Biomarker-Driven Treatments for Ovarian Cancer: Moving Beyond HRD?

Abstract

Ovarian cancer is a biologically heterogeneous disease, yet for more than two decades its standard treatment of surgery and platinum–taxane chemotherapy changed little. Only a modest additional benefit was seen with the addition of bevacizumab, a monoclonal antibody targeting vascular endothelial growth factor A (VEGF-A) in selected patients. The recognition that BRCA1/2 mutations impair the repair of DNA double-strand breaks by homologous recombination, and that inhibitors of PARP are highly active in this setting, established the first widely adopted biomarker-driven treatment in this disease. Homologous recombination deficiency (HRD) therefore remains the established predictive biomarker for first-line maintenance therapy, but it is imperfect: benefit is not confined to HRD-positive tumours, assays report historical genomic scars, and most patients ultimately relapse with platinum-resistant disease. This review examines how biomarker-driven treatment in ovarian cancer is developing beyond HRD. We appraise the evidence for strategies that are already clinically established, principally PARP inhibitors, the folate receptor α-directed antibody-drug conjugate mirvetuximab soravtansine, and, more recently, PD-L1-selected immune checkpoint inhibition in platinum-resistant disease and distinguish these from approaches that remain investigational, including antibody-drug conjugates directed at HER2, TROP2, CLDN6, B7-H3/H4 and CDH6, and agents exploiting cell cycle dysregulation and replication stress such as CCNE1-associated, WEE1 and ATR-directed therapies. We also consider the practical obstacles to implementation: tumour heterogeneity and evolution, the complementary roles of tissue and liquid biopsy, when molecular profiling should be repeated, assay standardisation and access, and the limitations of basket trial designs. Progress will depend on expanding the development of practical frameworks for biomarker testing, treatment selection and sequencing.

Introduction

Ovarian cancer is a leading cause of mortality in women worldwide and is the eighth most common cancer in women.¹ Early symptoms are often subtle and non-specific, so diagnosis is frequently delayed: over 75% of patients present with advanced disease, which is associated with high rates of recurrence.²

Ovarian cancer is a biologically heterogenous disease with the most common subtype being epithelial carcinoma and a small proportion comprising of non-epithelial subtypes.¹ Epithelial ovarian cancers are further divided by histological subtype, which include high-grade serous, low-grade serous, clear cell, endometrioid, and mucinous ovarian cancer.³ There is significant geographical variation in ovarian cancer incidence, with higher rates observed in Europe and North America and lower rates in parts of Africa.⁴

For many years, treatment options were largely uniform, consisting of primary surgery followed by platinum-taxane-based chemotherapy.⁵ Bevacizumab, a monoclonal antibody targeting VEGF-A, can also be added in some settings and continued as maintenance therapy.⁵ Response rates to first-line platinum-based chemotherapy are generally high, especially in serous carcinomas, the most common subtype.⁶ However, durable remission remains uncommon, particularly in patients presenting with advanced disease, and the majority will ultimately relapse.⁷

Initially, treatment was guided by clinical factors rather than tumour characteristics and did not account for the diverse tumour biology of ovarian cancer. The high rates of recurrence illustrate the limitation of a uniform approach in a biologically heterogeneous disease, and underline the need for biomarkers to guide and tailor treatment.

Biomarkers are measurable features that reflect tumour biology or response to treatment and can be used to guide clinical decision making.⁸ Examples include protein expression on tumour cells and gene alterations.⁹ The distinction between biomarker types is important and is used throughout this review. Prognostic biomarkers indicate the likely course of disease irrespective of treatment; predictive biomarkers identify patients likely to benefit from a specific therapy; and pharmacodynamic biomarkers demonstrate that a drug has engaged its target, without necessarily predicting clinical benefit.¹⁰ Only predictive biomarkers can direct the choice of a targeted therapy, and much of the difficulty in this field arises when a biological feature or a therapeutic target is assumed to be predictive before this has been demonstrated. Advances in molecular profiling have allowed for the identification of potential cancer biomarkers.¹¹

In ovarian cancer, the discovery of BRCA1 and BRCA2 mutations marked a significant shift in the treatment landscape towards precision medicine and represented one of the first examples of biomarker-driven therapy in this disease.

In current practice, biomarker testing follows a broadly established sequence. Germline BRCA1/2 testing is recommended for all patients with epithelial ovarian cancer, with somatic tumour testing in those without a germline pathogenic variant, and HRD assessment used to inform first-line maintenance decisions. These single-gene and genomic-scar assays are increasingly supplemented by broader next-generation sequencing panels and, in selected settings, by circulating tumour DNA. The strengths and limitations of each are considered in more detail later in this review.

