Vorinostat and isotretinoin with chemotherapy in young children with embryonal brain tumors: A report from the Pediatric Brain Tumor Consortium (PBTC-026)

Abstract

Brain tumors are the most common solid tumor in children and now the leading cause of pediatric cancer death.¹ Multimodal regimens, including high-dose craniospinal radiation, have proved curative for the majority of older children with embryonal tumors even in the case of metastatic disease.² However, neurocognitive toxicity of craniospinal radiation is unacceptably severe in young children.³ Therefore, intensive chemotherapy approaches are employed for young children with the goal of reducing, delaying, or eliminating radiation therapy.^4–8 This has been most successful in the case of localized stage M0 disease⁵ and young children with desmoplastic/nodular medulloblastoma (MB),^9–11 although the addition of high-dose therapy with stem cell rescue has improved the radiation-free survival of children with metastatic disease and non-desmoplastic histology.^7,12 Focal radiation following standard or high-dose chemotherapy for children with M0 disease may reduce focal recurrence without the neurocognitive impact of craniospinal radiation therapy.^13,14 While this “infant” approach to embryonal tumor treatment may reduce long-term toxicity, the acute toxicity of therapy is significant, including hematologic, infectious, gastrointestinal, liver, and pulmonary toxicity, and a toxic death rate of approximately 5% even with extensive supportive care and experience.^5,6,12,15 Novel approaches to therapy with non-overlapping toxicity are needed.

In the past decade, genomic studies have led to a wealth of information describing the molecular landscape of pediatric brain tumors.¹⁶ The application of molecular characterization has redefined pediatric CNS tumor types, including international consensus that MB is considered as 4 distinct molecular groups, specifically WNT-activated, SHH-activated, group 3, and group 4.^17,18 Approximately half of young children with MB have desmoplastic/nodular morphology and are SHH-activated.¹⁹ The age-specific distribution of MB molecular groups has further elucidated the poor survival of other young children with classic or large cell/anaplastic histology, who are most likely to have group 3, MYC-amplified tumors.²⁰ Young children with the diagnosis previously called supratentorial primitive neuroectodermal tumor (PNET) are more likely than older children to have embryonal tumor with multilayered rosettes (ETMR),¹⁷ molecularly characterized by amplification of a microRNA cluster on chromosome 19, and also with particularly poor prognosis.²¹

Vorinostat is a class I and II histone deacetylase (HDAC) inhibitor found to induce apoptosis in preclinical models of MB.²² In preclinical studies, vorinostat was found to have additive activity with isotretinoin and synergistic activity with cisplatin,²³ a standard chemotherapeutic agent used to treat embryonal tumors.^2,7,24 In preclinical studies using in vitro models of MB, courses as short as 4 days of oral isotretinoin and vorinostat prior to cisplatin were shown to be effective,²³ and this short course “priming” approach was chosen for clinical study during the intensive induction chemotherapy phase in order to limit toxicity.

A phase 1 trial of isotretinoin in pediatric patients established the maximum tolerated dose of 80 mg/m²/dose given twice daily,²⁵ and the addition of a 6-month maintenance phase in pediatric neuroblastoma has become standard of care since it was shown to improve survival over intensive chemotherapy alone.²⁶ The phase 1 consortium of the Children’s Oncology Group evaluated vorinostat alone and in combination with isotretinoin in children with relapsed cancers, and established a safe combination dose of 180 mg/m²/dose of vorinostat given daily 4 days per week with standard-dose isotretinoin given twice a day for 14 days every 28 days.²⁷

The PBTC-026 study investigated the safety and feasibility of administering vorinostat and isotretinoin with multiagent intensive chemotherapy and the prognostic value of histopathological and molecular characterization.

Methods

Patient Selection and Study Design

PBTC-026 was a prospective safety and feasibility study conducted by the Pediatric Brain Tumor Consortium (PBTC) (NCT00867178). The study was approved by the institutional review board of each participating institution and written informed consent was obtained from parents or legal guardians of each patient prior to study participation. Eligible patients were 2-48 months of age with newly diagnosed MB or supratentorial embryonal tumors, including pineoblastoma. Patients with stage M0 desmoplastic/nodular MB and atypical teratoid/rhabdoid tumor (ATRT) were excluded. Adequate pre-trial tumor material was required for central pathology review and correlative biology studies.

