Clinical application of PARP inhibitors and emerging strategies to overcome resistance: a pan-cancer perspective.

Introduction
In the field of oncology, therapeutic strategies aimed at targeting the acquired vulnerabilities arising from oncogenic events effectively eliminate cancer cells while preserving normal tissues. Among these strategies, PARP inhibitors (PARPi) have emerged as a transformative approach in oncological management [1]. Supported by numerous clinical trials, PARPi have received approval for use in multiple tumors characterized by deficiencies in homologous recombination repair (HRR), including ovarian, breast, prostate, and pancreatic cancers. However, similar to other targeted therapies, the development of acquired resistance significantly contributes to disease progression and undermines the sustained efficacy of PARPi, even when patients exhibit promising initial responses [2]. Therefore, elucidating the fundamental mechanisms underlying both PARPi activity and resistance is crucial for devising effective strategies to overcome this challenge.
PARP1 is the most extensively characterized member of the PARP family, responsible for regulating poly-ADP-ribosylation (PARylation), a post-translational modification, of itself and other target proteins [3]. PARP1 is pivotal in maintaining genomic integrity through various cellular processes, including the repair of single-strand breaks (SSBs), chromatin remodeling, replication fork (RF) stabilization, and the recognition of unligated Okazaki fragments (OFs) during DNA replication [4, 5]. Currently, six DNA damage repair pathways have been identified to address SSBs and double-strand breaks (DSBs). While PARP1 is central to SSB repair, the repair of DSBs relies on distinct mechanisms, predominantly homologous recombination (HR) and non-homologous end joining (NHEJ). HR is a high-fidelity pathway that becomes active specifically during the S and G2 phases due to the requirement for a sister chromatid template. In contrast, NHEJ is a faster, albeit more error-prone, mechanism that operates predominantly in the G1 phase. Beyond canonical proteins like Ku70/80, DNA-PKcs, Artemis, DNA pol λ/μ, DNA ligase IV-XRCC4, and XLF, the NHEJ machinery incorporates emerging factors including PAXX, TARDBP of TDP-43, IFFO1, ERCC6L2, RNase H2, and MRI/CYREN. Notably, MRI/CYREN exerts a cell-cycle-dependent dual function by promoting NHEJ in the G1 phase while suppressing it during S and G2 phases to prioritize HR-mediated repair [6]. However, as BRCA1 and BRCA2 are indispensable for HRR [7], homologous recombination deficiency (HRD) compels cells to resort to alternative repair mechanisms, thereby driving genomic instability and tumorigenesis. In this context, the concept of “BRCAness” was introduced to characterize tumors that share phenotypic traits with BRCA-deficient cells due to the loss of other HRR genes [8]. Mechanistically, HRD renders cancer cells highly dependent on compensatory pathways, particularly PARP-mediated signaling. Consequently, PARPi induce synthetic lethality in HRD tumors: by blocking SSB repair, PARPi treatment leads to an accumulation of DSBs that HR-deficient cells cannot effectively repair, ultimately triggering mitotic catastrophe [9].
Preclinical and clinical studies have identified numerous molecular alterations that drive resistance to PARPi. However, effective therapeutic strategies to overcome this resistance remain elusive. Through extensive investigation of the mechanisms of action of PARPi and the pathways leading to drug resistance, emerging hypotheses may provide potential avenues for reversing prevalent PARPi resistance. This review offers a comprehensive examination of landmark clinical trials across various cancer types, including ovarian, breast, prostate, and pancreatic cancers, to enhance our understanding of the clinical approvals related to PARPi. By comparing various predictive tools available for evaluating responses to PARPi, we aim to provide a comprehensive overview of their clinical applications and highlight novel strategies for monitoring acquired resistance. Furthermore, we summarize the diverse mechanisms underlying PARPi resistance and discuss potential approaches to address these significant challenges.

Clinical utilization of PARP inhibitors
Several clinical trials have demonstrated that tumors exhibiting HRD are susceptible to PARPi, the utilization of which significantly improves survival outcomes of various malignancies [10]. Based on the survival benefit, the Food and Drug Administration (FDA) has approved the utilization of PARPi in four malignancies, including ovarian, breast, prostate, and pancreatic cancers (Fig. 1) [11]. Fueled by numerous clinical trials, the clinical application of PARPi has expanded beyond metastatic therapy to include maintenance and perioperative settings. Furthermore, combining strategies with other therapeutic methods based on each mechanism broadens the indications for PARPi from homologous recombination (HR)-deficient to HR-proficient tumors. As depicted in Table 1 and Table S1, this review summarizes the clinical approvals of PARPi in four cancer types that have demonstrated the most significant benefits and highlights the promising advancements of PARPi in other tumors with potential for future application. Major clinical trials for novel and investigational PARPi are also summarized in Table 2.

Ovarian cancer
Ovarian cancer (OC) encompasses a heterogeneous group of malignancies with distinct biological behaviors. While Type I epithelial OCs are relatively indolent and genetically stable, typically arising from precursor lesions such as endometriosis or borderline tumors, Type II epithelial OCs are biologically aggressive, exhibiting a propensity for metastasis even from small-volume primary lesions. Notably, high-grade serous ovarian cancer, which accounts for approximately 75% of epithelial OCs, develops via the Type II pathway and is frequently characterized by TP53 and BRCA mutations [30]. Germline BRCA1/2 mutations (gBRCA1/2m) represent the strongest known genetic risk factors for epithelial OC, identified in 6–15% of diagnosed women [31]. Clinically, BRCA1/2 status serves as a critical biomarker for patient counseling. Although gBRCA1/2m carriers typically present with advanced stages and higher grades, they exhibit superior responses to platinum-based chemotherapy and improved survival compared to non-carriers. This specific defect in HR underpins the significant success of PARPi as maintenance therapies in advanced OC [32–35], distinct from the roles of chemotherapy and bevacizumab [36–38]. However, in contrast to their efficacy in the maintenance setting, PARPi do not improve survival outcomes compared to chemotherapy in post-line treatment settings. As evidenced by the phase 3 ARIEL4 (NCT02855944) [39] and SOLO3 (NCT02282020) [40, 41], PARPi may even exert detrimental effects on heavily pretreated patients, underscoring the complexity of resistance mechanisms in later treatment lines.
Given the challenges and potential toxicity encountered in later treatment lines, maximizing therapeutic efficacy in the early stages has become paramount. To date, four pivotal clinical trials have substantiated the utility of PARPi as standard-of-care first-line maintenance treatments, albeit with distinct therapeutic indication and specific agents. Specifically, olaparib was assessed in the SOLO1 trial (NCT01844986) [32, 33, 42], and in combination with bevacizumab in PAOLA-1 (NCT02477644) [43–46]. Parallelly, rucaparib was investigated within the ATHENA-MONO study (NCT03522246) [47], while niraparib served as the focus of the PRIMA trial (NCT02655016) [35, 48, 49].
The randomized, double-blind, phase 3 SOLO1 trial enrolled 391 patients with newly diagnosed advanced OC harboring a germinal or somatic BRCA1/2m (g/sBRCA1/2m) who responded to platinum-based chemotherapy as the first-line chemotherapy [33]. At a median follow-up of 5 years, the median progression-free survival (mPFS) was 56.0 months in the olaparib group versus 13.8 months in the placebo group [42]. This significant survival benefit of olaparib persisted even after a median follow-up of 7 years, with a median overall survival (mOS) significantly longer than placebo (not reached vs. 75.2 months, HR 0.55) [32]. Based on these compelling results, olaparib received FDA approval in 2018 as maintenance therapy for patients with advanced OC harboring deleterious or suspected deleterious g/sBRCA1/2m who achieved a complete or partial response to initial platinum-based chemotherapy.
Furthermore, the combination of olaparib with the anti-vascular endothelial growth factor (VEGF) monoclonal antibody bevacizumab expanded the indication of PARPi-based maintenance therapy to HRD tumors beyond g/sBRCA1/2m. This approach is rational because bevacizumab-induced inhibition of angiogenesis leads to hypoxia, downregulating HRR gene expression, thereby sensitizing tumors to PARPi-induced apoptosis [11]. In the PAOLA-1 trial (NCT02477644), 806 patients with advanced high-grade OC who achieved a complete or partial response to first-line platinum-based chemotherapy plus bevacizumab were enrolled [50]. Patients were randomly assigned to receive either olaparib plus bevacizumab or bevacizumab plus placebo as maintenance therapy [50]. This combination therapy significantly improved PFS and OS in HR-deficient patients, but not in the HR-proficient group [43, 45]. Moreover, rucaparib and niraparib significantly improved survival outcomes in patients with advanced OC, regardless of HRD status [35, 47, 49]. Based on the phase 3 PRIMA trial [35, 49], niraparib received FDA approval in 2020 for maintenance therapy in patients with advanced OC who have responded partially or entirely to initial platinum-based chemotherapy, irrespective of HRD status. Developing combination strategies and PARPi agents has broadened the clinical applicability of PARPi-based first-line maintenance therapy, extending its applicability from g/sBRCA1/2m to HRD and the general population.
Moreover, PARPi have received FDA approval as post-line maintenance therapy for recurrent, platinum-sensitive, advanced OC harboring BRCA1/2m. The phase 3 SOLO2 trial (NCT01874353) demonstrated that olaparib maintenance therapy significantly prolongs survival compared with placebo in patients who had recurrent, platinum-sensitive OC and harbored gBRCA1/2 m (mOS: 51.7 vs. 38.8 months, HR 0.74) [51, 52]. Furthermore, regardless of HRD status, rucaparib and niraparib have been demonstrated to improve the PFS in recurrent, platinum-sensitive OC [34, 53].

Breast cancer
Breast cancer (BC) is the most common malignancy and the second leading cause of cancer-related mortality in women globally [54]. BRCA1/2m is strongly associated with hereditary breast cancer, particularly in triple-negative breast cancer (TNBC) and early-onset BC [55]. PARPi-based strategies have demonstrated significant clinical benefit in advanced BC [56]. Moreover, PARPi have been further established as a novel therapeutic option in the perioperative setting, particularly in the adjuvant context [57, 58]. For patients with metastatic BC, the addition of PARPi to chemotherapy and immunotherapy exhibited superior clinical efficacy compared to conventional monotherapy [59–63]. However, in the neoadjuvant setting, the effectiveness of the combination strategy requires more robust supporting evidence [64, 65].
Olaparib and talazoparib first PARPi to received FDA approval for the treatment of metastatic BC with gBRCA1/2m, based on the findings of two clinical trials, OlympiAD (NCT02000622) [66, 67] and EMBRACA (NCT01945775) [68, 69]. For patients with metastatic HER2-negative BC harboring gBRCA1/2m, olaparib significantly prolonged mPFS compared with the physician’s choice of chemotherapy (7.0 vs. 4.2 months, HR 0.58) [66], with a trend towards improved mOS (19.3 vs. 17.1 months, HR 0.90) [67]. Similarly, talazoparib monotherapy significantly prolonged mPFS in patients with HER2-negative, locally advanced or metastatic BC and gBRCA1/2m (8.6 vs. 5.6 months, HR 0.54) [68]. However, the improvement in mOS between these two groups was not reach statistical significance (19.3 vs. 19.5 months, HR 0.85) [69]. The high prevalence of PARPi resistance may underlie the limited improvement in OS.
Hence, a combination strategy is being explored to enhance the clinical efficacy of PARPi. The phase 3 BROCADE3 trial, which evaluated a triplet regimen of veliparib, carboplatin, and paclitaxel, failed to demonstrate a statistically significant improvement in OS for patients with HER2-negative locally advanced or metastatic BC and gBRCA1/2m (mPFS: 14.5 vs. 12.6 months, HR 0.71; mOS: 33.5 vs. 28.2 months, HR 0.95) [59]. Beyond the PARPi-based chemotherapy, other early-phase clinical trials investigated combinations of PARPi with immune checkpoint inhibitors (ICIs) or other targeted therapies. For instance, the MEDIOLA trial (NCT02734004) evaluated olaparib plus durvalumab [60, 61], while the RADIOPARP trial (NCT03109080) assessed olaparib combined with breast radiotherapy [62, 63]. However, the survival benefit of these novel treatment approaches requires validation in larger, randomized clinical trials with directed comparisons.
While PARPi agents have demonstrated efficacy in metastatic BC, their integration into perioperative settings may benefit a broader patient population and improve long-term outcomes. Notably, olaparib has shown a significant survival benefit as an adjuvant therapy in preventing BC recurrence, highlighting its promise in this context. The OlympiA trial (NCT02000622) enrolled 1836 patients with high-risk, early-stage, HER2-negative BC harboring gBRCA1/2m who had previously undergone definitive local treatment and received neoadjuvant or adjuvant chemotherapy [57]. Participants were randomly assigned to receive olaparib or placebo as adjuvant therapy. After a 3-year follow-up, the invasive disease-free survival rate was 85.9% in the olaparib group versus 77.1% in the placebo group [57]. Similarly, the distant disease-free survival rate was significantly higher in the olaparib group compared with the placebo group (87.5% vs. 80.4%) [58]. Based on these results, the FDA approved olaparib as adjuvant treatment for patients with high-risk, HER2-negative early BC harboring deleterious or suspected deleterious gBRCA1/2m who have received prior neoadjuvant or adjuvant chemotherapy.
However, the efficacy of PARPi in the neoadjuvant setting remains limited, as revealed by several clinical trials. For instance, the single-arm, phase 2 NeoTALA trial (NCT03499353) evaluated the efficacy and safety of neoadjuvant talazoparib in patients with early-stage TNBC carrying gBRCA1/2m [70]. The pathologic complete response (pCR) rate was 49.2% in the intent-to-treat population, which did not meet the prespecified criteria for statistical significance [70]. Furthermore, when considering the entire TNBC patient population irrespective of BRCA1/2 status, adding PARPi did not improve the efficacy of standard neoadjuvant carboplatin-paclitaxel chemotherapy, supported by the phase 3 BrighTNess trial (NCT02032277) and the phase 2/3 PARTNER trial [64, 65].

