Rapid evidence review of the comparative effectiveness, harms, and cost-effectiveness of pharmacogenomics-guided antidepressant treatment versus usual care for major depressive disorder

Objective: This study aims to conduct an evidence review of the effectiveness, harms, and cost-effectiveness of pharmacogenomics-guided antidepressant treatment for major depressive disorder. Methods: We searched MEDLINE®, the Cochrane Central Registry of Controlled Trials, and PsycINFO through February 2017. We used prespecified criteria to select studies, abstract data, and rate internal validity and strength of the evidence (PROSPERO number CRD42016036358). Results: We included two randomized trials (RCT), five controlled cohort studies, and six modeling studies of mostly women in their mid-40s with few comorbidities. CNSDose (ABCB1, ABCC1, CYP2C19, CYP2D6, UGT1A1) is the only pharmacogenomics test that significantly improved remission (one additional remitting patient in 12 weeks per three genotyped, 95% CI 1.7 to 3.5) and reduced intolerability in an RCT. ABCB1 genotyping leads to one additional remitting patient in 5 weeks per three genotyped (95% CI 3 to 20), but tolerability was not reported. In an RCT, GeneSight (CYP2D6, CYPC19, CYP1A2, SLC6A4, HTR2A) did not statistically significantly improve remission, and evidence is inconclusive about its tolerability. Evidence is generally low strength because RCTs were few and underpowered. Cost-effectiveness is unclear due to lack of directly observed cost-effectiveness outcomes. We found no studies that evaluated whether pharmacogenomics shortens time to optimal treatment, whether improvements were due to switches to genetically congruent medication, or whether effectiveness varies based on test and patient characteristics. Conclusions: Certain pharmacogenomics tools show promise of improving short-term remission rates in women in their mid-40s with few comorbidities. But, important evidence limitations preclude recommending their widespread use and indicate a need for further research.