HRD remains the only genomic biomarker in routine use to select maintenance PARP inhibitor therapy in ovarian cancer, but it is an imperfect predictive biomarker. Benefit is not restricted to HRD-positive tumours, assays report historical genomic scars rather than contemporaneous repair capacity, and testing is technically complex and incompletely standardised. Approximately half of high-grade serous tumours are homologous recombination proficient and derive limited benefit, and most patients who do respond eventually relapse. The aim of this review is therefore to examine how biomarker-driven treatment in ovarian cancer is developing beyond HRD. We appraise the evidence for established and investigational strategies, distinguishing between the two throughout, and consider the practical challenges of implementation – including tumour heterogeneity, the choice and timing of molecular testing, and the increasingly complex problem of treatment sequencing when more than one targetable alteration may be present in the same tumour.

This is a narrative rather than a systematic review. Literature was identified through searches of PubMed using combinations of the terms “ovarian cancer”, “biomarker”, “homologous recombination deficiency”, “PARP inhibitor”, “antibody-drug conjugate”, “immune checkpoint inhibitor”, “replication stress” and “treatment sequencing”. Priority was given to randomised Phase III trials and registrational studies.

Emergence of Precision Therapy

BRCA1 and BRCA2 encode proteins that play a central role in the DNA damage response (DDR) pathway, particularly the repair of double-strand DNA breaks.^12,13 This pathway comprises multiple mechanisms that maintain genomic stability by recognising DNA damage and coordinating its repair.¹⁴ Homologous recombination repair (HRR) is the high-fidelity mechanism within the DDR pathway for repairing double-strand breaks.¹⁵ BRCA proteins are essential for homologous recombination repair,¹⁶ and BRCA1 and BRCA2 mutations therefore result in impaired DNA damage repair and genomic instability.

Homologous recombination deficiency (HRD) is the loss of effective homologous recombination repair.¹⁷ Recognition that this pathway plays an important role in the pathogenesis of ovarian cancer marked a significant turning point in its management. BRCA1 and BRCA2 mutations are recognised drivers, although other alterations may also be implicated and contribute to genomic instability.¹⁸ HRD results in the accumulation of characteristic “genomic scars”, permanent changes within the genome caused by defective DNA repair,¹⁹ which are detected using assays that determine HRD status. HRD is estimated to occur in up to 50% of high-grade serous ovarian carcinomas,²⁰ and therefore represents both a clinically relevant biomarker and a potential therapeutic vulnerability.

Identifying that many ovarian cancers are HRD positive led to therapies designed to exploit this deficiency, through the principle of synthetic lethality: a defect in either one of two genes has little effect on cell viability, whereas the combination of both defects causes cell death.¹⁸ In ovarian cancer, tumours exhibiting HRD become reliant on alternative DNA repair pathways because homologous recombination repair, mediated by BRCA1 and BRCA2, is defective. This creates a therapeutic vulnerability, in which those alternative pathways can be targeted to trigger tumour cell death.

This concept provides the basis for the development of poly(ADP-ribose) polymerase (PARP) inhibitors.²¹ PARP proteins are important in the DNA damage response pathway, particularly in repairing single-strand DNA breaks.²² PARP inhibitors block this repair, resulting in an accumulation of single-strand DNA breaks that ultimately lead to double-strand DNA breaks.²³ In HRD tumours this damage cannot be repaired, resulting in tumour cell death¹² (Figure 1). PARP inhibitors used in clinical practice include olaparib, niraparib and rucaparib. They were introduced as maintenance therapies given after a response to platinum-based therapy in recurrent ovarian cancer, where the presence of a BRCA mutation, or responsiveness to platinum therapy, were the key indicators of benefit.^24–27 Their development in front-line maintenance therapy introduced HRD as a biomarker to predict which patients are most likely to benefit.

Figure 1.

Synthetic lethality of PARP inhibition in homologous recombination-deficient ovarian cancer. PARP inhibition prevents the repair of single-strand DNA breaks, which are converted into double-strand breaks at the replication fork. In cells with intact homologous recombination these breaks are repaired, whereas in HRD cells DNA repair is impaired and the damage remains unrepaired, resulting in genomic instability and tumour cell death.

Abbreviations: DSB, double-strand break; HRD, homologous recombination deficiency; HRR, homologous recombination repair; PARP, poly(ADP-ribose) polymerase; SSB, single-strand break.

SOLO1 was a pivotal Phase 3 clinical trial which demonstrated the efficacy of the PARP inhibitors in newly diagnosed advanced ovarian cancer.²⁸ It found that maintenance olaparib significantly prolonged progression free survival in patients who had responded to first line platinum-based chemotherapy with a BRCA1/2 mutation. This suggested that patients with homologous recombination deficiency are likely to benefit from PARP inhibition, as BRCA1/2 mutations are key drivers of HRD. This trial was key in establishing PARP inhibitors as standard of care in the first line setting.

