Genetic engineering of adult subjects is currently not possible, especially with splicing
But it might be theoretically possible, with some huge challenges to overcome:
A retrovirus is a virus capable of modifying a host cell's genome. I think the current scientific cliff is left at this: The virus can either modify the genome to our purposes or replicate itself but not both. The goal is to get it to both modify a genome beneficially to our purposes and the cells as well as enable replication of the virus.
Otherwise it would require infeasible sums of engineered virus to modify one person.
The virus can be attacked by the immune system. Does no good if the virus is constantly eaten before it does its job. This can be overcome as some viruses can remain invisible to the immune system by the configuration of their protein shell. Its also problematic if the immune system kills the host in trying to kill the virus.
The virus mutates. Viruses are inherently genetically unstable and tend to mutate often. It becomes dangerous as hell if it mutates something dangerous with all the weapons its been granted. Its also dangerous if it becomes contagious infecting random people or creatures causing unknowable havoc.
Getting the Virus Everywhere is Hard Getting the virus into every cell is tricky as things reach different parts of the body at different rates. This also covers the fact that there are different types of cells which require different entry procedures for the virus to mutate. These problems come together in the most troublesome spot, the brain. Nerves are hard, both hard for viruses to enter (because they are less common) and are generally more protected (like the blood brain barrier). So getting the virus into everything uniformly can be difficult.
Adults are adults: Physical maturation is an important biological process regulated by hormones in the brain. This is important in where genes are inserted as well as the process for activating them. For instance if your re-engineering intended to grow a third eye you need someway to activate and more importantly shut off the growth process. You wouldn't want a thousand eyes. Currently this process isn't well understood either. It could be that you simply need to supply your patient hormonal treatments in order to progress the mutations construction.
Some Cells don't Divide Mutating a nerve would be tricky because they don't divide. That isn't to say nerves cant be enhanced because you can highjack something like a stem cell but how you manage modification of the brain can be enormously difficult and tricky. The slightest mistake could result in the permanent crippling or 'resetting'of the patient. Not to mention after you reset a person it might be possible to rehabilitate them either as the adult brain is no longer performing the same cognitive formatting as an infants. This challenge alone makes retroviral gene editing both dangerous and completely rules out neural modifications for at least the next 100 years if not the next millennium. It also limits the significance of mutations. Because general motor skills are hard wired during infancy, it would be impossible for an adult to grow 4 arms and then be able to learn how to use them.
Complexity of the mutation Since adults are fully grown trying to unleash massive rewrites would be dangerous. This might be conquerable through iterative editing but if you change too much too fast you run the risk of killing the host in an infinitesimal array of ways.