Survival With Maintenance Olaparib at a 7‐Year Follow‐Up in Patients With Newly Diagnosed Advanced Ovarian Cancer and a BRCA Mutation: The SOLO1/GOG 3004 Trial,” Journal of Clinical Oncology 41, no. 3 (2023): 609–617, 10.1200/jco.22.01549. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Ray‐Coquard I., Leary A., Pignata S., et al., “Olaparib Plus Bevacizumab First‐Line Maintenance in Ovarian Cancer: Final Overall Survival Results From the PAOLA‐1/ENGOT‐ov25 Trial,” Annals of Oncology 34, no. 8 (2023): 681–692, 10.1016/j.annonc.2023.05.005. [DOI] [PubMed] [Google Scholar]
- 12. Perrone C., Angioli R., Luvero D., et al., “Targeting BRAF Pathway in Low‐Grade Serous Ovarian Cancer,” Journal of Gynecologic Oncology 35, no. 4 (2024): e104, 10.3802/jgo.2024.35.e104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Bogani G., Moore K. N., Ray‐Coquard I., et al., “Incorporating Immune Checkpoint Inhibitors in Epithelial Ovarian Cancer,” Gynecologic Oncology 193 (2025): 30–40, 10.1016/j.ygyno.2024.12.011. [DOI] [PubMed] [Google Scholar]
- 14. Mutch D., Voulgari A., Chen X. M., et al., “Primary Results and Characterization of Patients With Exceptional Outcomes in a Phase 1b Study Combining PARP and MEK Inhibition, With or Without Anti‐PD‐L1, for BRCA Wild‐Type, Platinum‐Sensitive, Recurrent Ovarian Cancer,” Cancer 130, no. 11 (2024): 1940–1951, 10.1002/cncr.35222. [DOI] [PubMed] [Google Scholar]
- 15. Menon U., Karpinskyj C., and Gentry‐Maharaj A., “Ovarian Cancer Prevention and Screening,” Obstetrics and Gynecology 131, no. 5 (2018): 909–927, 10.1097/aog.0000000000002580. [DOI] [PubMed] [Google Scholar]
- 16. Miller R. E., Leary A., Scott C. L., et al., “ESMO Recommendations on Predictive Biomarker Testing for Homologous Recombination Deficiency and PARP Inhibitor Benefit in Ovarian Cancer,” Annals of Oncology 31, no. 12 (2020): 1606–1622, 10.1016/j.annonc.2020.08.2102. [DOI] [PubMed] [Google Scholar]
- 17. Chu Y. Y., Yam C., Yamaguchi H., and Hung M. C., “Biomarkers Beyond BRCA: Promising Combinatorial Treatment Strategies in Overcoming Resistance to PARP Inhibitors,” Journal of Biomedical Science 29, no. 1 (2022): 86, 10.1186/s12929-022-00870-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Li X. and Zou L., “BRCAness, DNA Gaps, and Gain and Loss of PARP Inhibitor‐Induced Synthetic Lethality,” Journal of Clinical Investigation 134, no. 14 (2024): e181062, 10.1172/jci181062. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Murai J. and Pommier Y., “BRCAness, Homologous Recombination Deficiencies, and Synthetic Lethality,” Cancer Research 83, no. 8 (2023): 1173–1174, 10.1158/0008-5472.Can-23-0628. [DOI] [PubMed] [Google Scholar]
- 20. Groslambert J., Schützenhofer K., Palazzo L., and Ahel I., “PARPs and ADP‐Ribosyl Hydrolases in Cancer Therapy: From Drug Targets to Biomarkers,” DNA Repair (Amst) 152 (2025): 103863, 10.1016/j.dnarep.2025.103863. [DOI] [PubMed] [Google Scholar]
- 21. Rusan M., Andersen R. F., Jakobsen A., and Steffensen K. D., “Circulating HOXA9‐Methylated Tumour DNA: A Novel Biomarker of Response to Poly (ADP‐Ribose) Polymerase Inhibition in BRCA‐Mutated Epithelial Ovarian Cancer,” European Journal of Cancer 125 (2020): 121–129, 10.1016/j.ejca.2019.11.012. [DOI] [PubMed] [Google Scholar]
