I think you have that backwards. 70% have the wild type gene. The population experienced 49% slowing of the decline rate.
About 30% have the mutant gene. In the 46 week trial the mutant group experienced a 25% slowing rate. That result was not stat sig in the small n population.
The OLE data showed some more differences between the wild and mutant types. The population in the OLE was a small subset of the original trial population.
The trial started with n=508. By the time baseline data was taken the n=462.
The trial ended at 48 weeks with n=320. Of that n=320, n=125 were placebo.
OLE started with n=257 made up of both treated and placebo cohorts that chose to go into the OLE trial.
The OLE ended with n= between 84 and 97 depending on which end point measure is being looked at.
So you can see the problems with the OLE data in terms of self selection, drop out bias, and small n.
That is not to say the OLE data is worthless. It is suggestive. Drawing hard conclusions is not warranted.