The merit of using HRD as an initial biomarker was demonstrated in the PAOLA1 trial, in which olaparib plus bevacizumab was compared with bevacizumab alone.²⁹ Randomisation was stratified by tumour BRCA mutation, and HRD testing was incorporated post-randomisation as a pre-planned exploratory outcome, allowing comparison of outcomes by HRD status among BRCA-mutated and BRCA wild-type tumours. In HRD-positive tumours, median progression-free survival was 37.2 months with olaparib plus bevacizumab compared with 16.6 months with bevacizumab alone, and favourable differences were seen in both BRCA-mutated and BRCA wild-type HRD-positive subgroups. Limited benefit was observed in HRD-negative tumours. Notably, some patients with a pathogenic BRCA mutation do not benefit from PARP inhibitors, which may depend in part on the location of the mutation,³⁰ an early indication that the relationship between HRD and response is not straightforward.

Whilst HRD status identifies patients more likely to benefit from PARP inhibition, the PRIMA trial, which used stratified randomisation based on HRD, showed that both HRD-positive and HRD-negative patients derive benefit.²⁶ In HRD-positive tumours median progression-free survival was 22.1 months with niraparib compared with 10.9 months with placebo, whereas in HRD-negative tumours the benefit was more modest at 8.1 vs 5.4 months. Similarly, ATHENA-MONO showed benefit from PARP inhibition regardless of HRD status, with a larger progression-free survival gain in HRD-positive tumours.³¹ Collectively these trials indicate that HRD-positive tumours derive the greatest benefit, and HRD remains the biomarker used in routine practice to inform first-line maintenance therapy. It nevertheless functions as an enrichment rather than a strictly binary predictive biomarker, since a negative result does not exclude benefit.

Limitations of HRD

Advances in biomarker development have changed the treatment landscape for tumours displaying homologous recombination deficiency. As a result, we can identify patients who are most likely to benefit from PARP inhibitors. However, trials show a mixed picture when we look at HRD negative disease. This highlights the uncertainty clinicians may face when making treatment decisions in this large heterogenous patient population.

This uncertainty extends to HRD-positive disease when the testing itself is considered. Tumour genetics are dynamic, but assays reflect only the molecular profile at the time the sample was taken,³² so results describe previous homologous recombination repair defects rather than current HRD status. In addition, not all tumours expected to respond to PARP inhibitor therapy do so, and prolonged maintenance treatment can promote resistance to subsequent chemotherapy.³² Resistance mechanisms include restoration of homologous recombination repair, where a further mutation in BRCA1/2 restores functional DNA repair, after which synthetic lethality is lost and the tumour becomes PARP-inhibitor resistant.³² HRD also encompasses many different underlying alterations, and testing is technically complex and not always standardised.³³ Together, these factors may explain the discordance between HRD status and observed treatment response in clinical practice.

These limitations have fuelled interest in alternative biomarker strategies that better reflect current homologous recombination repair status. RAD51 is a protein involved in homologous recombination with a key functional role in DNA damage repair,³⁴ and its value as a more precise biomarker is under investigation. Studies in other tumour types have shown that the presence of RAD51 foci in BRCA-mutated tumours is associated with PARP inhibitor resistance.³⁵ This offers a means of assessing current repair capacity rather than inferring it from genomic scars, although it remains investigational and still requires adequate, and often fresh, tissue.

Moving Beyond HRD

These challenges have stimulated a more biologically directed approach to the treatment of ovarian cancer, exploring alternative biomarkers to direct therapy. Immune checkpoint inhibitors, for example, were investigated extensively with limited success, largely because no robust biomarker identified the tumours in which they were likely to be effective – a position that has only recently begun to change. Other approaches target tumour cell surface markers with antibody-drug conjugates, where the target acts both as a therapeutic target and, potentially, as a predictive biomarker enabling more precise drug delivery.

Immune Checkpoint Inhibitors

Immune checkpoint inhibitors (ICIs) target regulatory pathways within the immune system, such as PD-1/PD-L1 and CTLA-4, which normally prevent immune activation.³⁶ Tumour cells exploit these pathways to suppress T-cell activity and avoid destruction; ICIs block these inhibitory signals so that the immune system can recognise and attack tumour cells.³⁶ Agents including pembrolizumab and ipilimumab have improved survival in melanoma and non-small cell lung cancer,^37,38 but this success was not initially replicated in ovarian cancer. Monotherapy trials did not demonstrate significant clinical benefit,^39–41 and combination strategies, including the addition of bevacizumab or PARP inhibitors, did not show a clinically meaningful survival benefit;^42–44 in ATHENA-COMBO, adding nivolumab to rucaparib maintenance did not improve progression-free survival compared with rucaparib alone.⁴⁵ This pattern was generally attributed to ovarian cancer behaving as a “cold” tumour with reduced T-cell infiltration^46,47 and a microenvironment rich in regulatory T cells (Treg) and tumour-associated macrophages (TAMs) that confers an immunosuppressive phenotype.⁴⁸ Interactions between T cells and the tumour microenvironment limit effective immune responses by preventing adequate T-cell infiltration⁴⁹ (Figure 2), and additional immune checkpoints may contribute, such that PD-1/PD-L1 and CTLA-4 blockade alone is insufficient.⁵⁰ Responses also vary by histological subtype, with some benefit reported in ovarian clear cell carcinoma,⁵¹ which may reflect differences in the immunosuppressive microenvironment described in this subtype.⁵² The Phase III ENGOT-ov65/KEYNOTE-B96 trial is an important exception, in which the addition of pembrolizumab to weekly paclitaxel with or without bevacizumab improved progression-free and overall survival in platinum-resistant recurrent disease.⁵³ Benefit was greatest in tumours expressing PD-L1, and the regimen has been approved on this basis for patients with a PD-L1 combined positive score of 1 or more. This is the first immunotherapy regimen to show an overall survival benefit in ovarian cancer, and it establishes PD-L1 as the first validated predictive biomarker for immune checkpoint blockade in this disease. It should be noted, however, that the benefit is confined to a specific chemotherapy backbone in the platinum-resistant setting, that PD-L1 expression had not predicted benefit in earlier ovarian cancer trials, and that tumour mutational burden and microsatellite instability remain of limited use here. Outside this indication, immune checkpoint inhibition in ovarian cancer remains investigational.

