Multi-Maintenance Olaparib Therapy in Relapsed, Germline BRCA1/2-Mutant High-Grade Serous Ovarian Cancer (MOLTO): A Phase II Trial

Ovarian cancer is the commonest cause of gynecologic cancer-related death in Europe and North America (1). Most women diagnosed with ovarian cancer present with advanced disease [International Federation of Gynecology and Obstetrics (FIGO) stage III or IV; ref. 2]. Thus, despite a good response to first-line multimodality therapy, most women develop recurrent disease within 3 years of their diagnosis, at which point cure is unlikely (3). Consequently, the 5-year survival for advanced ovarian cancer is approximately 30%. New therapeutic strategies are required to improve survival for women diagnosed with this often-fatal disease.

Platinum-sensitive, relapsed epithelial ovarian cancer is treated with rechallenge platinum chemotherapy (4). A single maintenance course of a PARP inhibitor (PARPi) improves progression-free survival (PFS) in platinum-sensitive, relapsed high-grade serous or endometrioid ovarian cancer (HGS/EOC), with the greatest improvement evident in tumors harboring a BRCA1/2 mutation (5–7). Although phase III trials have demonstrated highly impressive improvements in PFS using a single maintenance course of PARPi, none have met their defined secondary endpoint for prolongation of overall survival (OS; refs. 8–10). We hypothesized that OS could be improved with multiple sequential courses of maintenance PARPi.

Epithelial ovarian cancers harboring a germline BRCA1/2 mutation (gBRCAm) retain greater sensitivity to platinum chemotherapy compared with wild-type tumors, suggesting the possibility of coexistent retained sensitivity to PARPi (11, 12). Combination therapy with a PARPi and an anti-angiogenic agent is effective in HGS/EOC, regardless of BRCA1/2 genotype (13, 14). On the basis of these observations, we hypothesized that a second maintenance course of the PARPi, olaparib, with or without the anti-angiogenic agent, cediranib, would improve survival outcomes in relapsed, gBRCAm-HGS/EOC (15).

Both the use of platinum-based chemotherapy and PARPi increases the risk of therapy-related myeloid neoplasms (t-MN) such as myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML; refs. 16, 17). It was unknown whether PARPi retreatment would result in higher rates of t-MNs or be complicated by significant hematologic toxicity. We therefore conducted MOLTO, a phase II trial, to investigate the safety and efficacy of two sequential courses of maintenance olaparib in relapsed, gBRCAm-HGS/EOC.

There are no licensed biomarkers to guide the use of maintenance PARPi therapy in relapsed ovarian cancer. Preclinical data have shown that functional homologous recombination deficiency (HRD), according to Rad51 nuclear foci formation, predicts PARPi cytotoxicity in vitro (18). Genome-wide somatic copy-number alteration (SCNA) is a key genetic hallmark of high-grade serous ovarian cancer (HGSOC) and can be detected in cell-free circulating DNA (cfDNA; refs. 19, 20). BRCA1/2 reversion mutations are the best-studied mechanism of platinum and PARPi resistance in gBRCAm tumors (21). Thus, we interrogated longitudinal fresh tumor and liquid biopsies from patients enrolled in MOLTO to gain an understanding of the utility of these biomarkers to predict responses to platinum-olaparib treatment.

MOLTO was an investigator-led, single-center, nonrandomized, phase II trial. Eligibility criteria included relapsed, progressive, HGS/EOC of the ovary, fallopian tube, or primary peritoneum; age ≥18 years old; an Eastern Cooperative Oncology Group performance status of 0 to 2; a germline BRCA1/2 pathogenic or likely pathogenic variant (22); a life expectancy >12 weeks; at least one prior line of platinum chemotherapy to treat ovarian cancer; measurable disease by RECIST version 1.1 (23); adequate bone marrow (hemoglobin ≥90 g/L; platelet count ≥100 × 10⁹/L; absolute neutrophil count ≥1.5 × 10⁹/L), clotting (international normalized ratio <1.4 and activated partial thromboplastin ratio <1.4), liver function [total serum bilirubin ≤1.5 × upper limit of normal (ULN) and aspartate aminotransferase (AST) or alanine transaminase (ALT) ≤2.5 × ULN in the absence of liver metastases or ≤5 × ULN in the presence of liver metastases] and renal function (calculated creatinine clearance or corrected isotopic clearance measurement ≥51 mL/min); and an ability to swallow oral medication.