- 22. Tan S., Zhu F., Li Y., et al., “HELQ Upregulates PARP1 to Drive Platinum Resistance and Predict Therapeutic Response in Ovarian Cancer,” Translational Oncology 57 (2025): 102416, 10.1016/j.tranon.2025.102416. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Akashi H., Yachida N., Ueda H., et al., “SLFN11 Is a BRCA Independent Biomarker for the Response to Platinum‐Based Chemotherapy in High‐Grade Serous Ovarian Cancer and Clear Cell Ovarian Carcinoma,” Molecular Cancer Therapeutics 23, no. 1 (2024): 106–116, 10.1158/1535-7163.Mct-23-0257. [DOI] [PubMed] [Google Scholar]
- 24. Kim J. H., Kim S. I., Park E. Y., et al., “Impact of Postoperative Residual Disease on Survival in Epithelial Ovarian Cancer With Consideration of Recent Frontline Treatment Advances: A Systematic Review and Meta‐Analysis,” Gynecologic Oncology 179 (2023): 24–32, 10.1016/j.ygyno.2023.10.018. [DOI] [PubMed] [Google Scholar]
- 25. Li N., Zhu J., Yin R., et al., “Treatment With Niraparib Maintenance Therapy in Patients With Newly Diagnosed Advanced Ovarian Cancer: A Phase 3 Randomized Clinical Trial,” JAMA Oncology 9, no. 9 (2023): 1230–1237, 10.1001/jamaoncol.2023.2283. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Gonzalez R., Havrilesky L. J., Myers E. R., et al., “Cost‐Effectiveness Analysis Comparing “PARP Inhibitors‐For‐All” to the Biomarker‐Directed Use of PARP Inhibitor Maintenance Therapy for Newly Diagnosed Advanced Stage Ovarian Cancer,” Gynecologic Oncology 159, no. 2 (2020): 483–490, 10.1016/j.ygyno.2020.08.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Shi Y., Xiao D., Li S., Liu S., and Zhang Y., “Cost‐Effectiveness of Maintenance Niraparib With an Individualized Starting Dosage in Patients With Platinum‐Sensitive Recurrent Ovarian Cancer in China,” Frontiers in Pharmacology 14 (2023): 1198585, 10.3389/fphar.2023.1198585. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Li L., Gu Y., Zhang M., et al., “HRD Effects on First‐Line Adjuvant Chemotherapy and PARPi Maintenance Therapy in Chinese Ovarian Cancer Patients,” npj Precision Oncology 7, no. 1 (2023): 51, 10.1038/s41698-023-00402-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Li H., Liu Z. Y., Wu N., Chen Y. C., Cheng Q., and Wang J., “PARP Inhibitor Resistance: The Underlying Mechanisms and Clinical Implications,” Molecular Cancer 19, no. 1 (2020): 107, 10.1186/s12943-020-01227-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Zhao M., Qiu S., Wu X., et al., “Efficacy and Safety of Niraparib as First‐Line Maintenance Treatment for Patients With Advanced Ovarian Cancer: Real‐World Data From a Multicenter Study in China,” Targeted Oncology 18, no. 6 (2023): 869–883, 10.1007/s11523-023-00999-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Lin H., Ou Y. C., Fu H. C., et al., “Prognostic Nomogram for Advanced Epithelial Ovarian Cancer in the Era of Homologous Recombination Testing and PARP Inhibitor Therapy,” American Journal of Cancer Research 15, no. 5 (2025): 2439–2448, 10.62347/xytz1766. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Dias M. P., Moser S. C., Ganesan S., and Jonkers J., “Understanding and Overcoming Resistance to PARP Inhibitors in Cancer Therapy,” Nature Reviews. Clinical Oncology 18, no. 12 (2021): 773–791, 10.1038/s41571-021-00532-x. [DOI] [PubMed] [Google Scholar]
- 33. Sivakumaran T., Krasovitsky M., Freimund A., et al., “Treatment Patterns After Poly‐ADP Ribose Polymerase (PARP) Inhibitors in Epithelial Ovarian Cancer Patients,” International Journal of Gynecological Cancer 32, no. 7 (2022): 906–912, 10.1136/ijgc-2021-003009. [DOI] [PubMed] [Google Scholar]
