Adrabetadex efficacy and safety
6 evidence topics · 33 sources
Summary: externally controlled survival analysis in the prescribing information
The prescribing information establishes effectiveness in reducing the risk of mortality, in adult and pediatric patients who weigh 7 kg or greater with neurological manifestations onset prior to 6 years of age, on an integrated survival analysis of data from Trials 1, 2, and 3, an expanded access program, and Study 1, an observational, prospective, natural history study. Thirty-six treated participants (11, 2, 2, and 21 who started treatment in Trials 1, 2, 3, and the expanded access program) were matched on disease subtype, age of neurological symptom onset (±3 months), and baseline miglustat use to 26 external control participants from Study 1, drawn from 44 eligible; the weighted number of control participants was 36. In both groups the mean index age was 5.7 years (range 1 to 19 years), 67% had early infantile-onset NPC, and 50% received miglustat at baseline.
Deaths occurred in 12 (33.3%) treated and 24 (66.9%) external control participants. The hazard ratio for death from a weighted Cox regression model was 0.45 (95% CI 0.21, 0.95), which the label and the FDA describe as a 55% reduction in the risk of death. The FDA reported a median survival time of 8.5 years in the treated group and 4.9 years in the control group, and stated that the benefit was observed regardless of miglustat use. Median follow-up from the index age was 5.6 years in the treated group and 4.2 years in the external control group. Treated participants received adrabetadex for a median of 3.6 years (range 1.4 months to 8.7 years) at a median final dosage of 750 mg (range 300 to 1,050 mg). The manufacturer stated that the FDA determined that this comparison “served as the adequate and well-controlled study supporting approval”.
Summary: company-presented survival and disease progression analyses
- Survival analysis with broader external controls (cutoff January 10, 2025): 72 of 80 match-eligible treated participants (32 early infantile, 40 late infantile) from an expanded access program, phase 1/2a, and phase 2b/3 studies were matched with 119 of 255 eligible external controls (57 early infantile, 62 late infantile) from the NIH Natural History study and three other major disease databases or publications. Median treatment duration was 3.8 years (range 0.1 to 11.1 years). Deaths occurred in 17 (23.6%) treated and 44 (61.2%) control participants (HR 0.289; 95% CI, 0.141 to 0.593; P < 0.0001), which the company announced as a 71% reduction in the risk of mortality, with a 5-year survival rate of 84% for treated participants and 42% for external controls. Hazard ratios were 0.148 (early infantile), 0.344 (late infantile), 0.361 (with miglustat), and 0.139 (without miglustat); by comparator source they were 0.113 (French), 0.449 (0.214, 0.946; NIH), and 0.381 (0.128, 1.137; P = 0.1426; YODA).
- Earlier version of the same comparison (overall survival cutoff June 24, 2024): 66 treated participants and 110 matched controls, with 9 (13.6%) and 42 (63.3%) deaths and a hazard ratio of 0.162 (0.069, 0.380).
- Disease progression analysis: 79 treated participants contributed R4DNPCCSS scores. The annual rate of progression was 1.34 units/year before treatment and 0.76 units per year after treatment (difference 0.58 units/year; 95% CI, –0.98 to –0.18; P=0.0055), reported as a 43% reduction. The company's ACMG press release described the 43% reduction as “over 52 weeks compared with pre-treatment rates of disease progression”.
- Survival by baseline disease burden: 76 treated participants with neurological onset before 6 years were stratified by baseline R4DNPCCSS quartile and compared with an untreated NIH natural history cohort (N=44), with no matching procedure between the cohorts. No deaths were recorded among the 29 treated participants with minimal baseline burden during up to 11.3 years of treatment; in the severe quartile, 5-year KM-estimated survival was 50% compared with 17%.
- Prompt versus delayed treatment: among 25 participants treated for at least 5 years, the 9 who started within 1 year of neurological onset had numerically lower R4DNPCCSS progression than the 16 who started later; median baseline scores were 1.5 and 5.0.
Summary: VTS301 Parts A and B against sham
Parts A and B of VTS301 (Trial 1, NCT02534844) comprised a 52-week, randomized, double-blind, sham-controlled period that enrolled 56 treatment-naive participants aged 4 to 21 years with onset of neurological manifestations prior to 15 years of age. In Part A, 12 participants were randomized 1:1:1:1 to 900 mg, 1200 mg, or 1800 mg or a sham procedure; in Part B, 44 participants were randomized 2:1 to 900 mg or sham.