Other eligibility criteria included a Lansky Performance Score of at least 30. Organ function requirements included an unsupported absolute neutrophil count (ANC) of >1000/µL; unsupported platelet count of >100 000/µL; hemoglobin >8 g/dL; bilirubin, SGPT (ALT), and creatinine <1.5 times upper limit of normal or with GFR >70 mL/min/1.73 m². Subjects receiving any other anticancer or investigational agents, valproic acid within 2 weeks, or with significant unrelated systemic illness that would compromise the ability to tolerate protocol therapy were excluded.

Preoperative and postoperative cranial MRI with and without gadolinium was required. Spinal MRI with gadolinium was required either preoperatively or at least 2 weeks following surgery prior to initiation of treatment. Treatment was required to start within 31 days of definitive surgery.

Treatment

Patients received three courses of induction therapy (planned 21-day course) with vorinostat (days 1-4), isotretinoin (days 1-4), cisplatin (day 4), vincristine (days 4, 11, and 18), cyclophosphamide (days 5-6), and etoposide (days 4-6). Vorinostat was distributed by the Cancer Therapeutics Evaluation Program (CTEP) of the National Cancer Institute (NCI, IND #71976), and administered as a pediatric suspension (50 mg/mL) prepared by each institutional pharmacy. Isotretinoin and other medications were obtained from the commercial supply. Isotretinoin was prescribed under the Committed to Pregnancy Prevention Program (iPLEDGE) requiring provider and pharmacy registration. Medication diaries were used to collect compliance information for oral agents. Peripheral blood stem cells were harvested after each course of induction until adequate to provide support for three courses of consolidation therapy (28 days per course) with carboplatin and thiotepa (days 1-2, see details of treatment in Figure 1). Second look surgery at the end of induction therapy was encouraged for subjects with evidence of residual tumor.

Fig. 1

Treatment schedule. Induction therapy consisted of three 21-day cycles, including 4 days of vorinostat and isotretinoin priming for cisplatin (day 4), vincristine (day 4, 11, and 18), etoposide (days 4-6), and cyclophosphamide (days 5-6). Consolidation therapy consisted of three 28-day cycles of carboplatin and thiotepa (days 1-2) with stem cell infusion on day 4. Maintenance therapy consisted of twelve 28-day cycles of oral vorinostat and isotretinoin.

Open in new tabDownload slide

Subjects with stage M0 MB received focal radiation therapy (45 Gy if <18 months, and 54 Gy if >18 months of age) to the tumor bed, with the commencement of treatment within 7 weeks of the first day of course 3 consolidation therapy and prior to initiation of maintenance therapy. 3D conformal radiation therapy, intensity-modulated radiation therapy (IMRT), and proton beam radiation therapy (PRT) were permitted. Patients with supratentorial primary tumors or metastatic disease were treated with radiation therapy at the discretion of the treating physician. If treated, children with supratentorial tumors received a focal dose of 50.4 Gy.

Maintenance therapy, consisting of twelve 28-day courses of oral vorinostat and isotretinoin (dose and schedule per Figure 1), was to start 4 weeks after the end of radiation therapy, or 4 weeks after the start of final course of consolidation therapy for patients who did not receive radiation.

Vorinostat was provided as an investigational new drug (IND #71976) by the CTEP of the NCI. Other therapeutics and all supportive care medications were obtained by treating institutions through commercial supply.

Objectives and End Points

Three primary objectives were (1) to investigate the feasibility of administering vorinostat and isotretinoin over three courses of induction chemotherapy; (2) to describe the toxicity of this induction regimen; and (3) to investigate prognostic values of histopathologic classification and biologic markers.