Prostate cancer
Prostate cancer remains the most common noncutaneous malignancy in men, often associated with poor survival outcomes in metastatic cases [54, 71]. While androgen deprivation therapy (ADT) has long served as a cornerstone treatment, therapeutic paradigms have evolved. Currently, the combination of abiraterone acetate plus prednisolone with ADT is considered a new standard of care for patients with high-risk non-metastatic prostate cancer. In the metastatic setting, however, combining enzalutamide with abiraterone is not recommended for patients initiating long-term ADT. Crucially, the addition of abiraterone to ADT confers a clinically meaningful survival benefit that is maintained for over seven years [72]. Despite these advances, resistance frequently develops, leading to castration-resistant prostate cancer (CRPC) [73]. In this context, HRD has emerged as a critical therapeutic target. Specifically, patients harboring BRCA2 pathogenic sequence variants exhibit aggressive clinical features, including elevated serum prostate-specific antigen levels at diagnosis, a higher proportion of high Gleason score tumors, increased rates of nodal and distant metastases, and high recurrence rates [74]. These distinct phenotypic characteristics and the underlying genomic instability underscore the potential efficacy of PARPi [75]. Consequently, numerous clinical trials are actively investigating PARPi-based therapies for metastatic CRPC (mCRPC) [76–84], while early-phase trials are also exploring the efficacy of PARPi in high-risk biochemical recurrent (BCR) prostate cancer following radical prostatectomy [85].
For mCRPC patients with HRD, PARPi monotherapy has shown significant survival benefits, particularly in those who have progressed on androgen receptor pathway inhibitors (ARPIs). The phase 3 PROfound trial (NCT02987543) revealed that olaparib significantly improved survival compared with enzalutamide or abiraterone in mCRPC patients with HRD who progressed on ARPIs [76, 77]. Similarly, the phase 3 TRITON3 trial (NCT02975934) confirmed that rucaparib significantly prolonged imaging-based PFS compared to control medications in mCRPC patients with BRCA1/2m [78].
Furthermore, combining PARPi and ARPIs as first-line therapy for mCRPC patients significantly improves survival outcomes in the HRD subgroup, leveraging the synthetic lethality between these agents. Mechanistically, ARPIs downregulate HR proteins, thereby inducing a BRCAness phenotype and enhancing sensitivity to PARPi [11]. Conversely, PARPi trapping disrupts androgen receptor transcription, thereby enhancing ARPI activity [11]. Several clinical trials support this rationale. TALAPRO-2 (NCT03395197) demonstrated that talazoparib plus enzalutamide significantly improved radiographic PFS in mCRPC patients with HRD [79–81]. PROpel (NCT03732820) showed that olaparib plus abiraterone significantly prolonged PFS in the mCRPC subgroup with BRCA1/2m [82, 83]. MAGNITUDE (NCT03748641) revealed that niraparib plus abiraterone resulted in significant PFS and OS benefits in patients with BRCA1/2m and mCRPC [84].
Notably, a non-randomized controlled trial (NCT03047135) evaluated the efficacy and safety of olaparib monotherapy in patients with high-risk BCR prostate cancer after radical prostatectomy [85]. Patients with BRCA2 alterations achieved high and durable PSA50 response rates. Hence, further investigation of olaparib as a treatment strategy for patients with BCR prostate cancer is justified. Conversely, for BCR prostate cancer without HRD, this trial also revealed that olaparib lacked sufficient activity, suggesting it should not be considered for those patients [85].

Pancreatic cancer
Pancreatic cancer is characterized by its high aggressiveness and poor prognosis [86]. For patients with metastatic pancreatic cancer and gBRCA1/2m, olaparib is the only FDA-approved PARPi, although its survival benefit is limited. The POLO trial, a randomized, double-blind, phase 3 study, was the first to evaluate olaparib as first-line maintenance therapy in patients with metastatic pancreatic cancer who have not progressed within 16 weeks after platinum-based chemotherapy [87]. Compared to placebo, olaparib significantly prolonged median PFS (7.4 vs. 3.8 months, HR 0.53) but did not significantly improve mOS (19.0 vs. 19.2 months, HR 0.83) [87, 88]. These findings underscore the need for more effective maintenance strategies.
Notably, the NCT03404960 trial suggests that combining PARPi with anti-CTLA4 therapy is a promising approach for maintenance therapy in patients with platinum-sensitive pancreatic cancer [89]. The combination of niraparib and ipilimumab resulted in an mPFS of 8.1 months, a 6-month PFS rate of 59.6%, and a mOS of 17.3 months [89]. In contrast, the combination of niraparib and nivolumab yielded disappointing results (mPFS: 1.9 months; 6-month PFS rate: 20.6%) [89]. The survival benefit of this combination strategy needs further investigation in larger, randomized clinical trials.

Other cancer types
Despite the established therapeutic efficacy of PARPi in the four aforementioned malignancies, ongoing clinical investigations continue to explore their utility in additional tumor types, including urothelial cancer [90], melanoma [91], and colorectal cancer [92], albeit with limited success to date. Notable advancements have emerged in small cell lung cancer (SCLC) [93–95]. In particular, patients with platinum-responsive extensive-stage SCLC may benefit from niraparib as first-line maintenance therapy. The randomized, double-blind, phase 3 ZL-23060–005 trial enrolled 185 patients with extensive-stage SCLC who achieved complete or partial response to platinum-based first-line chemotherapy, randomizing them to receive niraparib or placebo as the maintenance treatment [93]. Although the trial did not meet its primary endpoint, niraparib modestly improved mPFS compared with placebo (1.54 vs. 1.36 months, HR 0.66) [93]. Furthermore, veliparib combined with temozolomide can prolong survival in SCLC with high SLFN11 expression [94]. Additionally, olaparib plus durvalumab demonstrated clinical activity in relapsed SCLC [95], offering a promising new therapeutic strategy for this highly aggressive malignancy.
For advanced or metastatic non-small cell lung cancer (NSCLC), the addition of PARPi to platinum-based chemotherapy has shown limited efficacy in improving survival outcomes [96]. Several clinical trials have demonstrated that veliparib in combination with carboplatin and paclitaxel does not exhibit significantly greater therapeutic efficacy than chemotherapy alone [97, 98]. Notably, however, in the LP52-positive subgroup of NSCLC patients, veliparib combined with chemotherapy substantially prolongs OS compared to chemotherapy alone [96, 97]. This underscores the critical role of biomarkers in identifying PARPi-sensitive subpopulations.
To summarize, PARPi have demonstrated substantial clinical efficacy in various malignancies. However, the optimal application of PARPi varies by cancer type and specific agent. Furthermore, beyond HRD status, novel biomarkers of PARPi sensitivity are also crucial to improving patient selection and survival outcomes. Notably, acquired resistance to PARPi remains a significant challenge [2, 99]. Although PARPi often induce promising initial responses, most patients develop resistance to these agents, leading to disease recurrence and limiting improvements in OS [67, 69, 87, 88, 100]. Several clinical trials utilized circulating tumor DNA (ctDNA) to monitor PARPi sensitivity, revealing that specific mutations occurred frequently and were associated with disease progression [101–104]. Consequently, overcoming primary and acquired resistance to PARPi remain a pressing challenge requiring urgent attention.

Working mechanism of the PARP inhibitor
PARPi resistance significantly impairs its clinical efficacy across various cancer types. A comprehensive understanding of their mechanism of action is crucial for elucidating resistance mechanisms and developing strategies to overcome them (Fig. 2). PARP family members, notably PARP1 and PARP2, are pivotal in DNA damage repair. Among them, PARP1 is extensively studied and is responsible for detecting and repairing SSBs [105, 106]. PARPi exploit the synthetic lethality between PARP1/2 and BRCAness, selectively targeting HRD tumor cells while sparing surrounding normal tissue. HRR is a preferred pathway for accurately addressing DSBs, in which BRCA1/2 plays a significant role [107]. Inhibition of PARP1/2 increases genomic instability, ultimately resulting in lethal DSBs for HR-deficient tumor cells. However, the precise mechanisms by which PARPi induces DSB accumulation remain a subject of ongoing investigation. Three core hypotheses have been proposed: PARP1/2 trapping, transcription-replication conflicts, and the replication gap model.
PARP trapping is widely recognized as the primary driver of PARPi-induced cytotoxicity [2]. Mechanistically, PARPi promote the entrapment of PARP1/2 on DNA via two synergistic pathways: the inhibition of enzymatic activity and the induction of conformational changes [108–110]. First, by blocking auto-PARylation, which involves the addition of negatively charged PAR chains required for PARP release, PARPi prevent dissociation from DNA [111]. Second, PARPi binding is thought to induce allosteric modifications that structurally reinforce the PARP-DNA interaction [108]. These trapped PARP-DNA complexes act as physical barriers that obstruct the recruitment of other DNA damage repair (DDR) factors [99] and stall replication forks, triggering severe DNA DSBs [106]. Crucially, the capacity to form these obstructive complexes underlies the significant differences in cytotoxic potency observed among various PARPi. Acting as physical obstructions, these trapped complexes disrupt DNA replication and repair processes far more severely than the mere inhibition of enzymatic activity. Among the PARPi evaluated to date, olaparib, niraparib, and rucaparib demonstrate substantially greater PARP-trapping capacity, estimated to be approximately 100-fold higher than that of veliparib [112]. This hierarchy of trapping potency correlates with clinical efficacy, where agents with enhanced trapping capacity, such as talazoparib and niraparib, exhibit superior cytotoxicity [109, 110]. Conversely, structural mutations in the PARP1 DNA binding domain that impair trapping have been shown to reduce the cytotoxicity of these potent agents [113].
Despite the trapping theory, the inhibition of PARP’s catalytic activity represents another key mechanism of PARPi action. This is supported by evidence that PARP1 depletion is lethal with HR deficiency, even without PARP1 entrapment [114]. Beyond its conventional role in SSBs repair, recent studies suggest that PARP1 is involved in other pathways related to genomic instability. Firstly, PARP1 is known to resolve accumulated R-loops, three-stranded nucleic acid structures formed during transcription, comprising RNA–DNA hybrids and a displaced non-template DNA strand [115]. Therefore, PARPi or PARP1 depletion induces genomic instability associated with unresolved R-loops. Secondly, although trapped PARP1 is believed to inhibit RFs progression [116], PARP1 also functions with TIMELESS and TIPIN to protect the replisome in the early S phase from transcription-replication conflicts [114]. Consistently, inhibiting transcription elongation in the early S phase renders HR-deficient cells resistant to PARPi [114], indicating that transcription-replication conflicts, rather than PARP trapping, are a major source of PARPi-induced DSBs accumulation.
Another hypothesis related to enzymatic catalytic inhibition is the replication gap model [117, 118]. This model centers on the role of PARP1 in processing OFs during lagging-strand maturation behind the RF [4, 119]. In this context, BRCA1/2 proteins are theorized to shield these nascent fragments from premature degradation, preventing the exposure of long single-stranded DNA (ssDNA) gaps behind the RF [120]. Therefore, PARPi inhibit the ligation of OFs, while BRCA deficiency induces the degradation of unprotected ssDNA behind the RF. This mechanism is further confirmed by the synthetic lethality between BRCA1/2 and polymerase POLθ [121], which mediates the fill-in process of post-replicative gaps [122–124].
Although the interaction with BRCA1/2 remains the archetype of synthetic lethality, the therapeutic application of PARPi has expanded to include a broad spectrum of genetic and metabolic alterations (Table S2). These alternative partners sensitize tumors to PARP inhibition by either recapitulating HR deficiency or by creating unique dependencies related to replication stress and genomic stability. The most direct extension of this model involves defects within the core DNA damage response architecture. Dysfunction in upstream sensors including ATM, ATR, CHK1/2, and NBS1, or downstream effectors such as RAD51 and PALB2, diminishes the cellular capacity to repair PARPi-induced strand breaks [125]. This effectively mimics the lethality seen in BRCA-deficient contexts. The Fanconi anemia pathway is also integral to this network, where defects in FANCC, FANCA, or FANCD2 hinder the repair of stalled replication forks and confer significant sensitivity to PARP inhibition [125].
Synthetic lethality also arises from the dysregulation of upstream pathways governing HR competence. Transcriptional control is exemplified by CDK12; its deficiency causes premature termination of long transcripts and selectively downregulates pivotal HR genes, including BRCA1, ATR, and FANCI [126]. This transcriptional suppression establishes a functional BRCAness state in models of prostate and ovarian cancer. Additionally, PTEN deficiency, which is common in sporadic tumors, represses RAD51 expression and recruitment, thereby generating a context-specific vulnerability to PARPi [127]. Beyond direct signaling defects, metabolic and splicing aberrations act as potent drivers of PARPi sensitivity. Mutations in IDH1/2 lead to the accumulation of 2-hydroxyglutarate (2-HG) [128]. This oncometabolite inhibits α-KG-dependent histone demethylases, resulting in the epigenetic repression of HR gene expression. Similarly, the recurrent SF3B1 K700E spliceosome mutation causes mis-splicing of DDR factors and promotes R-loop accumulation, which drives replication fork instability targetable by PARPi [129].
Recent evidence has further identified a mechanism of synthetic lethality driven by retrotransposon activity that is mechanistically distinct from classical HRD. In hematological malignancies, loss of the Polycomb group proteins ASXL1 and EZH2 results in the epigenetic reactivation of transposable elements. Reverse transcription of these elements generates toxic single-stranded DNA intermediates that rely on PARP1 for protection. Accordingly, PARP inhibition in ASXL1/EZH2-deficient cells induces cell death through excessive DNA damage linked to retro-transposition rather than HR defects [130]. These findings indicate that PARPi sensitivity is not defined solely by BRCA status but results from diverse molecular contexts that compromise genomic integrity.

