Phase I Trial of the PARP Inhibitor Olaparib and AKT Inhibitor Capivasertib in Patients with BRCA1/2- and Non–BRCA1/2-Mutant Cancers

PARP inhibitors are the first clinically approved drugs designed to exploit synthetic lethality in homologous recombination (HR)–deficient cells, demonstrating proof-of-concept activity in BRCA1/2-mutant cancers (1, 2). The PARP inhibitor olaparib (Lynparza; AstraZeneca) was the first in class to be approved by the FDA in the advanced recurrent ovarian cancer setting for women with BRCA1/2-mutant cancers (3). Olaparib has subsequently received FDA approval in the maintenance setting for recurrent platinum-sensitive ovarian cancers regardless of BRCA1/2 status and, most recently, in the first-line maintenance setting post–platinum-based chemotherapy both in women with germline or somatic BRCA1/2-mutated advanced ovarian cancer and in patients with germline BRCA1/2-mutated pancreatic cancer (4, 5). Olaparib is also FDA-approved for the treatment of patients with germline BRCA1/2-mutant, HER2-negative metastatic breast cancers (6, 7). In addition, there are now early clinical trial data from patients with tumors harboring other DNA repair aberrations (8).

Despite these broader indications, the greatest clinical benefit from PARP inhibitor monotherapy has been observed in the high-grade serous germline BRCA1/2-mutant ovarian cancer population (3). However, these patients almost inevitably develop PARP inhibitor resistance and disease progression. The utility of PARP inhibitor monotherapy in patients with different cancers harboring other DNA-repair aberrations is also limited by the emergence of drug resistance and generally short-lived antitumor responses (9). Even for patients with advanced solid tumors bearing deleterious BRCA1/2 mutations, response rates are between 30% and 60% depending on tumor type (3). There is therefore a major unmet need for novel antitumor strategies to increase both the proportion of patients with clinical benefit as well as the depth and duration of response for patients treated with PARP inhibitors (10). Such approaches include the development of rational combination strategies. Multiple preclinical studies have demonstrated synergistic antitumor activity with the combination of PARP and PI3K–AKT pathway inhibitors in both BRCA-deficient and BRCA-proficient cancer models (11–14). PI3K pathway inhibition has been shown to lead to suppression of BRCA gene transcription, which was accompanied by ERK phosphorylation. Overexpression of an active form of MEK1 was found to result in ERK activation and downregulation of BRCA1, resulting in HR deficiency and subsequent PARP inhibitor sensitivity, thus providing strong rationale for the development of this combination as an antitumor strategy (11). A phase Ib trial of olaparib in combination with the PI3Kα-specific inhibitor alpelisib demonstrated RECIST partial responses (PR) in 10 (36%) of 28 patients with ovarian cancer, providing early clinical proof of concept (15).

Several novel molecularly targeted agents against the PI3K–AKT pathway have now been developed, including the AKT inhibitor capivasertib (AZD5363; AstraZeneca; ref. 16). Capivasertib is a potent and selective ATP competitive inhibitor of all three isoforms of AKT, which is safe and well-tolerated in patients with advanced solid tumors, but with limited antitumor benefit as a single agent in early-phase clinical trials (17–19). This is hypothesized to be due to multiple factors, including the development of signaling cross-talk and disruption of feedback loops, leading to acquired resistance, supporting the use of combination strategies in molecularly defined patients for the optimal development of AKT inhibitors, such as with PARP inhibitors as described above (20, 21).

On the basis of these promising data, we conducted an investigator-initiated phase Ib clinical trial to determine the safety, tolerability, MTD, recommended phase II dose (RP2D), pharmacokinetics, pharmacodynamics, and preliminary antitumor activity of olaparib in combination with capivasertib in patients with advanced solid tumors. We also assessed a prospective intrapatient dose-escalation strategy where patients were permitted to prospectively increase doses of capivasertib after each cycle (if no grade 2 or worse toxicities were observed) in combination with a fixed olaparib dose. RP2D expansion cohorts were undertaken in patients with (i) germline BRCA1/2-mutant cancers and (ii sporadic cancers with DNA damage response (DDR) aberrations or molecular abnormalities along the PI3K–AKT pathway. Detailed analyses of archived and fresh sequential tumor biopsies, as well as targeted sequencing of serial cell-free DNA (cfDNA) samples, were conducted to identify determinants of response and resistance, including genomic factors and protein expression, and pharmacodynamic mechanism-of-action biomarker studies.

We enrolled 64 patients with advanced solid tumors from four major cancer centers in the United Kingdom into the dose-escalation (20 patients) or dose-expansion (44 patients) cohorts of this phase I trial. Characteristics of these patients are provided in Table 1. The most common tumor enrolled was advanced ovarian cancer (39% of patients); these patients had received a median of five prior therapies (range 1–12). The next most common tumor type was advanced breast cancer (28% of patients); these patients had received a median number of three prior therapies (range 2–10).

The prospective intrapatient dose-escalation strategy utilized during each dose schedule allowed rapid, seamless, and safe dose escalation, resulting in completion of the dose-escalation phases of two combination schedules over three dose levels in 7 months (Supplementary Fig. S1). Overall, only 10 patients were required to assess three different dose levels for each schedule and to establish both RP2Ds, respectively.

