Pooled phase I/II analysis versus external control
85 evidence topics · 20 sources
Objective
Pooled analysis protocols and prespecified external control comparison, September 2025
Comparison of Cohorts 1 and 2 with an Enroll-HD external control as the basis for a BLA
Updated external control comparison in the pooled protocols, September 2026
Location and study date
Data cutoff of the three-year analysis
Data cutoff of the four-year and updated 36-month analyses
Enroll-HD dataset releases used as external controls
Clinical trial sites of CT-AMT-130-01 and CT-AMT-130-02
Data cutoff of the 24-month interim analysis, July 2024
Data cutoff of the interim analysis, December 2023
Study design
Prespecified statistical analysis plan of the three-year analysis
Propensity score-matched external control reported as the prespecified primary comparison, September 2025
Statistical analysis plan as described in June 2025: propensity score-weighted primary analysis planned
Rationale for selecting Enroll-HD
Propensity score matching method and covariates
Statistical analysis plan of the 48-month analysis with an updated external control
Propensity-weighted TRACK-HD, TRACK-ON, and PREDICT-HD external control, July 2024
Criteria-matched TRACK-HD natural history cohort, December 2023
Eligibility criteria
Key eligibility criteria of the high-dose cohort in the three-year analysis
“Aged 25-65 years” (opens the source at this quote in a new tab)
“Diagnostic confidence level: 3 or 4” (opens the source at this quote in a new tab)
“Total functional capacity: 9-13” (opens the source at this quote in a new tab)
“HD medications stable prior 3 months” (opens the source at this quote in a new tab)
“Striatal volumes exceed specified minima” (opens the source at this quote in a new tab)
“Putamen: ≥ 2.5 cm3/side” (opens the source at this quote in a new tab)
“Caudate: ≥ 2.0 cm3/side” (opens the source at this quote in a new tab)
Selection of the Enroll-HD external control: exclusion criteria
Treatment
Treated participants by dose and imitation surgery controls, July 2024
Study outcomes
Rationale for using cUHDRS change at 36 months as the primary endpoint
Primary, secondary, and supportive endpoints of the three-year analysis
Primary endpoint of the 48-month analysis
Components of the cUHDRS as described by the sponsor
| Component | Domain described |
|---|---|
| Total Functional Capacity (TFC) | “Impairments in occupation, finance, domestic chores, daily living, and care level” |
| Total Motor Score (TMS) | “Impairment in motor function” |
| Symbol Digit Modalities Test (SDMT) | “Measure of attention, processing speed, and working memory” |
| Stroop Word Reading Test (SWRT) | “Measure of selective attention capacity and processing speed” |
Regulatory acceptance of cUHDRS as an intermediate clinical endpoint, December 2024
Sponsor statement on the sensitivity of the cUHDRS
Derivation of the cUHDRS from four UHDRS measures
“The overall pattern of the correlations and the PCA results indicated that TMS, TFC, SDMT, and SWR had significant overlap, but PBA-apathy did not.” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
“Note the negative sign for (scaled) TMS and the positive sign for all other variables.” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
“The observed difference in the S/N ratio between the cUHDRS and each individual component in TRACK-HD was numerically superior in all cases (e.g., 26%–38% larger than the best individual measure, which was the TMS at 24 and 36 months, respectively)” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
“The results (ICC2,1 = 0.96)9 showed excellent test-retest reliability of the cUHDRS.” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
Clinically efficacious slowing on the cUHDRS
“We have reached consensus that a slowing of clinical decline in the range of 20% to 30% annually on the cUHDRS (≈0.2–0.3 points per year), equivalent to ≈2.4 to 3.5 mo/y of time saved from progression, can be used to define a clinically efficacious treatment in this setting.” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
Correlation of the cUHDRS with imaging biomarkers