Exclusion criteria included patients considered a poor medical risk due to a serious uncontrolled medical disorder, nonmalignant systemic disease, or active uncontrolled infection including viral Hepatitis B and C or human immunodeficiency virus; symptomatic central nervous system metastases including leptomeningeal carcinomatosis; a resting electrocardiogram with a corrected QT interval >470 milliseconds on ≥2 time points within a 24-hour period or a family history of long QT syndrome; concomitant use of strong or moderate CYP3A inhibitors or inducers; another malignancy within the last 5 years except curatively treated nonmelanoma skin cancer, cervical carcinoma in situ, ductal carcinoma in situ, FIGO (2009) stage I grade 1 endometrial cancers, or another solid tumor including localized breast cancer and lymphoma; pregnant or breast feeding women; evidence of adverse events from prior treatment that had not resolved to grade ≤1 (except alopecia or grade 2 neurotoxicity); radiotherapy, chemotherapy, or tumor embolization within 21 days starting study treatment; major surgery within 2 weeks before starting study treatment; additional concurrent systemic anticancer treatment including maintenance bevacizumab. Patients with a somatic BRCA1/2 mutation were also excluded. One of the key issues we wished to address in MOLTO was the safety of retreatment with a PARPi. As such, it was important to look for off-target toxicities in organs harboring a mono-allelic BRCA1/2 mutation. These would not have been present in patients with a somatic BRCA1/2 mutation.

Specific exclusion criteria for cediranib included concurrent or a past history of malignant fistula; uncontrolled hypertension; sustained hypertension as defined by ≥3 readings of systolic blood pressure >140 mmHg or a diastolic blood pressure >90 mmHg; arterial thrombosis within 6 months of trial entry; clinically significant abnormalities that could increase the risk of gastrointestinal perforation; proteinuria >2 g/24 hours; a past history of VEGF inhibitor-associated reversible posterior leukoencephalopathy syndrome (RPLS).

All patients provided written informed consent prior to enrollment. The trial was adherent to the principles outlined in the Medicines for Human Use (Clinical Trials) regulations 2004 and International Council for Harmonisation Good Clinical Practice guidelines. The trial was conducted in compliance with the protocol, the Data Protection Act, the Declaration of Helsinki, Human Tissue Act (2004), the Research Governance Framework (2005), and other applicable regulations.

There were two entry points in the trial: Entry Point 1 (EP1) and Entry Point 2 (EP2). All patients were recruited prior to rechallenge platinum chemotherapy. Patients recruited at EP1 had never been treated with a PARPi. Patients recruited at EP2 had received a first maintenance course of olaparib immediately prior to trial enrollment. Patients commenced maintenance olaparib with or without cediranib following completion of four to six cycles of platinum chemotherapy and within 4 to 8 weeks of completing their last dose of platinum chemotherapy.

Dose levels for olaparib (150 or 100 mg tablets) were 300 mg twice daily (dose level 0), 250 mg twice daily (dose level −1), and 200 mg twice daily (dose level −2). Dose levels for cediranib (20 or 15 mg tablets) were 20 mg once daily (dose level 0), 15 mg once daily (dose level −1), and 15 mg once daily 5 days on and 2 days off (dose level −2). For maintenance olaparib with cediranib, the starting doses were olaparib 300 mg twice daily and cediranib 20 mg once daily. The treating physician could reduce the dose of olaparib and/or cediranib depending upon the adverse event encountered.

Olaparib was interrupted for common terminology criteria for adverse events (CTCAE) version 4.03 grade ≥3 anemia (hemoglobin <80 g/L), grade ≥3 neutropenia (neutrophil count <1.0 × 10⁹/L), grade ≥3 thrombocytopenia (platelet count <50 × 10⁹/L), grade ≥3 febrile neutropenia, an AST or ALT increase ≥3 × ULN, a total serum bilirubin increase ≥2 × ULN, or any adverse event that was considered by the investigator to warrant treatment interruption. Cediranib was interrupted for >2 g of proteinuria detected in a 24-hour collection or any adverse event that was considered by the investigator to warrant treatment interruption. Olaparib and cediranib could be interrupted for a maximum of 4 weeks and only restarted once the adverse event had returned to grade ≤1.

Olaparib was permanently discontinued following the diagnosis of MDS, AML, any other new primary malignancy, pneumonitis confirmed by CT, or any adverse event that was considered by the treating physician to warrant discontinuation. Cediranib was permanently discontinued following a diagnosis of persistent or severe hypertension despite optimal use of anti-hypertensive medications, gastrointestinal perforation, fistula, arterial thromboembolism, severe hemorrhage requiring medical intervention, radiologically confirmed RPLS or any adverse event that was considered by the investigator to warrant discontinuation. Olaparib and cediranib were permanently discontinued if a grade ≥2 adverse event had not returned to grade ≤1 within 4 weeks of treatment interruption. All protocol treatment was discontinued following clinical or RECIST-defined radiologic progression to platinum chemotherapy and the second maintenance course of olaparib ± cediranib, unacceptable toxicity, noncompliance, or withdrawal of consent.