The co-primary endpoints were the change in NPC-SS composite score (the sum of the ambulation, cognition, fine motor, and swallowing components) and the blinded clinician CGIC at Week 52. The mean change in composite score was 0.0 (SD 2.45) with adrabetadex (n = 38) and -0.1 (SD 1.55) with sham (n = 18). On the clinician CGIC, 16 of 38 adrabetadex and 9 of 18 sham participants were rated unchanged, and none was rated markedly improved. Mallinckrodt reported that the trial “did not show a statistically significant separation from placebo” and that neither arm showed the disease progression expected over 52 weeks; protocol amendment 15 states that the analysis “shows no benefit of adrabetadex on the pre-specified co-primary endpoints”. A post hoc analysis described stabilization in participants receiving miglustat and worsening in participants receiving adrabetadex without miglustat comparable to sham participants not on miglustat. Caregiver-rated CGIC responders numbered 23 of 38 with adrabetadex and 7 of 18 with sham, and the hearing domain score changed by a mean of 1.14 with adrabetadex and -0.22 with sham.
In a cerebrospinal fluid biomarker study of 34 participants, CSF 24(S)-OHC increased 27.7% from baseline at 52 weeks of treatment (n=22; P=0.0012), and calbindin D decreased 18.3% (n=24; P=0.0117) and FABP3 decreased 40.5% (n=24; P=0.0079); the authors defined nominal statistical significance as P<0.05 without adjustment for multiplicity. A YODA Project data request stated that the sham group had little progression.
Summary: open-label studies and expanded access
- Trial 1 Part C (NCT04958642): an open-label extension that enrolled 66 participants (35 treated in Parts A and B, 18 treatment-naive, and 13 from Trial 2) at 900 mg every two weeks. Following a determination of a negative benefit/risk balance on 20 January 2021, the objective was changed to longer-term safety and tolerability, and efficacy endpoints for Part C were not reported. No participant completed the study; termination by sponsor or regulatory authorities was recorded for 8 treatment-naive, 4 phase 1, and 13 Part A/B participants. Serious adverse events occurred in 9 of 18, 7 of 13, and 20 of 35 participants, with one death in the treatment-naive group (1/18) and one in the Part A/B group (1/35).
- Trial 3 (NCT03643562, VTS-270-001): a non-randomized, open-label, single-arm trial of 27 participants (9 treatment-naive, 18 previously treated through the expanded access program) aged 4 to 43 years, with a mean treatment duration of 75 (SD 36) weeks. Under the amended protocol, efficacy assessments were not collected or analyzed. Serious adverse events occurred in 5 of 9 and 9 of 18 participants, and no deaths were recorded.
- Trial 2 (NIH phase 1/2a, NCT01747135): a 78-week, non-randomized, open-label, single-center dose escalation trial of 14 participants aged 4 to 23 years at 200 to 1,200 mg once monthly, regimens the label does not recommend. Against a historical natural history cohort of 21 participants, 7 of 14 (50%) treated participants and 21 of 21 (100%) comparison participants showed disease progression on the NSS minus hearing. The authors list reliance on a historical comparison group and an ad-hoc efficacy analysis as limitations.
- Expanded access program: the label describes 85 participants aged 1 to 43 years (mean 12 years), 71 treatment-naive and 14 previously treated, with a mean treatment duration of 44 (SD 32) months, starting at 200 mg (less than 4 years of age) or 400 mg (4 years of age or older) every two weeks and increased to a maximum of 600 mg (7 to 12 kg) or 900 mg (more than 12 kg). Enrollment began in 2013 and required ineligibility for all currently enrolling clinical trials. A company analysis of the expanded access program (cutoff March 27, 2026; median time since first dose 52.8 months) reported TEAEs leading to death in 6 (7.0%), serious adverse events in 41 (47.7%), hearing impairment in 26 (30.2%), and TEAEs leading to discontinuation in 3 (3.5%).
- Biomarkers in long-term treated cohorts: in 10 individuals treated for an average of 7.2 years, mean CSF 24(S)OHC rose from 0.27 to 0.43 ng/ml 48 hours after dosing, a 1.6-fold increase (p=0.0289). In cohorts drawn from natural history, phase 1/2, phase 2/3, and expanded access protocols, intrathecal treatment was associated with a 19% reduction in CSF total Tau (FDR-adjusted p = 0.008; 101 individuals) in one analysis; another observed no protective effect on neurofilament light chain (OR=1.25, 95% CI 0.99 to 1.58) and increased NfL after adjustment for miglustat (OR=1.35, 95% CI 1.09 to 1.67).
