Multicenter Phase II Trial of the PARP Inhibitor Olaparib in Recurrent IDH1- and IDH2-mutant Glioma Open Access

Gliomas represent a heterogenous group of diseases that account for most primary brain tumors (1). The discovery of isocitrate dehydrogenase (IDH) 1 and IDH2 mutations (IDH1/2mt) in a subset of adult-type diffuse gliomas (2) and the observed improved prognosis associated with this phenotype (3) have led to the incorporation of this disease-defining molecular feature into the World Health Organization (WHO) classification of brain tumors (4). IDH1mts occur in more than 70% of what was formerly termed “low-grade” gliomas and up to 20% of “higher-grade” tumors; IDH2 mutations have been found in about 4% of gliomas (2). Bai and colleagues demonstrated that in gliomas, the IDH1mts initially present at diagnosis persist if the tumor recurs (5). The 2021 WHO classification of brain tumors now divides IDHmt tumors into oligodendrogliomas, which harbor a 1p/19q codeletion, and astrocytomas, typically characterized by ATRX mutations (6). On the basis of histologic features and expected prognosis, oligodendrogliomas can be further graded into WHO grades 2 and 3, and astrocytomas into WHO grades 2 to 4. IDHmt astrocytomas with CDKN2A/B homozygous deletion are graded as 4 regardless of histology. IDHmt gliomas with grade 4 histologic features are no longer termed glioblastoma, given the distinct biology and prognosis.

In general, first-line treatment for gliomas includes maximal safe resection which is often limited by tumor extent and location, adjuvant radiation, and alkylator-based chemotherapy. Chemotherapy options include temozolomide or combination of procarbazine, lomustine, and vincristine (7–11). While a better prognosis is observed in oligodendrogliomas, the disease invariably recurs. Hence novel therapy options are clearly needed.

In addition to its prognostic significance, IDH1/2mt is also a promising therapeutic target in the treatment of glioma. The normal function of IDH is to catalyze conversion of isocitrate to alpha-ketoglutarate in the citric acid cycle (12). IDH1/2mts induce aberrant formation of the oncometabolite 2-hydroxyglutarate (2HG), which is implicated in cancer progression through its inhibitory effect on alpha-ketoglutarate–dependent dioxygenases (13, 14). IDH has been successfully targeted in other malignancies that carry IDHmt. Currently, two drugs have been approved by the FDA: ivosidenib, approved for IDH1mt acute myeloid leukemia (AML) and cholangiocarcinoma, and enasidenib, approved in IDH2mt AML (15–17). Trials investigating ivosidenib and vorasidenib in IDHmt gliomas have been published recently (18, 19) and are also ongoing (20, 21).

Preclinically, IDH1/2mts induce a homologous recombination defect which renders tumor cells sensitive to PARP inhibitors (PARPi; ref. 22). This “BRCAness” phenotype, with sensitivity to PARPi, is due to a single alpha-ketoglutarate–dependent dioxygenase targeted by 2HG, KDM4A, which mediates homologous recombination (22). IDH1-dependent PARPi sensitivity has been shown in culture in patient-derived glioma cells as well as genetically matched tumor xenografts in vivo (22). This “BRCAness” was significant in IDH1/2mt cells and approached a 50-fold difference compared with IDH1 wildtype cells when exposed to olaparib (22). Olaparib, an orally bioavailable PARPi, has central nervous system (CNS) penetrance (23, 24) and is currently FDA approved for the treatment of several BRCA-mutated tumors (25). We report the results of a multicenter prospective phase II study investigating the role of olaparib monotherapy in recurrent or progressive IDH1/2mt gliomas.

Eligible patients had recurrent or transformed glioma that progressed despite standard therapy or for which no effective standard therapy existed with evidence of an IDH1/2mt associated with neomorphic activity of encoded proteins. On the basis of archived histology, tumors of all grades were eligible, but the presence of tumor contrast enhancement on T1 post-gadolinium MRIs performed prior to enrollment was required for all patients, as a surrogate marker of progression to a high-grade histology. Specific criteria for patient eligibility, tumor classification, and progression (4) can be found in Supplementary Data S1. The representativeness of study participants is discussed in Supplementary Table S1.

