Combining PARP with ATR inhibition overcomes PARP inhibitor and platinum resistance in ovarian cancer models

High-grade serous ovarian cancer (HGSOC) is commonly treated with surgical debulking followed by platinum-based chemotherapy, yet over 80% of patients recur and thousands die annually in the U.S. Targeted therapies such as PARP inhibitors (PARPi) have been approved in frontline maintenance and recurrent settings, but resistance emerges, limiting durability of responses. Approximately 50% of HGSOCs harbor homologous recombination (HR) defects (e.g., BRCA1/2), initially sensitizing tumors to platinum and PARPi; the remainder are HR-proficient, with about 40% showing CCNE1 amplification associated with poor outcomes and platinum resistance. The research question is whether co-targeting ATR, a key kinase activated by replication stress that stabilizes replication forks and enforces S/G2-M checkpoints, with PARP can overcome both acquired and de novo resistance to PARPi and platinum across diverse genetic contexts. The authors hypothesize that combined PARP and ATR inhibition (PARPi-ATRi) will exacerbate replication stress, increase DNA double-strand breaks, and induce tumor cell death independent of HR status, leading to durable therapeutic responses.

Documented resistance mechanisms to platinum include reduced drug accumulation, inactivation, enhanced DNA repair, and impaired apoptosis. PARPi resistance can be HR-dependent (e.g., secondary/reversion BRCA mutations; loss of 53BP1/RIF1/REV7/PTIP/Artemis/Shieldin) or HR-independent (drug efflux via MDR1/ABCB1, altered PARP activity or PARG loss, replication fork stabilization, RAS pathway activation, PI3K/AKT upregulation). ATR is activated by replication stress to stabilize forks and enforce checkpoints; selective ATR inhibitors (e.g., AZD6738, M6620) are in early-phase trials. Prior work showed PARPi activates ATR/CHK1 signaling and that ATRi can enhance PARPi efficacy in BRCA-deficient models, suggesting a rationale for combined PARP-ATR inhibition to overcome resistance.

• Cell models: Developed acquired PARPi-resistant BRCA2MUT (PEO1-PR, PR1, PR2) and BRCA1MUT (JHOS4-PR, PR1, PR2) lines via ~1.5 years of olaparib exposure; platinum-resistant BRCA2MUT lines (PEO1-CR, CR1, CR2) via >1 year carboplatin exposure. Assessed de novo resistance lines including PEO4 (BRCA2 reversion), Kuramochi (BRCA2 nonsense), UWB.289 derivatives (BRCA1+/− and 53BP1−/−). CCNE1-amplified platinum-resistant lines: OVCAR3, COV318, FUOV1; CCNE1 copy-normal control: OVKATE.
• Genomics and proteomics: Whole genome sequencing (~30x) of resistant and parental cell lines to identify mutations, copy-number alterations, and structural variants; whole-exome sequencing (~160x) of PDXs; reverse phase protein array (RPPA) profiling (>300 proteins) to assess pathway alterations.
• Biochemical assays: Western blotting for pATR, ATR, pCHK1, CHK1, MDR1, cleaved caspase-3; immunofluorescence for RAD51 foci in geminin-positive cells as an HR marker.
• Functional assays: Viability (MTT, 5 days), colony formation (crystal violet, ~10–13 days). Drug synergy quantified using coefficient of drug interaction (CDI = AB/(A×B)); CDI <1 indicates synergy (<0.7 significant).
• Cell cycle and DNA damage: Flow cytometry for BrdU incorporation and PI to quantify cell-cycle phases and Sub-G1; γH2AX in S-phase by flow cytometry after 24–36 h of treatment.
• Replication dynamics: DNA molecular combing after 30 min pretreatment with PARPi (AZD2281, 1 µM), ATRi (AZD6738, 1 µM), or combination, followed by sequential CldU and IdU pulses (15 min each) under continuous drug exposure; quantified fork speed (kb/min) and fork asymmetry (IdU track ratio).
• In vivo PDX studies: NSG mice bearing orthotopic patient-derived xenografts (PDXs). Models: WO-2 (gBRCA2MUT), WO-57 (PARPi-resistant gBRCA1 with BRCA1 reversion), WO-58 (gBRCA1 with CCNE1 amplification; PARPi-resistant), WO-19 (BRCA WT, CCNE1 amplified; platinum-resistant). For gBRCA PDXs, olaparib pretreatment until 2-fold tumor growth confirmed PARPi resistance, then randomized to: control; PARPi (olaparib 75–100 mg/kg/day PO, 5–6 days/week); ATRi (AZD6738 40–50 mg/kg/day PO, 5 days/week); combination (e.g., olaparib 50–75 mg/kg/day + AZD6738 50 mg/kg/day PO, 5 days/week). For WO-19, included carboplatin 30 mg/kg IP weekly. Tumor volumes by ultrasound; body weights weekly. Survival by Kaplan–Meier. Longitudinal tumor growth analyzed with linear mixed-effects modeling; synergy in vivo assessed via Bliss independence using TumGrowth slopes. IHC on tumor sections for pCHK1, γH2AX, and cleaved caspase-3 after 2 weeks of therapy.
• Statistics: One-way ANOVA with Tukey’s multiple comparisons for in vitro assays; Kruskal–Wallis for combing metrics; Mantel–Cox log-rank for survival. Biological replicates: ≥3 for in vitro; PDX group sizes typically 4–10 per arm.