During the dose-escalation phase of the 4-days-on, 3-days-off (4/3) schedule, doses of capivasertib were increased using the prospective intrapatient dose-escalation design from 320 and 400 mg to 480 mg twice a day with a fixed dose of olaparib at 300 mg twice a day (Supplementary Fig. S1). Of 10 patients treated in this 4/3 schedule, only one dose-limiting toxicity (DLT) of grade 3 maculopapular rash, typical of that observed with capivasertib and other AKT inhibitors, was observed at the highest dose assessed of capivasertib, 480 mg twice a day, with olaparib 300 mg twice a day (Table 2). The erythematous rash fully resolved after both capivasertib and olaparib were withheld, and no recurrent rash was observed after both drugs were restarted at a reduced dose of capivasertib 400 mg twice a day with olaparib maintained at 300 mg twice a day. At the dose level of capivasertib 480 mg twice a day with olaparib 300 mg twice a day administered in a 4/3 schedule, other non-DLT grade 3 toxicities were observed, including anemia (n = 1), vomiting (n = 1), and diarrhea (n = 2).Because of the DLT of grade 3 rash, non-DLT grade 3 adverse events, and chronic low-grade adverse events, for example, fatigue and anemia, observed outside the DLT period of 21 days at the dose of capivasertib 480 mg twice a day with olaparib 300 mg twice a day, the Safety Review Committee (SRC) established the dose level of olaparib 300 mg twice a day with capivasertib 400 mg twice a day as the RP2D for the 4/3 schedule.

For the 2-days-on, 5-days-off (2/5) schedule of capivasertib with olaparib 300 mg twice a day, dose escalation proceeded through dose levels of 480, 560, and 640 mg twice a day with olaparib 300 mg twice a day. In the 10 patients treated on the 2/5 schedule, no DLTs were observed, and the highest dose level of capivasertib at 640 mg twice a day with olaparib 300 mg twice a day was selected as the RP2D. In view of similarities in overall safety, tolerability, and DLT rates, the SRC elected to explore both the 4/3 and 2/5 schedules of capivasertib in the dose-expansion phase.

The most common all-grade treatment-emergent adverse events (TEAE) observed for all patients across both dose schedules were gastrointestinal (GI) toxicities, including nausea [67% (grade 3–4, 4%)], diarrhea [55% (grade 3–4, 6%)], and vomiting [41% (grade 3–4, 5%)], as well as fatigue [51% (grade 3–4, 5%); Table 2; Supplementary Tables S1 and S2]. Other significant grade 3–4 toxicities included grade 3 anemia (10%) on the 4/3 schedule. Overall, the 4/3 schedule appeared to be less well tolerated than the 2/5 schedule, as exhibited during dose escalation: six grade 3 TEAEs [anemia and diarrhea (n = 2 each); rash and vomiting (n = 1 each)] were observed with the 4/3 schedule, and only three grade 3 TEAEs [liver transaminitis, fatigue, and hyperglycemia (all n = 1)] in the 2/5 schedule. No drug-related grade 4–5 toxicities were observed in either schedule.

Dose escalation of capivasertib showed dose-dependent increases in pharmacokinetic exposures (Fig. 1A and B; Supplementary Tables S3 and S4). The pharmacokinetic profile and overall concentration–time profile of capivasertib and olaparib were similar to that previously observed in single-agent studies, with no significant interactions identified.

Pharmacodynamic studies in platelet-rich plasma (PRP) showed significant decrease in Ser9 GSK3β phosphorylation post treatment at all doses in the escalation and expansion phases (Fig. 2A and B), confirming target modulation by capivasertib. Phosphorylated ERK expression levels assessed with IHC increased in fresh tumor biopsies collected on cycle 1 day 15 compared with baseline samples in 6 of 8 patients, while remaining unchanged in 1 patient and decreasing in another patient (Fig. 2C). At the same timepoint, BRCA1 expression decreased in paired fresh tumor biopsies obtained from all 8 patients (Fig. 2D).

The antitumor activity of the combination of capivasertib and olaparib is detailed in Table 3, Fig. 3A–C, and Supplementary Table S5. Of the 56 patients who were evaluable for antitumor response, 19 (34%) patients had RECIST PRs and/or tumor marker response [Gynaecologic Cancer InterGroup (GCIG) CA-125 response or Prostate Cancer Clinical Trials Working Group 2 (PCWG2) PSA response]. Fourteen (25%) patients achieved RECIST PRs (12 confirmed and 2 unconfirmed). In addition, 11 (20%) patients had RECIST stable disease (SD) for at least 4 months (SD ≥ 4 months), giving a clinical benefit rate (CBR) of 44.6% (95% confidence interval, 31.3–58.5). Of these 56 evaluable patients, 16 (29%) patients were treated on study for more than 6 months, whereas 7 (13%) patients were treated on trial for more than 1 year.

Among the 25 (44.6%) patients who achieved clinical benefit [RECIST complete response (CR)/PR or SD ≥ 4 months], 14 (56%) patients had germline BRCA1 or BRCA2 mutations [ovarian cancer (n = 7), breast cancer (n = 5), and castration-resistant prostate cancer (CRPC; n = 2); Table 3; Fig. 3A–C; Supplementary Table S5]. Seven of the remaining patients had pathogenic DDR or PI3K pathway aberrations detected, whereas 3 patients did not, and 1 patient did not have available tissue for next-generation sequencing (NGS) testing.