The primary objective of this trial was to estimate the overall response rate (ORR) of olaparib in patients with IDH inhibitor naïve IDH1/2mt glioma. Secondary objectives were to estimate the distribution of progression-free survival (PFS), the overall survival (OS) and the duration of response in this population. Safety and tolerability of olaparib were also evaluated.

Treatment cycles were 28 days. Patients were administered olaparib in tablet form orally 300 mg twice daily continually, irrespective of food intake. No premedications were required. Toxicities were graded using the NCI Common Terminology Criteria for Adverse Events version 5.0. Retreatment criteria can be found in Supplementary Data S2. Patients were monitored for disease status using Response Assessment in Neuro-Oncology (RANO) criteria with MRI performed every 8 weeks. Because of preliminary findings suggesting prolonged stable disease (SD) in some patients, an independent review was commissioned. Results were compared with those reported by local investigator. Any discordance was reported and investigated.

Written informed consent was obtained from enrolled patients. Institutional Review Board approvals of the protocol and consent forms were obtained from all sites. Protocol design and conduct complied with all applicable regulations, guidance, and local policies. The study was conducted in accordance with the Declaration of Helsinki. The trial was conducted under an NCI-sponsored Investigational New Drug application (www.ClinicalTrials.gov: NCT03212274). Olaparib was supplied by the Division of Cancer Treatment and Diagnosis of the NCI.

The power analysis and sample size estimation were completed using the Bayesian adaptive trial design method. We modeled the probability of ORR using Beta distributions. The prior distribution is Beta(0.1, 0.9), which is fairly pessimistic; the expected value is 0.1, reflecting the response rate based on previous experiences. We estimated the ORR using the posterior distribution and reported its median and 90% credible interval (see protocol for details). It was predetermined that the trial would be terminated if there was a lack of response, that is, the probability of the ORR being greater than 0.1 was 50% or less. Because the ORR in the trial was zero, the trial was terminated. The SD rate and the progressive disease (PD) rate were reported, and the 90% credible intervals were estimated using the posterior distribution. SD was defined as lack of progression at first restaging after two cycles.

For secondary objectives, the median PFS and OS were estimated using Kaplan–Meier estimate with 95% confidence intervals (CI). The CIs were based on Greenwood formula for variance. Possible risk factors for PFS and OS of WHO grade and CDKN2A alteration were compared using the log-rank test. For the secondary objectives, all tests were two sided. Statistical analyses were performed using the R 4.1.1 software.

The data generated in this study are available upon request from the corresponding author.

Sixteen patients were enrolled. Fifteen patients were evaluable for response and toxicity. One patient was found to be ineligible after signing consent (Fig. 1). Gliomas were further divided according to WHO 2021 grading system by the following subtypes: grades 2 and 3 astrocytomas (n = 7), grade 4 astrocytoma, formerly known as gliblastoma (n = 5), and grades 2 and 3 oligodendrogliomas (n = 3) (Table 1). IDHmt was confirmed for each patient, either through prior documentation or testing during enrollment period. Tumors from 13 of 15 patients were found to have mutations that led to the R132H neomorphic phenotype in the IDH1 gene (one tumor also contained a variant mutation of uncertain significance in IDH2) and 2 of 15 patients had tumors which harbored neomorphic mutations in the IDH2 gene (R172G and R172K; Table 2). CDKN2A copy-number loss was documented in 5 patients (Table 2). Patient demographics and prior treatments are summarized in Table 1. All patients had evidence of progression at enrollment.

There were no unexpected toxicities reported and most toxicities were grade 1 or 2. The most common toxicities were nausea (67%) and fatigue (47%; Table 3). One patient with a history of ulcerative colitis reported grade 3 diarrhea that resulted in a dose reduction to 250 mg twice daily after the second cycle. The patient was taken off study during cycle 5 due to continued drug-related diarrhea. One patient developed grade 3 hypertension that was possibly attributed to olaparib. Onset was after patient stopped treatment at cycle 12 and subsequently improved at follow-up visit. Finally, grade 3 lymphopenia was reported in 1 patient.

All 15 patients were evaluable for response (Table 4). Nine (60%, 90% credible interval: 37–76) patients achieved SD and 6 (40%, 90% credible interval: 19–58) patients progressed. For patients with SD, the median number of progression-free months was 5.53 (range, 2.83–12.13). Four patients had SD for > 6 months (Table 4). The remaining 6 patients experienced PD after two cycles with a median of 1.82 months on treatment (range, 1.8–1.9).