Mechanisms of PARP inhibitor resistance
Although PARPi have exhibited great clinical responses in various cancer types, acquired resistance occurs frequently, leading to disease relapse and poorer survival outcomes. A comprehensive understanding of the mechanisms underlying this resistance is essential for re-sensitizing patients and improving the therapeutic efficacy of PARPi (Fig. 3). Based on the synthetic lethality of PARPi, five mechanisms related to acquired resistance are demonstrated (Table 3). Mechanistically, PARP1 alternation and HRR restoration potentially disturb the synthetic lethality. The compensatory enhancement of genomic stability through DDR protein overexpression and RF stabilization also confers resistance to PARPi. Additionally, the upregulation of drug efflux pumps represents another common resistance mechanism. Moreover, inherent characteristics of tumors, such as an immunosuppressed tumor microenvironment (TME) and moderate hypoxia, are also implicated in the acquired resistance of PARPi.

PARP1 signal pathway alteration
PARPi-induced downregulation of PARP1 signal and DNA trapping of PARP1 represent key elements of PARPi cytotoxic activity. Consequently, the structural alterations of PARP1, which disturb PARPi-induced inhibition and PARP trapping, lead to acquired resistance. The restoration of PARylation activity is also crucial for PARPi resistance.

Mutation of PARP1
The modification of drug targets is crucial in the development of therapeutic resistance. Specific mutations can reduce the affinity of PARP1 for PARPi or maintain its inherent catalytic activity. Through CRISPR screening in TNBC cells, several mutations within the PARP1, such as p.43delM and N329Q, impair the trapping function of PARPi and confer resistance [113]. Moreover, in a patient with OC harboring primary resistance to olaparib, the loss of trapping efficacy of PARPi due to the R591C mutation on PARP1 underscores the clinical significance of this resistance mechanism [113].

Downregulation of PARP1 trapping
Since the entrapment of PARP1 at SSBs is recognized as a key mechanism of PARPi, factors that influence the trapping process can lead to PARPi resistance. Notably, PARylation is the primary driver of PARP1 release from these lesions, driven by the electrostatic repulsion between the auto-modified PARP1 and histones [111]. Recent findings indicate that mutating the residues targeted by PARylation on PARP1 enhances trapping and increases cellular sensitivity to PARPi [190]. Additionally, histone PARylation has been implicated as a critical determinant of PARP1 dissociation, contributing to cellular resistance against PARPi [191]. Furthermore, PARylation is dynamic and reversible, governed by the rapid turnover mediated by PAR glycohydrolase (PARG), which degrades PAR chains. The inactivation of PARG has been shown to partially restore PAR chain formation, thereby mitigating PARP1 trapping and potentially leading to PARPi resistance [131]. Moreover, ARH3, another enzyme involved in PAR removal, also appears to contribute to PARPi resistance [132].
Despite the modification of PARPi, other factors are related to the DNA trapping of PARP1. Elevated levels of KAT6A have been associated with PARPi resistance in ovarian cancer, likely through stabilizing a KAT6A-PARP1–APEX1 complex, which reducing the amount of PARP1 trapping on DNA. Similarly, HMGB3, known for its ability to bind and bend DNA structures [133], also leads to PARPi resistance by interacting with PARP1 to reduce DNA trapping. The overexpression of HMGB3 is observed in high-grade serous ovarian carcinoma (HGSOC) and is related to decreased olaparib-induced DNA damage. Mechanistically, PARP1 is a binding partner of HMGB3, and this interaction is strengthened following DNA damage. The interplay between PARP1 and HMGB3 may alter the DNA-binding kinetics of PARP1, thereby reducing its trapping at the sites of DNA damage.

Restoration of the homologous recombination repair
HRD is the main indication of the utilization of PARPi, according to the synthetic lethality hypothesis that the accumulation of DSBs resulting from unrepaired SSBs results in cell death. BRCA1 and BRCA2 are key factors of HRR. BRCA1 binds to DSBs and facilitates the resection of DNA ends, and BRCA2 mediates the loading of RAD51 recombinase onto the ssDNA termini generated by resection, which is crucial for sister chromatid exchange during HR [192, 193]. Therefore, factors that affect the HRR determine the therapeutic efficacy of PARPi, including the reactivation of BRCA1/2, restoration of end resection, and upregulation of the recombinase RAD51. Additionally, the recruitment of HRR proteins to DNA break lesions and the modification and remodeling of chromatids are crucial elements impacting the efficacy of HRR.

Reactivation of BRCA1/2
The reconstitution of BRCA1/2 function via secondary mutation stands as the predominant mechanism underlying resistance to PARPi in clinical practice. The prevalence of this resistance mechanism varies across tumor types [134], occurring 10% − 40% of OC [99], 60% of metastatic BC [138], and up to 79% of CRPC [194]. Notably, frameshift deletions are crucial for BRCA2 functional restoration, with 60% of these deletions flanked by DNA microhomologies, which indicates the involvement of POLQ-mediated repair [134]. Moreover, splice site mutations in BRCA1 are also implicated in PARPi resistance. Approximately 30% of pathogenic gBRCA1m occur in exon 11. Hence, BRCA1 splice isoform △11 can confer resistance by excising the mutation-containing exon, yielding truncated, partially functional proteins [135].
In BRCA1/2 wild-type or heterozygous tumors, the re-expression of epigenetically silenced BRCA1/2 genes also contributes to PARPi resistance. A study utilizing patient-derived xenografts (PDXs) in HGSOC found that the hypermethylation of BRCA1 promoters predicts rucaparib sensitivity, whereas heterozygous methylation is linked to resistance [136]. Moreover, the loss of methylation in other HRR proteins’ promoters has also been identified as a mechanism leading to PARPi resistance [137].

Restoration of the end resection
The end resection process at DSBs is critical for HRR. The ssDNA generated through excision recruits RAD51 recombinase to facilitate HRR, while the 53BP1 and RIF1 proteins localized at DSBs recruit the Shieldin complex to inhibit end resection and promote non-homologous end-joining (NHEJ) [195, 196]. Consequently, restoring end resection is a crucial mechanism in BRCA1-deficient tumors.
The disruption of the 53BP1–RIF1-Shieldin axis has been shown to mitigate the impact of BRCA1 loss on genomic stability partially [138, 139]. Additionally, alterations in the regulatory factors of the 53BP1–RIF1-Shieldin complex can influence HRR activity. For instance, DYNLL1 induces the dimerization of 53BP1 and limits the end-resection activity of MRE11. The loss of DYNLL1 or its transcriptional activator, ATMIN, will enhance end resection in BRCA1-deficient cells, resulting in resistance to PARPi [141–143]. The TLK is crucial for the precise localization of 53BP1, and TLK deficiency induces PARPi resistance in TNBC and OC cell lines with BRCA1 mutations [144]. Meanwhile, factors that stimulate end resection, such as the TRIP13 ATPase, which disturbs the formation of the Shieldin complex, can also restore HR function and lead to PARPi resistance [145]. TIRR impairs 53BP1 function by obstructing its recruitment to DSBs, thereby reconstituting HR in BRCA1-mutant cells [146]. Additionally, the CST complex is downstream of the 53BP1–RIF1–Shieldin complex. The loss of CST components can restore the end resection and induce PARPi resistance in BRCA1-deficient cells [140].
Beyond the 53BP1–RIF1-Shieldin axis, other DDR proteins also affect the end resection process. Firstly, the phosphorylation of CtIP mediated by the Syk is a key signal of end resection. ATM kinase detects DSBs and can activate Syk and other DDR proteins responsible for recruiting Syk to DNA break lesions, thereby enhancing end-resection [147]. Secondly, ERCC6L2 and Ku’s complex are crucial components of NHEJ, inhibiting the assembly of the HRR complex and repairing end excision. The depletion of ERCC6L2 restores end resection [142], while the overexpression of miR-622 downregulates Ku complex expression and promotes HRR [148]. Similarly, CXorf56 disturbs the recruitment of Ku70 to DSBs by interacting with the Ku70 DNA-binding domain. The dissociation of Ku70 fosters the assembly of the RPA32/BRCA2/RAD51 complex at DNA lesions, thereby enhancing HRR in TNBC cells [149]. Thirdly, HELB inhibits end resection at DSBs by suppressing the activity of nucleases. The loss of HELB can partially restore HRR and confer resistance to PARPi [150]. Notably, none of these mechanisms leads to PARPi resistance in BRCA2-deficient cells, given the essential role of BRCA2 in downstream resection during HRR.

Upregulation of recombinase RAD51
The assembly of the RAD51-ssDNA complex is pivotal in HRR by scanning for homologous sequences and facilitating the invasion of DNA strands [193, 197, 198]. Recently, evidence has demonstrated that PARPi-induced DNA trapping of PARP1 disrupts the stability of RAD51 filaments [152]. Consequently, the upregulation of RAD51 function is intricately linked to PARPi resistance. For instance, the transcription factor KLF5 forms a regulatory complex with EHF and ELF3, which subsequently binds to the RAD51 promoter to augment its transcription [153]. This mechanism has been implicated in conferring PARPi resistance in OC. Additionally, the haploinsufficiency of ZNF251 is also linked to olaparib resistance, likely by enhancing RAD51-mediated HRR in BRCA1-mutated tumors [154]. The deletion of E2F7 has also been identified to upregulate RAD51, leading to resistance to PARPi in BRCA2-deficient cells [155, 156].

Remodeling and modification of the chromosome
Chromatin modification and remodeling are critical for the recruitment of HR factors. CHD1 is a nucleoprotein that modulates the accessibility of chromatin at DSBs. SOSTDC1 can prevent CHD1 degradation, potentially facilitating HRR by altering the chromatin landscape around DSBs [157]. Mechanistically, in response to DNA damage, SOSTDC1 translocates to the nucleus in an importin-α-dependent pathway and forms a complex with CHD1 to prevent its ubiquitination and degradation. Additionally, SOSTDC1 depletion compromises HRR and enhances the sensitivity of TNBC to olaparib in vivo.
Phosphorylation of histone H2AX to form γH2AX is another pivotal early event in DDR, marking DNA lesions and initiating HRR. Although ATM kinase has traditionally been regarded as the principal mediator of H2AX phosphorylation [158], emerging evidence suggests that STK39 is a novel facilitator of this essential signaling cascade [159]. ATM phosphorylates STK39 following DNA damage, potentiating its association with the Mre11-Rad50-Nbs1 (MRN) complex and subsequent chromatin recruitment. This process enables STK39 to augment H2AX phosphorylation, thereby enhancing HRR. Notably, STK39 is significantly overexpressed in pancreatic adenocarcinoma (PAAD) tissues. Additionally, the combination of STK39 inhibitor with PARPi demonstrates synergistic potential in inhibiting and reversing PAAD tumor growth.

