The promise of treatment matched to individual biology has been made for two decades. It has been delivered in specific areas and not generally, and the distinction is worth knowing.
Where it works: cancer
Sequencing a tumour identifies which mutations are driving it, and some of those mutations have drugs that target them specifically. A patient whose tumour carries a particular alteration may respond dramatically to a drug that does nothing for a histologically identical tumour without it.
This is genuine personalisation and it has changed outcomes in several cancers.
Where it works: drug metabolism
People vary genetically in how quickly they process particular drugs. A standard dose can be ineffective in a fast metaboliser and dangerous in a slow one.
Testing for the relevant variants before prescribing certain medications is established practice in some health systems and absent in others, which is a straightforward gap.
Where it has not delivered
Common complex diseases. Heart disease, diabetes and most psychiatric conditions involve many genes each contributing a small amount, interacting with environment and behaviour.
Risk scores built from those can be calculated. They predict poorly at the individual level, and knowing your score rarely changes what you should do, because the advice is the same regardless.
The direct-to-consumer problem
Tests sold directly to the public report risk variants without clinical context. That produces anxiety about findings that mean little and false reassurance about the absence of findings a limited test could never have found.
What to expect
Continued expansion in oncology, gradual adoption of pharmacogenomic testing, and continued overselling of everything else.