Integrated externally controlled survival analysis
56 evidence topics · 12 sources
Objective
Objectives of the company-presented survival analysis with broader external controls
Objectives proposed in the 2021 data request for external controls
Objective of the NIH natural history study used as the external control
Location and study date
Data cutoff of the company-presented survival analysis, January 2025
International registry and French cohort contributing external controls
Database lock of the prior sponsor's studies, December 2021
Data cutoff of the company-presented survival analysis, September 2025, superseded in February 2026
Study design
Integrated survival analysis as the basis of effectiveness in the label
FDA determination that the externally controlled comparison was the adequate and well-controlled study
Externally controlled analysis submitted as a single adequate and well-controlled study
Design rationale stated in the 2021 data request
Company-presented survival analysis with broader external controls: data sources
Company-presented survival analysis with broader external controls: target trial emulation
Company-presented disease progression analysis in treated participants
Company-presented survival analysis matching, September 2025, superseded in February 2026
Eligibility criteria
Neurological onset before 6 years of age in the label analysis
Company-presented survival analysis with broader external controls: eligibility for matching
Inclusion and exclusion criteria proposed in the 2021 data request
Treatment
Treatment exposure and dosage in the label analysis
Company-presented survival analysis with broader external controls: dosing across source studies
Company-presented survival analysis with broader external controls: concomitant NPC therapies
Treatment definition proposed in the 2021 data request
Study outcomes
Rationale for using overall survival as the primary endpoint
Overall survival definition in the label analysis
Survival as an outcome in very young children with the fastest progression
Age of neurological onset as the predictor of survival
Secondary measure in the company-presented analysis: rescored 4-domain NPC Clinical Severity Scale
Statistical analysis description
Number of participants (Planned and analyzed)
Treated and external control participants in the label analysis
Company-presented survival analysis with broader external controls: 72 treated and 119 matched controls
Company-presented survival analysis with 66 treated and 110 matched controls, September 2025, superseded in February 2026
Description of analysis sets
Index age, weighting, and statistical models in the label analysis
Company-presented survival analysis with broader external controls: statistical models and censoring
Company-presented survival analysis with broader external controls: matching with replacement and robust variance
Company-presented survival analysis with broader external controls: prespecified subgroup analyses
| Subgroup analysis | Definition |
|---|---|
| Baseline miglustat use | “Participants with and without baseline miglustat use” |
| Source of external control data | “NIH, YODA, and French cohorts considered separately” |
| Age of neurological onset | “Early and late I-NPC subgroups considered separately” |
| Standard-of-care therapy over time | “Participants born before 2010 and born in 2010 or later” |
Company-presented survival analysis with broader external controls: sensitivity analyses
Survival model proposed in the 2021 data request
Results
Participant disposition
Source studies of the treated participants and follow-up in the label analysis
Company-presented survival analysis with broader external controls: movement of participants between source studies
Baseline characteristics
Baseline characteristics of treated and matched external control participants in the label analysis
Company-presented survival analysis with broader external controls: baseline characteristics of 72 treated and 119 matched controls
| Characteristic | Adrabetadex (N=72) | External control (N=119) |
|---|---|---|
| Age at onset under 2 years, n/N (%) | “32/72 (44)” | “57/119 (48)” |
| Age at onset 2 to under 6 years, n/N (%) | “40/72 (56)” | “62/119 (52)” |
| Median age at neurological onset, months (Q1, Q3) | “25 (0, 68)” | “24 (0, 68)” |
| Overall miglustat use, n/N (%) | “52/72 (72)” | “72/119 (61)” |
| Gastrostomy tube, n/N (%) | “11/50 (22)” | “6/26 (23)” |
| Respiratory therapy, n/N (%) | “9/50 (18)” | “2/26 (8)” |
| Race White, n/N (%) | “65/72 (90)” | “22/26 (85)” |
| Race other, n/N (%) | “7/72 (10)” | “4/26 (15)” |
| Region Europe, n/N (%) | “11/72 (15)” | “90/119 (76)” |
| Region North America, n/N (%) | “61/72 (85)” | “27/119 (23)” |
| Region South America, n/N (%) | “0 (0)” | “2/119 (2)” |
Efficacy results
Overall survival in the label integrated survival analysis
| Outcome | LEMBAZ (N=36) | External control (N=36) |