Enhancement of tumor genomic stability
Genomic instability is the core element of PARPi-induced cytotoxicity in tumors with HRD. In addition to their conventional functions in HR, BRCA1/2 proteins are essential for preserving RF integrity during replicative stress resulting from PARPi [199, 200]. The nascent strands will be annealed on fork stalling to form a structure resembling a one-ended DSB. Without the protection of RAD51, several nucleases, such as MRE11 and Mus81, have been implicated in the degradation of nascent DNA at stalled forks [160, 163, 201]. Meanwhile, recent studies also highlight the involvement of PARP1 in the ligation of OFs [4], which are ssDNA segments formed during replication. Factors suppressing replication-associated ssDNA gaps have the potential to prevent the degradation of RF. Beyond the RF, the upregulation of other alternative DDR pathways can repair the damage and avoid cell death resulting from genomic instability.

Stabilization of RF
PARPi resistance has revealed distinct pathways in BRCA1- and BRCA2-deficient tumors, highlighting the critical role of RF stabilization [161, 164]. Inhibiting the recruitment of nucleases to stalled RF is the primary mechanism for RF stabilization. For instance, PTIP-MLL3/4 methyltransferase promotes MRE11 recruitment by modifying H3K4me1/3 at stalled RF and facilitating nucleosome remodeling factor CHD4 [160]. The loss of PTIP or CHD4 results in RF protection and resistance to PARPi in BRCA1/2-deficient cells [161]. Moreover, the UFMylation of PTIP by UFL1 is also crucial, as UFL1 loss or UFM1 protease UFSP2 overexpression confers resistance by protecting nascent DNA strands from degradation [162]. Similarly, EZH2 mediates MUS81 nuclease recruitment through H3K27me, contributing to RF degradation [163]. Disruption of the EZH2/MUS81 axis also promotes PARPi resistance in BRCA2-deficient cells [163].
Additionally, SNF2-family RF remodelers, such as SMARCAL1, ZRANB3, and HLTF, facilitate RF reversal, generating one-ended DSBs, subsequently initiating MRE11-dependent degradation of nascent DNA [164]. Inactivation of these remodelers can stabilize RFs in BRCA1/2-deficient cells [164]. Conversely, phosphorylation of H2AX promotes RF degradation by inhibiting CtIP-mediated protection in BRCA1/2-deficient tumors. H2AX depletion restores RF stability and confers PARPi resistance [165].
Fanconi anemia (FA) proteins collaborate with BRCA1/2 proteins to maintain genomic integrity through RF stabilization. For instance, FANCD2 is confirmed to limit RF progression and promote Polθ recruitment at DSBs to initiate alternative end-joining repair [202, 203]. However, FANCM has been revealed to confer PAPRi resistance by promoting SSB repair independently of RF protection. Mechanistically, FANCM counteracts 53BP1 to inhibit its function of ssDNA protection and facilitate the ligation of the ssDNA gap mediated by PRIMPOL complex [204]. FANCM depletion exacerbates DNA damage in the second S-phase by accumulating unrepaired SSBs, while 53BP1 deletion restores resection and mitigates PARPi sensitivity [204].

Upregulation of DNA damage response
For HR-deficient tumors, the upregulation of alternative DDR pathways acts as a compensatory mechanism to maintain genomic stability. In BRCA2-deficient cells, SMO is identified as a driver of PARPi resistance through GLI1-mediated upregulation of DDR factors, including HR pathway, such as the FANC family and BRCA1/2, as well as nucleotide excision repair [205]. Combined inhibition of PARP and GLI1 provides a potential synergistic strategy to overcome drug resistance [205]. In olaparib-resistant prostate cancer cell lines, IGFBP3 is identified to confer PARPi resistance through enhancement of NHEJ [166], which is an alternative repair pathway of DSBs and depends on DNA-PKcs and Ku70/80 heterodimer [167]. Mechanistically, IGFBP3 promotes the activation of the EGFR and DNA-PKcs [168]. The reduction of IGFBP3 increases DNA damage and ultimately sensitizes cells to olaparib [168]. Additionally, NLRP4 enhances DNA repair efficiency in response to olaparib, while complementing NLRP4 restores DNA repair capacity and results in olaparib resistance [170].
Analysis of ctDNA in patients undergoing olaparib maintenance therapy for OC revealed a strong association between the MRE11:p.K464R mutation and olaparib resistance [169]. As a critical component of the MRN complex, MRE11 promotes the recruitment of DDR factors to DSBs and the initiation of HRR [206, 207]. The MRE11:p.K464R mutation can inevitably disturb the MRN complex’s formation and function while augmenting its interaction with RAD50/RPS3 to enhance NHEJ [169]. Hence, this mutation effectively reduces DNA damage in HR-deficient tumors and confers resistance to olaparib.

Upregulation of drug efflux pumps
Upregulation of drug efflux pumps represents a common mechanism of drug resistance, and PARPi are no exception. Multidrug Resistance protein 1 (MDR1), which is encoded by the ABCB1 gene, plays a crucial role in resistance to various chemotherapeutic agents. This resistance is mediated by the efflux of drugs from the intracellular compartment, preventing their accumulation and effectiveness. Notably, recent studies have identified olaparib and rucaparib as substrates for MDR1, while veliparib and niraparib are relatively poor substrates for this transporter [208–212]. Experimental mouse models also illuminate that overexpression of ABCB1 can lead to resistance to olaparib, particularly after prolonged exposure [171]. Although the clinical implications of this mechanism in PARPi resistance remain unclear, the significance of ABCB1 upregulation through fusion with an enhanced promoter in chemotherapy resistance has been observed in OC [172]. This finding suggests that prior chemotherapy treatment may predispose patients to PARPi resistance. Additionally, the enhanced expression of other drug transporters, such as ABCC1 and ABCG2, has been implicated in talazoparib resistance [173], which can be effectively countered by combining their inhibitors [174].
Interestingly, PARP1 has been suggested to interact directly with the promoter of DOT1L, thereby stimulating transcription independently of its enzymatic activity [175]. The DNA trapping induced by PARPi further strengthens this interaction, leading to overexpression of DOT1L, which promotes H3K79me2 and mediates PLCG2 and ABCB1 transcription. Consequently, the PARP1–DOT1L-PLCG2/ABCB1 axis represents an important contributor to PARPi resistance. Moreover, combining DOT1L inhibitors with PARPi has exhibited synergistic cytotoxicity in both cell line-derived xenografts (CDXs) mouse models and patient-derived organoids (PDOs), underscoring a promising therapeutic strategy for overcoming ABCB1-mediated resistance to PARPi.

Inherent characteristics of tumors related to PARPi resistance
Hypoxia is a hallmark feature of the TME, which plays dual roles in PARPi resistance. Severe hypoxia disturbs the transcription of DDR proteins, while moderate hypoxia reduces the ROS-mediated DNA damage. Moreover, the prolonged use of PARPi triggers the accumulation of DNA fragments in the cytoplasm of tumors. The DNA sensor cGAS can detect cytoplasmic nucleic acid and trigger the translocation of STING from the endoplasmic reticulum to the Golgi apparatus [213, 214]. Subsequently, activated STING initiates the transcription of interferons and other cytokines by recruiting TBK1 to phosphorylate IRF3 [215, 216]. Emerging evidence suggests that hypoxia and inflammation may be intricately linked to cell survival and PARPi resistance.

Hypoxia and autophagy
The extent of hypoxia influences the efficacy of PARPi. Moderate hypoxia (2% oxygen) is revealed to promote PARPi resistance by reducing ROS-induced DNA damage regardless of HRR status [182]. In comparison, severe hypoxia (<0.5% oxygen) exhibits synergistic killing function with PARPi through disturbing DDR proteins transcription [183]. Moreover, the selective elimination of hypoxic tumor cells enhances the antitumor response compared with PARPi alone, without increasing toxicity in normal tissue [182].
Despite the DNA damage, ROS also triggers autophagy. Similarly, autophagy exerts dual effects on PARPi efficacy. The long non-coding RNA PART1 enhances sensitivity to PARPi in OC by ensuring the maintenance of mitophagy and mitochondrial quality [184]. Mechanistically, PART1 interacts with PHB2, a mitophagy receptor on the inner mitochondrial membrane [185]. This interaction can prevent PHB2 protein degradation, thereby facilitating PINK1-Parkin-dependent autophagy [186]. Notably, disruption of the PART1–PHB2 axis in preclinical models effectively confers resistance to PARPi through compromised mitochondrial quality control to induce cellular senescence [184]. On the other hand, another study reported that PINK1 overexpression contributes to PARPi resistance in prostate cancer [187]. NLRP4 has also been identified to promote olaparib resistance in PAAD by triggering mitochondrial ROS-driven autophagy, without affecting DNA damage [170].

Cell inflammation induced by the PARP inhibitor
The cytoplasmic accumulation of DNA segments induced by PARPi initiates an inflammatory response, which has been shown to promote tumor growth and drug resistance. In OC cells, high cGAS expression is significantly related to enhanced tolerance to olaparib, via the TBK1–IRF3 pathway, leading to type I interferon production [176]. This cascade further activates the NF-κB and IL-6-STAT3 pathways, promoting resistance to PARPi [176]. Additionally, Adora2b expression is upregulated in olaparib-resistant cells [177]. The overexpression of Adora2b mediated by NF-κB mediates resistance by sensing adenosine signals, thereby promoting tumor cell growth and migration via IL-6-STAT3 signaling [177]. Furthermore, other factors that influence NF-κB signal also result in PARPi resistance. PBK is implicated in the phosphorylation and nuclear translocation of TRIM37, thereby activating the NF-κB pathway [178]. PBK inhibition reverses resistance to olaparib [178]. Similarly, ATF3 also contributes to PARPi resistance, whereas its inhibition suppresses the NF-κB pathway and restores sensitivity [179]. Additionally, inflammatory signals can also promote tumor metastasis. For instance, the interferon-stimulated gene LY6E has been shown to confer PARPi resistance by shaping an immunosuppressive TME, which fosters progression and invasion [180]. LCP1 also results in olaparib resistance by activating the JAK2/STAT3 pathway and related epithelial-mesenchymal transition [181].
Interestingly, the suppression of inflammation is also linked to PARPi resistance. Low interferon-gamma response scores are demonstrated to correlate with poor responses to neoadjuvant PARPi therapy in BC [217]. Additionally, MiR-181a promotes PARPi resistance through the downregulation of the STING signal and pro-inflammatory cytokines. Extracellular vesicles can also transfer miR-181a to naïve cells and confer PARPi resistance [217]. The loss of IRF9 in BC cell lines is also associated with increased resistance to PARPi [218].

Other mechanisms of PARPi resistance
Some mechanisms, involving chromatin accessibility, cell cycle checkpoints, DNA damage response, among others, have been implicated in promoting PARPi resistance, although relevant research is limited. Firstly, PARPi-resistant tumors exhibit significant upregulation of CYP1B1 [219]. Mechanistically, CYP1B1 is associated with Linker histone H1.4, whose downregulation is associated with increased chromatin accessibility [220] and higher cell viability after olaparib treatment [219]. The loss of SLFN11 also compromises the ability to maintain a prolonged S-phase arrest following PARPi exposure, thereby contributing to the development of PARPi resistance [188, 189].
Additionally, the deubiquitination of BARD1 by USP15 is crucial for maintaining BRCA1 at DSB sites, leading to PARPi resistance [221]. CDK18 activates ATR by facilitating its interactions with Rad9 and ETAA1, promoting HRR and inducing resistance to PARPi [222]. Furthermore, the suppression of PARP activity has been shown to induce ferroptosis by downregulating glutathione biosynthesis, a process modulated by SLC7A11. Consequently, disruption of the ferroptosis signaling cascade can confer resistance to olaparib [223]. Polyploid giant cancer cells (PGCCs) can evade senescence through non-canonical modes of division, such as multipolar endomitosis [224]. They are formed via olaparib-induced endoreplication and generate mitotically competent daughter cells, contributing to acquired PARPi resistance [225]. Targeting PGCCs with mifepristone enhances the efficacy of olaparib in naïve tumors and reverses resistance in PARPi-refractory models [225].
To conclude, the emergence of resistance represents the major challenge in PARPi clinical utilization, resulting in disease progression and diminished survival benefits. Elucidating the mechanisms of PARPi resistance is the fundamental approach to broadening or refining their clinical application. In summary, restoring the HRR pathway is a primary mechanism of resistance to PARPi, as the repaired DSBs disrupt the synthetic lethality induced by PARPi. Back-in-frame reverse mutations of BRCA1/2 are frequently observed in patients undergoing PARPi treatment. Additionally, the enhancement of tumor genomic stability is a critical factor frequently observed in clinical settings. For example, the stabilization of RF and the compensatory upregulation of other DDR pathways significantly diminish the PARPi-induced cytotoxicity. Despite these DDR-related mechanisms, inherent tumor characteristics that upregulate survival signals, such as hypoxia and inflammation, play essential roles in PARPi resistance. However, due to the complexity of cellular signaling, these mechanisms are still under investigation and exhibit inconsistencies. Furthermore, alterations in PARP1 and related PARylation signaling pathways can also lead to PARPi resistance, although with a lower frequency than other mechanisms. Similarly, the upregulation of drug efflux pumps is a general resistance mechanism that also occurs with PARPi.
Nevertheless, some mechanistic studies focus on preclinical experiments, which may not fully reflect clinical conditions and can result in inconsistent conclusions. Moreover, investigating mechanisms alone cannot influence clinical therapeutic decisions for patients using PARPi, hindering the translation from mechanistic insights to clinical application. For instance, effective detection methods are needed to monitor PARPi sensitivity and identify the specific mechanisms responsible for resistance in individual patients. Research aimed at directly reversing PARPi resistance will be essential to establishing strategies for overcoming this challenge.