“Worse performance in the cUHDRS over time correlated with longitudinal volume decreases in the occipito-parietal cortex and centrum semiovale, whereas lower baseline scores correlated with decreased volume in the basal ganglia and surrounding WM.” (opens the original PubMed page at this quote in a new tab)View highlighted excerpt for this quote
“The cUHDRS correlates with imaging biomarkers and tracks atrophy progression in HD supporting its biological relevance.” (opens the original PubMed page at this quote in a new tab)View highlighted excerpt for this quote
Key secondary endpoint: Total Functional Capacity
“It assesses capacity as opposed to actual performance and consists of a 5‐item interview between a clinician, and the patient and a person familiar with the patient's functioning.” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
“Extensive data from multiple observational studies and clinical trials suggest sensitivity to change over time.” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
“There appears to be a ceiling effect for early stage HD and a floor effect for late stage HD.” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
“Recommendation: The UHDRS‐TFC is suggested for assessing severity of limitation in functional capacity in HD, because it lacks core clinimetric data, namely, test–retest reliability and internal consistency to reach a “recommended” status.” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
Total Functional Capacity as a measure of daily function: investigator statement
Secondary endpoints: Symbol Digit Modalities Test and Stroop Word Reading Test
“(1) the symbol digit modalities test (SDMT/sampled cognitive domains: visual attention, working memory, symbolic encoding, and psychomotor speed), in which the total score corresponds to the number of correct responses within 90 seconds” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
“(2) the Stroop color word interference test (sampled cognitive domains: psychomotor speed, cognitive flexibility, response inhibition, and selective attention), which includes 3 conditions (word reading, color naming, interference), each one of which is scored as the total number of items accurately completed within 45 seconds” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
Supportive endpoint: cerebrospinal fluid neurofilament light chain
Natural history of cerebrospinal fluid neurofilament light chain in the HD-CSF cohort
“mHTT in CSF, as well as NfL in CSF and plasma, all rose detectably within participants over 2 years.” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
“On the basis of simulations of the models in Fig. 2 (D to F), CSF NfL rose in manifest HD by 98.85 pg/ml per year, in preHD by 79.16 pg/ml per year, and in controls by 20.05 pg/ml per year (Fig. 2J).” (opens the original PMC page at this quote in a new tab)View highlighted excerpt for this quote
Natural history comparator for cerebrospinal fluid neurofilament light chain, July 2024
Statistical analysis description
Number of participants (Planned and analyzed)
Participants and external controls in the three-year analysis
Participants and external controls in the four-year and updated 36-month analyses
Participants with data at each visit in the four-year analysis, high dose
| Group and endpoint | Baseline | 12 months | 24 months | 36 months | 48 months |
|---|---|---|---|---|---|
| AMT-130 high dose (cUHDRS and TFC) | “17” | “17” | “15” | “15” | “12” |
| Enroll-HD B, cUHDRS | “1337” | “953” | “760” | “690” | “625” |
| Enroll-HD B, TFC | “1337” | “967” | “772” | “704” | “641” |
| Enroll-HD A, cUHDRS (post hoc) | “619” | “484” | “398” | “355” | “349” |
| Enroll-HD A, TFC (post hoc) | “619” | “492” | “401” | “360” | “364” |
Participants in the 24-month interim analysis, July 2024
Participants in the interim analysis, December 2023
Description of analysis sets
Derivation of the Enroll-HD external control arm analysis set
| Step | Participants |
|---|---|
| Diagnosis of HD in ENROLL-HD | “N = 19,542” |
| Met AMT-130 eligibility criteria | “N = 6,596 (33.8%)” |
| Did not meet AMT-130 eligibility criteria | “N = 12,946” |
| Excluded from the analysis set | “N = 3,326” |
