Pharmacogenomics in psychiatric practice has a defined but limited clinical place. The genuine actionable variants — CYP2D6 and CYP2C19 polymorphisms predicting altered metabolism, HLA-B*1502 for carbamazepine in specific populations, a few others — are clinically meaningful in specific scenarios. The broader commercial panels claiming to predict drug response across psychiatric medications frequently produce data without commensurate clinical actionability. The discipline is to use the testing where it changes treatment decisions and to avoid the testing where it produces noise.
CYP2D6 and CYP2C19 are the clinically meaningful pharmacokinetic variants. CYP2D6 metabolizes many antidepressants (paroxetine, fluoxetine, venlafaxine, tricyclics) and antipsychotics; poor metabolizers may have elevated drug levels and side effects, ultrarapid metabolizers may have inadequate exposure on standard doses. CYP2C19 metabolizes citalopram, escitalopram, sertraline; similar implications. The testing provides genuinely actionable information about dose adjustments and agent selection in patients with multiple unexplained treatment failures or severe side effects on standard doses.
HLA-B*1502 for carbamazepine in selected populations. Patients of Asian descent (particularly Han Chinese, Thai, others) have elevated frequency of HLA-B*1502, which is associated with substantially increased risk of Stevens-Johnson syndrome and toxic epidermal necrolysis on carbamazepine. Testing before initiation in these populations is FDA-recommended. The clinical actionability is clear — positive test contraindicates carbamazepine.
The broader pharmacodynamic panels. Commercial panels (GeneSight, Genomind, others) include pharmacodynamic markers (HTR2A, SLC6A4, others) with claims about drug response prediction. The clinical evidence for these is weak; the prediction accuracy is poor in independent validation studies. The panels produce abundant data that does not robustly translate to clinical decisions. The patient considering these panels deserves honest framing about the limited actionability beyond CYP testing.
When to order pharmacogenomic testing. Multiple unexplained treatment failures across appropriate drug trials. Severe side effects on standard doses. Family history of similar pattern (poor or rapid metabolism). Specific clinical scenarios where CYP testing changes decisions. Not routine practice; not before initial treatment trial; not as substitute for adequate trial of standard treatment. The testing is selective use for the scenarios where it genuinely changes care. The discipline is to use pharmacogenomic testing where it has actionable yield (CYP variants in unexplained failure or severe side effects, HLA-B*1502 in indicated populations) and to decline the broader panel use that produces data without clinical decisions.