Olaparib for childhood tumors harboring defects in DNA damage repair genes: arm H of the NCI-COG Pediatric MATCH trial

Abstract

Background

The National Cancer Institute-Children’s Oncology Group Pediatric Molecular Analysis for Therapy Choice (MATCH) precision oncology platform trial enrolled children aged 1-21 years with treatment-refractory solid tumors and predefined actionable genetic alterations. Patients with tumors harboring alterations in DNA damage repair (DDR) genes were assigned to receive olaparib.

Methods

Tumor and blood samples were submitted for centralized molecular testing. Tumor and germline sequencing were conducted in parallel. Olaparib was given twice daily for 28-day cycles starting at a dose 30% lower than the adult recommended phase 2 dose (RP2D). The primary endpoint was the objective response.

Results

Eighteen patients matched (1.5% of those screened) based on the presence of a deleterious gene alteration in BRCA1/2, RAD51C/D, or ATM detected by tumor sequencing without germline subtraction or analysis of loss of heterozygosity (LOH). Eleven (61%) harbored a germline mutation, with only one exhibiting LOH. Six patients enrolled and received the olaparib starting dose of 135 mg/m²/dose. Two participants were fully evaluable; 4 were inevaluable because <85% of the prescribed dose was administered during cycle 1. There were no dose-limiting toxicities or responses. Minimal hematologic toxicity was observed.

Conclusion

Most DDR gene alterations detected in Pediatric MATCH were germline, monoallelic, and unlikely to confer homologous recombination deficiency predicting sensitivity to olaparib monotherapy. The study closed due to poor accrual.

ClinicalTrials.gov Identifier

NCT03233204. IRB approved: initial July 24, 2017.

This article reports clinical trial results of the National Cancer Institute-Children’s Oncology Group Pediatric Molecular Analysis for Therapy Choice precision oncology platform trial, which enrolled patients aged 1-21 years with treatment-refractory solid tumors and predefined actionable genetic alterations. Those with tumors harboring alterations in DNA damage repair genes were assigned to receive olaparib.

Lessons Learned.

• Germline and somatic loss of function alterations in DNA damage repair (DDR) genes such as BRCA1/2, RAD51C/D, and ATM are infrequent in refractory pediatric malignancies, generally monoallelic and likely do not confer susceptibility to olaparib monotherapy.

• Alternative biomarkers of potential response to agents targeting DDR/response pathways in patients in pediatric care are required.

• In some cases, genetic biomarkers validated in adult malignancies may be insufficient to predict response in pediatric cancers.

Discussion

Olaparib is a potent orally bioavailable small-molecule inhibitor of poly(ADP-ribose) polymerase (PARP), an essential enzyme for single-strand DNA break repair by base excision (BER). By virtue of “synthetic lethality,” adult malignancies including breast, ovarian, pancreatic, and prostate cancer with germline and somatic BRCA1/2, RAD51C/D, and ATM inactivating mutations and resultant homologous recombination deficiency (HRD) are susceptible to PARP inhibition with olaparib monotherapy.¹ DNA damage repair (DDR) gene alterations are uncommon in pediatric cancer, and it remains unclear whether they confer vulnerability to PARP inhibition.² In this phase II study, we selected patients to receive olaparib monotherapy based on the validated genetic markers of DDR in adult malignancy detected with tumor-only sequencing with germline testing performed and reported separately. Actionable tumor alterations were identified in 5 DDR genes in 18 patients: BRCA2 (n = 8), BRCA1 (n = 2), ATM (n = 6), RAD51D (n = 2), and RAD51C (n = 1). In total, 11/18 variants were found to also be present in the germline: 4/8 BRCA2, 2/2 BRCA1, and 5/6 ATM. Retrospective analysis revealed tumor loss of heterozygosity (LOH) in only one patient; in all other cases, the variant allele frequency of somatic mutations was consistent with monoallelic loss. Of the 18 matched patients, 9 were ineligible for treatment, 3 declined assignment, and 6 enrolled (Figure 1). Treated patients received a dose 30% below the adult recommended phase 2 dose (RP2D) as stipulated by the operating rules for Pediatric Molecular Analysis for Therapy Choice (MATCH) for agents not previously tested in children. The only patient with biallelic tumor loss of ATM was removed from the study following an acute allergic reaction (hives) to the first dose of olaparib recurring upon rechallenge; 3 others received <85% of the agent during cycle 1 (early progression, patient withdrawal, and dosing error; each n = 1); 2 completed the first cycle with no DLT. Minimal hematologic toxicity was observed, including one patient with grade 3 lymphopenia, and 3 with grade 1/2 anemia. There were no responses. The median (range) number of cycles received was 1. Due to small patient numbers, no conclusions can be made regarding the efficacy of olaparib monotherapy in children. Additionally, the patients enrolled in this study may have had insufficient exposure to olaparib based on the starting dose, limited pharmacokinetics, and a recently published pediatric phase I study suggesting that even at olaparib doses equivalent to the adult RP2D, children attained lower peak plasma concentrations and AUC.³ Critically, in contrast to adult malignancies, the genomics of the pediatric tumors evaluated in the National Cancer Institute (NCI)-Children’s Oncology Group (COG) Pediatric MATCH trial and other recent studies confirm that the majority of DDR pathway gene mutations in children are monoallelic and unlikely to confer a sufficient exploitable dependence on PARP inhibition. Alternative biomarkers of potential response to agents targeting DDR/response pathways in children are needed.