Three RECIST-evaluable patients harboring tumors with both DDR-related and PI3K pathway mutations achieved RECIST PRs (Fig. 3B). The mutations for these 3 patients were: (i) germline ERCC2 mutation, somatic PIK3CA mutation, and PTEN mutations and PTEN IHC loss; (ii somatic BRCA2 mutation and PTEN IHC loss; and (iii) germline BRCA2 and PIK3CA mutations, respectively. Eight of 22 (36.4%) patients with tumors harboring only DDR-related mutations achieved RECIST PR, including those with BRCA1 (n = 5), BRCA2 (n = 2), and PALB2 (n = 1) mutations. A further 8 patients had a best response of SD ≥ 4 months. Among those patients with tumors harboring only PI3K pathway mutations (n = 5), there was one objective response in a patient with a tumor found to have a PTEN mutation. In patients with tumors harboring neither PI3K pathway nor DDR-related mutations (n = 25), there were two RECIST PRs.

Among 25 patients with BRCA1/2-mutant cancers [20 with germline BRCA1/2 mutations, 5 with somatic BRCA1/2 mutations; breast cancer (n = 7), ovarian cancer (n = 15), and CRPC (n = 3)], 22 patients had RECIST measurable disease; 16 (72%) of these 22 patients achieved clinical benefit with the combination of olaparib and capivasertib.

A total of 18 patients with advanced breast cancer were enrolled onto the study, 8 (44%) of whom achieved clinical benefit (Supplementary Table S5). Five (71.4%) of 7 patients with BRCA1/2-mutant breast cancer had clinical benefit; 4 had RECIST PR and 1 had SD of 19.4 weeks, with a median duration of response of 39.1 weeks (range: 14.9–80.9). Two of these responders with clinical benefit were platinum-resistant. None of the responding patients with advanced breast cancer had prior therapy with PARP or PI3K pathway inhibitors.

There were a total of 25 patients with advanced ovarian cancer, 11 of whom achieved clinical benefit (Supplementary Table S5). Seven (63.6%) of these 11 patients with germline BRCA1/2-mutant ovarian cancer achieved clinical benefit for a median duration of response of 24 weeks (range 11.3–115.0); 6 of these 7 patients were platinum-resistant. Four other patients with advanced ovarian cancer who also achieved clinical benefit included those with tumors harboring (i) somatic BRCA1, TP53, and AR mutations, (ii somatic BRCA2 and TP53 mutations, (iii) somatic PTEN, KRAS, and SMARCA4 mutations, and (iv) somatic TP53 mutation.

Of 4 patients with advanced CRPC, 3 had germline BRCA1/2 mutations, of whom 2 achieved clinical benefit (Supplementary Table S5). None of the 4 patients had received prior platinum-based chemotherapy.

Thirteen patients had prior exposure to PARP inhibitors, 5 of whom had clinical benefit on this combination study (Supplementary Table S6). One of the patients, who had high-grade serous ovarian cancer (HGSOC) with a somatic BRCA2 mutation, achieved RECIST PR and GCIG CA-125 response lasting 31 weeks, whereas another patient with platinum-resistant germline BRCA2-mutant HGSOC with somatic TP53 mutation and somatic BRCA1 variant of unknown significance achieved a GCIG CA-125 response and RECIST SD lasting 56 weeks. Another patient, who had previously received a PARP inhibitor, had platinum-resistant HGSOC harboring a germline BRCA2 mutation and achieved a GCIG CA-125 response and RECIST SD on this trial lasting 115 weeks. Only 1 patient had previously received a PI3K pathway inhibitor prior to this clinical trial; she was a patient with advanced peritoneal mesothelioma who had previously achieved a RECIST PR on a single-agent PI3K pathway inhibitor prior to eventually progressing. On this clinical trial, she achieved a CA-125 response by GCIG criteria and durable RECIST SD lasting 84 weeks before progression.

A total of 157 cfDNA samples were serially collected from 41 patients for analysis on a targeted NGS panel. Of these patients, at least one mutation was detected in baseline cfDNA samples from 38 (93%) patients. Of 39 patients where both tumor and cfDNA samples were available for analysis, mutation status at baseline was concordant between tumor and cfDNA samples in 34 (87.2%) patients. All germline and somatic mutations detected through germline and/or tumor testing were detected in cfDNA. The most common mutations detected in cfDNA included TP53 [n = 26 (63.4%) patients], BRCA2 [n = 11 (26.8%)], BRCA1 [n = 7 (17.1%)], KRAS [n = 4 (9.8%)], PIK3CA [n = 3 (7.3%)], ARID1A [n = 3 (7.3%)], and PTEN [n = 2 (4.9%)].

The cfDNA allele frequencies of somatic mutations decreased in selected responding patients and increased upon disease progression (Supplementary Fig. S2A and S2F). In patients with known germline mutations, for example BRCA1/2 mutations, falls in the cfDNA allele frequencies of germline mutations toward 50% were observed as they responded to trial therapy (Supplementary Fig. S2C and S2D). Of 20 patients harboring germline and/or somatic BRCA1/2 mutations with available cfDNA sampling, 5 patients were found to have BRCA1/2 reversion mutations at disease progression in their end-of-treatment cfDNA samples (Supplementary Table S7). One of these patients with advanced ovarian cancer, who had early disease progression after 4 weeks on trial, had a tumor somatic BRCA1 mutation (c.329insA, p.K110fs*4) in archived tissue, but was found to have a secondary BRCA1 mutation deletion restoring the original reading frame (c.335_338delATAA) in her cfDNA sample collected at baseline on cycle 1 day 1 and at disease progression.

