Predicting antidepressant response – problems and promises

Early effective treatment is important

Precision medicine may seem a long way off in the clinic. The Diagnostic and Statistical Manual of Mental Disorders relies on symptom criteria for diagnosis, whilst clinicians are aware of the many different combinations of symptoms and responses to treatment seen in their patients. Psychiatric disorders such as major depressive disorder (MDD) are heterogenous¹, especially once genetic risk variants² and comorbid conditions³ are considered.

Suboptimal treatment response prolongs the period of depression

Early effective treatment is important as a suboptimal response prolongs the period of depression. This is associated with worse outcomes including increased risk of relapse⁴ and decreased ability to work⁵. Depression also worsens outcomes of many general medical conditions⁶ with the associated personal and societal costs.

Can we predict antidepressant response?

Knowledge of the MDD subtype can already help to guide treatment, for example adding an antipsychotic or electroconvulsive therapy for psychotic depression. The holy grail is being able to predict which is the best first line antidepressant for a particular patient and their constellation of symptoms, comorbidities and risk factors rather than resort to trial and error.

Certain clinical and demographic factors have been shown to predict antidepressant response

Data from the STAR*D study (Sequenced Treatment Alternatives to Relieve Depression)⁷ showed that only 33% of patients achieved remission after first line treatment. The remission rate increased with subsequent steps, but there was diminished effectiveness and increased treatment intolerance with each treatment level.

Certain clinical and demographic factors have been shown to predict antidepressant response including depression severity⁸, gender⁷, and childhood trauma⁹, but there is little known about the mechanisms or modification targets.

Pharmacogenomics in psychiatry

Precision psychiatry would allow the treatment most likely to have the highest drug effect based on genetic profiling and stratification. This is being used in other areas of medicine, including breast cancer treatment based on oestrogen receptor and human epidermal growth factor receptor 2 status¹⁰. Pharmacogenetic variability that may determine the dose-effect relationship includes both pharmacokinetic (mutations in genes encoding metabolic enzymes and transporters) and pharmacodynamic (mutations in genes encoding receptors and other molecular drug targets) variability.

Antidepressant response maybe associated with norepinephrine transporter and corticotropin-releasing factor system genetic polymorphisms

Studies have suggested that antidepressant response maybe associated with norepinephrine transporter¹¹ and corticotropin-releasing factor system¹² genetic polymorphisms. Commercial tests are available to guide drug response in various psychiatric conditions based on serotonin transporter and calcium channel mutations.

The future

Although specific genetic variants have been associated with conditions such as MDD, Prof Nemeroff suggested there is still insufficient data to support the widespread use of combinatorial pharmacogenetic testing in clinical practice¹³. The FDA have issued a warning regarding the use of genetic tests to predict patient response to specific medications¹⁴, although some specific examples such as use of cytochrome P-450 status to aid psychiatric drug dose adjustment maybe useful¹⁵.

Still insufficient data to support the widespread use of combinatorial pharmacogenetic testing

Prof Nemeroff concluded by discussing the use of functional imaging¹⁶ and neurophysiology¹⁷in this area, moving from focusing on neurotransmitters to a circuit-based model. Electroencephalography is showing some promise on both the group and single-subject level to predict antidepressant response versus placebo¹⁸, allowing the balance between mechanistic information and clinical utility.