In the whole study population, the median PFS was 3.63 months (95% CI, 1.87–5.53; Fig. 2A), median OS was 20.73 months (95% CI, 13.77–not reached [NR]; Fig. 2B), median number of cycles completed was 4 (range, 2–14), and the median number of months on treatment was 3.7 (range, 1.8–12.1). The study did not meet the prespecified response-based threshold for moving to stage 2.

We compared PFS and OS of patients with WHO grade 2 or 3 lesions based on archived tissue from diagnosis, to those with grade 4 by histology or CDKN2A alteration. The median PFS of patients with grade 2 and/or 3 was 5.23 months (95% CI, 3.63–9.2) while those with grade 4 was 1.8 months (95% CI, 1.77–NR; log-rank test P = 0.0013 Fig. 2C). The median OS of patients with grade 2/3 disease was 29.5 months (95% CI, 19.87–NR) vs. 15.17 months (95% CI, 9.1–NR) for grade 4 tumors (P = 0.21; Fig. 2D).

Reevaluation of imaging of all patients by a blinded NCI-designated neuroradiologist was performed. Two discordant results were identified: one patient's tumor had progressed after six cycles of treatment but was kept on for four additional cycles by the enrolling site. The corrected months of progression-free disease was reported in the above data. The second patient with discordant results had been taken off treatment due to PD after five cycles by their treating physician. However, reassessment of their scans demonstrated SD.

Olaparib monotherapy was well tolerated in patients with recurrent or progressive contrast-enhancing IDH1/2mt glioma but did not meet prespecified response-based activity criteria for moving to stage 2 of the trial. However, clinically significant prolonged SD was observed in a relatively large proportion of patients, particularly those with WHO grades 2 and 3 histologies, suggesting olaparib monotherapy could be of clinical benefit in select patients. Grade 4 tumors, as defined by the 2021 WHO classification using histology or presence of CDKN2A deletion, derived minimal or no benefit from this drug demonstrating the utility of this new classification for possible future patient stratification and trial design.

PARP is important in the repair of single-strand breaks (26). In PARP inhibited cells, single-strand breaks accumulate and become more lethal double-strand breaks in the process of DNA replication. Cells with BRCA1 or BRCA2 mutations have defects in homologous recombination repair of double-strand DNA breaks. Patients with BRCA1- and BRCA2-mutated tumors treated with PARPis often have dramatic treatment responses as the synthetic lethality induced by the PARPi leads to cell death within the tumor (27, 28). Favorable responses to PARPi have also been seen in tumors with DNA repair alterations other than BRCA mutations (29, 30). Despite preclinical susceptibility of IDH1/2mt tumors to PARPi, our trial did not meet the prespecified endpoint of ORR to unequivocally demonstrate biological activity deriving from the “BRCAness” effects associated with 2-HG. However, the PFS data and disease stability observed in a relatively large proportion of patients (60%) raises the question of whether PARPi could still be of clinical benefit in these patients, in the absence of measurable tumor shrinkage. While our findings may suggest potential benefit in patients with lower grade histologies or at earlier stages of disease progression, further investigation is required as such phenotypes are inherently less aggressive. Other trials have studied PARPi in patients with gliomas, but their inclusion criteria likely excluded most patients with IDH1/2mt and/or other homologous recombination defects (31, 32). In our trial, selection of patients based on contrast enhancement was an attempt to focus on the patients with IDHmt tumors of a higher grade or more aggressive behavior. However, the molecular heterogeneity of patients included in our study, patients’ enrollment at varying stages in their tumor's natural history, the small number of patients with oligodendroglioma, and differences in timing of collection of molecular information (at diagnosis vs. at time of recurrence) limit the conclusions that can be drawn from our data. Regardless, it is clear from the results of this trial that single-agent olaparib is not a sufficient treatment for grade 4 gliomas.