Figure 2.

The immune-cold tumour microenvironment in ovarian cancer. Reduced effector T-cell infiltration, together with an abundance of regulatory T cells (treg) and tumour-associated macrophages (TAMs), creates an immunosuppressive microenvironment that limits the activity of immune checkpoint inhibitors.

Abbreviations: CTLA-4, cytotoxic T-lymphocyte-associated protein 4; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; TAM, tumour-associated macrophage; Treg, regulatory T cell.

Surface Markers

Antibody-drug conjugates (ADCs) have emerged as a promising therapeutic strategy in ovarian cancer, particularly in the context of chemotherapy resistance and the limitations of conventional cytotoxic therapy. ADCs consist of a monoclonal antibody linked to a cytotoxic payload; the antibody recognises a cell surface marker on the tumour cell and, following internalisation, delivers the payload intracellularly.⁵⁴ This targeted delivery enhances antitumour activity while limiting systemic toxicity compared with traditional chemotherapy (Figure 3). ADCs are of particular interest in homologous recombination-proficient tumours, which derive limited benefit from PARP inhibition. In this class the surface antigen serves simultaneously as therapeutic target and candidate predictive biomarker, although, with the exception of folate receptor α, the predictive value of these antigens has not been formally established.

Figure 3.

Mechanism of action of antibody-drug conjugates in ovarian cancer. A monoclonal antibody binds a tumour cell surface antigen; the antibody-drug conjugate is then internalised and the cytotoxic payload released within the cell, causing tumour cell death while limiting systemic exposure. Membrane-permeable payloads may additionally kill adjacent antigen-negative cells through a bystander effect.

Abbreviations: ADC, antibody-drug conjugate.

A promising target for such drugs is folate receptor α. Ovarian cancer cells have been found to express high amounts of this surface marker.⁵⁵ Mirvetuximab soravtansine is an ADC targeting folate receptor α (FRα) which was initially licensed in the USA on the basis of a Phase II trial.⁵⁶ The first Phase III trial, FORWARD I, compared mirvetuximab soravtansine with investigator’s choice chemotherapy in platinum-resistant disease and did not improve progression-free survival, although secondary endpoints suggested improved outcomes in patients with high FRα expression.⁵⁷ The subsequent Phase III MIRASOL trial, restricted to patients with high FRα expression in platinum-resistant ovarian cancer, demonstrated improved progression-free and overall survival compared with chemotherapy.⁵⁸ FRα is currently the only surface marker in ovarian cancer that functions as a validated predictive biomarker linked to an approved therapy, and the contrast between FORWARD I and MIRASOL illustrates a wider principle: an ADC may fail in an unselected population yet succeed when the target is used to select patients, and the threshold of expression used matters as much as the target itself.

With the encouraging results seen with mirvetuximab, several other anti-folate ADCs have entered clinical trials, and attention has also turned to other surface markers. Human epidermal growth factor receptor 2 (HER2) is overexpressed in many tumours including breast cancer, where HER2-directed therapies are a standard of care,⁵⁹ and is overexpressed in a proportion of high-grade serous ovarian cancers and ovarian clear cell carcinomas.⁶⁰ The DESTINY-PanTumor02 Phase II trial evaluated the HER2-directed ADC trastuzumab deruxtecan (T-DXd) across multiple tumour types in patients with HER2 expression, including ovarian cancer that had progressed on standard treatment.⁵⁹ Clinical activity was seen across several tumour types, with an objective response rate of 45% in the ovarian cohort, and responses correlated with HER2 overexpression. This supports evaluation of HER2 as a therapeutic target in the later phase trials now under way, although the data derive from a single-arm basket study and HER2 is not yet an established predictive biomarker in ovarian cancer.