In this study, we have shown that the novel combination of olaparib and capivasertib is well tolerated at biologically effective doses that achieve clinical benefit, including durable responses, in patients with a range of treatment-refractory cancers, including both germline BRCA1/2-mutated tumors and sporadic cancers harboring actionable somatic alterations. Antitumor responses were also observed in patients who had previously developed disease progression on PARP and PI3K pathway inhibitors.

Two different intermittent schedules of capivasertib were assessed to determine differences in safety, tolerability, pharmacokinetics–pharmacodynamics activity, and antitumor responses in combination with olaparib, by comparing a high dose of capivasertib given over a shorter 2/5 schedule versus a lower dose of capivasertib over a longer 4/3 schedule. Overall, this combination was generally well tolerated on both schedules; treatment-related toxicities were reversible and were mainly GI-related, including diarrhea, mucositis, nausea, and anorexia. Such potential overlapping GI toxicities were effectively managed with simple supportive measures, such as antiemetics and antidiarrheals, when indicated clinically. No DLTs were observed on the 2/5 schedule, whereas one DLT of reversible grade 3 rash was observed at the highest tested dose of 480 mg twice a day of capivasertib with 300 mg twice a day of olaparib in the 4/3 schedule. Dose-proportional pharmacokinetics were observed, and proof-of-mechanism pharmacodynamic studies confirmed AKT pathway modulation across dose levels in PRP. Two different combination RP2Ds were established: 400 mg twice a day of capivasertib with 300 mg twice a day of olaparib for the 4/3 schedule, and 640 mg twice a day of capivasertib with 300 mg twice a day of olaparib for the 2/5 schedule.

This study employed a prospective intrapatient dose-escalation trial design, which enabled the rapid completion of dose-escalation phases of two different combination schedules, each involving three dose levels, within 7 months and requiring only a total of 10 patients in each schedule. Apart from optimizing speed and minimizing patient numbers, this prospective intrapatient dose escalation has benefits over established phase I escalation strategies, including minimization of patient numbers receiving subtherapeutic drug doses, and safe optimization of drug exposures at an individual level to ensure maximal blockade of critical targets for combination strategies.

In preclinical studies, inhibition of the PI3K pathway has been shown to lead to upregulation of poly-ADP-ribosylation and phosphorylation of H2AX, indicating increased DNA damage in cells (12). The accumulation of unrepaired DNA double-stranded breaks in BRCA-deficient cells in turn makes them exquisitely sensitive to PARP inhibition, potentially accounting for the synergistic effects seen in combined olaparib and capivasertib treatment. Apart from germline BRCA-mutant breast cancers, preclinical studies have also demonstrated that PI3K inhibition in triple-negative breast cancer (TNBC) models drives ERK-dependent activation of the ETS transcription factor, which suppresses BRCA gene transcription, causing a deficiency of HR activity and PARP inhibitor sensitivity (11). Correlative tumor studies from this study have shown increase in tumor phosphorylated ERK expression associated with decrease in BRCA1 expression levels, supporting this hypothesis and rationale for this novel combination (14). Given that ERK phosphorylation may potentially have negative protumor consequences as well, future studies should investigate whether this may represent a compensatory response that blunts antitumor efficacy.

The combination of olaparib and capivasertib on both 4/3 and 2/5 schedules have shown evidence of antitumor activity, with clinical benefit observed in 44.6% of evaluable patients on study. Fourteen (56%) of these 25 responding patients harbored germline BRCA1/2 mutations, whereas 8 (32%) of 25 had relevant somatic aberrations, including BRCA2, PIK3CA, PTEN, and PALB2 mutations (Fig. 3A). All 3 patients with tumors harboring both DDR-related and PI3K pathway–related aberrations had RECIST PRs, 8 of 23 evaluable patients with tumors harboring only DDR-related tumor mutations achieved RECIST PR, and 1 of 5 patients deficient for PTEN had a RECIST PR (Fig. 2B and C). Importantly, clinical benefit was also observed in patients with neither PI3K pathway– nor DDR-related tumor mutations detected, as well as in patients who had previously progressed on PARP or PI3K pathway inhibitors. Though based on small numbers, these findings are in keeping with preclinical studies, which indicate that efficacy is not necessarily confined to tumors with actionable mutations such as BRCA1/2 mutations (11, 12). Regardless, a suitably powered randomized trial will be necessary to formally determine whether there is synergistic patient benefit in the clinic with this capivasertib and olaparib combination.

In this study, concordance in the detection of selected mutations between cfDNA and tumor was 87.2%, supporting the use of cfDNA for the contemporaneous molecular profiling of patients. This high concordance may be associated with the large proportion of patients with germline mutation cancers included in this trial. Similar to previous studies (22), the cfDNA allele frequencies of somatic mutations decreased in selected responding patients and increased upon disease progression. Falls in the allele frequencies of germline mutations toward 50% were observed as they responded to trial therapy, suggesting elimination of the tumor clone (22). The development of BRCA1/2 reversion mutations was observed at disease progression in cfDNA from 5 (25%) of 20 patients with BRCA1/2-mutant cancers with available cfDNA for analysis (Supplementary Table S7). This finding supports BRCA reversion as a likely resistance mechanism in PARP inhibitor–based therapies, including regimens such as this olaparib–capivasertib combination, and advocates the use of serial cfDNA sampling longitudinally in detecting the emergence of such aberrations. The patient with a BRCA1 reversion mutation, detected in cfDNA at baseline and again at early disease progression, had primary resistance to this combination, suggesting that the detection of such aberrations in cfDNA should also be considered as part of screening tests prior to PARP inhibitor–based therapies (22, 23).