Figure 1.

CONSORT diagram. *Germline alteration.

Trial Information
Disease	Advanced solid tumors including non-Hodgkin lymphomas, CNS tumors, and histiocytoses that harbor deleterious genetic alterations in the DDR pathway
Stage of disease/treatment	Relapsed or refractory/olaparib
Prior therapy	Patients must have fully recovered from the acute toxic effects of all prior anticancer therapy and must meet the following minimum duration from prior anticancer directed therapy prior to enrollment; if after the required timeframe, the numerical eligibility criteria are met, eg, blood count criteria, the patient is considered to have recovered adequately Cytotoxic or myelosuppressive chemotherapy: ≥21 days after the last dose (42 days if prior nitrosourea) Anticancer agents not known to be myelosuppressive: ≥7 days after the last dose of agent Antibodies: ≥21 days must have elapsed from the infusion of the last dose of antibody, and toxicity related to prior antibody therapy must be recovered to grade ≤1 Corticosteroids: if used to modify immune adverse events related to prior therapy, ≥14 days must have elapsed since the last dose of corticosteroid Hematopoietic growth factors: ≥14 days after the last dose of a long-acting growth factor (eg, pegfilgrastim) or 7 days for short-acting growth factor Interleukins, interferons, and cytokines (other than hematopoietic growth factors): ≥21 days after the last administration Stem cell infusions (with or without total-body irradiation [TBI]): ◦ Allogeneic (nonautologous) bone marrow or stem cell transplant, or any stem cell infusion including donor lymphocyte infusion (DLI) or boost infusion: ≥84 days after infusion and no evidence of graft-versus-host disease (GVHD) ◦ Autologous stem cell infusion including boost infusion: ≥42 days Cellular therapy: ≥42 days after the completion of any type of cellular therapy (eg, modified T cells, natural killer [NK] cells, dendritic cells, etc.) Radiation therapy (XRT)/external beam irradiation including protons: ≥14 days after local XRT; ≥150 days after TBI, craniospinal XRT or if radiation to ≥50% of the pelvis; and ≥42 days if other substantial bone marrow (BM) radiation. Note: radiation may not be delivered to “measurable disease” tumor site(s) being used to follow response to subprotocol treatment Radiopharmaceutical therapy (eg, radiolabeled antibody, iobenguane I-131 [131I-MIBG]): ≥42 days after systemically administered radiopharmaceutical therapy Patients must not have received prior exposure to olaparib, veliparib, niraparib, rucaparib, talazoparib, or other PARPi
Type of study	Phase II/signal finding
Primary endpoint	Objective response rate (ORR; complete response + partial response)
Secondary endpoints	Progression-free survival (PFS; defined as the time from the initiation of protocol treatment to the occurrence of any of the following events: disease progression or disease recurrence or death from any cause. PFS along with CIs was estimated using Kaplan-Meier method) Safety and tolerability Pharmacokinetics
Investigator’s analysis	Study closed early due to poor accrual

Additional details of endpoints or study design

Patient enrollment in the NCI-COG Pediatric MATCH screening protocol was required. Tumor and blood samples were obtained and the results of centralized tumor evaluation determined if the patient’s tumor had an actionable Mutation of Interest (aMOI) for which subprotocol H was appropriate. The treatment assignment to MATCH subprotocol H was communicated to the enrolling institution via the COG treatment assignment mechanism and a reservation to subarm H was secured. Reservations were withdrawn by the institution if the patient indicated they did not intend to consent to participation or the site investigator indicated the patient would never be eligible.

Steady-state pharmacokinetics were assessed on cycle 1, day 8. The area under the plasma concentration-time curve (AUCSS) was determined using the linear trapezoidal approximation through the day 8 24-hour dosing interval. The plasma concentration 24 hours after administration of the day 8 dose was estimated to be equivalent to the predose concentration based on the assumption of steady-state pharmacokinetics for olaparib. Steady-state clearance (CLSS) was calculated as dose/AUCSS.