Recently, preliminary results of the OLAGLI trial investigating olaparib in 35 patients with recurrent IDH1/2mt high-grade contrast-enhancing gliomas were presented at American Society of Clinical Oncology 2021 (33). That study found a 6-month PFS rate of 31%, which did not meet prespecified threshold for significance, but 2 patients (5%) had a PR and 14 patients (40%) achieved SD. The median duration of response in the combined PR and SD patients was 9 months. The median OS was 15.9 months (33). Similar to our study, these preliminary results suggest that there may be patients who experience clinically meaningful benefit from single-agent PARPi through disease stability, although olaparib's activity as a single agent as measured by radiographic responses and tumor shrinkage is modest.

An important concern in the development of novel therapies for CNS tumors is the ability of novel agents to penetrate through areas of intact blood–brain barrier to address infiltrative disease. CNS penetration of olaparib in humans has been shown in a phase I study combining temozolomide with olaparib in patients with relapsed glioblastoma; olaparib, given prior to surgery, was detected in 71 of 71 tumor core specimens from 27 patients (34). Detected drug levels were greater than 100 nmol/L, the concentration determined by in vitro experiments to be needed for maximal PARP inhibition (35). Trials are ongoing investigating novel PARPis with potential for even greater brain penetration (NCT00687765, NCT03914742).

Identifying mutations that further define distinct subgroups within IDHmt glioma for updated grading and selection of patients who may benefit from PARPi remains of high interest. Homozygous deletion of CDKN2A has been noted to be an adverse prognostic factor associated with worse PFS and OS in patients with IDH1/2mt disease (36, 37) and per the 2021 WHO classification of tumors confers grade 4 status (6). In our study, 4 of 5 patients with known CDKN2A alteration progressed on first restaging. Our findings therefore seem to support the inclusion of CDKN2A status as a grade-defining molecular finding for the grading of gliomas, as acknowledged in this new classification of gliomas by the WHO. A limitation in our observation is that CDKN2A copy-number status was assessed by next-generation sequencing (NGS) performed locally, and the distinction between homozygous and heterozygous deletion was not uniformly provided. While in IDH wildtype glioblastoma CDKN2A deletions are nearly universally homozygous, we acknowledge that in some patients with IDH1/2mt gliomas, the CDKN2A deletion as determined by NGS may not be homozygous. Therefore, it is possible, but unlikely, that the 2 patients with poor outcomes that were reclassified as grade 4 per WHO 2021 may not meet criteria for such, should other methods for CDKN2A evaluation be utilized.

In addition to PARPi, a significant focus of research in IDH1/2mt gliomas is on direct inhibition of IDHmt proteins. In a phase I trial, 30 patients with recurrent or progressive IDH1/2mt enhancing glioma received vorasidenib, an oral, brain-penetrant inhibitor of IDH1/2mt. ORR was 0% but 56.7% of patients achieved SD. The median PFS was 3.6 months, similar to our trial (19). A window-of-opportunity trial of vorasidenib and the IDH1 inhibitor ivosidenib randomized patients with recurrent nonenhancing WHO 2016 grade 2 or 3 IDH1mt glioma who were undergoing craniotomy to receive either ivosidenib, vorasidenib, or no treatment for 4 weeks preoperatively. Postoperatively, they were randomized to ivosidenib or vorasidenib. Both drugs were found to be CNS penetrant (brain:plasma ratio 0.16 for ivosidenib and 2.4 for vorasidenib) and both lowered 2HG levels compared with untreated samples (21). ORR was 30.8%. Switching to a focus on earlier stage disease, the phase III INDIGO trial is randomizing IDH1/2mt residual or recurrent grade 2, nonenhancing glioma to receive vorasidenib or placebo, with primary endpoint of PFS (38).

There are numerous future study directions in this population. Trials investigating timing of IDH-targeted therapy with IDH inhibitors or PARPi in the first-line (NCT03581292) or recurrent/relapsed setting and timing with respect to surgery will be important. An ongoing study is investigating the effect of niraparib prior to surgery in IDHmt glioma (NCT05076513). Studies that compare responses in patients with contrast-enhancing and non–contrast-enhancing tumors will also provide information on which populations are most likely to benefit. Outcomes of treatment with PARPi in patients who have received prior treatment with IDH inhibitors are currently being investigated in a phase II clinical trial (NCT03212274). Finally, it is possible that PARPi may require combination with additional therapies to improve activity and induce tumor shrinkage. A single-center retrospective study reported the results of patients with glioma, some with IDH1/2mt, who were treated with combination olaparib and temozolomide. Four of the 8 patients with grade 2 or 3 IDHmt disease had complete or partial response. Consistent with our findings, no responses were seen in patients with grade 4 IDHmt glioma. This trial suggests promise of olaparib in combination but is limited by its retrospective design (39). Other ongoing trials investigating combination therapy include PARPi + PD-L1 inhibitor (NCT03991832), PARPi + temozolomide (NCT02152982, NCT04552977, NCT01390571, NCT04910022, NCT03581292, NCT03749187, NCT03914742, NCT03212742), and PARPi + temozolomide + radiotherapy (NCT04614909). Taken together, this body of work will establish whether PARPi will eventually play a meaningful role in the treatment of gliomas.

