When Genetic Testing Is Unproven: The Case of Depression Treatments

Patients are increasingly requesting commercially available genetic testing. What is the level of evidence supporting such requests?

Genetic Testing Panels

A 2017 systematic review identified five studies on the impact of genetic testing panels on clinical outcomes in major depression,1 all at least partially industry-funded and focusing on panels of genes (usually <10) involved in pharmacokinetics (drug metabolism) and pharmacodynamics (drug site of action; e.g., receptors). The studies were relatively small (sample sizes for genetic studies ordinarily need to be in the thousands to ascertain statistical significance). Three were open and uncontrolled: Thus, it remained possible that patients who knew they were getting genetic testing were more likely to have a placebo response while doctors inflated the benefit (i.e., exaggerated patients' responses) to support the hypothesis.

Only two studies used randomized, double-blind, prospective methods; one showed no significance. In the other, focusing simply on “how to dose antidepressants,” researchers examined variants in pharmacokinetic genes (cytochrome P450 [CYP] 2D6 and CYP 2C19 and blood–brain barrier transporters ABCC1 and ABCB1) in ≈150 depressed patients. The study did not specify how testing guided physicians' prescribing. Its findings — 72% remission rate in patients whose physicians used the tests to guide their prescriptions versus 28% in patients whose physicians did not — seem unlikely because virtually all depression medication and psychotherapy studies report remission rates in the 30% to 45% range.

Assay Validity vs. Clinical Utility

Despite these meager results, the FDA has approved several genetic tests. What does this mean? FDA approval requires only “assay validity” — that the test measures what it says it is measuring. There is no requirement that a test shows clinical validity or clinical utility (i.e., actually resulting in better patient outcomes). Despite FDA approval of assay validity only, these tests are being heavily marketed as if they had been shown to improve patient outcomes.

In fact, more-powerful and well-designed studies have failed to show that current pharmacogenetics techniques can predict antidepressant outcome. The largest report, a meta-analysis2 of three large treatment trials (STAR*D, GENDEP, and MARS) involving 2256 patients with major depression, combined data on genetic variation across the genome into a “polygenic risk score” that accounted for at most 1% to 2% of the variance in response. Covering vastly more genes than today's commercial tests, this analysis revealed much less explanatory power than clinicians have been led to believe.

A subsequent, more sophisticated analysis3 using data from STAR*D and another treatment study estimated — under the most ideal circumstances (not achievable with current techniques) — a potential 42% genetic contribution to antidepressant response; however, an editorialist4 noted that in practical terms, the ability of genes to explain response based on this analysis would be “an order of magnitude smaller” — around 4% (surprisingly close to the 1%–2% figure of the previous study). Future studies will require much larger sample sizes plus advanced analytic, computational models.

Unfortunately, relying on genes controlling the CYP metabolic system to identify “fast” and “poor” metabolizers of specific drugs seems less clinically useful than one might anticipate. In the GENDEP study of 223 depressed patients treated with nortriptyline or escitalopram,5 the significant correlations found between genotypes for CYP2D6 and CYP2C19 (enzymes that metabolize these medications) and antidepressant serum concentrations were quite modest; neither genotype nor serum levels correlated with clinical response. Many factors beyond genetic control of drug metabolism affect blood levels, including comorbid medical illnesses, adherence, other medications, chronicity of treatment, gene and second messenger interactions, induction of one metabolic pathway when another is suppressed, genes' disparate actions in peripheral blood versus brain tissues, and actions of medications on dimensions of depression (e.g., anhedonia) rather than the full syndrome.6 Further, epigenetic alteration of gene expression determines whether and how much specific genes are actually expressed.