To be eligible to receive a first course of maintenance Olaparib, a patient had to have RECIST complete or partial response (CR/PR) to their first course of rechallenge platinum on the trial (hereon referred to as “platinum 1”). To be eligible to receive a second course of maintenance olaparib ± cediranib, a patient had to have RECIST CR/PR or stable disease to rechallenge platinum following progression on their first course of olaparib (hereon referred to as “platinum 2”). For those patients treated with platinum chemotherapy following progression on a first course of olaparib, the platinum-free interval (PFI) determined whether a patient received cediranib as part of their second course of maintenance therapy. If the PFI was ≥6 months, the patient would receive a second maintenance course of olaparib monotherapy. If the PFI was <6 months, the patient received a second maintenance course of olaparib with cediranib.

All patients recruited to MOLTO had cross-sectional imaging of the abdomen and pelvis ± thorax (preferably CT, but MRI was permitted) as part of radiologic tumor assessment, within 28 days prior to receiving each course of platinum chemotherapy (hereon referred to as the “pretreatment scan”). Although RECIST measurable disease was required for trial entry, those patients enrolled at EP1 who went on to receive platinum 2 did not require RECIST measurable disease prior to platinum 2 to continue on trial. Further cross-sectional imaging, using the same imaging modality used for the pretreatment scan, occurred following completion of a course of platinum chemotherapy (hereon referred to as the “end of platinum scan”), every 12 weeks during olaparib ± cediranib for the first 24 months and every 24 weeks thereafter whilst receiving maintenance therapy.

A washout period of at least 28 days was scheduled between the first course of olaparib and platinum 2. Active surveillance could take place following RECIST progression on the first course of olaparib and prior to receiving platinum 2. This accounted for any patient who had RECIST-defined radiologic progression, but no evidence of clinical progression during their first course of olaparib. During the active surveillance phase of the trial, patients were assessed in clinic every month and CT/MRI scans were scheduled every 12 weeks. An earlier CT/MRI scan could be performed following the onset of symptoms of clinical progression. An asymptomatic rise in CA125 could trigger an earlier CT/MRI scan, however the patient was not offered platinum 2 unless they had clinical or radiologic evidence of progression.

Patients could undergo secondary cytoreductive surgery following progressive disease on their first course of olaparib after agreement by the local multidisciplinary team. In such cases, platinum 2 had to commence within 4 to 8 weeks of secondary cytoreductive surgery.

The co-primary outcomes were the proportion of patients who received a second course of olaparib (feasibility proportion) and the proportion of patients who received a second course of olaparib for ≥6 months (6-month maintenance proportion). A patient was evaluable for the co-primary outcomes if they received at least one dose of platinum 2.

Secondary outcomes included PFS, time to first subsequent therapy (TFST), PFS2, OS, and objective responses to platinum 1 and platinum 2. PFS was defined as the time interval between cycle 1 day 1 of the first maintenance course of olaparib to the date of clinical or RECIST-defined progression or death, whichever occurred first. Time to first subsequent therapy was defined as the time interval between cycle 1 day 1 of the first maintenance course of olaparib to cycle 1 day 1 of platinum 2 or death, whichever occurred first. PFS2 was defined as the time interval between cycle 1 day 1 of the first maintenance course of olaparib to the date of clinical or RECIST-defined progression or death following platinum 2, whichever occurred first. OS was defined as the time interval from cycle 1 day 1 of the first maintenance course of olaparib to the date of death.

Objective responses were defined according to RECIST version 1.1 and Gynecological Cancer Intergroup (GCIG) CA125 criteria (23, 24). The RECIST response to platinum 1 and 2 was determined by comparing the pretreatment scan with the end of platinum scan. To be evaluable for GCIG CA125 response to platinum 1 and platinum 2, a patient was required to have a pretreatment CA125 level 2 × ULN (normal range 0–30 U/mL) within 28 days prior to cycle 1 day 1 of platinum.

All adverse events were graded according to the NCI CTCAE version 4.03 preferred terms and system organ classifications. All grade ≥2 adverse events that occurred during olaparib ± cediranib were included in the safety analysis. All serious adverse events that occurred during the trial were included in the safety analysis. Adverse events of special interest included AML, MDS, other new primary malignancy and pneumonitis.