| External control arm analysis set | “N = 3,270 (49.6%)” |
| Propensity-score matched to High-Dose cohort | “N = 940” |
Analysis method for the primary and key secondary endpoints
Primary and post hoc comparisons in the 48-month analysis
Safety analysis set and adverse event definitions, June 2025 data cutoff
Results
Participant disposition
Lumbar puncture declined at 36 months
Baseline characteristics
Demographic and baseline characteristics of the high-dose cohort and the propensity score-matched external control
| Characteristic, mean (SD) | AMT-130 high dose (N=17) | PSM external control, Enroll-HD (N=940) |
|---|---|---|
| Sex, males (%) | “47.1” | “55.6” |
| Age | “45.8” | “45.2” |
| CAG repeats | “42.4” | “42.8” |
| CAP100 score | “86.2” | “86.8” |
| DCL = 3, 4 (%) | “35.3, 64.7” | “30.5, 69.5” |
| PIN score | “0.77” | “0.81” |
| cUHDRS | “14.9” | “14.7” |
| TFC | “12.2” | “12.1” |
| SDMT | “46.1” | “45.3” |
| SWRT | “89.9” | “87.6” |
| TMS | “12.1” | “11.6” |
| HD-ISS Stage 2, 3 (%) | “47.1, 52.9” | “51.6, 48.4” |
| Region; No. America, Other (%) | “58.8, 41.2” | “28.9, 71.1” |
Baseline scores of the treated cohorts, December 2023
Efficacy results
Primary endpoint at 36 months: cUHDRS, high dose
Key secondary endpoint at 36 months: TFC, high dose
Other secondary endpoints at 36 months: SDMT, high dose
Other secondary endpoints at 36 months: SWRT and TMS, high dose
Results at 36 months, low dose
| Endpoint | AMT-130 low dose (N=12), LSM change (SE) | Enroll-HD PSM external control, LSM change (SE) | External control N | LSM difference [95% CI] | Slowing of disease progression (%) | P-value |
|---|---|---|---|---|---|---|
| cUHDRS | “-1.65 (0.411)” | “-1.72 (0.151)” | “N=383” | “0.07 [-0.75, 0.88]” | “3.9” | “0.871” |
| TFC | “-0.33 (0.296)” | “-1.04 (0.120)” | “N=392” | “0.71 [0.12, 1.31]” | “68.1” | “0.019” |
| SDMT | “-6.44 (1.532)” | “-3.35 (0.503)” | “N=387” | “-3.09 [-6.14, -0.05]” | “-92.3” | “0.046” |
| SWRT | “-3.67 (4.134)” | “-5.20 (1.373)” | “N=387” | “1.44 [-7.45, 10.33]” | “27.7” | “0.751” |
| TMS | “8.64 (2.039)” | “5.61 (0.688)” | “N=392” | “3.02 [-1.23, 7.28]” | “-53.9” | “0.163” |
Dose response and sensitivity analyses at 36 months
Supportive endpoint at 36 months: cerebrospinal fluid neurofilament light chain
Updated 36-month analysis: cUHDRS and TFC, high dose (n=15)
Updated 36-month analysis: other secondary endpoints, high dose
| Endpoint | AMT-130 high dose (N=17), LSM change | Enroll-HD B (N=1337), LSM change | Slowing of disease progression | P-value |
|---|---|---|---|---|
| SDMT | “0.67” | “-2.78” | “Slowed HD progression by 124% based on SDMT” | “p=0.040” |
| SWRT | “1.32” | “-5.75” | “Slowed HD progression by 123% based on SWRT” | “p=0.047” |
| TMS | “1.93” | “4.77” | “Slowed HD progression by 59% based on TMS” | “p=0.114” |
Primary analysis at 48 months: cUHDRS and TFC, high dose (n=12)
Primary analysis at 48 months: other secondary endpoints, high dose
| Endpoint | AMT-130 high dose (N=12), LSM change | Enroll-HD B, LSM change | Enroll-HD B N | Slowing of disease progression | P-value |
|---|---|---|---|---|---|
| SDMT | “-2.75” | “-3.43” | “Enroll-HD B (N=634)” | “Slowed HD progression by 20% based on SDMT” | “p=0.731” |
| SWRT | “0.95” | “-7.53” | “Enroll-HD B (N=633)” | “Slowed HD progression by 113% based on SWRT” | “p=0.006” |
| TMS | “4.82” | “5.10” | “Enroll-HD B (N=631)” | “Slowed HD progression by 5.6% based on TMS” | “p=0.908” |
Post hoc analysis at 48 months against the prior Enroll-HD external control
| Endpoint | AMT-130 high dose (N=17), LSM change | Enroll-HD A (N=619), LSM change | Slowing of disease progression |
|---|---|---|---|
| cUHDRS | “-0.90” | “-1.94” | “54% slowing of disease progression at 48 months (p=0.041” |
| TFC | “-0.39” | “-1.24” | “68% slowing of disease progression at 48 months (p<0.001” |
Cerebrospinal fluid neurofilament light chain through 48 months, both doses
Interim results at 24 months against the propensity-weighted external control, July 2024
Interim results against the criteria-matched TRACK-HD cohort, December 2023
Health-related quality of life and patient-reported outcomes
No evidence found.
