No, this is simply not true. Unless you also mean “cures the disease, but kills the patient for some other reason” and other net negative outcomes.
Most of those studies are on tissue samples, or rodent analogs. That’s not the same as live trials and there is no basis for claiming they’d work in situ.
There's further no guarantee that the rat survives, or mates, or becomes emperor of China for that matter. What is reproducible is what is described in the paper, not what you might wish would follow.
Reproducibility is a very specific requirement. It requires that, following the procedure as outlined, you will find similar data. It has nothing to do with application, or utility, or anything else. It is simply that your results, for the same experiment, will be similar.
Few early-stage trials make it to market, but that's an entirely different matter. The treatment they developed, and published in a highly regarded journal, almost certainly works as described. That's a very different beast from a treatment that will be effective in humans to the high standards we hold modern drugs to.