This Phase II trial was designed to determine the response rate of capsular fenretinide in children with refractory/resistant high-risk neuroblastoma, based on activity observed against neuroblastoma in pre-clinical models (5
) and the Phase I CCG 09709 study (25
). The trial utilized a novel design evaluating response in two different cohorts based on tumor sites at entry. It was hypothesized that fenretinide activity may differ against mass disease (stratum 1) versus disease limited to MIBG avid sites and bone marrow metastases (stratum 2). Stratum 2 patients have traditionally not been eligible for Phase II studies, which utilized the RECIST criteria (28
). The RECIST criteria have not been shown to be associated with outcome, and may not be applicable to neuroblastoma, where bone and bone marrow are the most frequent and often only sites of relapse (36
). In addition, agents with modest systemic toxicity are potentially suitable for treating minimal residual disease, non-measurable by RECIST criteria, that remains after completing front-line therapy. Responses in MIBG avid lesions were defined using the Curie scoring method (29
), which has been validated for bone metastases and has shown prognostic value (37
). Preliminary data from the COG A3973 trial for newly diagnosed high-risk neuroblastoma suggest that the Curie score at the end of induction chemotherapy is prognostic for EFS (38
). Bone marrow response is difficult to quantify, due to patchy involvement. Neuroblastoma patients may also have minimal residual marrow disease (<5% tumor) variably detected on serial sampling. This study defined only complete response, SD, or PD in bone marrow. An ongoing retrospective study in the New Approaches to Neuroblastoma Consortium (NANT) will evaluate if these response criteria correlate with PFS and OS.
Neither stratum had sufficient responses to meet protocol criteria for efficacy. One of 59 (1.7%) patients with MIBG avid bone sites had a partial response. However, 13/59 (22%) patients had prolonged SD for ≥4 (median: 15; range 4–45+) courses and one eventually achieved a CR on further compassionate fenretinide therapy. Among 13 patients with SD, the high median Curie score and large median longest dimension of 4 cm at study entry may indicate that prolonged SD was possible in patients with significant tumor burden. However, we cannot definitively conclude that the prolonged SD observed is any different from the natural history without any therapy in this diverse patient population.
The two strata design based on tumor sites at protocol entry may not be superior to a single strata design to determine efficacy based on response. Two other COG Phase II studies (40
) using this same design found responses which met the statistical endpoint for efficacy in Stratum Two only. Additional studies are needed to resolve this issue. The two strata approach provides a framework for future clinical trials to better define the impact of disease burden on assessing drug activity in recurrent neuroblastoma.
The unplanned comparisons of survival by stratum, sites of tumor, or age were underpowered. However, marrow disease at entry was associated with significantly lower PFS and OS. The COG 09709 Phase I study of fenretinide also found this association with shorter time to progression in patients with bone marrow disease at entry (25
). These data suggest that tumor sites may affect outcome after salvage therapy with novel agents. Patients with persistent refractory tumor (documented by histology) had significantly higher PFS and OS than patients with a history of prior relapse. Among patients with prolonged SD, 7 had prior relapse and 6 had refractory disease. The patient population on this study was heterogeneous in terms of sites of tumor, and whether they had recurrent progressive disease after prior responses or were primarily refractory to therapy; these factors will be critical to consider in future Phase II study design, since they potentially affect response rate and/or time-to-progression endpoints. Further data with larger numbers of patients are required to test these hypotheses.
Fenretinide steady-state pharmacokinetics confirmed Phase I CCG 09709 data (25
) that steady state drug concentrations in the range associated with in vitro
activity are achievable. However, intracellular biodistribution of fenretinide is complex, and much higher concentrations may be required in patient plasma than in cell culture to achieve cytotoxic intracellular drug concentrations. While it was recommended that the drug be given with high fat meals known to increase fenretinide bioavailability (42
), wide inter-patient variability may be due to diet variations and/or incomplete disintegration of the gelatin capsules. Patients received 5–14 capsules per dose, which was challenging to administer to young children. Poor bioavailability of the capsular formulation may have limited efficacy.
Systemic toxicity was minimal. The death from hepatic failure was attributable to tumor progression. While the Phase I 09709 studies reported three reversible cases of pseudotumor cerebri at three dose levels (25
), none occurred on this study. Despite significant retinol depletion, only 1/6 patients over 18 years reported nyctalopia. There were no cases in younger patients, which may be due to under-reporting.
Novel formulations of 4-HPR which optimize pharmacokinetics and feasibility of administration in children are currently being tested in pediatric and adult Phase I trials. A 4-HPR formulation packaged in LYM-X-SORB™ (LXS™) (43
), a lipid matrix technology powder, was tolerated in doses up to 2210 mg/m2
/day without dose-limiting toxicity in an ongoing NANT trial (44
). Mean peak plasma levels were 15–20 µM versus 6–9 µM with the capsular formulation. An absorption plateau was observed, as seen with the capsule formulation. An intravenous 4-HPR emulsion formulation is also being tested in ongoing adult cancer trials and a pediatric neuroblastoma (NANT) trial, with clinically tolerable peak plasma levels up to 50 µM in adults (45
). Results from these studies will help to determine if higher fenretinide plasma levels are tolerable and associated with improved anti-tumor activity.
The cumulative data with fenretinide support activity of this agent against neuroblastoma. The capsule formulation utilized in this study was suboptimal due to poor bioavailability, and difficulty administering to children, and is not recommended for future trials. However, this study provides important clinical response data for comparison to data obtained in future trials of novel fenretinide formulations that can achieve higher drug exposures that will be necessary to define the role of fenretinide in therapy for high-risk neuroblastoma.
Statement of translational relevance
This phase II study assessed the activity of a capsular formulation of fenretinide in refractory/recurrent neuroblastoma. A novel study design utilized two strata: 1) RECIST-defined measurable disease by CT/MRI scans and 2) disease evaluable by non-RECIST methods, i.e. bone marrow morphology and semi-quantitative scoring of I-131MIBG avid disease. Other novel variables were identified that affected time to progression: history of prior relapse versus resistant tumor and tumor sites at study entry. Identification of variables affecting time to progression is critical for design of future studies using this endpoint. This study assessed the utility of a multi-strata phase II trial evaluating agents with minimal systemic toxicity but also minimal activity against mass disease and demonstrated sufficient activity of fenretinide at modest systemic exposures to justify ongoing trials of novel fenretinide formulations with higher bioavailability. Importantly this study documents responses and time to progression in both strata for comparison to ongoing and future phase II clinical trials in recurrent neuroblastoma.