Drug Information
Generic/working name	Olaparib
Company name	AstraZeneca
Drug type	Small-molecule inhibitor
Drug class	PARP inhibitor
Planned dose levels (mg/m2/dose)
−1	100 (max 175 mg)
1a	135 (max 225 mg)
2	175 (max 300 mg)
Route	Oral tablet
Schedule of administration	Twice daily (BID) continuously

^aOnly dose level 1 was tested: 135 mg/m²/dose BID (max 225 mg/dose)

Patient Characteristics
Number of patients
Male	4
Female	2
Age: median (range)	10 years (range 9-20)
Performance status: Lansky/Karnofsky
100
90	3
80
70	2
60	1
Cancer types or histologic subtypes
Rhabdomyosarcoma	3
CNS tumors—astrocytoma	1
Ewing sarcoma	1
Neuroblastoma	1

Primary Assessment Method
Title	NCI-COG Pediatric MATCH —phase II subprotocol of olaparib in patients with tumors harboring defects in DDR genes
Number of patients screened	18 (matched)
Number of patients enrolled	6
Number of patients evaluable for toxicity	6 (2 evaluable for DLT)
Number of patients evaluated for efficacy	6
Evaluation Method	RECIST 1.1
Response assessment, PD	3 (50%)
Not evaluated	3 (50%), 2 patients came off treatment early for reasons other than PD without having had an assessment and 1 patient (treated for 3 cycles then withdrew) did not have a confirmation scan
Median duration assessments, PFS	1.5 months (95% CI: 0.4-NA)
Median duration assessments, OS	2.5 months (95% CI: 0.8-NA)
Duration of treatment	1.1 months

Pharmacokinetics and Pharmacodynamics
Dose level	Dose of drug 1 (mg/m2)	Dose of drug 2	Number of patients participating in PK analysis	Cmax (μg/mL), mean ± SD	T max (h), range	AUC 0-12 (h*μg/mL), mean ± SD	T ½ (h). mean ± SD	CIearance at steady state (L/h/m2) mean ± SD
1	135		3	7.5 ± 2.7	1-4	48.8 ± 27.8	5.6 ± 4.6	4.1 ± 3.6

Assessment, Analysis, and Discussion
Completion	Study completed
Investigator’s assessment	Study closed early due to poor accrual

The Pediatric NCI-COG MATCH (NCT03155620) trial is a national precision medicine platform for children, adolescents, and young adults up to 21 years of age with relapsed and refractory cancers sponsored by the NCI and conducted by the COG. Centralized next-generation sequencing facilitated enrollment of patients whose tumors harbored predetermined actionable genetic alterations onto one of multiple phase II, histology agnostic subarms of matched molecularly targeted therapy⁴ Developed in parallel with the adult NCI-MATCH trial (NCT02465060), the study was designed to study single molecularly targeted agents, leveraging the clinically validated cancer gene panel assay generated for the overarching NCI sponsored precision oncology initiative,⁵ with gene-targeted agent pairs chosen based on pediatric gene (target) frequency and predetermined levels of preclinical and clinical evidence for the agent class, predominantly derived from the adult experience.⁶ Germline sequencing was also performed, but only somatic results without germline subtraction or real-time determination of tumor LOH were used for treatment assignment.

Olaparib is an oral PARP inhibitor capable of trapping inactivated PARP at single-strand nicks, blocking BER, leading to the collapse of DNA replication forks, accumulation of DNA double-strand breaks, and dependence on homologous recombination (HR). In BRCA1 or BRCA2 mutant cells, HR is already impaired and PARP inhibition leads to synthetic lethality.^1,7,8 In 2014, Olaparib was FDA approved as monotherapy for the treatment of patients with advanced, heavily pretreated ovarian cancer who carry deleterious germline BRCA1/2-mutations based on a single arm phase II study demonstrating an ORR of 31%.^7,9 Similar response rates to monotherapy were seen in patients with breast cancer with germline BRCA1/2-mutations,⁸ castration-resistant prostate with BRCA2 inactivation and ATM loss,¹⁰ and using a different PARP inhibitor, platinum-sensitive high-grade ovarian carcinoma with RAD51C mutation.¹¹ The impressive single-agent efficacy in adult studies, as well demonstrated anti-tumor activity of a PARP inhibitor in 3 pediatric preclinical models, including exquisite sensitivity in a tumor with PALB2 mutation,¹² formed the basis for the inclusion of olaparib as subarm H in the pediatric MATCH. However, at the time the study was conceived, the prevalence of somatic LOH in tumors of children and adolescents with germline BRCA1 or BRCA2 mutations was unknown, and uncertainty existed as to whether the monoallelic loss would be sufficient to confer the HRD phenotype and sensitivity to PARP inhibitory monotherapy.¹³

