Recent advances in ovarian cancer have expanded not only the number of available therapies but also the complexity of deciding how they should be sequenced. The 2026 National Comprehensive Cancer Network (NCCN) guidelines reflect this shift by reorganising first-line maintenance around breast cancer susceptibility gene 1/2 (BRCA1/2) and homologous recombination status, narrowing later-line poly(ADP-ribose) polymerase inhibitors (PARPi) use and introducing biologically informed options for platinum-resistant disease. The key question is no longer simply whether an actionable biomarker exists, but when it was assessed, how prior therapy altered tumour biology, and how each decision affects subsequent options. Here, dynamic treatment sequencing means iterative reassessment of tumour biology, prior treatment exposure, response duration, toxicity and surgical opportunity at each disease transition, rather than a one-time biomarker–drug match (figure 1).
A practical framework for dynamic treatment sequencing in ovarian cancer. ADC, antibody–drug conjugate; CPS, combined positive score; ctDNA, circulating tumour DNA; FRα, folate receptor alpha; GLS, glutaminase; HR, homologous recombination; HRD, homologous recombination deficiency; HRP, homologous recombination proficient; IDS, interval debulking surgery; MMR/MSI, mismatch repair/microsatellite instability; PARPi, poly(ADP-ribose) polymerase inhibitor; PDS, primary debulking surgery; TMB, tumour mutational burden.
PARP inhibitors and Post-PARP recurrence
PARPi remain central to maintenance therapy, but their use is becoming more selective. Long-term follow-up from SOLO1 and PAOLA-1 supports durable benefit in BRCA1/2-mutated or homologous recombination-deficient (HRD) tumours, whereas benefit is smaller in homologous recombination-proficient (HRP) disease.1 2 HRD should not, however, be treated as a binary gatekeeper for PARP inhibition. Genomic-scar assays may not reflect functional homologous recombination at treatment, and resistance can emerge despite an HRD-positive result.3 4 Functional assays such as RAD51 foci may complement genomic testing, although their clinical role remains under evaluation.5 HRD is therefore better viewed as an estimate of benefit rather than a rigid threshold. In HRP tumours or when HRD status is unknown, treatment should be individualised according to residual disease, platinum response, recurrence risk, toxicity and accessibility.
First-line PARP exposure is also reshaping recurrence. The NOVA and ARIEL3 trials established a progression-free survival (PFS) benefit for PARPi maintenance after platinum-sensitive recurrence, but mature overall survival (OS) data have made PFS alone insufficient for decision-making.6 7 In BRCA wild-type, platinum-sensitive recurrent disease, PARPi are no longer routine maintenance options in the updated NCCN Guidelines. Evidence for PARP rechallenge remains limited, and the PARPi-free interval may also provide additional context for retreatment decisions, although no validated threshold defines benefit.8
Treatment decisions should also incorporate prior PARPi and bevacizumab exposure, histology, resectability and updated biomarkers. Reassessment is most relevant when results could alter management; repeat tissue sampling may reassess targets such as FRα, while circulating tumour DNA may help capture emerging genomic alterations.9
Platinum-resistant disease: mechanism-Based treatment
Platinum-resistant ovarian cancer illustrates the transition from empiric chemotherapy rotation to mechanism-driven treatment. The 2026 NCCN guideline adds nab-paclitaxel plus relacorilant as a preferred regimen after up to three prior lines and prior bevacizumab. Relacorilant inhibits glucocorticoid receptor-mediated survival signalling and chemotherapy tolerance, and ROSELLA demonstrated improvements in PFS and OS.10 This broadens targeted therapy beyond genomic alterations to include stress responses and microenvironment-mediated protection.
The FRα-directed antibody–drug conjugate (ADC) mirvetuximab soravtansine established a complementary protein-expression-guided paradigm. MIRASOL improved objective response rate, PFS and OS compared with chemotherapy in FRα-high platinum-resistant disease.11 Nevertheless, FRα expression may vary across sites and over time, while ADC efficacy can be limited by impaired internalisation or lysosomal trafficking, linker stability, payload-specific resistance, bystander effects and drug efflux. ADC selection should consider biopsy site, testing time, expression intensity and the need for reassessment.
Immunotherapy: Progress with important limits
Ovarian cancer is not intrinsically immune-sensitive, and immunotherapy requires cautious interpretation. KEYNOTE-B96 showed that pembrolizumab plus weekly paclitaxel, with or without bevacizumab, improved outcomes, particularly among patients with PD-L1 CPS-positive platinum-resistant disease, extending median PFS and OS by 1.1 months and 4.2 months, respectively.12 The benefit may reflect antigen release from paclitaxel, improved immune-cell trafficking with antiangiogenic therapy and reinvigoration of T-cell responses by PD-1 blockade. PD-L1 CPS is clinically actionable but biologically incomplete because it does not capture antigen-presentation capacity, suppressive myeloid states, tertiary lymphoid structure organisation, interferon-related programmes or spatial immune exclusion.13 Single-cell and spatial studies suggest that tumour-cell major histocompatibility complex (MHC) class II expression, macrophage-driven exhaustion, T-cell localisation, myeloid networks and exhaustion niches may ultimately be more informative as composite biomarkers than as isolated markers.14 15 KEYNOTE-B96 should therefore be viewed as proof of concept rather than the endpoint of immunotherapy development.
Beyond current guideline-supported therapies, investigational approaches including CAR-NK cell therapy, IL-12–based strategies and glutaminase inhibition may eventually enter the treatment sequence.16–18 Their future position should depend on biomarker-defined selection, prior treatment exposure, resistance evolution and evidence of durable clinical benefit.
Histology-specific management and precision surgery
Ovarian cancer comprises biologically distinct malignancies. Low-grade serous, clear cell, mucinous and small cell carcinoma of the ovary, hypercalcaemic type, should not be managed as variants of high-grade serous carcinoma. Each has distinct molecular drivers, natural history and therapeutic vulnerabilities; histology is therefore the first biomarker guiding molecular testing and treatment.
As systemic options expand, surgery must increasingly be judged within the broader treatment sequence, balancing the likelihood of complete gross resection and durable benefit against operative morbidity and delay of subsequent therapy. Lymphadenectomy in ovarian neoplasms (LION) trial showed no survival benefit from systematic lymphadenectomy in patients with complete gross resection and clinically negative nodes, but demonstrated increased surgical morbidity.19 DESKTOP III further supports secondary cytoreduction in rigorously selected patients when complete gross resection is highly achievable.20 Precision surgery should maximise net clinical benefit rather than operative extent.
Overall, the 2026 NCCN framework reframes precision oncology as longitudinal treatment planning. At diagnosis, histology, BRCA1/2 and HRD status and resectability guide primary therapy and maintenance; after PARPi exposure, recurrence should trigger reassessment of platinum-free and PARPi-free intervals, surgical opportunity and evolving biomarkers; and in platinum-resistant disease, FRα, PD-L1, HER2 and other actionable features may redirect treatment towards ADCs, immunotherapy-based combinations, mechanism-based therapy or clinical trials. Precision management therefore requires repeated reassessment across the treatment trajectory, with each decision chosen not only for immediate disease control but also for its impact on subsequent treatment options.