Trophoblast cell surface antigen-2 (TROP2), a transmembrane glycoprotein overexpressed in many malignancies and correlated with poorer prognosis, is also being investigated in ovarian cancer.^61,62 Early phase trials are evaluating TROP2-targeted ADCs, including the Phase II TROPION trial of datopotamab deruxtecan (Dato-DXd)⁶³ and the Phase II TROPiCS-03 trial of sacituzumab govitecan.⁶⁴ Initial results are encouraging but larger studies are needed. Emerging evidence from other tumour types suggests that TROP2 expression may correlate with response to TROP2-directed therapy,⁶⁵ but its role as a predictive biomarker in ovarian cancer has not been established.

ADCs targeting many other surface markers are undergoing clinical trials in ovarian cancer, as summarised in Table 1. These include claudins, transmembrane proteins that are integral components of tight junctions and help maintain epithelial and endothelial barriers.⁶⁶ Claudin expression is dysregulated in ovarian cancer, and claudin-6 (CLDN6) is found on ovarian cancer cells with minimal expression in normal adult tissue, making it an attractive target.⁶⁶ TORL-1-23, an ADC targeting CLDN6, is being assessed in the CATALINA-2 trial in CLDN6-positive platinum-resistant disease,⁶⁷ and first-in-human Phase I results have demonstrated promising antitumour activity.⁶⁸ CLDN6 may therefore have a role as both a therapeutic target and a biomarker, although randomised data are required to validate this.

Table 1.

Emerging Biomarkers and Targeted Therapies in Ovarian Cancer

Biomarker	Biological Feature	Therapy Example	Clinical Relevance
HRD	Defective homologous recombination repair	PARP inhibitors (olaparib, niraparib)	Current clinical standard for first-line maintenance treatment
FRα	Cell surface marker expression	ADC (mirvetuximab soravtansine)	Approved treatment selection biomarker
HER2	Cell surface marker expression	ADC (trastuzumab deruxtecan)	Phase II trials and ongoing Phase III trial evaluation
TROP2	Cell surface marker expression	ADC (datopotamab deruxtecan)	Phase II trials evaluating ADCs in ovarian cancer and other solid tumours
CLDN6	Cell surface marker expression	ADC (TORL-1-23)	Phase II trials evaluating target in ovarian cancer
B7-H3, B7-H4	Cell surface marker expression	ADC (ifinatamab deruxtecan and GSK5733584)	Phase I/II clinical trials
CDH6	Cell surface marker expression	ADC (raludotatug deruxtecan)	Phase II clinical trials
CCNE1	Cell cycle dysregulation	CDK2 inhibitors (INCB123667), PKMYT1 inhibitors (RP-6306)	Phase I trials in CCNE1-amplified tumours
ATR, WEE1	Replication stress response	ATR inhibitors (ceralasertib), WEE1 inhibitors (adavosertib, azenosertib)	Phase I/II clinical trials
TP53 Y220C	TP53 mutation with altered protein function	p53 reactivator (rezatapopt)	Phase II basket trials

Further potential therapeutic targets are B7-H3 and B7-H4, transmembrane proteins overexpressed in ovarian cancers and linked to poor prognosis.⁶⁹ They belong to the B7 ligand family, which has important roles in immune regulation and tumour progression.⁶⁹ Several ADCs targeting B7 family proteins are under investigation in early phase trials. BEHOLD1 is a Phase I trial evaluating a B7-H4-targeting ADC in advanced platinum-resistant ovarian cancer and endometrial cancer.⁷⁰ An ADC targeting B7-H3 has shown encouraging preliminary results in extensive-stage small cell lung cancer, and IDeate-PanTumor02 is an early phase basket trial assessing its efficacy in multiple solid tumours including ovarian cancer,⁷¹ although data in ovarian cancer remain limited and both targets should be regarded as preclinical to early phase at this stage.

Early results for raludotatug deruxtecan (R-DXd) for use in ovarian cancer are also promising.⁷² Raludotatug deruxtecan is an ADC targeting CDH6 which is overexpressed in epithelial ovarian cancers. Cadherin-6 (CDH6) is a transmembrane protein which has been implicated in tumour metastases and poor prognosis.⁷³ The Phase II REJOICE study demonstrated early efficacy of R-DXd in platinum resistant ovarian cancer and will continue to be evaluated in later phase studies.⁷²

The identification of these surface markers also raises important questions regarding biomarker testing and treatment selection. Most are detected using immunohistochemistry (IHC), which assesses protein expression within tumour tissue. As more antibody-drug conjugates enter clinical development, uncertainty will arise over which therapy should be prioritised, particularly in tumours expressing several targetable markers. Testing methods are evolving in parallel: multiplex immunohistochemistry (mIHC) allows several biomarkers to be assessed within a single tissue sample,⁷⁴ but although widely used in research it is not yet adopted in standard care because of technical complexity and cost.⁷⁵ Biomarker testing to identify surface markers is therefore becoming central to the selection of targeted therapy in ovarian cancer.

Emerging Precision Therapies

Beyond surface-marker-directed therapies, there is increasing interest in targeting alternative mechanisms of genomic instability, particularly in tumours that are homologous recombination proficient and therefore derive limited benefit from PARP inhibition. Mechanisms of cell cycle dysregulation and replication stress are being explored as potential therapeutic targets. It should be emphasised that the markers discussed in this section are currently biological features and therapeutic targets rather than validated predictive biomarkers, and none is yet used to select treatment outside a clinical trial.