Following an initial safety phase consisting of the evaluation of the first 6 patients for the absence of dose-limiting toxicity (DLT) during the first cycle of induction, the feasibility phase was defined by the proportion of patients who were able to complete all three induction courses and remain eligible to begin consolidation therapy within 98 days. To describe toxicity, adverse events were recorded according to the NCI Common Terminology Criteria for Adverse Events (CTCAE) version 4.0, including attribution to vorinostat. Prospective central pathology review confirmed the histologic diagnosis prior to initiation of therapy. Retrospective molecular analyses were performed using Illumina 850K methylation platform for classification of all tumors in comparison with a reference cohort as previously described.²⁸

Secondary objectives included (1) response rate in patients with measurable residual disease; (2) disease-specific progression-free survival (PFS) and overall survival (OS). Retrospective central radiology review was conducted to confirm response and progression events. Response was according to modified RANO criteria, considered complete response (CR) if no evaluable disease, partial response (PR) if greater than 50% decrease in tumor area measured by the greatest 2-dimensional area, and stable disease (SD) if less than 50% decrease or less than 25% increase in the tumor area. Progressive disease (PD) was defined as 25% or greater increase in 2-dimensional tumor area or any new area of disease, including development of positive cerebrospinal fluid cytology. An additional exploratory neuroradiologic analysis was conducted to focus on the evaluation of post-radiation treatment changes, including white matter changes, enhancement, or hemorrhage.

Statistical Considerations

The safety phase consisted of a three-plus-three design with one potential dose de-escalation. If no more than one DLT was observed in the first 6 patients, then the starting dose level would be used for the feasibility phase. The feasibility objective proposed that induction therapy could be delivered to at least 80% of patients with recovery to meet the criteria to start consolidation therapy within 98 days. In an intent-to-treat analysis, all eligible patients who received at least one dose of protocol treatment were counted, and patients who went off treatment for any reason before completion of induction were counted as failure toward the feasibility objective. Simon’s 2-stage optimal design with 10% type 1 error and 80% power was used to assess feasibility rates with 60% deemed as unacceptable and 80% as desirable. The first stage would accrue up to 11 patients; if more than 7 patients met feasibility criteria, the trial would continue to accrue up to 31 eligible patients. It would be considered feasible to deliver therapy if more than 21 patients met feasibility criteria. The study was monitored by an independent data safety monitoring committee, which also monitored for successful stem cell harvest and for high-grade toxicities.

The investigation of prognostic values of histopathological classification and biologic markers in the context of a feasibility study was intended to be hypothesis-generating; the biology section of protocol was amended prior to analysis to incorporate rapidly evolving molecular approaches. Survival analyses included Kaplan-Meier estimates of PFS (time from the treatment start date until the earliest date of progression, second malignancy, or death from any cause) and OS (treatment start date until death). Patients without events were censored at their last follow-up. Patients who received craniospinal irradiation (CSI) were censored at the date radiation commenced for PFS analyses.

Results

Thirty-three patients were enrolled between October 2009 and July 2014. One was ineligible due to the unavailability of required frozen tissue, and one withdrew prior to beginning protocol therapy. Demographics of 31 eligible and evaluable subjects are detailed in Table 1. There was a male predominance (61%) with a median age of 26 months at the time of diagnosis (range 6-46 months). Institutional histologic diagnosis was MB in 20, PNET in 9, and ependymoblastoma and pineoblastoma in 1 each. Central pathology diagnosis confirmed MB in 19 and identified one high-grade glioma in patients with institutional diagnosis of MB. For non-MB tumors, central pathology confirmed ETMR in 5, CNS embryonal tumor in 5, and identified one high-grade glioma. Nine of 20 (45%) of the MB patients and 3 of 11 (27%) non-MB patients had metastatic disease at diagnosis.

The 230 mg/m²/day dose of vorinostat was considered safe after the first 6 evaluable patients completed the induction phase without DLT. Feasibility data were available for all 31 patients. All patients completed successful stem cell harvest. Of these, 24 completed the induction phase and started consolidation therapy within the 98-day pre-defined feasibility window. Seven feasibility failures were due to disease progression (n = 4), consolidation starting outside the 98-day window (n = 2), and toxicity (intracranial hemorrhage, n = 1).

Sixteen patients with M0 disease underwent gross total resection at the time of diagnosis and were therefore not evaluable for response. Of these, 11 remained without evidence of disease, 5 had recurrence 2.3-18 months from therapy start; 4 died of disease 10.3-35.5 months from therapy start; and 1 child died of pulmonary toxicity 6.3 months from therapy start. Fifteen patients with metastatic disease (n = 12) and/or less than gross total resection (n = 9; 1 near total resection, 4 subtotal resection, 4 biopsy alone) were evaluable for response. Response to induction therapy was CR in 2; PR in 4; SD in 6; and PD in 3. Response to consolidation for the remaining 12 evaluable patients without PD in induction was CR in 4, PR in 2, SD in 5, and PD in 1. Response to maintenance therapy for the remaining 11 patients without PD prior to starting maintenance was CR in 5, PR in 1, SD in 2, and PD in 3.