The low incidence of somatic and germline alteration in DDR genes in pediatric cancer was anticipated to be a challenge for accrual to this trial. Of note, PALB2 mutation could not be detected by the MATCH assay, and eligible actionable mutations were limited to deleterious alterations in BRCA1, BRCA2, ATM, RAD51C, and RAD51D. Over 5 years with 1206 tumors screened, only 18 (1.5%) patients were matched to the olaparib arm, which was in line with the previous estimated DDR gene frequency of 1%-3%.^2,13 By contrast, the AcSe-eSMART pediatric precision oncology initiative in Europe (NCT02813135) studies combination therapy approaches and uses a different testing/assignment approach wherein a multidisciplinary molecular tumor board stratifies patients to one of several arms aimed at classes of genomic alterations.^14-16 For alterations in DDR, eSMART references a broader panel of qualifying genes and more flexible general rules for “enrichment” using alterations primarily supported by preclinical data, and anecdotal experience when clinical data are not readily available. For the first arm D, using olaparib and irinotecan, not all patients needed a putative biomarker, and 70 patients were enrolled over 6 years, determining dose, safety, and suggesting the overall lack of appropriate predictive biomarkers of response.^17,18 The subsequent arm N studying olaparib in combination with the ATR inhibitor ceralasertib has already completed phase I and has expanded to 2 cohorts studying various biomarkers of HR deficiency and replication stress, with some early signal of efficacy.¹⁹

Our trial of olaparib monotherapy did not demonstrate signs of efficacy in the very limited number of subjects enrolled. One explanation might be insufficient drug exposure to effectively inhibit PARP based on a starting dose below the adult RP2D. A recent pediatric phase I study of olaparib suggests that even at doses equivalent to the adult RP2D of 300 mg BID (187.5 mg/m²/dose BID), children attained lower peak plasma concentrations and AUC, with exposure more closely approximating that seen in adults at doses of 200 mg BID 3. The limited data (n = 3) in this study, however, found mean peak plasma concentrations and AUC approximately 20% lower than those found for the adult RP2D.²⁰ It is also likely that the pediatric tumors evaluated in this study, all but one of which harbored monoallelic defects in DDR pathway genes, were not sufficiently dependent on PARP to create an exploitable vulnerability to olaparib monotherapy. It has long been assumed that adult patients with cancer with BRCA1 or BRCA2 germline mutation will harbor locus-specific LOH within their tumors, thereby conferring HRD or BRCAness. Evidence suggests, however, that while predominant in breast and ovarian cancer, somatic LOH is not universal, and the absence of LOH may be a biomarker of primary resistance to PARP inhibition.²¹ The absence of LOH is increasingly recognized as a characteristic in pediatric malignancies in the setting of pathogenic germline mutation,²² and in a recent study, only the exceptionally rare childhood tumor with biallelic BRCA2 inactivation produced the predictive HRD-associated mutational signature 3 associated with BRCAness.²³ Finally, emerging data from the AcSe-eSMART arms and the growing body of research specific to pediatric malignancy suggests that childhood cancers may be more reliant on ATR or Wee-1 DDR mechanisms and that intrinsic replication stress, domesticated transposable elements, recombination-mediated alternative lengthening of telomeres, and compound biomarkers such as 11q loss may be more relevant predictive biomarkers for DDR-targeted therapies in pediatric oncology.²⁴

In summary, suboptimal predictive enrichment biomarkers derived from adult malignancy, potentially inadequate drug exposure, and poor accrual limited the utility of this pediatric phase II trial of olaparib. Future pediatric precision oncology trials should incorporate rapid dose optimization strategies and move swiftly to combination therapy where single-agent activity is less likely. This strategy will be pursued in the forthcoming ECOG-ACRIN-NCI ComboMATCH (NCT05564377), which plans to include children in biologically relevant subarms.

Conflicts of interest

J.L.G.B. reported grant support from NCI P30 CA008748, NCI P50 CA217694 (research funding); Jazz Pharmaceuticals (consulting/advisory relationship, compensated; institutional research funding); Springworks, Merck and Pfizer (DSMB, uncompensated); BMS (consulting/advisory relationship, uncompensated) and Eisai (consulting/advisory relationship, uncompensated; institutional research funding); Lilly, Loxo-Oncology, Cellectar, and Bayer (institutional research funding). J.M.R. reported grant support from NCI P30 CA15083 (research funding); Elucida Oncology (consulting/advisory relationship). K.A.J. reported consulting/advisory relationship with Bayer, Inhibrx, and Illumina. The other authors indicated no financial relationships.

Data availability

The data underlying this article are available in the article and in the ClinicalTrials.gov PRS system (protocol registration and results system) and can be accessed with ClinicalTrials.gov ID: NCT03233204.