An example of a promising strategy is targeting CCNE1. CCNE1 amplification is associated with refractory disease and with resistance to platinum-based combination chemotherapy.⁷⁶ CCNE1 encodes cyclin E1, which binds cyclin-dependent kinase 2 (CDK2) to initiate DNA synthesis, and its overexpression is associated with poor prognosis.⁷⁷ CCNE1-amplified tumours remain capable of homologous recombination and account for approximately 20% of high-grade serous ovarian cancers;⁷⁸ cyclin E1 overexpression increases replication stress and greater reliance on cell cycle checkpoint mechanisms.⁷⁹ This defines a subgroup deriving limited benefit from both standard chemotherapy and PARP inhibition, in whom alternative strategies are needed. Early phase trials of selective CDK2 inhibitors suggest a reasonable safety profile and preliminary efficacy.⁸⁰ PKMYT1 inhibition offers a further synthetic lethal approach: PKMYT1 is a protein kinase on which CCNE1-amplified tumours depend to prevent premature cell division, and its inhibition can lead to tumour cell death.⁸¹ RP-6306, a PKMYT1 inhibitor, is being investigated in advanced disease including ovarian cancer.⁸² At present CCNE1 is best regarded as a prognostic feature and a therapeutic target under investigation; its value as a predictive biomarker has not been established in prospective randomised trials.

Replication stress is increasingly recognised as a key feature of cancer, arising when tumour cells replicate DNA in a rapid and dysregulated manner with accumulation of DNA damage.^83,84 In response, cells activate checkpoint pathways that temporarily halt cell cycle progression to allow time for repair.⁸⁴ Cancer cells with high levels of replication stress become increasingly dependent on these checkpoints for survival, which represents a vulnerability that can be exploited.⁸⁵ Key regulators of this response include ataxia telangiectasia and Rad3-related protein (ATR) and WEE1, which coordinate cell cycle arrest and facilitate DNA repair in the presence of replication stress.⁸⁶

WEE1 kinase plays an important role in cell cycle progression and DNA repair and is overexpressed in many tumours.⁸⁷ In a randomised Phase II trial, the WEE1 inhibitor adavosertib (AZD1775) combined with gemcitabine improved progression-free and overall survival compared with gemcitabine alone in platinum-resistant or refractory high-grade serous ovarian cancer,⁸⁸ although significant haematological toxicity has limited its further development and application. WEE1 inhibition is thought to be enhanced in tumours harbouring TP53 mutations, since loss of p53 and WEE1 inhibition both disrupt key cell cycle checkpoints, leading to synthetic lethality and tumour cell death.⁸⁹ A Phase II study of adavosertib plus carboplatin in patients with TP53-mutated ovarian cancer refractory or resistant to first-line therapy reported an overall response rate of 43%.⁹⁰ However, a substantial proportion of patients did not benefit, and since TP53 is mutated in the great majority of high-grade serous tumours it identifies almost the entire population rather than a selected subgroup; TP53 mutation alone is therefore unlikely to serve as a useful predictive biomarker.⁸⁹ Azenosertib, an oral WEE1 kinase inhibitor, showed favourable antitumour activity in an early phase trial in metastatic gynaecological malignancies, particularly in CCNE1-positive platinum-resistant ovarian cancer, and this cohort is being investigated further in Phase II trials.⁹¹

In addition to WEE1 inhibition, targeting ATR has emerged as a strategy in tumours characterised by replication stress. ATR is a key regulator of the DNA damage response, activated primarily by damage arising during replication,⁹² and has an important role in stabilising replication forks and preventing the accumulation of DNA damage.⁹² Its inhibition disrupts these processes, leading to replication fork collapse and tumour cell death,⁹³ and tumours with high levels of replication stress, including those with CCNE1 amplification, may be particularly dependent on ATR signalling.⁹⁴ Ceralasertib (AZD6738), an oral selective ATR inhibitor, has been evaluated in patients with advanced solid tumours;^95,96 a Phase I study of ceralasertib with carboplatin demonstrated manageable toxicity, predominantly haematological, with evidence of antitumour activity across multiple tumour types.⁹⁵ In the Phase II ATARI trial, ceralasertib as monotherapy and in combination with olaparib demonstrated only modest clinical activity in patients with relapsed gynaecological cancers, including ovarian cancer.⁹⁷ These findings support further evaluation but also illustrate the central difficulty: reliance on the ATR pathway is not determined by a single mutation, and no validated biomarker yet identifies the patients most likely to benefit.⁹⁸

The ATARI trial also provides a promising treatment strategy for rarer ovarian cancers with ARID1A mutations such as ovarian clear cell carcinomas (OCCC). ARID1A alterations are frequently observed in ovarian clear cell carcinoma and have been linked to resistance to conventional therapies,⁹⁹ highlighting an area of unmet clinical need. In tumours with ARID1A alterations, impaired DNA damage repair results in increased reliance on ATR signalling to maintain genomic stability,⁹⁹ making ATR inhibition another reasonable target. This suggests that ARID1A may help identify tumours that are more likely to respond to ATR inhibition. However, ARID1A alterations are mainly seen in specific histological subtypes such as clear cell carcinoma, which limits their applicability to the broader ovarian cancer population.