Compliance with oral agents during induction therapy was excellent. One patient did not receive either vorinostat or isotretinoin during induction cycle 3 as cisplatin was discontinued due to toxicity of hearing loss; 2 patients discontinued study treatment due to disease progression. The remaining 28/31 patients received 100% of prescribed vorinostat; 20/31 received 100% of isotretinoin and 8/31 missed at least one dose of isotretinoin but received >90% of prescribed isotretinoin in induction phase of therapy. Of the 19 patients who received maintenance therapy, 6 subjects received 100%, 2 subjects received >100% due to parental error, and 6 subjects received 90%-100% of prescribed vorinostat. For isotretinoin, 3 patients received 100% and 9 patients received 90%-100%. During maintenance therapy, 1 patient held one cycle of isotretinoin due to pancreatitis, and another held and subsequently discontinued isotretinoin due to renal dysfunction with increased creatinine.

There was clinically significant toxicity observed as expected for the backbone of intensive cytotoxic therapy. All grade 3-5 adverse events which were experienced in more than one (>5%) subject are summarized in Table 2. There were 12 deaths observed, 9 due to primary CNS tumors. One patient had a cause of death reported as indeterminant between disease vs radiation necrosis; 1 patient died of a secondary malignancy consisting of glioblastoma in the prior radiation field 55 months after primary tumor diagnosis; and 1 patient died of pulmonary toxicity immediately following the third cycle of consolidation therapy.

Twelve patients (12/31 = 39%) experienced relapse or progression, 7 (7/20 = 35%) with MB, and 5 (5/11 = 45%) with supratentorial tumors. Time to relapse or progression event ranged from 1 to 19 months from treatment start. Relapse was during induction phase in 5; in consolidation in 2, and during or just following maintenance in 5 patients. One patient with MB was taken off a study on day 4 of therapy due to a CNS hemorrhagic event and subsequently experienced both local and metastatic relapse. The remaining 6 patients with MB relapse all had a metastatic pattern of relapse, including isolated positive CNS cytology in 2. Site of relapse for supratentorial tumors was local in 3, and both local and metastatic in 2.

Of the 11 MB patients with localized M0 disease, 2 progressed prior to completion of consolidation therapy, and the remaining 9 received focal radiation therapy as mandated on the study. Of these, 8 of 9 (89%) received proton radiation therapy and 1 patient received 3D conformal photon radiation. On central neuroradiology review, 6 of these 9 M0 MB patients (67%) had imaging changes deemed to be consistent with radiation treatment-related changes, including white matter changes (n = 5), enhancement (n = 3), or hemorrhage (n = 2). Only one patient with M0 MB was reported to have clinical symptoms of radiation necrosis, which was the patient who received 3D conformal radiation and had both white matter changes and enhancement on central imaging review. Focal radiation following consolidation was given prior to relapse at the discretion of the treating institution to an additional 3 patients with metastatic MB and 4 patients with supratentorial tumors. Two additional patients with supratentorial ETMR received focal radiation following local relapse. Craniospinal radiation was given following consolidation and prior to relapse in one patient with pineoblastoma who was over the age of 3 years at diagnosis. Four additional patients received craniospinal radiation following relapse or progression. Eight patients received no radiation therapy (Table 3).

Survival is reported ±standard error. With a median follow-up of 4.9 years (range 2.5-6.5 years), the 5-year PFS and OS for all 31 patients was 55% ± 15% and 61% ± 13%, respectively; 55% ± 17% and 65% ± 15% for MB (n = 20). There was no significant difference observed in either PFS or OS related to M-Stage for the entire cohort (P = .72), or for the subset with MB (5-year PFS 64% ± 17% for 11 M0, 44% ± 33% for 9 M+, P = .44).

