Deramiocel

HOPE-3 randomized 106 participants, of whom 90 (84.9%) were non-ambulatory. The published percentage-change analysis favoured deramiocel, whereas FDA’s analysis using the prespecified plan did not demonstrate a significant difference in the primary endpoint. FDA also identified statistical changes without an updated SAP. The published key secondary LVEF result was not statistically significant. These findings require the analysis scale, analysis population, and statistical plan to remain distinguishable. The Patient Global Impression of Severity confidence interval included no difference. The 12-month controlled comparison and its predominantly non-ambulatory population do not establish effects on long-term survival or applicability across all DMD stages.

HOPE-2 enrolled 20 participants rather than the original target of 84; an amended protocol limited enrollment to 20. Allocation was 8 to CAP-1002 and 12 to placebo, and the FDA primary efficacy analysis included 19 participants. The publication reported a significant percentile-rank result, whereas FDA’s original-scale analysis had a confidence interval crossing zero and a nominal p value of 0.13. FDA identified changes to the statistical plan after unblinding. The SAP described subsequent efficacy p values as nominal. The abstract and full-text results report different confidence intervals for the percentile difference. The small sample, incomplete follow-up, and disagreement between analyses limit precision and confidence in the treatment-effect estimate.

The extension enrolled 13 participants who had completed the parent trial; all received deramiocel, and 9 continued to month 60. The absence of a concurrent randomized control group, selection of prior-trial completers, and attrition limit causal interpretation. External-comparator analyses depend on modeling and cross-study comparability. The five-year comparator estimate is a projection, not an observed five-year randomized comparison. The earlier poster described the 3.46-point and 7.19-point estimates as yearly declines, whereas the later presentation labels them as three-year modeled changes; these descriptions should not be combined as if they used the same time scale. Off-treatment comparisons used linear interpolation, and analyses were exploratory and unadjusted for multiplicity.

HOPE-OLE treated 8 participants who had previously been allocated to usual care in HOPE-Duchenne. It was an open-label extension without a concurrent randomized comparator. The regimen comprised two intravenous administrations of 75 million cells, with the repeat dose at month 3, rather than the 150-million-cell regimen used in HOPE-2 and HOPE-3. The small, selected population and different exposure limit its contribution to efficacy assessment and extrapolation to later dosing regimens.

HOPE-Duchenne was an open-label phase I/II trial with 25 participants and a single intracoronary administration rather than the repeated intravenous regimen used in later trials. The upper-limb responder analysis and definition of the lower-functioning subgroup were post hoc. Statistical testing did not adjust for multiplicity, so the nominal questionnaire and functional findings should not be interpreted as independently confirmed effects. Its size, open-label design, and different route and dosing schedule limit extrapolation to the proposed repeated-infusion treatment.