Five (71.4%) of 7 patients with germline BRCA1/2-mutant breast cancers achieved clinical benefit. In the phase III OlympiAD trial, where patients with germline BRCA1/2-mutant HER2-negative metastatic breast cancer were randomized to receive olaparib or single-agent chemotherapy of the physician's choice, the response rate was 59.9% in the olaparib group and 28.8% in the standard therapy group (6). Although based on small numbers, our study provides preliminary clinical data that supports the combination of capivasertib with olaparib as a rational strategy to potentially improve patient benefit beyond that of single-agent olaparib. The addition of capivasertib to first-line paclitaxel chemotherapy for TNBC was also shown to lead to significantly longer progression-free survival (PFS) and overall survival (OS) in a randomized, double-blind, placebo-controlled phase II trial, particularly in patients with tumors harboring PIK3CA/AKT1/PTEN alterations (24). In addition, the FAKTION phase II trial showed that the addition of capivasertib to fulvestrant in patients with endocrine-resistant, advanced estrogen receptor–positive breast cancer also resulted in significantly longer PFS and an observed OS improvement of approximately 6 months, although this was not statistically significant (37% OS data maturity; ref. 25). Phase III trials of capivasertib-based combinations are planned or ongoing.

We observed clinical benefit in 7 of 10 patients (median duration of response for these responders was 24 weeks, range 11.3–115.0) with germline BRCA1/2-mutant ovarian cancer, of which 6 were platinum-resistant. Overall, 6 (24%) of 25 patients with advanced ovarian cancer achieved RECIST PR, whereas 5 (20%) had RECIST SD. An ongoing phase Ib/II trial of olaparib and capivasertib in patients with advanced ovarian, endometrial, and triple-negative breast cancers used the combination RP2D established in this trial of 400 mg twice a day of capivasertib with 300 mg twice a day of olaparib in the 4/3 schedule. Preliminary results from this trial have shown a RECIST PR rate of 7 (24%) of 30 patients, including RECIST PRs in 4 of 8 patients with recurrent endometrial cancer, 2 patients with TNBC, and 1 patient with ovarian cancer, as well as RECIST SD ≥ 4 months in 6 additional patients (26). This compares with 10 (36%) of 28 patients with ovarian cancer achieving RECIST PR, and 14 (50%) attaining RECIST SD in the phase Ib trial of olaparib in combination with the α-specific PI3K inhibitor alpelisib (15).

In summary, capivasertib in combination with olaparib was well tolerated in both 4/3 and 2/5 schedules. Blockade of AKT led to a downregulation of pSer9 pGSK3β, indicating target modulation, while increased pERK and decreased BRCA1 expression provided translational mechanistic insights into potential synergistic activity between capivasertib and olaparib. Antitumor activity was observed more frequently in patients with either germline BRCA1/2-mutant cancers or sporadic cancers with somatic aberrations along the DDR and PI3K/AKT pathways. There were also preliminary signals of clinical benefit in platinum-resistant HGSOC and in patients who had received prior PARP inhibitors. Our results support the development of the combination of olaparib and capivasertib as a promising strategy that warrants further exploration in future clinical trials.

This investigator-initiated study (ClinicalTrials.gov: NCT02338622) was designed by T.A. Yap and J.S. de Bono, with support from AstraZeneca, and conducted in accordance with the provision of the Declaration of Helsinki and Good Clinical Practice guidelines. Dose escalation was conducted at the Royal Marsden NHS Foundation Trust (RM; London, United Kingdom), and the dose-expansion phase also involved University College London Hospital (London, United Kingdom), Northern Centre for Cancer Care (Newcastle, United Kingdom), and Cambridge University Hospitals NHS Foundation Trust (Cambridge, United Kingdom). The Central London Research Ethics Committee (REC) approved the protocol (REC reference 14/LO/0103). The trial was cosponsored by the Institute of Cancer Research (ICR; London, United Kingdom) and RM, and centrally managed by the Drug Development Unit (DDU; Investigator Initiated Trials Team) at the ICR/RM. Funding was provided by the Experimental Cancer Medicine Centers (ECMC) network, National Institute of Health Research, Cancer Research UK (CRUK), and AstraZeneca.

Eligible patients had histologically confirmed advanced solid tumors refractory to standard therapies and Eastern Cooperative Oncology Group performance status (ECOG PS) 0–1. Complete eligibility criteria are available in the Supplementary Data. Dose-expansion cohorts mandated patients with germline BRCA1/2-mutated tumors (cohort A) or germline BRCA1/2 wild-type patients with sporadic tumors likely to harbor HR defects or demonstrating somatic aberrations known to result in a hyperactivated PI3K–AKT pathway (or defective DNA repair; cohort B). The dose-escalation cohort was enriched for patients with characteristics mandated in expansion cohorts A and B, but this was not a requirement. Written informed consent was obtained from all patients.

