[dekhn]

I dedicated a significant fraction of my life to the idea that gene therapy would become a viable medical treatment. It seems unlikely to really become a high-impact treatment.

Being able to change DNA in cells is one thing; actually being able to show your treatment does what it's supposed to and didn't have negative effects is hard. There are a limited number of diseases where gene therapy should work great, but there are a wide range of others where it won't, until solve multiple grand-challenge class problems.

[tropshop]

Would you elaborate on what makes a disease a good fit and these grand-challenge class problems?

[dekhn]

A disease is a good fit if you can just inject the therapy into a localized region (say, the eye, or an organ) and the treatment works for a reasonably long period of time (months+). Typically another requirement is that the target is a defective gene where the phenotype can be repaired through addition of a "corrected" form of the gene, without the defective gene needing to be removed. This is the case in X-linked retinitis pigmentosa, where the genetic cause of the disease has been understood for some time, it's relatively simple underlying mechanism (so we think), and you can deliver the medicine to the retina with periodic injections.

Treating a disease where you have to remove an inserted retrovirus from a large number of freely circulating or "hidden" cells (which is the case in HIV) is far more challenging- you need a way to recognize the cells of interest, access all of them, and get 100% transversion. All without causing negative side effects.

[gmarx]

We learned early on that one of the difficulties in eliminating HIV is that it hides in the nervous system and can reemerge at any time. This article show promise in that it can reduce HIV viral load during active shedding. I think it is less likely that it could ever eliminate HIV entirely (i.e. cure)

[dekhn]

which is why press releases like this aren't really accurate and create false expectations