The aim of the trial was to determine the feasibility of delivering two successive courses of maintenance olaparib in women diagnosed with relapsed gBRCAm-HGS/EOC. The co-primary endpoints were first, the proportion of patients who received a second maintenance course of olaparib and second, of those, the proportion of patients who received a second maintenance course of olaparib for ≥6 months. The point estimates of co-primary endpoints were supplemented with an exact binomial confidence interval (CI) having at least 95% coverage. For the co-primary endpoints, it was predicted that over a 24-month recruitment period, 26 participants would be enrolled for an estimated 20 evaluable patients. No statistical calculation was performed to determine the sample size because at the time the trial was set up, no similar trial had been conducted and so no safety calculation could be used to design MOLTO.

Categorical data were reported as a percentage (95% CI) or number (percentage). Continuous data were reported as median, mean, and range. Time-to-event data were calculated using the Kaplan–Meier method. All statistical analyses were performed using R.

The trial was completed on December 31, 2021. The final trial database lock was April 7, 2022. The study was registered with ClinicalTrials.gov, NCT02855697.

Fresh tumor biopsies were taken prior to commencing each course of platinum chemotherapy. Liquid biopsies were taken prior to commencing each course of platinum and olaparib ± cediranib. A detailed description of the molecular assays used in MOLTO is provided in the Supplementary Methodology. Next-generation sequencing was used to detect BRCA1/2 and TP53 mutations in germline DNA, tumor DNA, and cfDNA (25). De novo somatic mutations in germline BRCA1 or BRCA2 were called as reversion mutations if they were: (i) a base substitution that changed a nonsense mutation to a missense mutation or (ii) an insertion/deletion that restored the open-reading frame (26). Ultra-low-pass whole-genome sequencing was used to detect genome-wide SCNA in cfDNA (20). Primary cultures were generated from fresh tumor biopsies to test functional HRD using the Rad51 ex vivo immunofluorescence assay (18, 27).

The data generated from this trial are available within the article and its Supplementary Data. Any data that are not included in the article or supplementary files can be requested from the Manchester Clinical Trials Unit or the chief investigator (GCJ). Data will be shared to researchers if the submitted methodologic proposal is considered appropriate and achievable by the chief investigator (GCJ). All individual participant data shared will be anonymized.

The trial was funded by AstraZeneca UK (ESR-15–10650). The funder had no role in study design, data collection, analysis, interpretation, or writing of the manuscript. The corresponding author (RDM), trial statistician (WDJR), and chief investigator (GCJ) had full access to all data from the trial.

Between May 26, 2017, and September 9, 2019, 28 patients were enrolled in MOLTO and 27 started treatment (Fig. 1). One patient enrolled at EP1 was found to be ineligible prior to commencing treatment due to the lack RECIST measurable disease and was excluded. Thus, 17 patients were treated at EP1 and 10 at EP2. Demographic data are presented in Table 1. At the scheduled date of trial closure, 25 patients had discontinued trial treatment due to clinical or radiologic progression, withdrawal from trial or death. The remaining 2 patients were exceptional responders to olaparib, having received their first course for over 24 months without disease progression. The median follow-up was 22.8 months (range: 1.6–50.6 months, 23 patients) including patients enrolled at EP1 (median: 30.6 months; range: 9.5–50.6 months, 13 patients) and EP2 (median: 15.3 months; range: 1.6–40.2 months, 10 patients). The 4 patients who did not receive a first course of olaparib were excluded from the follow-up analysis.

At the scheduled date of trial closure, 19 patients had received at least one dose of platinum 2 and were evaluable for the co-primary outcomes. Of these 19 patients, 12 (63%; 95% CI, 38%–84%) received a second course of olaparib (Fig. 1). Of these 12 patients, 4 (33%; 95% CI, 10%–65%) received a second course of olaparib for ≥6 months (all olaparib monotherapy). Four of 19 (21%; 95% CI, 6%–46%) patients commencing platinum 2 received ≥6 months of olaparib retreatment.

Almost all patients received six cycles of platinum doublet chemotherapy for platinum 1 (25/27, 93%) and platinum 2 (19/19, 100%). The RECIST and GCIG CA125 responses were lower for platinum 2 compared with platinum 1 (Table 2). The RECIST CR/PR to platinum 1 and platinum 2 were 76% (13/17) and 50% (8/16), respectively (Fig. 2). The GCIG CA125 response to platinum 1 and platinum 2 were 93% (14/15) and 64% (9/14), respectively (Fig. 2).

The secondary efficacy outcomes were calculated for the 19 patients eligible for platinum 2 following progression on their first course of olaparib (Supplementary Fig. S1). The median PFS was 8.7 months (95% CI, 4.7–12.2 months, 19 events). The median TFST was 11.8 months (95% CI, 6.9–18.5 months, 19 events). The median PFS2 was 17.4 months (95% CI, 13.6–24.7 months, 19 events). The median OS was 30.5 months (95% CI, 17.5–43.6 months, 14 events).