This was an open-label, multicenter phase Ib trial assessing the combination of olaparib and capivasertib in patients with advanced solid tumors. The primary objectives were to determine the safety and tolerability of olaparib in combination with capivasertib, and to establish a MTD and/or RP2D of this combination. Secondary objectives included the characterization of pharmacokinetic and pharmacodynamic profiles of both agents in combination. Exploratory objectives included the assessment of preliminary antitumor activity of the combination and evaluation of putative predictive biomarkers of response and resistance. Study conduct was overseen by an SRC, comprising the chief investigator, principal investigator or delegate from each investigational site, DDU pharmacovigilance officer or delegate, CR-UK's Drug Development Office medical advisor or delegate, an observer from AstraZeneca, a Clinical Trials Manager or delegate, an independent senior ECMC network clinician, and a Royal Marsden Hospital representative who was independent of the study team.

This phase I trial utilized a prospective intrapatient dose-escalation trial design, where doses of capivasertib were prospectively escalated in each patient in combination with a fixed continuous dose of olaparib (Supplementary Fig. S1; Supplementary Methods).

Toxicities and laboratory variables were assessed using the Common Terminology Criteria for Adverse Events version 4.0. Patients had safety evaluations weekly and tumor response assessments after every three treatment cycles, using CT scans evaluated by RECIST version 1.1. As appropriate, different tumor markers were used to assess the effects of study treatment on respective tumor types; for example, serum CA-125 was assessed in patients with ovarian cancer according to GCIG criteria or serum PSA levels were assessed in patients with CRPC according to PCWG2 criteria.

Pharmacokinetics modeling was conducted using a noncompartmental extravascular model for plasma with Phoenix WinNonLin Software version 64 (Pharsight). Pharmacodynamics biomarker analysis of pSer9 GSK3β was undertaken on PRP where available, using assays validated to Good Clinical Practice standards on the MesoScale Discovery (MSD) technology platform (Supplementary Methods; ref. 27). ERK expression was assessed using IHC conducted on formalin-fixed, paraffin-embedded (FFPE) tissue sections (Supplementary Methods). BRCA1 expression was assessed using BRCA1 pan-nuclear IHC staining (Supplementary Methods).

Targeted NGS studies were conducted on patients with available tumor tissue at the ICR (London, United Kingdom; Supplementary Methods); libraries were constructed with the use of GeneRead DNA seq Panel (Qiagen) and run on a MiSeq Sequencer (Illumina). Whole-exome sequencing was also performed on germline and tumor DNA of responding patients using Illumina HiSeq 2500 in paired-end mode. Variants were annotated as pathogenic if considered “pathogenic” or “likely pathogenic” according to the ClinVar database (28) and/or ACMG classification using VarSome (29). PTEN loss was assessed by IHC performed on FFPE tissue sections from archived tumors where available, as described previously (30).

Analysis was conducted after all the patients had received one cycle of treatment and had completed their last study visit conducted 28 days after the last dose of combination treatment. Patients were evaluable for antitumor efficacy assessment if they had a baseline RECIST assessment and at least one post-baseline RECIST assessment. Under the assumption that the true underlying CBR was 20%, with anticipated recruitment of 40 patients in the expansion phase, there would be a < 1% chance of seeing no responses, and a > 99% chance of observing two or more responses.