Y. Shyr reports grants from NIH/NCI during the conduct of the study; grants from NIH outside the submitted work. S. Boerner reports grants from NCI-CTEP and Rising Tide Foundation during the conduct of the study. D. Durecki reports grants from NIH/NCI UM1CA186689 during the conduct of the study. J. Drappatz reports other from Pfizer, GSK, Gilead, and Vertex outside the submitted work. V. Puduvalli reports personal fees and other from Servier and Novocure, other from Karyopharm, and personal fees from Orbus therapeutics during the conduct of the study; other from Amarin and Gilead outside the submitted work. F.S. Lieberman reports grants from ABTC during the conduct of the study; grants from Novocure, Chimerix, Blue Diamond Therapeutics, and Abbvie outside the submitted work. J. Gonzalez reports other from AstraZeneca outside the submitted work. P. Giglio reports grants from NCI during the conduct of the study; other from Vanguard Funds; grants from BioMimetix, Denovo Biopharma, Institut de Recherches Internationales Servier, Novocure, and Prelude Therapeutics outside the submitted work. R.S. Bindra reports a patent to 62/344,678 pending. A. Omuro reports grants from Arcus Biosciences; personal fees from Pyramid, Merck, Kiyatec, and Ono outside the submitted work. P. LoRusso reports other from AbbVie, Agios, Five Prime, GenMab, Halozyme, Roche-Genetech, Genentech, CytomX, Takeda, Sotio, Cybrexa, Agenus, Tyme, IQVIA, TRIGR, Pfizer, ImmunoMet, Black Diamond, Glaxo-Smith Kline, QED Therapeutics, AstraZeneca, EMD Serono, Shattuck, Astellas, Salarius, Silverback, MacroGenics, Kyowa Kirin Pharmaceutical, Kineta, Zentalis, Molecular Templates, ABL Bio, SK Life Science, STCube Pharmaceuticals, Bayer, I-Mab, Seagen, imCheck, Relay Therapeutics, Stemline, Compass BADX, Mekanist, Mersana Therapeutics, BAKX Therapeutics, Scenic Biotech, Qualigen, Roivant, and NeuroTrials outside the submitted work. No disclosures were reported by the other authors.

K. Fanucci: Data curation, formal analysis, visualization, writing-original draft, writing-review and editing. M.J. Pilat: Conceptualization, data curation, formal analysis, writing-original draft, writing-review and editing. D. Shyr: Formal analysis, visualization, writing-review and editing. Y. Shyr: Conceptualization, formal analysis, visualization, methodology, writing-review and editing. S. Boerner: Conceptualization, resources, investigation, project administration, writing-review and editing. J. Li: Formal analysis, writing-review and editing. D. Durecki: Resources, data curation, project administration, writing-review and editing. J. Drappatz: Conceptualization, data curation, writing-review and editing. V. Puduvalli: Resources, data curation, writing-review and editing. F.S. Lieberman: Resources, data curation, writing-review and editing. J. Gonzalez: Resources, data curation, writing-review and editing. P. Giglio: Resources, data curation, writing-review and editing. S.P. Ivy: Conceptualization, project administration, writing-review and editing. R.S. Bindra: Conceptualization, funding acquisition, writing-original draft, writing-review and editing. A. Omuro: Formal analysis, supervision, validation, writing-original draft, writing-review and editing. P. LoRusso: Conceptualization, resources, data curation, supervision, funding acquisition, validation, investigation, methodology, writing-original draft, writing-review and editing.

The authors would like to thank CTEP, The Catherine Ivy Foundation, and The Rising Tide Foundation for their support of this project through grants to P. LoRusso.