Measurements of PARPi resistance
Assessing the sensitivity of tumors to PARPi is essential for optimizing effective treatment strategies (Fig. 4). The cytotoxic effects of PARPi arise from the impaired repair of DSBs due to intrinsic HRD present in tumors. While pathogenic mutations in BRCA1 and BRCA2 have demonstrated promising predictive capabilities and are approved as companion tests for PARPi indications, the limited prevalence of gBRCA1/2m, approximately 1.5% for prostate cancer, 7% for breast cancer and pancreatic cancer, and 15% for ovarian cancer, has prompted investigations into PARPi sensitivity in patients exhibiting an HRD-like phenotype independent of BRCA1/2 mutations. Tumors harboring pathogenic alterations in other HRR genes, such as PALB2 and RAD51C/D, have shown sensitivity to PARPi therapy [226]. However, given the complexity inherent in HRD mechanisms and the dynamic nature of HRD status, precise pretreatment predictive approaches are necessary to refine patient selection processes [227–229]. Furthermore, acquired resistance frequently develops during PARPi treatment; thus, sensitive monitoring methodologies are essential for timely adjustments to therapeutic regimens. Current methodologies primarily focus on three key aspects: (i) Genomic profiling of HRR pathways: this entails detecting deleterious or suspected deleterious mutations in genes that are directly or indirectly involved in HRR pathways such as BRCA1/2, ATM, and CHEK2; (ii) HRD-related genomic features: this involves quantifying the accumulation of genomic scars and mutational signatures resulting from HRD; (iii) Functional HRR assays: this primarily involves the detection of RAD51 foci to evaluate real-time HRR activity dynamically.

Genomic profiling of HRR pathways
HRD may result from various alterations in the multistep HR process, including the recognition of DSBs mediated by the MRN complex, ATM-activated recruitment of BRCA1, and BRCA2-mediated template searching. In multiple cancer types, germline or somatic mutations in BRCA1/2 serve as primary biomarkers guiding the utilization of PARPi [230]. Clinical trials have shown that patients with BRCA1/2 mutations consistently derive substantial clinical benefit from PARPi treatments [33, 66–69]. Germline pathogenic variants not only inform therapeutic and prognostic decisions but also facilitate familial risk stratification due to their heritable nature. Tumors may also acquire somatic BRCA1/2 mutations during tumorigenesis and progression [34, 53, 231, 232]. Consequently, established clinical guidelines recommend BRCA1/2 testing in all newly diagnosed patients with non-mucinous ovarian cancer [233, 234]. However, the predictive value of BRCA mutations for PARPi sensitivity remains incomplete; a subset of BRCA wild-type tumors also exhibits favorable responses to PARPi therapy [235]. This highlights the necessity to refine predictive biomarkers further to optimize patient selection for PARPi treatment.
Beyond BRCA1/2 mutations, HRD can arise from germline or somatic mutations in other genes associated with HRR, including RAD51C, RAD51D, PALB2, and BRIP1 [234]. Theoretically, HRD underlies the principle of synthetic lethality that PARPi exploit. However, clinical evidence suggests that the utility of non-BRCA HRR gene mutations for predicting responses to PARPi remains limited. The phase 2 CLIO trial showed that the predictive reliability of non-BRCA HRR mutations is inadequate in first-line PARPi treatment for OC [236]. Only 2 of 10 patients harboring RAD51 mutations exhibited confirmed responses, while no responses were observed in other non-BRCA HRR mutation carriers [236]. Conversely, other studies indicate potential predictive value of non-BRCA HRR mutations regarding response to PARPi. A post-hoc analysis of the phase 2 clinical trial Study-19 identified 21 OC patients with mutations in non-BRCA DDR genes, such as BRIP1, CDK12, RAD54L, and RAD51B, who derived benefits comparable to those with BRCA1/2 mutations [237]. Similarly, additional research has reported partial responses to rucaparib among 4 out of 5 OC patients harboring RAD51C mutations [231]. These conflicting findings may arise from challenges in distinguishing pathogenic mutations from benign single-nucleotide polymorphisms or variants of unknown significance. Therefore, HRR mutations may not accurately reflect HRD status, which is critical for predicting PARPi response. Moreover, the rarity of non-BRCA HRR mutations often leads researchers to categorize them collectively, despite differing predictive abilities among specific mutations [235]. Thus, standardized genomic classification and functional validation of HRD are imperative to resolve these discrepancies and advance precision PARPi therapy.
Although the impact of deleterious BRCA1/2 mutations on the efficacy of PARPi is well established, the clinical significance of epigenetic modifications in HRR genes is still under investigation. Methylation of the promoters of BRCA1 and RAD51C diminishes gene expression, thereby inducing HRD [238]. However, clinical studies that incorporate HRR gene methylation into predictions regarding responses to PARPi have produced conflicting results [231, 235, 239, 240], potentially limited by current detection methods for homozygous hypermethylation [238, 239]. The loss of methylation in a single BRCA1 copy can restore HRR functionality and confer resistance to PARPi [136], suggesting that high-precision quantification of methylation levels may enhance predictive accuracy. For example, utilizing quantitative droplet digital PCR to measure BRCA1 promoter methylation levels in OC patients, a study found that homozygous methylation closely correlates with genomic instability and HRD status [241]. This study defines high degrees of methylation (>70%) as indicative of homozygous silencing, and identified 15 patients with homozygous BRCA1 methylation from 79 BRCA wild-type OC patients [241]. Consequently, accurate testing for epigenetic modification holds significant promise for expanding the clinical application of PARPi treatments, which warrants rigorous validation through further clinical studies across diverse tumor types.

HRD-related genomic features
Accumulation of DNA alterations results from defects in the DNA repair machinery, while the mutational features of HRD can provide insight into HRD status at the phenotypic level. Tumors harboring HRD are characterized by genomic instability, including copy number variations (CNVs), insertions and deletions, single base substitutions, and structural variants (SVs) [235]. Notably, large-scale genomic defects arising from impaired DNA repair mechanisms, also called ‘genomic scars’, can be classified into three signatures of HRD-related genomic alterations: loss of heterozygosity (LOH), telomeric allelic imbalance (TAI), and large-scale state transitions (LST). LOH is characterized by the irreversible loss of one parental allele at a specific chromosomal locus. The length of the LOH and the percentage of genomic LOH serve as indicators of HRD status; specifically, an LOH size exceeding 15 Mb and a genomic LOH percentage greater than 14% are significantly associated with HRD status [231, 242]. TAI is defined as the number of subtelomeric regions exhibiting allelic imbalance (either copy loss or copy gain) without crossing the centromere, while the quantity of TAI is correlated with impaired DNA DSBs repair mechanisms. LST refers to chromosomal breaks occurring between adjacent regions larger than 10 Mb, which may include deletions, inversions, and translocations. Tumors are classified as HRD-positive if they exhibit at least 15 LSTs when classified as near-diploid, or more than 20 LSTs if classified as near-tetraploid [243]. Based on the superior efficacy of combinational prediction, the Genomic Instability Score (GIS) was calculated as the unweighted sum of LOH, TAI, and LST. This composite score has demonstrated a performance that surpassed that of each individual signature when utilized in isolation.
Moreover, different HRD testing varies in the specific genomic features measured and the methodologies used to establish HRD-defining thresholds [234]. This heterogeneity limits their validation across different tumor types and clinical settings. Consequently, the European Society for Medical Oncology recommends concurrent assessment for BRCA mutations and HRD status where feasible, to facilitate therapeutic decision-making [234]. Currently, two FDA-approved commercial assays based on the genomic scars are utilized for evaluating HRD status and predicting PARP inhibitor response: Myriad’s MyChoice CDx and FoundationOne CDx [244]. Myriad’s MyChoice CDx quantifies genomic instability using GIS [245] and defines HRD status by either a deleterious BRCA1/2 mutation or a GIS exceeding a threshold of 42, as established in trials such as PAOLA-1 [50]. Alternatively, FoundationOne CDx BRCA LOH detects BRCA1/2 mutations and evaluates the percentage of the genome affected by LOH. Tumors with ≥ 16% LOH are classified as “LOH-high”, a metric validated in the ARIEL3 trial [34]. However, the two genomic scar assays are limited in identifying BRCA wild-type patients who will benefit from PARPi therapy. A retrospective analysis of Study 19 indicated that although GIS moderately stratifies the BRCA1/2 wild-type population into higher and lower benefit groups, it does not adequately define an HRR-proficient subgroup that derives no benefit from PARPi [237]. Beyond the two HRD tests, an additional machine learning algorithm that integrates BRCA1/2 status and CNV signatures, referred to as HR-SC, has been developed by utilizing samples from OC patients recruited in two international clinical trials [246]. HR-SC demonstrates low failure rates of 6.4% in PAOLA-1 and 4% for MITO16A, suggesting clinical feasibility for predicting HRR status. Furthermore, another model based on a convolutional neural network, termed ‘GIInger’, which calculates the genomic instability index (GII) from low-pass whole-genome sequencing data, exhibits comparable predictive capabilities for HRD status [247]. Notably, the GIInger demonstrates a remarkable ability to differentiate among BRCA wild-type patients [247]. Individuals identified as HRD-positive by GIInger from BRCA1/2 wild-type participants in the PAOLA-1 trial exhibited significantly prolonged PFS following treatment with PARPi. Moreover, DirectHRD, a genomic scar-based classifier for HRD derived from whole-genome sequencing (WGS), exhibits the sensitivity necessary for identifying HRD scars in samples with low tumor purity, particularly in the context of liquid biopsies [248]. DirectHRD exclusively employs a highly specific type of HRD scar, small deletions accompanied by microhomology, and integrates these elements within a probabilistic framework. In a cohort of 90 ctDNA samples, DirectHRD attained an area under the curve of 0.87 and demonstrated efficacy in detecting HRD at tumor fractions as low as 1% [248].
In addition to large genomic rearrangements, specific mutational signatures exhibit significant promise in evaluating HRD status. The single-base substitution signature 3 (SBS3) is identified through analysis of base-pair change profiles in tumor samples derived from targeted gene panels and comprises a pattern of 96 possible SBS types. SBS3 is strongly associated with germline mutations in BRCA1/2, as well as other known HRR genes such as PALB2 and RAD51 [249]. Patients who do not harbor mutations in recognized HRR genes but display SBS3 may also benefit from therapy that targets the selective vulnerability inherent to HRD tumors [249]. For instance, a high burden of SBS3 has been correlated with increased sensitivity to PARPi treatment in BC organoids [250].
However, the relatively ambiguous nature of the signature concept presents a challenge in establishing a definitive diagnostic threshold for HRD status, thereby limiting its clinical application [235]. Additionally, targeted gene panels often lack specificity and can lead to false-positive results. In contrast, while WGS boasts high accuracy, it largely remains confined to research settings due to the associated high costs and logistical challenges.
To address these limitations, whole-exome sequencing (WES) data integrated with machine learning techniques show substantial potential for HRD assessment and are becoming more accessible in clinical oncology. A prime example of this approach is the HRDetect algorithm, which was specifically developed to identify BRCA1/2 deficiencies [251]. This algorithm synthesizes information across all four mutation classes and strategically assigns differential weightings to six distinct genomic features. These features include indels such as microhomology-mediated deletions, SBSs encompassing Signatures 3 and 8, SVs including rearrangement Signatures 3 and 5, as well as CNVs, notably the HRD score derived from Myriad’s MyChoice HRD test [251]. By employing a probabilistic threshold of 70%, the HRDetect assay demonstrated significantly superior performance compared to established genomic scar measures such as the GIS [235, 251].
Another tool, HRProfiler, developed from WES data, enables robust detection and quantification of HRD-associated mutational signatures [252]. By utilizing a predefined signature specific to HRR, HRProfiler employs advanced computational methods to evaluate the relative contribution of this HRR-deficient signature within the overall mutational landscape of tumor samples. This quantitative assessment is especially beneficial for predicting tumor response to PARPi therapy, particularly in cases where conventional biomarkers such as BRCA1/2 mutations are lacking.
Nevertheless, an inherent limitation of all genomic signature or scar assays is their reflection of genomic mutations indicative of prior HRD status. Consequently, these assays do not provide direct insights into the current functional state of the HRR pathway, as HRR function may be restored through various mechanisms.