T.A. Yap reports grants and non-financial support from AstraZeneca during the conduct of the study; grants, personal fees, and non-financial support from AstraZeneca outside the submitted work; research support (to institution) from Artios, Bayer, Clovis, Constellation, Cyteir, Eli Lilly, EMD Serono, Forbius, F-Star, GlaxoSmithKline, Genentech, ImmuneSensor, Ipsen, Jounce, Karyopharm, Kyowa, Merck, Novartis, Pfizer, Ribon Therapeutics, Regeneron, Repare, Sanofi, Scholar Rock, Seattle Genetics, Tesaro, and Vertex Pharmaceuticals, and the following consultancies: Almac, Aduro, AstraZeneca, Atrin, Axiom, Bayer, Bristol-Myers Squibb, Calithera, Clovis, Cybrexa, EMD Serono, F-Star, Guidepoint, Ignyta, I-Mab, Janssen, Merck, Pfizer, Repare, Roche, Rubius, Schrodinger, Seattle Genetics, Varian, and Zai Lab. R. Kristeleit reports personal fees from AstraZeneca (conference travel, consultancy, and advisory boards), GlaxoSmithKline (conference travel, consultancy, and advisory board), and Basilea (consultancy); grants and personal fees from Clovis (conference travel, fellow grant, consultancy, and advisory boards), and Merck (trial grant and consultancy), and personal fees from Sierra Oncology (conference travel and consultancy) outside the submitted work. J.S.J. Lim reports personal fees from Pfizer (honorarium and consulting), AstraZeneca (honorarium and consulting), and Novartis (honorarium and consulting) outside the submitted work. R. Miller reports personal fees and non-financial support from AstraZeneca, personal fees from Tesaro-GlaxoSmithKline and Roche, grants and personal fees from MSD, and personal fees from Clovis Oncology outside the submitted work. D. Nava Rodrigues reports other support from AstraZeneca (shareholder) during the conduct of the study. N.C. Turner reports grants and personal fees from AstraZeneca, Merck Sharpe & Dohme, Pfizer, and Roche/Genentech, personal fees from Bristol-Myers Squibb, Lilly, Novartis, Bicycle Therapeutics, Taiho, Zeno Pharmaceuticals, and Repare Therapeutics, non-financial support from Bio-Rad, grants from Clovis, and non-financial support from Guardant Health outside the submitted work. F.I. Raynaud reports other support from The Institute of Cancer Research (involved in the development of AKT inhibitors leading to AZD5363) outside the submitted work. S. Decordova reports grants from CRUK [grant number (C309/A11566) support of the Clinical PD Biomarker Group Laboratory equipment servicing and maintenance] during the conduct of the study as well as other support from Boston Pharmaceutical Inc. (salary and research funding), Menarini Ricerche (salary and research funding), AstraZeneca (salary and research funding), and CRUK (salary and research funding) outside the submitted work. K.E. Swales reports grants from CRUK [CRUK (grant number C347/A18077) funds Karen Swales post and CRUK (grant number C309/A11566) support of the Clinical PD Biomarker Group Laboratory equipment servicing and maintenance]during the conduct of the study as well as other support from AstraZeneca (research funding), Bayer (research funding), Boston Pharmaceuticals Inc. (research funding), Sierra Oncology (research funding), Menarini Ricerche (research funding), and Merck (research funding) outside the submitted work, and reports employment with The Institute of Cancer Research, which is involved in the development of PI3K, HSP90, HDAC, AKT, ROCK, RAF, CHK1, and HSF1 inhibitors. L. Finneran reports grants from AstraZeneca (grant provided to institution for database and statistical support for this project) during the conduct of the study as well as grants from AstraZeneca (grant provided to institution to support salary) outside the submitted work. E. Hall reports non-financial support from AstraZeneca (study drug supplied by AstraZeneca), grant funding to support central trial management and running costs at ICR-CTSU during the conduct of the study, as well as grants from Kyowa Hakko UK (educational grant to ICR to support central trial costs at ICR-CTSU for the BOXIT trial in intermediate/high risk non-muscle invasive bladder cancer) and Alliance Pharma (was Cambridge Laboratories; educational grant to ICR to support central trial costs at ICR-CTSU for the BOXIT trial in intermediate/high risk non-muscle invasive bladder cancer), grants and non-financial support from Merck Sharpe & Dohme [educational grant to host institution (ICR) to support central trial costs at ICR-CTSU and IMP supplies for an investigator-initiated prostate cancer study], Bayer Healthcare Pharmaceuticals Inc. (educational grant to ICR to support central trial costs at ICR-CTSU for an investigator-initiated cancer study), AstraZeneca (educational grant to ICR to support central trial costs at ICR-CTSU for investigator-initiated prostate cancer studies), Accuray Inc. (grant received by institution as contribution to the central trial management of the PACE clinical trial of SBRT in prostate cancer), Aventis Pharma Limited (Sanofi; educational grant to ICR to support central trial costs at ICR-CTSU for investigator-initiated prostate and penile cancer studies), and Janssen-Cilag (grant received by institution as contribution to support trial health economic evaluation in an academic trial in prostate cancer), and grants from Varian Medical Systems Inc. (grant received by institution as contribution to support central trial costs at ICR-CTSU for an academic prostate radiotherapy trial) and Roche Products Ltd. (grant received by institution as contribution to support central trial costs at ICR-CTSU for an academic bladder cancer trial) outside the submitted work. P. Rugman reports other support from AstraZeneca (employee of AstraZeneca) outside the submitted work. J.P.O. Lindemann reports other support from AstraZeneca (employee) during the conduct of the study. A. Foxley reports other support from AstraZeneca (employee) during the conduct of the study. C.J. Lord reports grants and personal fees from AstraZeneca, Merck KGaA, and Artios, personal fees from Syncona, Sun Pharma, Gerson Lehrman Group, Vertex, and Ono Pharma, personal fees and other support from Tango (stock), and other support from Ovibio (stock) outside the submitted work; he has a patent for PARP inhibitors and stands to gain as part of the ICR (rewards to inventors) scheme issued and with royalties paid from AstraZeneca. U. Banerji reports other support from The Institute of Cancer Research (employee of The Institute of Cancer Research, which is involved in the development of PI3K, HSP90, HDAC, AKT, ROCK, RAF, CHK1, and HSF1 inhibitors) during the conduct of the study as well as grants from AstraZeneca [phase I IIT (TAX-TORC)], Onyx [phase I IIT (ONX-0801)], BTG International (additional funding ONX-0801), Carrick Therapeutics [phase I IIT CT900 (previously ONX-0801)], Chugai Pharma Japan (phase I IIT RO5126766), and Verastem/Chugai Pharma Japan [phase I IIT (FRAME)], personal fees from Astellas (Precision Med - Advanced Prostate Cancer Mtg, London, United Kingdom, November 2018), Novartis (Transl Clin Oncology Adv Bd, London, United Kingdom, July 2018), Karus Therapeutics (Clin Adv Bd, London, United Kingdom, March 2018), Phoenix Solutions (GI Adv Bd, London, United Kingdom, December 2, 2017), Eli Lilly (Eur Digestive Oncology Res Forum, London, United Kingdom, November 2017), Janssen (Business Dev Clin Panel, Philadelphia, October 2019), and Boehringer-Ingelheim [BETi in NUT carcinoma (2019–2020), Boston, MA, October 2019], and other support from Bayer (return travel to attend BAY19829 Investigator Mtg, Houston, January 2020) outside the submitted work. R. Plummer reports other support from Cancer Research UK (trial costs) during the conduct of the study. B. Basu reports other support from GenMab (consultancy), Roche (advisory board), and Eisai Europe Limited (advisory board), grants from Celgene (investigator initiated trial), non-financial support from Bayer (travel and registration for Congress), and other support from Eisai Europe Limited (speakers bureau) outside the submitted work. J.S. Lopez reports grants from AstraZeneca during the conduct of the study as well as Roche Genentech, grants and personal fees from Basilea, and grants from Genmab outside the submitted work. Y. Drew reports grants and personal fees from AstraZeneca outside the submitted work. J.S. de Bono reports grants from AstraZeneca (trial funded by AstraZeneca) during the conduct of the study as well as grants, personal fees, and non-financial support from AstraZeneca (advisory board), grants from Pfizer (advisory board), personal fees from Merck Serono (advisory board), MSD (advisory board), Daiichi (advisory board), Sanofi Aventis (advisory board), Bayer (advisory board), Janssen (advisory board), Amgen (advisory board), and Genentech Roche (advisory board) outside the submitted work; ICR has a patent on PARPi for DNA repair cancers issued, and ICR has a commercial interest in capivasertib issued; and he has served as the chief investigator of trials with olaparib as well as capivasertib and the AKT inhibitor ipatasertib. No potential conflicts of interest were disclosed by the other authors.