• Establishment of resistance and genomic changes: PARPi-resistant BRCA2MUT (PEO1-PR) and BRCA1MUT (JHOS4-PR) cells showed 13-fold and 8-fold increases in PARPi IC50, respectively, vs parental. Platinum-resistant PEO1-CR had a 4.6-fold increase in carboplatin IC50. De novo resistance examples included Kuramochi (PARPi IC50 ~12 µM) and UWB.289/BRCA1+/− (~83-fold higher IC50 vs parental). Resistant lines acquired multiple mutations and increased chromosomal rearrangements; many alterations affected DNA repair (e.g., HERC2, STAG2, PRKDC, EXO1, APEX2, EZH2, XRCC6; in BRCA1MUT JHOS4-PR: XPC amplification, DCLRE1C LOH). MDR1 upregulation occurred in PEO1-PR and PEO1-CR.
• Elevated ATR/CHK1 signaling in resistance: Resistant and CCNE1AMP lines exhibited higher basal pATR/pCHK1 than parental. PARPi further induced pCHK1 in parental but less so in resistant cells; ATRi suppressed pCHK1 across models.
• ATRi monotherapy modest activity: AZD6738 (<1 µM) reduced viability modestly; acquired PARPi-resistant lines trended more sensitive than parental but differences were generally not significant.
• Synergy of PARPi-ATRi in vitro: Combination significantly reduced viability and colony formation versus monotherapies across parental, acquired and de novo PARPi-/platinum-resistant, and CCNE1AMP models (P values commonly <0.0001). Colony formation synergy was strong with CDI <0.6 across lines. OVKATE (CCNE1-normal) was comparatively insensitive.
• Cell-cycle effects: In parental BRCA1/2MUT and CCNE1AMP lines, PARPi increased G2/M fraction; ATRi abrogated G2/M arrest. In acquired resistant BRCA1/2MUT cells, PARPi had minimal G2/M effects, suggesting other mechanisms dominate synergy.
• HR modulation: PARPi induced RAD51 foci in several resistant models (PEO1-PR, UWB/BRCA1+/−, PEO1-CR, PEO4, OVCAR3, FUOV1), indicating HR restoration; JHOS4-PR and Kuramochi did not form RAD51 foci despite resistance. ATRi significantly suppressed PARPi-induced RAD51 foci in models where it occurred (P<0.001).
• DNA damage: γH2AX-positive S-phase cells increased more with PARPi-ATRi than ATRi alone across most resistant and CCNE1AMP models (P<0.001), but not in OVKATE.
• Replication stress: ATRi reduced fork speed (P<0.0001); PARPi-ATRi further reduced speed vs ATRi alone in resistant and CCNE1AMP models (e.g., PEO1-PR, JHOS4-PR P<0.0001; PEO1-CR P=0.02; OVCAR3 P=0.04; parental PEO1 P=0.03). ATRi increased fork asymmetry in all lines; combination further increased asymmetry in resistant and CCNE1AMP models (P values ≤0.005). Fork asymmetry synergy CDI: PEO1-PR 0.7, PEO1-CR 0.8, JHOS4-PR 0.7, OVCAR3 0.8.
• Apoptosis: Combination increased apoptosis beyond ATRi alone (e.g., PEO1-PR 1.8×, P=0.002; other lines P<0.001) and elevated cleaved caspase-3 in resistant models (e.g., JHOS4-PR 13.9×, P=0.0069).
• In vivo efficacy (PDX): • WO-2 (gBRCA2MUT, acquired PARPi-resistant): PARPi-ATRi induced near-complete regression, improved median OS >16× vs control (P=0.0011), >5× vs PARPi (P=0.0007), >3× vs ATRi (P=0.0007); 50% complete responses; some dose reductions for weight loss (>15%) with recovery thereafter. • WO-57 (PARPi-resistant gBRCA1 with BRCA1 reversion): Combination improved tumor regression and median OS 2.6× vs control (P=0.0118), 1.7× vs PARPi (P=0.0292), 1.9× vs ATRi (P=0.0078); 40% partial responses; synergy index 0.80; well tolerated without dose reductions. • WO-58 (gBRCA1 with CCNE1 CN=7, PARPi-resistant): Combination increased OS vs PARPi (P<0.0001); tumor regression 3.2× greater; partial response rate 40% vs 12.4% with ATRi alone; synergy index 0.89; well tolerated. • WO-19 (BRCA WT, CCNE1 CN>10, platinum-resistant): Combination improved tumor regression vs PARPi (P=0.0003) and median OS 2.5× vs control (P=0.0142), 2.1× vs carboplatin (P=0.0206), 2.1× vs PARPi (P=0.0462); 40% complete responses; synergy index 0.94; weights stable.
• Pharmacodynamic biomarkers in PDXs: PARPi induced pCHK1 that was abrogated by ATRi; γH2AX and cleaved caspase-3 increased with combination, mirroring in vitro mechanisms.