Summary: safety, including the boxed warning and lumbar puncture procedure risks
The label carries a boxed warning for hearing impairment, including permanent hearing loss, and calls for a comprehensive audiology evaluation before treatment; the FDA states that hearing should be evaluated regularly throughout treatment and that dose adjustments may be needed. Permanent hearing impairment ranged from mild to profound, occurred in some participants with normal hearing at baseline, and predominantly affected higher frequencies (at least 2,000 to 8,000 Hz). With upward titration guided by audiometry in Trial 3 and the expanded access program, 37% (22/59) of participants experienced hearing impairment. The label lists no contraindications.
The safety population comprised 141 participants, including 80 with neurological symptom onset before 6 years of age. In Trial 1 Parts A and B, adverse reactions in participants receiving up to 900 mg (N=32) and sham (N=18) included hearing impairment in 23 (71.9%) and 8 (44.4%), ataxia in 22 (68.8%) and 5 (27.8%), fatigue in 18 (56.3%) and 5 (27.8%), vomiting in 17 (53.1%) and 3 (16.7%), back pain in 16 (50.0%) and 3 (16.7%), headache in 14 (43.8%) and 3 (16.7%), and pyrexia in 13 (40.6%) and 3 (16.7%). Ataxia started within 3 days of administration in 61% (107/174) of events and fatigue in 91% (87/96), with a median fatigue duration of 2 days. Aspiration pneumonia occurred in 9% (3/32) of treated and 6% (1/18) of sham participants; in Part C, one participant had fatal aspiration pneumonia after receiving LEMBAZ with sedation, and the label states the cause is unclear. Participants given 1,200 mg or 1,800 mg in Part A had a higher rate of ataxia and fatigue and greater severity of hearing impairment, ataxia, and fatigue than those given 900 mg.
Serious adverse events in Parts A and B are reported in two forms. The ClinicalTrials.gov record gives 16/29 (55.17%) with 900 mg and 3/15 (20.00%) with sham in Part B, and a poster of the Parts A and B safety population gives non-fatal treatment-emergent serious adverse events in 20 (53%) of 38 adrabetadex and 4 (22%) of 18 sham participants.
LEMBAZ is administered intrathecally over 1 to 2 minutes by, or under the supervision of, a healthcare provider experienced in performing lumbar punctures; the label advises considering local anesthetic and sedation, considering imaging when access is difficult, and keeping the patient in the Trendelenburg position for approximately 40 minutes afterward. At Rush University Medical Center, adverse events followed 3.3% of 2935 lumbar punctures in 59 participants (headache 1.2%, back pain 1.3%, vomiting 1.0%, and nausea 0.3% of LPs). Serious adverse events in Part C included spinal subdural haematoma (1/18), spinal subarachnoid haemorrhage (1/35), bacterial meningitis (1/35), staphylococcal meningitis (1/35), and device dislocation (2/35); device implantations were terminated in August 2018 after reports of catheter tip migration. In the NIH phase 1/2a trial, post lumbar puncture headache occurred in 64% (9/14) of participants.
Summary: prior sponsor's benefit-risk determination
In November 2018, Mallinckrodt announced that the registration trial did not show a statistically significant separation from placebo; the FDA had indicated at a Type A meeting in August 2018 that its view on potential approvability would be based on the totality of data. In the three months ended December 27, 2019, Mallinckrodt recognized a full impairment of $274.5 million on its IPR&D asset related to VTS-270, primarily driven by continued regulatory challenges. Effective 20 January 2021, Mallinckrodt recommended that treatment be discontinued as soon as possible, stated that a comprehensive evaluation of all available clinical data “shows no clear evidence of benefit of adrabetadex in the treatment of neurologic symptoms of NPC”, and concluded that “the benefit/risk balance is negative”. Treatment was allowed to continue up to 20 October 2021; continuation after 21 June 2021 required that the investigator assess the participant as benefiting and that the participant understand the risks, including hearing loss. Part C of VTS301 and VTS-270-001 were terminated by the sponsor. Beren, through its subsidiary Mandos, acquired the program from Mallinckrodt in 2021, and a later announcement stated that agreement had been reached with the FDA to continue the US Expanded Access Program beyond October 20th, 2021.