It would be much more difficult, expensive and time consuming to create biological replacements than it would be to create neural transduction-controlled mechanical replacements. A biological replacement for a 20 year old, would essentially take 20 years to 'grow', resulting in a 20 year old replacemet for a now 40 year old recipient. The problems of rejection and reduced neural connectivity from transplantation would still be an issue, as well as a morality issue of whether or not human biological replacements should even be created.
On the other hand, a mechanical replacement, controlled by a nueral transduction interface (think of an implantable Arduino, whose signal inputs are powered by the very nerve fibers that were used to articulate the lost limb. The chance for rejection is greatly reduced ( only a small chip that translates nerve impulses to and from the replacement would actually be in contact with biological tissue ) The time frame would be greatly reduced, ( a mechanical replacement could be fabricated in a few short months as opposed to many years to grow a biological replacement ) So it's not a question of if and when, but one of 'why even bother?'