Pharmacogenomics of ketamine: A systematic review

Ketamine is a stereoselective NMDA receptor antagonist long used as a dissociative anaesthetic and more recently investigated for pain management and treatment‑resistant depression (TRD). Randomised clinical trials and meta‑analyses have shown rapid antidepressant and anti‑suicidal effects, but interindividual variability in therapeutic response and treatment‑emergent adverse events remains poorly understood. Because ketamine administration is resource intensive, identifying predictors of response or risk prior to treatment could improve patient selection, dosing and cost‑effectiveness. Pharmacogenomics is a promising avenue to personalise drug selection and dosing by linking genetic variation to pharmacokinetics and pharmacodynamics, but available evidence for ketamine has been inconsistent and limited in size and scope. Meshkat and colleagues set out to systematically identify and synthesise published studies that evaluated genetic predictors of ketamine (and esketamine) efficacy and adverse effects. The primary aim was to catalogue candidate genes and allelic variants implicated in ketamine response or side effects, assess the strength and consistency of reported associations, and highlight gaps requiring further research to determine clinical validity of pharmacogenomic testing for ketamine treatment planning.

The investigators performed a systematic review in line with PRISMA guidelines. Searches were run on PubMed, Scopus, and Web of Science from database inception to July 2021, supplemented by Google Scholar searches and manual reference‑list screening. No language or publication‑date limits were imposed. Search strategies combined MeSH terms and keyword searches for ketamine and pharmacogenomics/pharmacogenetics, and also targeted searches pairing ketamine with candidate gene families (for example CYP450, BDNF, glutamate, opioid, GABA, opioid). Eligibility required (1) administration of ketamine or esketamine to human participants, (2) reporting of genotyping results, and (3) outcomes measuring therapeutic effects or treatment‑emergent adverse events. Case reports were eligible; excluded were non‑peer‑reviewed abstracts, in vitro and animal studies, systematic reviews and unpublished datasets. Two reviewers independently extracted predefined data fields from each included paper (year, sample size, age, dose and route, genes/variants, genotyping method, outcomes and definitions, ethnicity, main findings, and underlying condition). Discrepancies were resolved by discussion and the authors attempted to obtain missing information from study authors when necessary. Risk of bias for randomised trials was assessed using the Cochrane Collaboration tool across domains including randomisation, deviations from intended interventions, missing outcome data, outcome assessment and selective reporting. A PRISMA flow process was used: 2,991 records were identified, duplicates removed, titles/abstracts screened, and 24 full texts reviewed. Twelve studies met inclusion criteria and were retained for qualitative synthesis. The authors judged heterogeneity across studies—in design, populations, genes examined and outcomes—too great to permit quantitative meta‑analysis.