Functional assays to assess HRD status
Functional assays offer a dynamic readout of the actual, current status of the HRR pathway by measuring the capability of cells to repair DSBs. Consequently, these assays present significant potential for monitoring both primary and acquired resistance to PARPi, independent of the underlying mechanisms that restore the HRR pathway. The RAD51 foci formation assay is a widely utilized experimental approach for estimating subnuclear RAD51 levels through immunostaining of formalin-fixed, paraffin-embedded (FFPE) sections, typically conducted without external induction of DNA damage [253–255]. RAD51 protein is recruited to DSB sites by the BRCA1/PALB2/BRCA2 complex, and the presence of RAD51 foci reflects activation within the HRR machinery. The HRD status derived from this assay has been shown to correlate with responses to PARPi therapy, indicating considerable promise in predicting patient sensitivity to treatment. A study utilizing breast cancer PDXs demonstrated that the RAD51 score was highly discriminative between PARPi sensitivity and resistance, notably outperforming traditional genomic tests [254]. Furthermore, in patients undergoing neoadjuvant chemotherapy, applying genomic scar-based HRD assessments such as Myriad’s Genomic Instability Score may be limited due to low tumor cellularity (<30%) [255]. In such scenarios, a functional HRR assay based on RAD51 foci formation can effectively determine HRD status and predict overall survival following PARPi treatment in recurrent ovarian cancer settings [255]. A preclinical validation study confirmed that assessing RAD51 foci exhibited superior performance compared to mutation detection and genomic-based HRD analysis. PDX models representing various tumor types, including 103 breast cancers, 4 ovarian cancers, and 2 pancreatic cancers, were employed to reproduce patient-specific HRD statuses and treatment responses; this validated a cut-off for the RAD51 score at 10%. The accuracy rates for predicting PARPi response were found to be 95% for the RAD51 test compared with 67% for HRR gene mutations and 71% for genomic HRD analysis, respectively [256]. Additionally, the presence of γH2AX foci and 53BP1 foci serves as a valuable indicator of DSBs. In contrast, BRCA1 foci have demonstrated potential predictive value for tumor response to olaparib in ovarian cancer PDX models [257].
To note, the presence of phospho-RPA2–T21 (p-RPA2) foci as a biomarker for replication stress demonstrates predictive capability in response to PARPi therapy among HR-proficient ovarian cancer [258]. Immunofluorescence analysis of p-RPA2 foci was conducted on FFPE high-grade serous carcinoma samples. These samples were classified as p-RPA2-high if more than 16% of the cells exhibited two or more p-RPA2 foci according to automated analysis [258]. The p-RPA2 assay effectively predicted survival outcomes in patients receiving PARPi treatment, emphasizing the significance of incorporating replication stress markers such as p-RPA2 alongside HRD status during therapeutic planning.
Nevertheless, the routine clinical application of the functional foci formation assay remains uncertain due to technical discrepancies, such as variations in detection methods and differences in foci determination [243]. Currently, no commercial RAD51 foci assays are recommended for clinical use; instead, this technique is employed in clinical trials that predominantly involving ex vivo induction of DNA damage, which limits its practical utility in standard clinical settings. To note, for tumors characterized by low proliferation rates, determining HRD status through the RAD51 formation assay proves impractical [259], leading to a high failure rate [260]. Furthermore, it is noteworthy that the presence of RAD51 foci can persist even when tumors exhibit HRD as a result of ATM alterations [259].

Other approaches for predicting PARPi sensitivity
Beyond conventional tumor biopsies, emerging strategies such as liquid biopsy (utilizing ctDNA) and PDOs have demonstrated substantial potential in predicting PARPi sensitivity [102, 261]. Notably, ctDNA analysis provides a non-invasive method to monitor acquired PARPi resistance by detecting molecular mechanisms that restore HRR functionality. For instance, a longitudinal study collected serial whole-blood samples from patients with BRCA1/2m OC prior to the administration of PARPi and then serially every 3 months until disease progression [102]. ctDNA extracted from these samples underwent sequencing with a 531-gene panel encompassing key resistance mechanisms [102]. Continuous monitoring of ctDNA revealed multiple acquired resistance pathways, yielding critical insights for the design of post-progression therapies and enhancements in survival outcomes. Therefore, predictive models that integrate resistance mechanisms identified through sequential ctDNA monitoring may serve as early indicators of impending disease progression. However, it is important to consider the risk of false-negative results arising from low levels of ctDNA, as well as the potential for false-positive results resulting from clonal hematopoiesis of indeterminate potential.
Furthermore, PDOs enable rapid functional testing for PARPi sensitivity [261–263]. A study developed a platform to functionally profile DNA repair in short-term OC PDOs [261]. Results showed that regardless of DNA repair gene mutational status, functional HR deficiency in organoids associated with PARPi response [261]. Functional analysis of other epithelial OC PDOs also identified specific resistance mechanisms in a patient with progressive disease, revealing potential strategies to overcome PARPi resistance [262]. This highlights how functional organoid assays complement genomic analysis in refining PARPi sensitivity predictions, though clinical validation remains essential. Additionally, short-term PDOs function as highly effective in vitro models for the direct assessment of drug resistance and the identification of potent therapeutic combinations, as they faithfully replicate primary tumors’ phenotypic and genetic features [264].
Interestingly, a panel of ribosomal genes has been identified as predicting a novel synthetic lethal interaction in HR-proficient patients [226]. PARPi selectively eliminates cells with high expression of eight panel genes, mediated by ribosomal stress-induced apoptosis via DNA damage signaling [226]. This phenomenon, alongside ATM signaling-induced pro-survival HRR, constitutes a novel model governing cell fate upon DNA damage. Consequently, evaluating the ribosomal gene panel and HR status could enhance the precision of predicting clinical responses to PARPi in first-line maintenance settings, validated by a cohort of 58 OC patients [226]. All PARPi-sensitive patients within this cohort had either an HR-deficiency or relatively high expression of genes in this ribosomal panel. This panel may therefore identify PARPi-benefiting patients lacking HRD. Furthermore, melanoma cells characterized by high basal PARP1 expression demonstrate significantly increased cell death following treatment with PARPi. This phenomenon is attributed to enhanced PARP1 trapping in comparison to cells exhibiting low levels of PARP1 expression [265]. PARP1 expression may thus serve as a predictive biomarker, particularly relevant for late-stage metastatic melanoma patients who exhibit heightened PARPi sensitivity due to elevated PARP1 [265]. However, the specific threshold of PARP1 expression required to develop a clinical predictive tool of PARPi sensitivity remains undefined, representing a critical knowledge gap in clinical translation. Furthermore, the relationship between PARP1 expression and PARPi efficacy necessitates additional validation across various cancer types, particularly in tumors with FDA-approved indications for PARPi use.
Additionally, due to the complex and varied mechanisms underlying the action of PARP inhibitors and the emergence of drug resistance, leveraging deep learning or other artificial intelligence (AI) models for efficacy prediction may serve as a valuable supplementary tool in identifying potential beneficiaries. Notably, morphological features observed in hematoxylin and eosin (H&E)-stained pathological images, including high tumor cell density, prominent nucleoli, tissue necrosis, distinctive laminated fibrosis, and evidence of tumor infiltration, serve as critical discriminative markers between HRD and HR-proficiency [266]. Given the substantial time and financial resources associated with conventional molecular testing methods, readily available H&E-stained slides present a promising alternative for predicting HRD status in clinical practice. AI-based approaches have increasingly been employed to automatically extract diagnostic features from whole-slide images (WSIs) for HRD prediction; however, the accuracy of existing models remains suboptimal [266–271]. While conventional attention-based multiple instance learning methods often fail to capture the global context of WSIs, recent advances have demonstrated improved performance. For instance, the Sufficient and Representative Transformer algorithm achieved superior HRD prediction in breast cancer patients (AUC: 0.887 ± 0.034) [272]. Similarly, a deep learning-based classifier, termed the Ensemble model, is designed for ovarian cancer to effectively stratify HRD status by analyzing tumor regions in WSIs (AUC: 0.769) [273]. Nevertheless, most current models primarily rely on HRD scores and do not comprehensively assess PARPi sensitivity or the associated survival benefits. Future AI-driven pathological prediction models that directly incorporate PARPi responsiveness as an endpoint could significantly enhance clinical utility by reducing costs and improving feasibility. Furthermore, due to the black-box nature inherent in artificial intelligence research, the biological interpretability of the predicted features involved in the PARPi resistance process necessitates further investigation. Additionally, its actual implementation in clinical practice will require an extended period of exploration and validation.
To conclude, evaluating PARPi sensitivity and monitoring acquired resistance necessitates a comprehensive assessment of HRD status, encompassing underlying gene mutations and modifications, HRD-related genomic features, and functional assays. Regarding primary resistance, deleterious BRCA1/2m and other HR gene mutations currently serve as the primary clinical indicators for PARPi utilization, as endorsed by treatment guidelines. However, the clinical sensitivity and specificity of relying solely on these mutations are limited. Therefore, standardized genomic classification and functional validation of HRD status are imperative for advancing precision PARPi therapy. Furthermore, predictive models focusing on genomic scars indicative of lost HR function, such as Myriad’s MyChoice CDx, FoundationOne CDx, and GIInger, show promise in identifying BRCA1/2 wild-type patients who may benefit from PARPi treatment. Additionally, genomic mutational signatures associated with HRD, such as HRDetect, which is assessed via whole-genome sequencing, demonstrate superior predictive performance compared to genomic scars alone. However, a key limitation of both genomic scars and mutational signatures is their reflection of historical HRD events, which hinders their utility for real-time evaluation of PARPi sensitivity. Notably, functional assays offer a dynamic assessment of HR pathway activity. Those measuring RAD51 foci in easily accessible FFPE samples appear particularly feasible. Moreover, RAD51 foci analysis exhibits superior predictive ability compared to assessments based solely on HR gene mutations or genomic HRD analysis. Nonetheless, the clinical utility of RAD51 foci assessment requires further validation through large-scale clinical trials.
Monitoring acquired PARPi resistance shares fundamental principles with primary resistance assessment. A key advantage lies in ctDNA analysis, as liquid biopsies offer a non-invasive and convenient method for detecting underlying resistance mechanisms. Serial ctDNA monitoring thus holds significant promise for prospectively predicting acquired resistance and guiding the development of counteractive therapeutic strategies. However, the clinical adoption of ctDNA-based monitoring is hindered by challenges related to accessibility and economic cost, with the cost-effectiveness of these biomarkers remaining unclear.
Consequently, current clinical studies predominantly rely on disease progression or recurrence as markers of PARPi resistance. Such reliance reflects the lack of validated dynamic biomarkers. This reliance often limits the design of timely interventions and can impact patient survival outcomes. To address this, several combination strategies are being explored to reduce the risk of acquired resistance and improve patient survival.

Strategies to overcome PARPi resistance
Elucidating the mechanisms underlying PARPi resistance is fundamental to developing effective strategies to overcome it. Several therapeutic strategies have been designed to exploit emergent vulnerabilities associated with underlying mechanisms to reverse PARPi resistance. The primary strategies focus on three key aspects corresponding to resistance mechanisms: targeting alterations in the PARP1 signaling pathway, inhibiting the restoration of HRR, and reducing genomic stability (Fig. 5). Concurrently, combining PARPi with other therapeutic modalities, including antiangiogenic agents, chemotherapy, and ICIs, enhances cytotoxicity and mitigates resistance development (Table S4–S6). Furthermore, developing next-generation PARPi, featuring unique structural modifications and alternative mechanisms of action, such as more effective PARP-DNA complex trapping or targeting alternative DNA repair pathways, holds promise for circumventing existing resistance mechanisms and improving long-term clinical outcomes (Table 2).

Targeting acquired vulnerability
Specific mechanisms underlying PARPi resistance, notably alterations in PARP signaling and the restoration of HRR, can induce acquired vulnerabilities that present opportunities for re-sensitization [274]. For instance, in BRCA2-deficient tumor cells, the loss of PARG activity increases PARylation signaling, conferring PARPi resistance [131]. However, this specific context renders the cells susceptible to alternative therapeutic strategies. Inhibition of EXO1/FEN1 in these cells leads to the accumulation of unresolved OFs, which persist as ssDNA gaps and prove selectively lethal to these resistant cells [275]. These findings underscore that targeting the specific molecular weaknesses acquired during resistance development holds considerable promise for overcoming PARPi resistance and improving therapeutic outcomes.