The study demonstrates that increased ATR/CHK1 signaling is a hallmark of PARPi and platinum resistance and of CCNE1 amplification-associated replication stress. While ATRi alone has limited efficacy, combining with PARPi produces synergistic cytotoxicity across diverse resistant contexts by converging on replication fork stability and DNA damage response. Mechanistically, in many resistant models where HR function is reacquired (RAD51 foci present), ATRi suppresses RAD51 recruitment, thereby creating an HR-deficient state and amplifying PARPi-induced cytotoxicity. In models without RAD51 restoration (e.g., JHOS4-PR, Kuramochi), synergy still occurs via enhanced replication fork slowing and asymmetry, leading to γH2AX accumulation and apoptosis, suggesting additional ATR-PARP interdependencies at replication forks. In CCNE1-amplified tumors, inherent replication stress and checkpoint dependency make them particularly susceptible to dual inhibition, with combination therapy disrupting G2/M checkpoint control, impairing fork progression, and triggering apoptosis. The robust and durable responses in multiple clinically relevant PDXs, including those with BRCA reversions and CCNE1 amplification, underscore the translational potential of PARPi-ATRi to overcome prevalent resistance mechanisms and improve survival.

Combining PARP and ATR inhibition overcomes both acquired and de novo resistance to PARP inhibitors and platinum in ovarian cancer models, including BRCA1/2-mutant with reversion mutations and CCNE1-amplified tumors. The combination increases replication fork stalling and collapse, augments DNA double-strand breaks, suppresses RAD51-dependent HR where restored, and triggers apoptosis, yielding marked tumor regressions and significant survival benefits in PDXs. These data support clinical translation of PARPi-ATRi (e.g., AZD6738 plus olaparib) in recurrent ovarian cancer following PARPi or platinum resistance. Future work should optimize dosing schedules and manage toxicity, identify predictive biomarkers (e.g., baseline pCHK1, RAD51 foci dynamics, CCNE1 status), delineate mechanisms in RAD51-negative resistant models, and evaluate durability and resistance to the combination in clinical trials.

• Mechanistic heterogeneity: Not all resistant models restored RAD51 foci; in such cases, precise mechanisms of synergy remain undefined.
• Preclinical scope: Findings are based on cell lines and PDX models; clinical efficacy and safety require validation.
• Toxicity/dosing: Some PDXs required combination dose reductions due to weight loss, indicating a potentially narrow therapeutic window.
• Biomarker uncertainty: BRCA reversion status alone did not predict resistance mechanisms or response; robust predictive biomarkers need development.
• Sample sizes and diversity: Although multiple models were used, the number per genotype/context is limited; broader genetic diversity and longitudinal resistance to the combination were not fully assessed.