Fresh frozen tumor tissue was obtained from 30 of 31 subjects. Molecular analysis was obtained for 26 patients as shown in Table 3 and Figure 2. Molecular diagnoses largely confirmed and refined histologic diagnosis, although it is noted that one very young subject diagnosed with MB was determined by molecular subgrouping to have ETMR, and 2 subjects with supratentorial tumors were found to have group 3 MB tumors.

Fig. 2

Molecular classification. (A) Classification of tumors (n = 26) by methylation analysis; (B) PBTC-026 methylation classification compared to a reference cohort of 2801 CNS tumors; (C) Heatmap showing differential methylation signatures according to molecularly classified tumor groups; (D) Copy number summary plots of SHH MB, group 3 MB, and ETMR tumor groups; (E) Representative genome-wide copy number profiles of SHH MB, group 3 MB, and ETMR tumors. Abbreviations: ETMR, embryonal tumor with multilayered rosettes; SHH MB, sonic hedgehog medulloblastoma.

Open in new tabDownload slide

Considering survival by integrated molecular diagnosis (Figure 3), 5-year PFS and OS was 42% ± 13% and 58% ± 19% for group 3 MB; 80% ± 25% for SHH MB; 33% ± 19% for ETMR; and 100% for the 3 patients with other molecular diagnoses. The group 3 MB molecular group included 2 patients with supratentorial tumors. The one patient with SHH MB who relapsed was the patient removed from protocol therapy due to CNS hemorrhage.

Fig. 3

(A) Progression-free survival (PFS) and overall survival (OS) for all 31 eligible patients; (B) PFS and OS for 20 patients with medulloblastoma; (C) PFS for 26 patients by molecular classification; (D) OS for 26 patients with molecular classification.

Open in new tabDownload slide

Isochromosome 17 and MYC amplification were observed in a small subset of MB patients in this study (n = 5 with iso17, 3 of which also had MYC amp), and all five of these patients experienced metastatic relapse of disease. Two of these patients are long-term survivors after salvage therapy, including craniospinal radiation, both of whom were over the age of 36 months at the time of relapse.

Discussion

This study demonstrated the safety and feasibility of incorporating non-cytotoxic therapy with vorinostat and isotretinoin into an intensive chemotherapy backbone in young children with embryonal tumors. We report survival outcomes of a high-risk group of patients with a molecularly characterized disease that is comparable or superior to previously reported studies.

As children with localized desmoplastic/nodular MB were excluded, it is not surprising that only 5 patients on this study were found to have SHH MB, only one with typical desmoplastic/nodular morphology, four of which had metastatic disease, and none of whom relapsed on therapy. This finding supports the consideration of therapy reduction strategies for SHH MB without TP53 mutation, including those with metastatic disease. However, careful consideration should be given to therapy substitution rather than reduction alone following 2 recent studies in M0 nodular desmoplastic MB where an unacceptable relapse rate was seen with reduced intensity therapy even in this “lower risk” SHH MB subgroup.^29,30

The majority of MB patients in this study had group 3 MB. The 5-year PFS for group 3 MB of 42% ± 13% without craniospinal radiation observed in this cohort compares favorably to 8.3% PFS reported in the contemporary SJYC07 study,³⁰ and is similar to 49% ± 14% PFS reported in a retrospective cohort of 16 children with group 3 MB treated with a similar intensive sequential high-dose chemotherapy backbone, some along with other non-uniform intensification strategies, including high-dose methotrexate, intrathecal therapy, maintenance therapy, and adjuvant radiation.³¹

This is the first study using a high-dose chemotherapy strategy to prospectively collect tissue for molecular characterization; unfortunately, radiation therapy and molecular characterization data are not available for the CCG-99703⁷ dose-finding study on which the chemotherapy backbone of this study was based. Overall, the 55% ± 17% and 65% ± 15% EFS and OS for 20 MB patients on this study appears comparable to the 50.1% ± 11.8% and 60.6% ± 11.6% reported for 18 non-desmoplastic patients treated on CCG-99703, as well as 26% ± 6% and 53% ± 7% for 52 patients with classic MB treated on the HeadStart III³² study.