Switching to an alternative PARPi
The distinct chemical structures and pharmacological effects of different PARPi suggest that switching to an alternative PARPi may offer a promising strategy to overcome acquired resistance [276]. While olaparib and rucaparib are identified as MDR1 substrates, veliparib and niraparib demonstrate limited affinity for this efflux transporter [208–212]. If resistance arises from P-gp overexpression, which reduces intracellular drug concentrations, transitioning to PARPi agents unaffected by this transport mechanism may reverse resistance and improve clinical outcomes [277]. However, given the uncertain clinical implications of drug efflux upregulation and other potential resistance mechanisms underlying PARPi resistance, the “PARPi-switching” strategy may have limited long-term efficacy in clinical settings.
Meanwhile, developing next-generation PARPi represents a promising approach to expand therapeutic options and delay resistance onset. A notable example is AZD5305, a novel PARPi with high selectivity for PARP1. Compared to conventional PARPi, AZD5305 exhibits optimized target specificity (minimal PARP2 interaction), improved efficacy metrics, and a favorable safety profile. In HRD PDX models, AZD5305 alone or with carboplatin or ceralasertib showed potent antitumor activity and delayed resistance [278]. Early clinical validation comes from the phase 1/2a PETRA trial, where AZD5305 achieved a 28% objective response rate in PARPi-pretreated patients with HRD-associated metastatic malignancies spanning ovarian, breast, pancreatic, and prostate carcinomas [279].
Additionally, PARP7 inhibitors have been shown to restore type 1 interferon signaling responses, providing a new path for cancer therapy [280]. Although no specific PARP7 inhibitors are approved, recent advances highlight novel compounds and formulations with potent PARP7 inhibitory activity [280]. Notably, thioparib, a next-generation PARPi with high affinity for PARP1, PARP2, and PARP7, demonstrates robust antitumor activity against both PARPi-sensitive and resistant HRD cells in vitro and vivo [281]. Furthermore, two novel anti-resistance PARPi, YCH1899 and B1, have been identified in PARPi-resistant cell lines [282, 283]. YCH1899 maintains cytotoxicity in PARPi-resistant cells with restored BRCA1/2 function or 53BP1 loss, exhibits acceptable pharmacokinetic properties in rats, and shows dose-dependent antitumor activity in olaparib- and talazoparib-resistant xenograft models [282]. The compound B1, a novel 4-hydroxyquinazoline derivative, exhibits superior cytotoxicity in primary PARPi-resistant cells, dose-dependently suppresses intracellular PAR formation, and enhances γH2AX aggregation [283]. Mechanistic studies reveal that B1 induces intracellular ROS formation and mitochondrial membrane depolarization, thereby increasing apoptosis and cytotoxicity [283]. These potent PARPi that overcome resistance warrant further investigation.
Notably, the natural product nimbolide represents a distinct strategy for inducing PARP1 trapping while concurrently promoting the recruitment of PARylation-dependent DNA repair factors [284]. Unlike conventional PARPi, nimbolide inhibits the E3 ligase activity of RNF114, a protein recruited to DNA lesions that targets PARP1 for degradation upon genotoxic stress sensing [284]. The blockade of this degradation pathway by nimbolide prevents the removal of PARP1 from DNA lesions, resulting in significant PARP1 trapping. Furthermore, nimbolide demonstrates synthetic lethality with BRCA1/2m cells, effectively overcoming both intrinsic and acquired resistance to PARPi in both in vitro and in vivo models [284]. This unique mechanism underscores the potential of nimbolide as a promising therapeutic agent to enhance the efficacy of current PARPi strategies.

Inhibiting the HRR pathway
The restoration of HRR is a key mechanism underlying the acquired resistance of PARPi. Strategies to downregulate HRR pathways are therefore promising for sensitizing tumors to PARPi. Firstly, approaches to reduce the expression of HR genes demonstrate the potential. In prostate cancer, androgen receptor antagonists can suppress the expression of HR genes. Combining PARPi with androgen receptor inhibitors is a potential approach to overcome resistance [285]. Given the critical role of histone deacetylases (HDACs) in regulating the expression of HRR-related genes and promoting the accurate assembly of HRR-directed sub-nuclear foci, HDAC inhibitors have emerged as a promising agent to enhance the antitumor effect of PARPi by blocking the HRR pathway. HDAC inhibition has been shown to increase DNA damage and enhance PARPi sensitivity in PTEN-positive TNBC and prostate cancer [286, 287]. Secondly, other strategies disrupting HRR signaling also show promise. For instance, GCH1 is an HRR-associated gene aberrantly enriched in BC and OC following niraparib treatment via JAK-STAT signaling. Flow cytometry in vitro and PDX models in vivo have confirmed that GCH1 inhibitors significantly enhance PARPi-mediated tumor killing [288]. Furthermore, PLK1 is a master regulator of mitosis, which enhances HRR and promotes G2/M checkpoint recovery. Combined PARPi with PLK1 inhibitor has resulted in superior tumor growth inhibition and significantly prolongs survival in olaparib-resistant PDXs with BRCA1m [289]. PFKFB3 is a pivotal regulator of glycolysis, whose activity is significantly elevated in OC tissues and associated with PARPi resistance [290]. Mechanistically, PFKFB3 interact with replication protein A3 involved in HRR. The enhanced cytotoxicity of the PFKFB3 inhibitor is validated by an in vitro study, where both genetic and pharmacological inhibition of PFKFB3 sensitized OC cells to PARPi, impairing HR repair and increasing DNA damage [290]. Furthermore, the feasibility of a combined PFKFB3 inhibitor with olaparib is supported by the in vivo evidence, where this strategy significantly inhibits tumor growth and induces apoptosis in OC xenografts without exacerbating adverse effects [290]. Meanwhile, a genome-wide CRISPR knockout screen has identified UBA1 as a target to overcome the primary PARPi resistance in HR-proficient OC cells. Both silencing and pharmacological inhibition of UBA1 impede HRR. In vivo, PDXs have confirmed the survival benefit and safety of combining PARP and UBA1 inhibition, establishing it as a promising strategy to extend PARPi indication [291].
Additional approaches targeting the RAD51-mediated process also effectively overcome PARPi resistance. DDB2 is a DNA damage recognition factor, and its reduction delays HRR by triggering RAD51 proteasomal degradation, which increases cellular sensitivity to PARPi in TNBC [292]. Meanwhile, in prostate cancer, deletion of MMS22L hypersensitized cells to PARPi by disrupting RAD51 loading through a TP53-dependent manner [155]. Conversely, PG545 reduces RAD51 expression in an autophagy-dependent manner [293]. Interestingly, PG545 can also inhibit the endocytosis of DEK, a heparan-sulfate proteoglycan-interacting DNA repair protein, sequestering it in the TME. Thus, nuclear DEK loss further impairs HRR, and PG545 synergizes with PARPi in OC cells with either primary or acquired resistance [293].
Interestingly, 2-HG, which is traditionally recognized as an oncometabolite, paradoxically induces HRR defects and render tumor cells susceptible to PARPi [294]. Liposomes co-loaded with veliparib, and 2-HG exhibit enhanced antitumor activity in both PARPi-resistant BRCA1/2m and wild-type tumors. Moreover, this co-delivery approach has been shown to augment cytotoxic T-cell function by activating the STING pathway and downregulating PD-L1 expression through 2-HG-induced hypermethylation [294].

Upregulating genomic instability
Several strategies to sensitize tumor cells to PARPi have emerged by disturbing genomic stability. Enhancing DNA damage or downregulating DDR factors both upregulate genomic instability, thereby bypassing the restoration of HRR, and overcoming PARPi resistance. In addition to chemotherapy and radiotherapy, other approaches to induce DNA damage overload restored HRR or enhanced DDR processes, subsequently reversing PARPi resistance. DNPH1 is a nucleotide salvage factor, and its inhibition enhances PARPi sensitivity by promoting the accumulation of cytotoxic 5-hydroxymethyl-deoxyuridine (hmdU). Subsequently, SMUG1-dependent excision of genomic hmdU induces RF collapse, DNA break formation, and apoptosis [295]. The efficacy of hmdU and DNPH1 inhibition has also been shown to reverse PARPi resistance in BRCA1-deficient tumors, underscoring the critical role of genomic hmdU in determining PARPi efficacy [295]. Furthermore, the upregulation of replication stress presents additional targets for reversing PARPi resistance. For instance, the combination of ATR and AKT inhibitors induces R loop-mediated replication stress by disrupting the AKT1–DHX9 interaction, which is upregulated to solve the replication stress in PARPi-resistant BRCA1/2m OC [296]. This strategy is further confirmed to reduce tumor growth and prolong survival in PARPi-resistant OC animal models [296].
The accumulation of ssDNA gaps is also lethal for PARPi-resistant cancer cells. For instance, DNA polymerase epsilon (Polε) is a critical replication factor for activating CMG helicase and synthesizing leading strands. The loss of the accessory Polε subunits POLE3 and POLE4 has been shown to sensitize tumors markedly to PARPi, through the accumulation of replication-associated DNA gaps formed during the repriming process mediated by PRIMPOL [297]. Interestingly, the sensitization effect observed in POLE3/POLE4 knockout cells persists without 53BP1, a factor often implicated in PARPi resistance [297]. Although this strategy has demonstrated strong efficacy against key resistance mechanisms, its clinical translation requires further validation in preclinical in vivo models. Moreover, the CRISPR screen demonstrates that OC and BC cell lines, which harbor HR defects, highly depend on the activity of USP1 [298]. The efficacy of USP1 inhibitors in HRD tumor models is correlated with ssDNA gaps and RF instability [298, 299]. Adding USP1 inhibitors to PARPi also exhibits significant synthetic lethality in PARPi-resistant tumors, by promoting ssDNA accumulation [299]. For instance, combining PARPi with KSQ-4279, the first selective USP inhibitor, was well-tolerated and inhibited tumor growth across multiple patient-derived PARPi-resistant models [298]. Meanwhile, USP1 inhibition also regulates PARP1 signaling by removing its K63-linked polyubiquitination, subsequently modifying PARP1 DNA trapping and PARylation activity [300]. Hence, combined blockade of USP1 and PARP1 is a compelling therapeutic strategy for reversing PARPi resistance [300].
Recently, ruthenium polypyridyl complexes (RPCs) have emerged as promising antitumor candidates due to their DNA binding properties, which enable RPCs to interfere with DNA replication or transcription in distinct mechanisms compared to cisplatin [301, 302]. Combining specific RPCs and PARPi demonstrates efficacy in olaparib-resistant BRCA1/2 wild-type BC cell lines via enhanced DNA damage and resultant apoptosis [302]. While enhanced DDR activity is a pivotal mechanism driving resistance to PARPi, combining DDR inhibitors with PARPi offer a promising strategy to overcome resistance. Indisulam acts as a molecular glue, facilitating the interaction between the splicing factor RBM39 and the E3 ubiquitin ligase DCAF15 [303]. This interaction leads to RBM39 polyubiquitination, degradation, and subsequent RNA splicing defects. Notably, RBM39 loss induces splicing alterations in critical DDR genes in OC, thereby increasing sensitivity to various PARPi. Co-administration of indisulam significantly enhances the efficacy of olaparib in mouse models bearing PARPi-resistant tumors [303].