Safety results
Safety summary at the three-year data cutoff
Timing and resolution of drug-related serious adverse events
Adverse events in Cohorts 1 to 3, June 2025 data cutoff
Reclassified serious adverse event and most common adverse events, June 2025 data cutoff
Safety summary at the four-year data cutoff
Adverse events in all 51 treated participants, June 2026 data cutoff
Serious adverse events across both studies, December 2023
Study limitations
Summary
- External comparison: efficacy is assessed against a propensity score-matched external control from the Enroll-HD natural history data set. The Form 10-K states that the analyses rely on external historical controls and that these may not be accepted as a basis for regulatory approval. After a Type A meeting in January 2026, the FDA stated that it cannot agree that these data are sufficient as primary evidence of effectiveness. One commentary noted that the evidence of efficacy “depends entirely on the ENROLL-HD controls being well-matched”.
- Open-label treatment: the same commentary noted that most participants were in open-label arms, so a placebo effect is possible. An analysis of the GENERATION HD1 placebo arm found a placebo response on the cognitive scores (SDMT and SWR), which also influences the cUHDRS.
- Matching covariates: MRI data were not collected in Enroll-HD, so striatal volume, a predictor of decline and an eligibility criterion of the phase I/II studies, could not be matched. The sponsor analysis used baseline age, sex, CAP100, DCL, TFC, TMS, SDMT, and SWR and reports that other covariates can substitute for striatal volume. A comment submitted to ICER listed limits of the natural history comparison, including bias from selection into a surgical trial versus a registry, different visit schedules, and very small numbers of treated patients at longer follow-up.
- Small numbers: 12 high-dose participants contributed 36-month data to the three-year analysis, 15 to the updated 36-month analysis, and 12 to the 48-month analysis; CSF NfL was available for 11 high-dose participants at 36 months in the three-year analysis.
- Missing data and changes in the control: missingness in the updated Enroll-HD matched controls reached 53% at 48 months, controls with longer follow-up declined more slowly, and decline in Enroll-HD B was 20% slower than in Enroll-HD A at 48 months. Against Enroll-HD B the 48-month cUHDRS result did not reach statistical significance (p=0.144); the result against Enroll-HD A comes from a post-hoc sensitivity analysis. A news report on the four-year analysis noted that the natural history database is updated, which “can alter results”.
- Analysis choices and nominal p-values: the June 2025 release described a propensity score-weighted primary analysis, while the September 2025 release reports the propensity score-matched comparison as the prespecified primary analysis. The 24-month comparison with a propensity-weighted external control of 154 participants was a post-hoc analysis. Several p-values are nominal, which the Form 10-K states regulators may weigh less than standard p-values.
- Reporting: the results are reported in company press releases, presentations, and SEC filings. One commentary noted that individual data points are not shown, error bars are not defined, data from imitation (sham) procedure participants are not shown, and there are no data on mutant HTT lowering.
Sponsor-disclosed risks of external controls, post hoc analyses, and nominal p-values
Missing data and survivor bias in the updated Enroll-HD external control
Absence of striatal volume data in Enroll-HD
Effect of volumetric criteria on 3-year decline in TRACK-HD, TRACK-ON, and PREDICT-HD
| Analysis | Volumetric criteria applied (N = 232) | Volumetric criteria not applied (N = 442) |
|---|---|---|
| cUHDRS mean change from baseline to 3 years, unweighted | “-1.4” | “-1.7” |
| cUHDRS mean change from baseline to 3 years, PS weighted | “-1.4” | “-1.4” |
| TFC mean change from baseline to 3 years, unweighted | “-1.0” | “-1.2” |
| TFC mean change from baseline to 3 years, PS weighted | “-0.9” | “-0.9” |