There were no new safety signals for olaparib retreatment. The grade ≥2 adverse events reported during the first and second courses of olaparib are shown in Table 3. The most common grade ≥2 adverse events during olaparib retreatment were anemia, nausea, and fatigue. No cases of MDS, AML, other new primary malignancy, or pneumonitis were reported during the median follow-up of 22.8 months (range: 1.6–50.6 months). Two deaths occurred as a result of trial treatment, both during platinum 2. One patient died from acute coronary syndrome and one patient died from non-neutropenic sepsis. All serious adverse events reported during the trial are shown in Supplementary Table S1.

A post hoc analysis was performed to investigate the duration of each maintenance course of olaparib. There was a significant difference between the mean duration of treatment of the first and second courses of olaparib (t test P < 0.001). The mean duration of treatment of the first course of olaparib was 12.1 months [standard deviation (SD): 9.8 months; interquartile range (IQR), 5.5–17.6 months, 23 patients]. The reasons for discontinuing the first course of olaparib included radiologic progression (20 patients) or noncompliance with the trial protocol (1 patient). The mean duration of treatment of the second course of olaparib was 4.4 months (SD: 2.7 months; IQR, 2.6–6.1 months, 12 patients). All patients discontinued the second course of olaparib due to clinical or radiologic progression (12 patients). There was no difference in mean duration of treatment of the second course of olaparib in patients with RECIST stable disease versus CR/PR following platinum 2 (3.6 months vs. 4.3 months, respectively, t test P = 0.25). There was no difference in mean duration of treatment of the second course of olaparib in patients who received their first course of olaparib for 6 to 12 months versus >12 months (3.5 months vs. 5.5 months, respectively, t test P = 0.73). The mean total duration of exposure of olaparib was 14.4 months (SD: 10.5 months; IQR, 5.7–21.6 months, 23 patients).

Tumor biopsies were tested for functional HRD, somatic TP53 mutations, and BRCA1/2 reversion mutations. Liquid biopsies were tested for genome-wide SCNA, somatic TP53 mutations, and BRCA1/2 reversion mutations.

Primary cell cultures were attempted for 28 fresh tumor biopsies (15 pre-platinum 1 and 13 pre-platinum 2). Of these, functional homologous recombination repair (HR) status, based on Rad51 nuclear foci formation, was successfully determined in 19 (68%). Eight (42%) primary cultures were reported as functional HRD and 11 (58%) were reported as HR proficient (HRP; Supplementary Fig. S2). For the two patients with longitudinal primary cultures (patient 12 and 15), pre-platinum 1 and 2 biopsies were both reported as HRP in patient 12, and for patient 15 the pre-platinum 1 biopsy was reported as HRP, whereas the pre-platinum 2 biopsy was reported as HRD. There was no correlation between functional HR repair status and PFS (Supplementary Table S2).

Liquid biopsies were taken prior to commencing both courses of platinum and olaparib ± cediranib (Supplementary Table S3). Cell-free circulating DNA was extracted from 55 liquid biopsies. The cfDNA yield ranged from 1.6 to 172.5 ng/mL. Thirty-seven (67%) and 12 (22%) liquid biopsies contained >5 and >10 ng/mL of cfDNA, respectively. There was no difference between the cfDNA yields from plasma samples acquired at each time point (Supplementary Fig. S3). There was no correlation between cfDNA yield and serum CA125 levels (Pearson correlation, P = 0.74). There was a significant positive correlation between cfDNA yield and tumor volume (Pearson correlation, P < 0.0001). Genome-wide SCNA was detected in 24 (44%) liquid biopsies (Supplementary Fig. S4). Genome-wide SCNA was most frequently detected prior to platinum chemotherapy, as opposed to olaparib [61% (22/36) vs. 11% (2/19), Chi squared test P < 0.001]. There was no correlation between SCNA status and PFS (Supplementary Table S2). Further analysis showed no correlation between individual bioinformatic algorithms used to determine SCNA status (Moran statistics and Z-score) and PFS.

Tumor DNA from 31 biopsies were genotyped for BRCA1/2 mutations (9 archival pre-platinum 1, 10 fresh pre-platinum 1, and 12 fresh pre-platinum 2). All gBRCAm were detected in tumor DNA extracted from solid tumor biopsies at a variant allele frequency (VAF) of ≥50% consistent with bi-allelic loss-of-function mutations. No BRCA1/2 reversion mutations were detected in any tumor biopsy. Further analysis of all 12 pre-platinum 2 biopsies using a minimum VAF cut-off of ≥1% did not detect any BRCA1/2 reversion mutations.