As we have gained a greater understanding of the distinct molecular groups of MB and other rare embryonal tumors which were not understood or evaluated at the time of initial studies of high-dose therapy,^6,7 the major limitation of this and all other prospective studies of radiation avoidance strategies to date is small size and therefore limited power to evaluate key efficacy questions within clinically relevant molecularly defined disease groups. The single-arm feasibility design of this study does not facilitate the evaluation of the efficacy of the components of therapy separately, specifically whether the survival may be improved by the incorporation of the novel agents. Further study will be needed to evaluate the efficacy of novel agents in MB therapy, and due to the rarity of this disease, it will be critical to conduct pooled analyses and efficacy studies through larger international cooperative efforts.

This study mandated focal radiation therapy for M0 stage MB, while other approaches incorporate radiation at the time of relapse.^5,10 It is notable that the two group 3 MB patients who are long-term survivors after relapse both experienced relapse prior to planned focal radiation and received craniospinal radiation. While it is likely that much of the long-term neurocognitive damage from craniospinal radiation therapy may be avoided with administration to a limited focal field,³ future studies of complete radiation avoidance may be considered to further reduce rare but serious side effects of focal radiation, including radiation necrosis and secondary malignancy, both of which were observed in this small study.

The principal goal of radiation avoidance in young children is to preserve neurocognitive development,³ but neurocognitive outcomes were not collected in the context of this study, and thus it is a limitation that we rely on reduced radiation as a surrogate endpoint of success. Clinically significant associations between radiation-induced white matter changes and increased survival but also poorer neurocognitive outcomes have been reported in a larger cohort.³³ The acute toxicity of both induction and consolidation phases of this intensive regimen should not be minimized, as rare toxic deaths due to sepsis or pulmonary toxicity like the one observed on this study continue to be observed on similarly intensive regimens including recently reported ACNS0333¹⁵ and HeadStart III³² studies.

Molecular characterization refined the clinical diagnosis in most cases but did result in a discordant diagnosis for 1 child diagnosed with MB found to have ETMR and 2 children with supratentorial embryonal tumors whose tumors were molecularly characterized as group 3 MB. Molecularly defined group 3 MB in the supratentorial compartment is an interesting phenomenon, and it remains to be seen whether these rare tumors should be clinically considered group 3 MB or a distinct entity. These children with discordant diagnoses where the youngest three enrolled on study. We also describe a 1-year-old child with a molecular diagnosis of RELA-fusion positive ependymoma, which is an unexpected age to find this entity, and no histologic features of ependymoma were identified on central pathology review. While anecdotal in the small group in this study, an integrated molecular diagnosis may be of particular importance for the youngest children with undifferentiated tumors.

This study demonstrates the feasibility of incorporating novel agents into an intensive cytotoxic therapy regimen at the time of diagnosis in young patients with high-risk embryonal tumors. Collaborative efforts are underway to pool and analyze demographic, molecular and treatment data from multiple clinical trials to determine the best approach on which to build future efficacy studies.

Funding

This work was supported by the National Institutes of Health (2UM1CA081457, P30 CA008748 to I.J.D.), The American Syrian and Lebanese Associated Charities (ALSAC), The Brain Tumour Charity (Clinical Biomarkers Award to P.A.N.), and St. Baldrick’s Foundation (Arceci Innovation Award to P.A.N.).

Acknowledgments

The preliminary results of this study were presented in abstract form at the 3rd Biennial Pediatric Neuro-Oncology Basic and Translational Research Conference—San Diego, CA, USA, May 7-8, 2015 (Neuro-Oncology, Volume 17, Issue suppl. 3).

Conflict of interest statement. The authors have no conflicts of interest related to this research to disclose.

Authorship statement. Clinical trial design: J.R.G, T.J.M., and J.M.O.; Clinical trial supervision (PI or co-PI): J.R.G., T.J.M., S.E.S.L., and L.K.; Biologic correlative study design: S.E.S.L., L.K., and P.A.N.; Central radiologic review: T.Y.P.; Central pathologic review: D.W.E.; Data analysis: S.E.S.L, M.K., J.I.H., K.S.S., J.E.H., A.O.T., and P.A.N.; Manuscript draft: S.E.S.L, M.K., A.O., and P.A.N.; Manuscript review and editing: all authors; Supervision: I.J.D., M.F., A.O., and P.A.N.; Funding acquisition: I.J.D., M.F., A.O., and P.A.N.

References