Further to the targeting of TP53 loss through synthetic lethality, there is increasing interest in directly targeting specific TP53 mutations. TP53 mutations are a defining feature of high-grade serous ovarian cancer, with studies reporting mutation rates of 96%.¹⁰⁰ The Y220C mutation is a rare alteration that creates a small binding pocket within the p53 protein, causing thermal instability at physiological temperatures and impairing DNA binding and tumour suppressor function.¹⁰¹ Rezatapopt (PC14586) is a first-in-class small molecule designed to bind this pocket, stabilise the mutant protein and restore tumour suppressor function.¹⁰² Early phase data from the ongoing Phase II PYNNACLE basket trial have demonstrated encouraging activity in advanced solid tumours harbouring the TP53 Y220C mutation, including responses in ovarian cancer.^103,104 These remain early data from a non-randomised study, and the rarity of the alteration highlights wider challenges around patient identification and the feasibility of delivering targeted therapy to a very small proportion of patients.

Current Challenges

Despite the growing number of potential biomarkers in ovarian cancer, translating biomarker-directed therapy into routine clinical practice remains challenging.¹⁰⁵ Encouraging early clinical activity does not establish that a marker will function as a reliable predictive biomarker in routine care. Ovarian cancer is highly heterogeneous, with genetic differences within and between patients that may also differ between primary and recurrent disease,¹⁰⁶ and tumour growth is dynamic, influenced by interactions with the microenvironment and by treatment pressures over time.¹⁰⁷ A single tissue sample may therefore not represent the biology of the whole tumour, and a biomarker identified at diagnosis may not predict treatment response at relapse. Because biomarker assessments are often performed on archival tissue, results may not reflect real-time changes.¹⁰⁵ This applies not only to emerging targets but to HRD-guided treatment itself, since HRD status reflects tumour biology at the time of sampling and does not account for subsequent evolution.³³

In addition to biological complexity, there are practical challenges to biomarker testing in routine practice. A key consideration is the distinction between germline and somatic testing. Germline mutations are inherited alterations present in all cells, whereas somatic mutations arise within tumour cells and are not inherited;¹⁰⁸ both are important for establishing cancer risk and informing treatment. Current guidance advises germline BRCA1/2 testing for all patients with epithelial ovarian cancer and somatic tumour testing in those without a germline pathogenic variant.¹⁰⁹ In practice, somatic testing is frequently performed on tumour tissue alone, and identified variants may be somatic or underlying germline alterations, which are difficult to distinguish without matched normal DNA.¹¹⁰ This may require additional testing and lengthen clinical pathways.

Genomic profiling methods are also changing. Next-generation sequencing (NGS) allows the analysis of multiple genes simultaneously from a single tumour sample.¹¹¹ As increasingly rare molecular alterations are identified, such as TP53 Y220C, single-gene testing is unlikely to detect them, whereas NGS can identify uncommon but potentially actionable mutations.¹¹² There is therefore growing interest in incorporating NGS-based genomic profiling into routine care, particularly as the number of potential therapeutic targets continues to expand.

NGS nevertheless requires careful clinical interpretation, as not all detected variants will be clinically relevant or actionable.¹⁰⁸ Assay standardisation is a further obstacle: HRD platforms differ in algorithm and threshold and are not fully harmonised, so results are not always interchangeable between laboratories, while functional assays such as RAD51 foci detection may offer a more dynamic assessment of homologous recombination repair but remain limited by the availability of suitable tumour tissue.¹¹³ Access is also uneven, since comprehensive genomic profiling, multiplex immunohistochemistry and functional assays are not uniformly funded or available, and turnaround times may exceed the interval within which a treatment decision must be made. A biomarker supported by convincing trial data may therefore still be difficult to deploy in routine practice.

Given these limitations, there is increasing interest in circulating tumour DNA (ctDNA) as a minimally invasive means of capturing real-time tumour biology. ctDNA is a blood-based biomarker that detects somatic alterations and may enable monitoring of treatment response and early detection of recurrence,¹¹⁴ allowing repeat sampling in a way that tissue biopsy does not. However, sensitivity may be low in early disease and reliability varies between tumour types,¹¹⁵ largely because plasma levels depend on tumour shedding, which limits detection in patients with low-volume or early stage disease.¹¹⁶ Tissue and plasma are therefore best regarded as complementary rather than interchangeable: tissue remains the reference standard for HRD assessment and for immunohistochemically defined surface markers, whereas ctDNA is better suited to serial monitoring and to capturing heterogeneity that a single biopsy would miss. When profiling should be repeated is not established, although progression on a biomarker-directed therapy is the most logical point at which to reassess.