Liquid biopsies taken prior to platinum 2 were analyzed for BRCA1/2 reversion mutations. To ensure a high confidence of reporting true positive BRCA1/2 reversion mutations in cell-free circulating tumor DNA (ctDNA), we tested cfDNA from liquid biopsies containing a detectable somatic TP53 mutation. This strategy was possible because the concordance rate between somatic TP53 mutations in tumor DNA and cfDNA was high (76%). Of the 14 liquid biopsies taken from patients prior to receiving platinum 2, one contained a putative BRCA2 reversion mutation (Supplementary Fig. S5). In this putative BRCA2 reversion mutation, a thymine nucleotide had been inserted 13 base pairs downstream (BRCA2 c.8310_8311insT) from the original germline variant (BRCA2 c.8297del), thereby restoring the open reading frame (p.Thr2766AsnfsTer11 → p.Thr2766_Ala2770delinsAsnLeuLeuLysPro). Analysis of the sequencing data showed that both variants were in cis. This patient died during platinum 2 from acute coronary syndrome. Therefore, we were unable to determine whether her response to platinum 2 or olaparib retreatment was affected by the presence of the BRCA2 reversion mutation.

At the time when MOLTO was set up, the pivotal olaparib phase II trial, Study 19, described one of the most impressive PFS hazard ratios reported in ovarian oncology, yet no OS benefit was reported (28–31). Study 19 had randomized patients with platinum-sensitive, relapsed HGSOC to receive either a single maintenance course of olaparib monotherapy or placebo. We hypothesized that retreatment with a PARPi would be needed to improve OS, but there were concerns that retreatment might incur severe toxicities and/or may be ineffective (15, 32). We therefore used co-primary endpoints that included safety, which we termed feasibility, and efficacy for which we used a 6-month PFI as the working definition on the grounds that this approximated to retained platinum-sensitivity (33). All participant's recruited to MOLTO had been diagnosed with gBRCAm-HGSOC and were predicted to retain platinum sensitivity despite prior platinum exposure (11, 12). We had anticipated that by allowing patients to receive olaparib retreatment following RECIST CR/PR or stable disease following platinum 2, most patients would be eligible for, and therefore potentially benefit from, a second course of maintenance olaparib (34).

The results of the co-primary outcomes in MOLTO show that rechallenge olaparib can be administered to women diagnosed with relapsed, gBRCAm-HGSOC. Almost two-thirds of patients rechallenged with platinum received a single dose of olaparib re-treatment and one-third of these patients were re-treated for ≥6 months. However, it is notable that only one-fifth of patients who received platinum rechallenge were re-treated with olaparib for ≥6 months, restricting this therapeutic strategy to a minority of patients, therefore limiting its clinical value. Moreover, because MOLTO did not include a placebo arm, it is unknown whether olaparib retreatment provided any therapeutic benefit after platinum rechallenge (35).

It is important to interpret the safety data in MOLTO within the context of the duration of olaparib therapy as well as the length of follow-up. No cases of MDS, AML, or other primary malignancy were reported following a mean total treatment duration of 14.4 months and a median overall follow-up of 22.8 months. These time periods were relatively short. In the phase III trial, SOLO2/ENGOT Ov-21, 8% of women diagnosed with gBRCAm-HGSOC developed MDS/AML after an average of 29.1 months of a single course of olaparib and 65.7 months of follow-up (8). Moreover, 56% of MDS/AML cases reported in SOLO2 occurred after the safety follow-up cut-off, suggesting a relatively long latency period is required to develop t-MNs after starting PARPi. Participants in MOLTO may therefore have lacked sufficient followed-up for t-MNs to develop (15).

The finding of modest efficacy for olaparib rechallenge (mean duration of olaparib retreatment was 4.4 months) in MOLTO aligns with recently presented data from the phase IIIb trial, OReO/ENGOT Ov-38 (36). In this trial, patients diagnosed with nonmucinous epithelial ovarian cancer who had progressed on any prior PARPi received olaparib retreatment (300 mg twice daily) or placebo. In the subgroup of tumors harboring a BRCA1/2 mutation (n = 74) retreated with olaparib, the median PFS was 4.3 months. The median PFS in the placebo group (n = 38) was 2.8 months (HR, 0.57; 95% CI, 0.37–0.87; P = 0.022; ref. 36). There were notable differences in the eligibility criteria used in OReO and MOLTO. In OReO, patients had to have received a first course of olaparib for ≥18 months or ≥12 months following their first or later lines of platinum chemotherapy, respectively. Also, to be eligible for a second course of olaparib in OReO, patients required RECIST CR/PR to rechallenge platinum. In contrast, in MOLTO, no prior longevity of olaparib treatment was mandated and patients could have RECIST CR/PR or stable disease prior to receiving olaparib retreatment. In MOLTO, 50% of patients retreated with olaparib had received their first course of PARPi for <12 months and 25% had RECIST stable disease to platinum 2. These data suggest that the cohort recruited to MOLTO had more chemotherapy-resistant disease than those recruited to OReO. Despite these differences, the finding of only modest efficacy for olaparib retreatment was similar in both MOLTO and OReO.