T. Yap: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing. R. Kristeleit: Resources, supervision, investigation, project administration, writing-review and editing. V. Michalarea: Resources, writing-review and editing. S.J. Pettitt: Resources, data curation, formal analysis, validation, investigation, visualization, methodology, writing-original draft, writing-review and editing. J.S.J. Lim: Resources, writing-original draft, writing-review and editing. S. Carreira: Resources, data curation, formal analysis, supervision, validation, investigation, visualization, writing-review and editing. D. Roda: Resources, data curation, formal analysis, validation, investigation, writing-review and editing. R. Miller: Resources, project administration, writing-review and editing. R. Riisnaes: Resources, data curation, formal analysis, supervision, validation, investigation, visualization, writing-review and editing. S. Miranda: Resources, data curation, formal analysis, validation, investigation, writing-review and editing. I. Figueiredo: Resources, data curation, formal analysis, validation, investigation, writing-review and editing. D. Nava Rodrigues: Resources, data curation, formal analysis, validation, investigation, writing-review and editing. S. Ward: Data curation, investigation, writing-review and editing. R. Matthews: Data curation, investigation, writing-review and editing. M. Parmar: Resources, data curation, software, formal analysis, supervision, validation, investigation, visualization, writing-original draft, project administration, writing-review and editing. A. Turner: Supervision, project administration, writing-review and editing. N. Tunariu: Resources, data curation, formal analysis, supervision, investigation, visualization, writing-review and editing. N. Chopra: Resources, data curation, formal analysis, validation, investigation, visualization, writing-original draft, writing-review and editing. H. Gevensleben: Resources, data curation, formal analysis, validation, investigation, visualization, writing-review and editing. N.C. Turner: Resources, data curation, supervision, validation, investigation, writing-review and editing. R. Ruddle: Resources, data curation, software, formal analysis, validation, investigation, visualization, writing-original draft, writing-review and editing. F.I. Raynaud: Resources, data curation, software, formal analysis, supervision, validation, investigation, visualization, writing-original draft, writing-review and editing. S. Decordova: Resources, data curation, formal analysis, validation, investigation, visualization, writing-review and editing. K.E. Swales: Resources, data curation, software, formal analysis, supervision, validation, investigation, visualization, writing-original draft, writing-review and editing. L. Finneran: Resources, data curation, formal analysis, writing-review and editing. E. Hall: Data curation, software, formal analysis, supervision, validation, investigation, writing-review and editing. P. Rugman: Resources, writing-review and editing. J.P.O. Lindemann: Resources, writing-review and editing. A. Foxley: Resources, writing-review and editing. C.J. Lord: Resources, data curation, formal analysis, supervision, validation, investigation, visualization, methodology, writing-original draft, writing-review and editing. U. Banerji: Resources, writing-review and editing. R. Plummer: Resources, methodology, writing-review and editing. B. Basu: Resources, data curation, supervision, investigation, writing-original draft. J.S. Lopez: Resources, data curation, supervision, writing-original draft, project administration, writing-review and editing. Y. Drew: Resources, data curation, formal analysis, supervision, writing-original draft, writing-review and editing. J.S. de Bono: Conceptualization, resources, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing.

The authors would like to thank and acknowledge all patients for taking part in this study, their caregivers, and the trial research nurses, data managers, and clinical coordinators. Funding for this academic study was provided by AstraZeneca through the Cancer Research UK Experimental Cancer Medicine Centre (ECMC) Combinations Alliance. The authors acknowledge the ECMC (London – The Institute of Cancer Research, London - University College London, Cambridge, and Newcastle Centres), National Health Service (NHS) funding to the National Institute for Health Research (NIHR) Biomedical Research Centres at the Royal Marsden NHS Foundation Trust and The Institute of Cancer Research and University College London, NIHR Cambridge Clinical Research Facility, and Cancer Research Technology Limited. Capivasertib (AZD5363) was discovered by AstraZeneca after a collaboration with Astex Therapeutics (and its collaboration with The Institute of Cancer Research and Cancer Research Technology Limited). U. Banerji is a recipient of an NIHR award (RP-2016-07-028). The authors thank Dr. Filip Janku, MD, PhD (The University of Texas MD Anderson Cancer Center, Houston, TX) for helpful discussions on cfDNA studies.