Trial design imposes its own limitations, particularly when generating evidence for rarer gene alterations or tumour subtypes. Many of these promising strategies rest on data from basket trials, which group patients by shared molecular alteration rather than tumour histology,¹¹⁷ allowing biomarker-driven therapies to be assessed efficiently across tumour types.¹¹⁸ Their central assumption – that a shared molecular alteration predicts a shared treatment response – is not always correct, since tumour lineage, heterogeneity and prior treatment all influence outcome;¹¹⁹ the tumour-agnostic structure also requires coordination across multiple specialties, moving away from disease-specific trials.¹²⁰ Results may be driven by activity in a single tumour type, and analyses of individual cohorts are often limited by small patient numbers, reducing reliability and making it difficult to determine whether treatment effects are meaningful.¹¹⁹ Adaptive features such as parallel cohorts and protocol modification during delivery add further complexity, which regulators including the FDA and MHRA have identified as a risk to data integrity.¹²¹ Bayesian approaches that share information across tumour types can partly mitigate small sample sizes,¹²² but do not resolve the underlying problem of heterogeneity. These constraints are most acute for rarer ovarian subtypes such as clear cell carcinoma, where conventional trials are not feasible and the evidence supporting biomarker-directed treatment will inevitably remain less robust.

Treatment Selection and Sequencing

Looking forward, the treatment landscape in ovarian cancer continues to evolve, raising important questions regarding appropriate treatment sequencing. With the emergence of multiple surface markers and profiling methods, challenges arise in determining which biomarker driven therapy should be prioritised, particularly when tumours express more than one targetable alteration.

In BRCA-mutated disease, PARP inhibitors are used early in the treatment pathway, whereas most emerging targets have been studied in later line settings, and there is limited evidence to guide optimal sequencing. With expanding molecular assessment, it is unclear whether genomic profiling should be repeated at progression. Archival tumour samples may not reflect current tumour biology following treatment pressures, and repeat biopsy may be required to confirm that a particular surface marker remains relevant at the time of the treatment decision, although liquid biopsy is likely to provide a non-invasive means of doing so.

In addition, the degree of biomarker expression may influence treatment response. Emerging data suggest that ADC outcomes may depend on target expression, although this has not been consistent across trials. The improved outcomes observed with mirvetuximab soravtansine in patients with high FRα expression suggest a role for prioritising treatment according to high target expression. The subsequent use of biomarker-driven therapy also requires consideration: for FRα-directed therapy, persistent surface marker expression could indicate continued clinical relevance, but evidence is currently insufficient to support serial use of agents against the same target, or to determine whether switching to an alternative target would be more effective.

As we enter an era with expanding potential therapeutic targets and diagnostic methods, further evidence and frameworks are required to guide treatment selection and sequencing.

Conclusion

The introduction of HRD testing and PARP inhibitors marked a significant step towards biomarker-driven therapy in ovarian cancer and remains the current clinical standard for selecting first-line maintenance treatment. The limitations of HRD as a predictive biomarker are now well recognised and have prompted the search for more precise and clinically applicable strategies in a biologically heterogeneous disease. Of the approaches reviewed here, only PARP inhibitors selected by BRCA or HRD status, folate receptor α-directed therapy with mirvetuximab soravtansine, and PD-L1-selected pembrolizumab in platinum-resistant disease are established in routine practice. The remainder – antibody-drug conjugates directed at HER2, TROP2, CLDN6, B7-H3/H4 and CDH6, and agents targeting CCNE1-associated replication stress, WEE1, ATR and mutant TP53 – are promising but investigational, and require prospective validation of both the therapy and the biomarker used to select for it before they can be adopted. The take-home message for clinicians is therefore twofold. First, HRD remains the biomarker that should guide first-line maintenance decisions today, and its imperfections should be understood rather than used to dismiss it. Second, future progress will depend less on the discovery of further candidate biomarkers than on validating those already identified and on developing practical frameworks for biomarker testing, treatment selection and sequencing that can be delivered within routine care.

Disclosure

Dr Liz Joseph has nothing to disclose. Dr Michael Flynn reports honoraria from CLOVIS Oncology, Eisai, AstraZeneca, GSK; meeting/travel support from AstraZeneca; data safety/advisory board participation from Epitopea, Abbvie, outside the submitted work. Professor Jonathan Ledermann reports grants or contracts from MSD/Merck, AstraZeneca; consulting fees from Flindr, Novocure, Abbvie/Immunogen, Merck/MSD, GSK, AstraZeneca; honoraria from Medison, GSK, AstraZeneca; data safety/advisory board participation from Sutro bio, Lilly, Genmab, Zentalis, AstraZeneca & Daiichi-Sankyo; leadership or fiduciary roles from Trustee Cancestry, Chair ESMO Gynae Clinical Practice Guidelines, Chair Trustees GTG-UK, outside the submitted work.