MOLTO provides important data about the treatment responses to rechallenge platinum in relapsed, gBRCAm-HGSOC. Until now, only retrospective studies had reported response rates of ≤40% (37–40). In contrast, MOLTO showed RECIST and GCIG CA125 responses of 50% and 64%, respectively. This finding of good responses to rechallenge platinum but relatively poorer responses to olaparib retreatment suggests a lost relationship between platinum and olaparib sensitivity following prior olaparib exposure (28, 41). A similar observation has been reported in a post hoc analysis of the phase III trial, SOLO2, in which olaparib maintenance appeared to be associated with shorter time to subsequent treatment in platinum-retreated patients (42). Future studies should aim to understand which PARPi resistance mechanisms retain sensitivity to platinum, thereby optimizing selection of patients more likely to benefit from rechallenge platinum.

MOLTO shows that a primary cell culture assay that reports functional HR repair status according to Rad51 nuclear foci formation can be incorporated in to a clinical trial. Previously reported trials investigating PARPi efficacy in HRD tumors have used mutational “scar” assays on fixed tumor tissue to determine HR repair status (43, 44). In MOLTO, 61% of fresh solid biopsies were expanded into living primary cultures and could be tested using the Rad51 assay. Ultimately, the success of any cell culture assay depends upon the ability to grow cancer cells from tissue biopsies. The failure to generate primary cultures in almost 40% of fresh tumor biopsies probably resulted from a paucity of viable cancer cells present in a single core biopsy. Further work is required to optimize culture conditions to improve the likelihood of generating primary cultures from small volumes of solid tumor tissue.

For those samples that were successfully tested using the Rad51 assay, we found no correlation between low levels of Rad51 nuclear foci and PARPi efficacy. The number of samples tested is too small to draw any clinically meaningful conclusions regarding the utility of HR repair function to predict PARPi responses. However, the fact that some BRCA1/2-mutant tumors had high levels of Rad51 nuclear foci following DNA damage suggests that Rad51 formation may not correlate accurately with functional HR status in primary explant cell cultures. In addition, it must be noted that fresh tumor biopsies were taken prior to platinum, as opposed to before olaparib. Thus, PFS in the biomarker study had to include platinum and olaparib efficacy to account for the timing of biopsies. When the Rad51 assay was originally developed, functional HRD correlated with in vitro sensitivity to PARPi alone (18). The reduction in tumor volume that occurs following platinum chemotherapy often prohibits tumor biopsies being taken prior to starting maintenance therapy. The in vivo relationship between functional HRD and PARPi efficacy may therefore be more appropriately assessed by investigating PARPi therapy, not maintenance therapy.

Cell-free DNA was tested in MOLTO for SCNA as a surrogate marker of genomic instability. We hypothesized that genomic instability could be tested using minimally invasive liquid biopsies (45, 46). Liquid biopsies were taken on multiple occasions prior to platinum and olaparib to understand the optimal timepoint for testing. MOLTO shows that there is a cohort of ovarian cancers where SCNA is detectable in cfDNA, and that SCNA-positive cfDNA is more likely to be detected prior to platinum, not maintenance PARPi. The reduction in the proportion of plasma samples with detectable SCNA between pre-platinum and pre-olaparib timepoints probably reflects the reduction in tumor volume following platinum. This observation is likely to prohibit the use of SCNA as a biomarker in trials investigating maintenance therapies. However, the lack of significant improvement in PFS following platinum-olaparib in patients with SCNA-positive versus SNCA-negative cfDNA should not halt further investigations of this novel biomarker (47). Indeed, to understand the potential of cfDNA-based SCNA, larger populations of patients must be tested including newly diagnosed women and those with BRCA1/2 wild-type.