- 34. Ma Y., Liu J., Li N., et al., “Efficacy of Poly (ADP‐Ribose) Polymerase Inhibitors Monotherapy and the Impact to Subsequent Platinum‐Based Chemotherapy in Breast Cancer Susceptibility genes1/2‐Mutated Ovarian Cancer Patients With Secondary Platinum‐Sensitive Relapse,” Journal of Ovarian Research 16, no. 1 (2023): 209, 10.1186/s13048-023-01283-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Kristeleit R. S. and Moore K. N., “Life After SOLO‐2: Is Olaparib Really Inducing Platinum Resistance in BRCA‐Mutated (BRCAm), PARP Inhibitor (PARPi)‐resistant, Recurrent Ovarian Cancer?,” Annals of Oncology 33, no. 10 (2022): 989–991, 10.1016/j.annonc.2022.08.003. [DOI] [PubMed] [Google Scholar]
- 36. Park J., Kim S. I., Jeong S. Y., et al., “Second‐Line Olaparib Maintenance Therapy Is Associated With Poor Response to Subsequent Chemotherapy in BRCA1/2‐Mutated Epithelial Ovarian Cancer: A Multicentre Retrospective Study,” Gynecologic Oncology 165, no. 1 (2022): 97–104, 10.1016/j.ygyno.2022.02.002. [DOI] [PubMed] [Google Scholar]
- 37. Nameki R., Chang H., Reddy J., Corona R. I., and Lawrenson K., “Transcription Factors in Epithelial Ovarian Cancer: Histotype‐Specific Drivers and Novel Therapeutic Targets,” Pharmacology & Therapeutics 220 (2021): 107722, 10.1016/j.pharmthera.2020.107722. [DOI] [PubMed] [Google Scholar]
- 38. Tonti N., Golia D'Augè T., Cuccu I., et al., “The Role of Tumor Biomarkers in Tailoring the Approach to Advanced Ovarian Cancer,” International Journal of Molecular Sciences 25, no. 20 (2024): 11239, 10.3390/ijms252011239. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Fraune C., Rosebrock J., Simon R., et al., “High Homogeneity of MMR Deficiency in Ovarian Cancer,” Gynecologic Oncology 156, no. 3 (2020): 669–675, 10.1016/j.ygyno.2019.12.031. [DOI] [PubMed] [Google Scholar]
- 40. Yin H., Liu Q., Feng S., et al., “Development of the Model for Predicting the Efficacy of Bevacizumab and Prognosis in Ovarian Cancer Using Deep Learning‐Based Pathomics Signatures,” Chinese Medical Journal 139, no. 5 (2026): 773–775, 10.1097/cm9.0000000000003823. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Wu Y., Zhang Q., Jiang W., Gao Y., Qu B., and Wang X., “CT‐Based Radiomics Predicts HRD Score and HRD Status in Patients With Ovarian Cancer,” Frontiers in Oncology 14 (2024): 1477759, 10.3389/fonc.2024.1477759. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1: Kaplan–Meier curves for PFS in the PARPi cohort. (A) Kaplan–Meier curves show PFS of patients in BRCA subgroups. (B) Kaplan–Meier curves show PFS of patients in residual disease subgroups. (C) Kaplan–Meier curves show PFS of patients in FIGO subgroups. (D) Kaplan–Meier curves show the PFS of patients in KELIM subgroups. (E) Kaplan–Meier curves show OS of patients in BRCA subgroups. (F) Kaplan–Meier curves show OS of patients in age subgroups.
Figure S2: Prognostic factor analysis stratified by BRCA status in PARPi‐treated ovarian cancer patients. (A) Forest plot of univariate Cox regression for PFS in the BRCA‐mutant subgroup. (B) Multivariate analysis of significant predictors in BRCA‐mutant patients. (C) Univariate analysis for PFS in the BRCA‐wild‐type subgroup. HGSOC, High‐grade serous ovarian cancer; FIGO, the International Federation of Gynecology and Obstetrics; NACT, neoadjuvant chemotherapy; last CT, last chemotherapy; CR, complete response; PR, partial response.
Table S1: Characteristics of PARPi‐treated ovarian cancer patients stratified by BRCA status.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.