|---|---|---|
| Number of deaths (%) | “12 (33.3%)” | “24 (66.9%)” |
| Median survival time in months | “102” | “59” |
| Hazard ratio for risk of death (95% CI) | “0.45 (0.21, 0.95)” | Reference |
Overall survival as stated by the FDA
Company-presented survival analysis with broader external controls: overall survival in 72 treated and 119 matched controls
| Population and comparator source | Deaths, adrabetadex (%) | Deaths, external control (%) | HR (95% CI) | P-value |
|---|---|---|---|---|
| Overall infantile | “17 (23.6%)” | “44 (61.2%)” | “0.289 (0.141, 0.593)” | “< 0.0001” |
| Early infantile | “9 (28.1%)” | “24 (76.1%)” | “0.148 (0.064, 0.343)” | “< 0.0001” |
| Late infantile | “8 (20.0%)” | “20 (49.2%)” | “0.344 (0.146, 0.812)” | “0.0457” |
| With miglustat | “8 (22.9%)” | “13 (37.9%)” | “0.361 (0.173, 0.752)” | “0.0205” |
| Without miglustat | “7 (33.3%)” | “19 (91.2%)” | “0.139 (0.049, 0.398)” | “< 0.0001” |
| French comparator | “9 (16.4%)” | “44 (80.7%)” | “0.113 (0.042, 0.301)” | “< 0.0001” |
| NIH comparator | “12 (33.3%)” | “30 (84.2%)” | “0.449 (0.214, 0.946)” | “0.0322” |
| YODA comparator | “8 (21.6%)” | “6 (16.4%)” | “0.381 (0.128, 1.137)” | “0.1426” |
Company-presented survival analysis with broader external controls: hazard ratio and five-year survival as announced
Company-presented survival analysis with broader external controls: consistency across sensitivity analyses
Company-presented disease progression analysis: annual rate of change on the rescored 4-domain NPC Clinical Severity Scale
Company-presented survival analysis with 66 treated and 110 matched controls, September 2025, superseded in February 2026
| Population and comparator source | Deaths, treated (%) | Deaths, control (%) | Hazard ratio (95% CI) | P-value |
|---|---|---|---|---|
| Overall infantile, pooled comparator | “9 (13.6%)” | “42 (63.3%)” | “0.162 (0.069, 0.380)” | “<0.0001” |
| Early infantile, pooled comparator | “6 (18.8%)” | “23 (72.8%)” | “0.109 (0.035, 0.333)” | “<0.0001” |
| Late infantile, pooled comparator | “3 (8.8%)” | “18 (54.3%)” | “0.148 (0.040, 0.546)” | “0.0036” |
| With miglustat | “2 (7.1%)” | “12 (41.8%)” | “0.097 (0.023, 0.406)” | “0.0005” |
| Without miglustat | “5 (21.7%)” | “18 (79.9%)” | “0.123 (0.036, 0.425)” | “<0.0001” |
| French comparator | “5 (9.8%)” | “42 (82.1%)” | “0.073 (0.023, 0.236)” | “<0.0001” |
| NIH comparator | “5 (15.2%)” | “24 (71.2%)” | “0.176 (0.060, 0.517)” | “<0.0001” |
| YODA comparator | “4 (11.4%)” | “6 (17.9%)” | “0.298 (0.082, 1.086)” | “0.1168” |
Health-related quality of life and patient-reported outcomes
No evidence found.
Safety results
No evidence found.
Study limitations
Summary: limitations of the integrated externally controlled survival analysis
- Design: the comparison is with external controls. A statistical methods poster on the analysis states that externally controlled analyses are susceptible to bias from differences in baseline prognosis, treatment initiation at different ages, calendar time and standards of care, and post-baseline therapies, and that target trial emulation identified four principal threats: prognostic factor confounding, post-baseline treatment confounding, immortal time bias, and temporal bias. Both the label analysis and the company analysis required each matched control to be alive at the treated participant's age at treatment initiation.
- Label analysis: all 26 external control participants came from Study 1 and were weighted to a number of 36 against 36 treated participants. Matching used disease subtype, age of neurological symptom onset (±3 months), and baseline miglustat use; 58% of treated and 33% of control participants were male, median follow-up from the index age was 5.6 and 4.2 years, and index ages ranged from 1 to 19 years. The label states that safety and effectiveness in reducing mortality with neurological onset at 6 years of age or older has not been established.
- Analysis with broader external controls: 72 treated participants were matched with 119 external controls drawn from 4 major disease databases or publications, and the poster states that comparability factors were not available from all sources.
- Variation in the estimate: the hazard ratio differed by comparator source (French 0.113; NIH 0.449, 95% CI 0.214 to 0.946; YODA 0.381, 95% CI 0.128 to 1.137, P = 0.1426) and between data cuts (0.162 with 66 treated participants at the June 24, 2024 cutoff; 0.289 with 72 at the January 10, 2025 cutoff). The label analysis reported 0.45 (0.21, 0.95). The statistical methods poster reported that hazard ratio estimates across its prespecified sensitivity analyses remained consistent with the primary analysis.
- Analysis plan: the YODA Project data request that proposed the external control analysis stated that “All planned statistical analyses are exploratory” and that no minimum sample size was determined a priori.