The MOLTO trial adds to the small number of clinical studies demonstrating that BRCA1/2 reversion mutations are detectable in ctDNA from women diagnosed with ovarian cancer (26, 48–50). The BRCA2 reversion mutation detected in MOLTO is likely to be a true positive result because the variant was detected at a high VAF in a plasma sample that contained a somatic TP53 mutation detected at a high VAF (26). Unfortunately, there was no solid tissue available to confirm the presence of the BRCA2 reversion mutation in solid tumor DNA. We were also unable to determine whether the presence of the BRCA2 reversion correlated with a poor response to retreatment platinum-olaparib because the patient died during platinum 2 (26, 48–50). We acknowledge that other PARPi resistance mechanisms have not been investigated in MOLTO other than BRCA1/2 reversions.

In conclusion, we report that olaparib retreatment in relapsed gBRCAm-HGSOC is safe but only modestly efficacious. Future studies should focus on determining the most appropriate therapy to use following the onset of PARPi resistance in ovarian cancer.

A.R. Clamp reports grants from AstraZeneca during the conduct of the study as well as personal fees from Clovis Oncology, MSD, GSK, and Immunogen and grants and personal fees from AstraZeneca, outside the submitted work. W.D.J. Ryder reports grants from AstraZeneca during the conduct of the study. A.D. Murphy reports nonfinancial support from MSD outside the submitted work. D.G.R. Evans reports personal fees from AstraZeneca during the conduct of the study. C. Dive reports grants and personal fees from AstraZeneca and Boehringer Ingelheim; grants from Amgen, Carrick Therapeutics, Merck AG, Taiho Oncology, Bayer, Roche, BMS, Novartis, Celgene, Epigene Therapeutics Inc., Angle PLC, and Menarini; and personal fees from Merck and GRAIL outside the submitted work. A.J. Wallace reports other support from AstraZeneca outside the submitted work. R.J. Edmondson reports personal fees from GSK and Clovis outside the submitted work. G.C. Jayson reports grants from AstraZeneca during the conduct of the study as well as grants from AstraZeneca outside the submitted work. No disclosures were reported by the other authors.

R.D. Morgan: Data curation, formal analysis, funding acquisition, investigation, methodology, writing–original draft, writing–review and editing. A.R. Clamp: Conceptualization, supervision, funding acquisition, investigation, methodology, writing–original draft. D.J. White: Data curation, validation, investigation, methodology, writing–original draft. M. Price: Data curation, validation, investigation, methodology, writing–original draft. G.J. Burghel: Data curation, validation, investigation, methodology, writing–original draft. W.D.J. Ryder: Data curation, validation, methodology, writing–original draft. R.D. Mahmood: Investigation, writing–review and editing. A.D. Murphy: Investigation, writing–review and editing. J. Hasan: Investigation, writing–review and editing. C.L. Mitchell: Investigation, writing–review and editing. Z. Salih: Investigation, writing–review and editing. C. Wheeler: Investigation, writing–review and editing. E. Buckley: Investigation, writing–review and editing. J. Truelove: Project administration, writing–review and editing. G. King: Data curation, validation, project administration, writing–review and editing. Y. Ainaoui: Data curation, validation, project administration, writing–review and editing. S.S. Bhaskar: Investigation, writing–review and editing. J. Shaw: Data curation, validation, investigation, methodology, writing–original draft, writing–review and editing. D.G.R. Evans: Investigation, writing–review and editing. B. Kilerci: Data curation, investigation, writing–review and editing. S.P. Pearce: Data curation, investigation, writing–review and editing. G. Brady: Supervision, validation, investigation, methodology, writing–review and editing. C. Dive: Supervision, validation, investigation, methodology, writing–review and editing. J.P.B. O'Connor: Data curation, validation, investigation, methodology, writing–original draft. A.J. Wallace: Data curation, validation, investigation, methodology, writing–original draft. D.G. Rothwell: Data curation, supervision, validation, investigation, methodology, writing–original draft. R.J. Edmondson: Data curation, formal analysis, supervision, validation, investigation, methodology, writing–original draft. G.C. Jayson: Conceptualization, data curation, formal analysis, supervision, funding acquisition, validation, investigation, methodology, writing–original draft.

The trial was funded by AstraZeneca UK (ESR-15–10650). The trial was sponsored by the Christie Foundation NHS Trust and was monitored by the Manchester Academic Health Science Centre Trials Co-ordination Unit on behalf of the sponsor. The trial was supported by the Manchester Cancer Research Centre Biobank. The trial was also supported by the United Kingdom Clinical Research Collaboration-registered King's Clinical Trials Unit at King's Health Partners, which is partly funded by the National Institute for Health Research (NIHR) Biomedical Research Centre for Mental Health at South London and Maudsley NHS Foundation Trust and King's College London and the NIHR Evaluation, Trials and Studies Coordinating Centre. The trial protocol and supporting documentation were reviewed by the North West Research Ethics Committee (16/NW/0825).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.