The current read
Constructive, but less so after the four-year update.
- AMT-130’s 48-month data weakened what had been an unusually clean disease-modification narrative. In 12 high-dose patients, cUHDRS showed 44% slowing versus the updated external control and was no longer statistically significant (p=0.144), down materially from the earlier three-year estimate. TFC held up considerably better, with 61% slowing of functional decline (nominal p=0.008).
- The result therefore looks more like increased uncertainty around effect size and external-control reliability than evidence that the treatment effect disappeared. That interpretation is supported by the updated 36-month cohort of all 15 high-dose patients, which showed 80% slowing on cUHDRS and 67% on TFC.
- The regulatory distinction matters. FDA previously agreed that the 36-month dataset could serve as the primary basis for the accelerated-approval BLA, and the four-year results were not part of the submitted application. The investment case has nevertheless become higher risk: a small treated cohort, substantial attrition in the external control and materially different effect estimates depending on timepoint and comparator make the durability of the true treatment effect harder to estimate than it appeared before September 29.