Combining PARPi with conventional therapeutic methods

Combining PARPi with antiangiogenic agents
Angiogenesis is a hallmark feature of TME, providing essential nutrients to support tumor cell proliferation [304]. Antiangiogenic agents exert dual synthetic lethal effects when combined with PARPi. On the one hand, these agents inhibit tumor angiogenesis, leading to hypoxia within the TME, which subsequently induces additional DNA damage [305]. On the other hand, the inhibition of angiogenesis downregulates HRR signaling, particularly reducing BRCA1/2 gene expression in tumor cells [306]. This dual mechanism thus reinforces the therapeutic rationale for combining PARPi with antiangiogenic strategies, as validated by numerous clinical trials.
Several representative trials have investigated the benefit of integrating small-molecule VEGF receptor inhibitor cediranib with PARPi. For patients with recurrent platinum-sensitive OC, combining olaparib with cediranib significantly prolongs mPFS compared to olaparib monotherapy (17.7 vs. 9 months, 10.4 vs. 8.2 months) [307, 308]. Notably, the phase 2 EVOLVE trial evaluated the efficacy of this combination in OC patients progression following prior PARPi therapy [100]. This study demonstrated that the activity of cediranib and olaparib varied according to the specific PARPi resistance mechanism, reporting an overall objective response rate (ORR) of 9% and a disease control rate of 68% [100]. Paris saponin VII (PS VII) has been found to partially reverse PARPi resistance in OC patients, expanding the scope to natural antiangiogenic compounds [309]. Mechanistically, PS VII appears to hinder glycolysis and angiogenesis by binding and stabilizing the expression of RORα [309], which inhibits ECM1 and interferes with the VEGFR2 signaling pathway.
Additionally, other representative approaches focus on bevacizumab, an antibody targeting the VEGF signal. In the first-line maintenance therapy for platinum-sensitive patients with OC, the phase 3 PAOLA-1/ENGOT-ov25 trial demonstrated superior survival outcomes of bevacizumab and olaparib group compared to bevacizumab monotherapy (mOS: 56.5 vs. 51.6 months), underscoring the value of this combination strategy [46]. Furthermore, the phase 2 AVANOVA2 trial confirmed the clinical benefit of combining niraparib with bevacizumab even in OC patients without BRCA1/2m, yielding significantly improved mPFS versus niraparib alone (Intention-to-treat (ITT): 11.9 vs. 5.5 months; BRCA1/2 wild-type: 5.9 vs. 3.1 months) [232]. This trial directly compared PARPi-antiangiogenic combination with an antiangiogenic agent with PARPi monotherapy, providing strong evidence for utilizing bevacizumab to reverse PARPi resistance. These findings collectively establish angiogenesis inhibition as the most clinically validated PARPi combination strategy to date.

Combining PARPi with cell-cycle checkpoint inhibitors
Cell-cycle checkpoint inhibitors are extensively utilized across multiple cancer types because these checkpoints are critical for maintaining genomic stability and ensuring the orderly progression of the cell cycle. Cell-cycle checkpoint kinases can be categorized into two groups based on their primary physiological roles. The DNA damage checkpoint pathway, involving key components such as ATM, CHK2, and TP53, primarily monitors genetic integrity and induces cell cycle arrest to allow for DNA repair [310]. However, these pathways are frequently impaired in tumor cells, permitting continuous proliferation. Conversely, many tumors exhibit elevated expression of DNA replication stress checkpoints, including ATR, CHK1, and WEE1, to mitigate replication stress and safeguard genome integrity [310]. Recent investigations highlight a promising therapeutic strategy: combining cell-cycle checkpoint inhibitors with PARPi. The selective inhibition of DNA replication stress checkpoints disrupts the acquired protection mechanism of genomic integrity exploited by tumors. This synergistic combination aims to induce replication catastrophe and selectively increase lethality in tumors, particularly those harboring DNA repair deficiencies or exhibiting resistance to PARPi monotherapy by accelerating cancer cells with replication stress into mitosis.
Notably, preclinical investigations indicate that the combination of PARPi and ATR-CHK1–CDC25A/C inhibitors holds significant therapeutic promise. Mechanistically, this pathway is essential for stabilizing RFs and facilitating RAD51 recruitment to the DSBs, which are substantial in the DDR process [305]. Several studies utilizing cell lines, PDXs, and organoids confirm that targeting the ATR-CHK1–CDC25A/C pathway overcomes acquired PARPi resistance in BRCA1/2m tumors [261, 311–314]. This combination strategy is further under evaluation in clinical trials for TNBC [315], OC [316], and SCLC [317], although the clinical efficacy of these trials has been modest. A phase 2 trial assessing the combination of olaparib and the ATR inhibitor ceralasertib in advanced TNBC reported a confirmed ORR of 17.1%, which did not meet predefined thresholds [315]. Another phase 2 trial compared the olaparib-ceralasertib combination with olaparib monotherapy in SCLC. Although both groups did not meet the predefined efficacy endpoint, the combination arm showed a more evident disease stabilization rate [317].
Interestingly, pharmacological inhibition of CDK12, achieved via the broad-spectrum inhibitor dinaciclib, has been demonstrated to reverse acquired PARP inhibitor resistance [318]. Mechanistically, CDK12 inhibition downregulates HR gene expression and impairs HRR functionality, thereby inducing a “BRCAness” phenotype. Consequently, dinaciclib synergized with olaparib across multiple TNBC cell lines, particularly in models with established resistance, a finding validated in PDXs where efficacy was observed in BRCA1 wild-type and olaparib-resistant models [318]. Parallelly, targeting the cell cycle regulator WEE1 represents another pivotal strategy [319]. WEE1 inhibition disrupts the G2/M cell-cycle checkpoint and compels cells to prematurely enter the S-phase [320]. This abrogation leads to excessive replication origin firing and the accumulation of lethal DSBs, ultimately triggering mitotic catastrophe in the context of HRD and concurrent PARP inhibition. Clinically, this combination has effectively overcome PARPi resistance in ovarian cancer and BRCA1/2 wild-type TNBC models [321–323], while also enhancing sensitivity in BRCA1/2 wild-type pancreatic cancer cells [324]. Beyond these targets, the inhibition of CK2 in RB1-deficient cells results in p130 degradation and S-phase accumulation, accelerating PARPi-induced mitotic cell death [325]. Similarly, a high-throughput screen identified the CDC7 inhibitor XL413, which synergizes with olaparib to induce robust DNA damage and a type-I interferon response via the cGAS/STING pathway [326].
Collectively, targeting cell-cycle regulators, particularly CDK12 or WEE1, emerges as a robust strategy to counteract PARPi resistance driven by HR restoration. However, a significant translational gap remains: while preclinical synergy is well-established, validation in large-scale clinical trials is pending. The complexity of the TME and the overlapping toxicities of these regimens pose substantial challenges to their clinical utility. Consequently, triple combination strategies are being explored as a promising avenue to optimize the therapeutic index and enhance patient safety.

Combining PARPi with immune checkpoint inhibitors
Immunotherapy, particularly ICIs, has demonstrated remarkable clinical efficacy across multiple malignancies. The therapeutic rationale for combining PARPi with ICIs (targeting CTLA-4 or PD-1/PD-L1) is partly grounded in the hypothesis that tumors harboring BRCA1/2 mutations or HRD display a higher neoantigen load compared to HR-proficient cancers, thereby eliciting a more robust anti-tumor immune response. Mechanistically, PARPi exert a complex dual regulatory effect on the tumor immune microenvironment [327, 328]. On one hand, BRCA1/2 deficiency and PARP inhibition trigger a STING-dependent innate immune response via the accumulation of cytosolic DNA fragments. This activation stimulates the production of type I interferons and pro-inflammatory cytokines, promoting cytotoxic T-cell infiltration and transforming immunologically “cold” tumors, such as ovarian cancer, into “hot” targets [329, 330]. On the other hand, PARP inhibition can concurrently promote immune evasion. Evidence suggests that PARPi inactivate GSK3 in a dose-dependent manner, leading to the upregulation of PD-L1 expression. Consequently, this upregulation suppresses T-cell activation and blunts the immune attack. Therefore, the combination of PARPi and ICIs exhibits a strong synergistic rationale: ICIs counteract the PARPi-induced PD-L1 upregulation while leveraging the enhanced immunogenicity driven by HRD and cGAS-STING activation, offering a novel strategy to overcome resistance [331].
Several clinical trials have evaluated the efficacy of combining PARPi with ICIs, although larger, multi-arm trials are still required. A phase 2 study investigated olaparib plus durvalumab for recurrent OC, demonstrating modest clinical activity (ORR: 14%; disease control rate: 71%) [332]. Two ongoing phase 3 trials are pivotal: ATHENA is assessing rucaparib monotherapy or in combination with nivolumab as first-line maintenance for OC [47], while MITO 33 compares niraparib and dostarlimab versus physician’s choice chemotherapy in recurrent OC [333]. These trials are expected to provide robust evidence for PARPi-ICIs combination strategies. The phase 2 MEDIOLA trial also reported promising antitumor activity for olaparib plus durvalumab in gBRCA1/2m metastatic BC (ORR 63.3%) [60]. Conversely, the phase 2 BAYOU trial in unselected metastatic urothelial carcinoma found that adding olaparib to durvalumab did not improve survival outcomes compared to durvalumab alone (mPFS: 4.2 vs. 3.5 months; mOS: 10.2 vs. 10.7 months; ORR: 22% vs. 14%) [334]. Notably, a phase 1b/2 trial demonstrated the promising activity of niraparib plus ipilimumab in advanced pancreatic cancer patients with stable disease after platinum-based chemotherapy, meeting the primary endpoint of 6-month mPFS [89]. This contrasts with the niraparib plus nivolumab arm, which yielded inferior PFS, highlighting the influence of ICIs types on clinical efficacy. However, the heterogeneity of trial outcomes underscores the importance of biomarker-driven patient selection.

Combining PARPi with chemotherapy or radiotherapy
Chemotherapy and radiotherapy are established therapeutic modalities for various types of cancer. Both induce extensive DNA damage, imposing significant stress on the DDR pathways. Meanwhile, the upregulation of DDR pathways is a key mechanism underlying resistance to PARPi, thereby providing a rationale for combination therapy. Hence, combining PARPi with these treatments enhances cytotoxicity and holds promise for overcoming PARPi resistance. Several ongoing clinical trials are evaluating the efficacy of such combination strategies. However, most of these studies primarily focus on establishing the superiority of combination therapy over standard chemotherapy alone [59, 64, 65, 96, 335]. Directly assessing the ability of these combinations to overcome PARPi resistance, using PARPi monotherapy as a control, remains insufficiently explored.

Future perspectives
Currently, PARPi have been approved to treat four cancers characterized by HRD. Expanding the application of PARPi to other types of cancers or different therapeutic contexts within these four tumor types holds promise. This expansion is constrained by primary resistance due to HRR proficiency. Localized approaches that mimic HRD status within tumors, such as local thermotherapy, which promotes BRCA2 degradation [336], can potentially transform the landscape of PARPi utilization. Furthermore, identifying novel biomarkers capable of accurately predicting patient responses to PARPi therapy is crucial. Beyond BRCA1/2 or HR gene mutations alone, genomic scars and signatures further elucidate overall HRR status and help identify patients who may benefit from PARPi even without identifiable HR gene mutations. Notably, functional assays represent a promising approach for selecting patients eligible for PARPi treatments. By utilizing routine FFPE samples, functional assays can effectively assess current functional HRR capabilities without incurring additional costs associated with genetic testing. Further large-scale validation studies on RAD51 foci are warranted to enhance their applicability in clinical practice.
Considering the acquired resistance that develops during PARPi therapy, the restored HR pathway, particularly reverse mutations in BRCA1/2, is the most frequently observed mechanism in clinical settings. Nevertheless, due to dynamic molecular alterations, no common resistance mechanism has been consistently identified across all patients progressing on PARPi treatment. Consequently, specific agents targeted to reverse acquired resistance may exhibit limited clinical efficacy across the entire patient population, since other alterations render such agents ineffective. Integrating PARPi into conventional therapeutic modalities of specific tumors appears more promising and beneficial. However, this field remains in its early stages. The only approved combination strategy thus far is olaparib combined with bevacizumab for platinum-sensitive OC patients who demonstrate HRD status in a first-line maintenance setting. Further efforts should be directed toward exploring combinations in other tumors with careful evaluation of specific combinatorial partners. Ultimately, translating findings from bench research to bedside application is crucial. Given the high costs associated with serial ctDNA analysis, there is an urgent need for novel, feasible monitoring methods to prospectively monitor the emergence of acquired PARPi resistance.

Conclusions
PARPi have demonstrated clinical efficacy across a diverse spectrum of cancers, including ovarian, breast, prostate, and pancreatic malignancies. In the context of OC, PARPi exhibit substantial effectiveness as first-line maintenance therapy. The therapeutic response is critically influenced by prior platinum sensitivity and the presence of BRCA1/2 mutations or HRD status. For BC with gBRCA1/2m, olaparib has received approval for use as an adjuvant treatment; meanwhile, other PARPi may be employed in managing metastatic BC. In cases of metastatic castration-resistant prostate cancer characterized by HR gene mutations, PARPi are utilized in post-line therapy, with combinations of PARPi and androgen receptor pathway inhibitors approved for first-line treatment.
However, the emergence of resistance underscores the urgent need for improved patient stratification through the identification of novel biomarkers that can accurately predict responses to PARPi. Establishing these innovative biomarkers will facilitate the expansion of existing clinical indications for PARPi and ultimately broaden the population of patients who may benefit. Additionally, understanding mechanisms underlying resistance is fundamental to devising therapeutic strategies aimed at overcoming these challenges. While several targeted therapies endeavor to exploit acquired vulnerabilities within tumor cells, combining PARPi with conventional treatment modalities appears both feasible and clinically promising. Future research efforts will concentrate on identifying optimal combinations of oncological therapies across various tumor types, to determine the most effective regimens while minimizing toxicity.

Electronic supplementary material
Below is the link to the electronic supplementary material.