Twelve studies involving multiple populations were included: 1,219 participants with TRD, 75 patients undergoing elective surgery who received ketamine anaesthesia, 49 patients treated for pain and 68 healthy volunteers. Methodologically, the set comprised eight randomised controlled trials, two open‑label studies, one non‑randomised clinical trial and one pilot study. Antidepressant efficacy was often measured with MADRS or HDRS; six studies used 0.5 mg/kg intravenous ketamine and four used 0.2 mg/kg or 0.5 mg/kg. Across the included reports, 18 genes were examined, with BDNF and CYP2B6 among the most frequently studied. Findings were mixed but several recurring signals were reported. BDNF (Val66Met, rs6265): Six studies assessed this polymorphism in TRD cohorts. Ping Su et al. (double‑blind RCT, n = 71) reported genotype frequencies of Val/Val 17%, Val/Met 56.3% and Met/Met 26.8% and found no overall association between rs6265 and antidepressant effect (p = 0.55). Reanalyses of that dataset by other investigators suggested antisuicidal and antidepressant benefits were greater in Val‑allele carriers: Chen et al. reported an antisuicidal effect in individuals with any Val allele (p = 0.012) but not in Met/Met (p = 0.215), and another reanalysis found Met/Met carriers were less likely to respond to low‑dose ketamine (p = 0.007). Laje et al. also reported rapid antidepressant effects in a separate TRD sample, but full genotype–outcome details were not clearly extractable from the text. CYP450 system, chiefly CYP2B6: Three studies examined CYP family variants. In 49 chronic pain patients receiving 24‑hour continuous subcutaneous ketamine, Li et al. reported allelic frequencies CYP2B6*1 71%, *5 3%, *6 24% and *7 1%; carriers of CYP2B6*6 had reduced steady‑state ketamine clearance and a higher incidence of treatment‑emergent adverse events (p < 0.001). By contrast, a TRD sample of 67 patients found no significant effect of multiple CYP genotypes on response or side effects (p = 0.88). Aroke et al. examined CYP2B6*6 in 75 elective surgery patients and found no association with ketamine‑induced emergence phenomena (EP) (p = 1.00); in that cohort 40% experienced EP measured with the CADSS. OPRM1 (mu‑opioid receptor, A118G; rs1799971): Two large clinical trials found no influence of this SNP on antidepressant or antisuicidal effects. Saad et al. analysed a phase‑3 esketamine nasal spray trial in 406 TRD patients and found no genotype effect on MADRS reduction (p = 0.52). Grunebaum et al. studied 80 TRD patients receiving 0.5 mg/kg IV ketamine and reported no treatment‑by‑genotype interaction on suicidal ideation or depressive scores at 24 hours (SSI p = 0.554; POMS p = 0.554; HDRS‑17 p = 0.439; HDRS‑24 p = 0.582). GRIN (NMDA receptor subunits): One study evaluated GRIN2B variants (rs1019385, rs1806191) in 75 surgical patients and found no association with emergence phenomena after ketamine (p = 1.00). Separately, some GWAS signals implicated GRIN family genes in antidepressant response (see below). Norepinephrine transporter (NET, rs28386840): In a double‑blind trial of 68 healthy participants, Liebe et al. reported that NET rs28386840 genotype, sex and baseline blood pressure predicted greater acute hypertensive responses to 0.5 mg/kg IV ketamine; T‑allele carriers reached maximal systolic blood pressure earlier than A homozygotes and the genotype predicted increased cardiovascular sequelae (p = 0.030). Genome‑wide association studies (GWAS): Two GWAS reports were described. One analysis (small sample sizes reported as 65 and 326 in the extract, but the extraction is unclear) identified multiple SNPs mapped to genes including NTRK2, GRIN2A/B, GRIN3A, BDNF and CYP2B6 that associated with antidepressant response at various early and later timepoints at nominal significance (examples: p ≈ 2.5×10−3, p ≈ 8.3×10−3). A separate GWAS by Gua et al. on 326 TRD patients applied genome‑wide markers to investigate acute antidepressant response and dissociative side effects; that study reported no association between the candidate genes they examined (SEC11A, RASGRF2, ROBO2, SPRED2 and others) and ketamine antidepressant effect or treatment‑emergent adverse events. Overall, the included studies reported mixed and sometimes contradictory results. Recurrent, but preliminary, associations were that BDNF Val allele carriers tended to show greater antidepressant and antisuicidal responses, CYP2B6*6 was linked to reduced ketamine clearance and more adverse events in one study, and NET variation predicted acute cardiovascular responses. Many null findings were also reported across CYP isoforms, OPRM1 and GRIN variants in some cohorts.

Meshkat and colleagues interpret the body of evidence as suggestive but inconclusive: BDNF Val allele (Val66Met, rs6265) emerged as the most consistent signal linking genotype to clinical outcome, with Val carriers showing greater antidepressant and antisuicidal effects in several reanalyses, while Met/Met carriers appeared less likely to respond to standard low‑dose ketamine. The authors relate this to biological plausibility: BDNF is essential for synaptic plasticity and ketamine‑induced increases in BDNF and synaptogenesis have been implicated in its mechanism of action; Met allele effects on BDNF trafficking and TrkB binding may blunt these processes and thus attenuate clinical response. Regarding pharmacokinetics, the review highlights CYP2B6*6 as a candidate predictor of reduced ketamine clearance and higher rates of adverse events in at least one study, noting that CYP2B6 is a key hepatic enzyme for ketamine metabolism and that the *6 allele is a common loss‑of‑function variant. The authors also note inconsistencies: other CYP2 and CYP3A genotypes were not consistently associated with response or side effects across studies, and individual disease states (for example obesity, inflammation, diabetes) may modulate CYP expression and complicate genotype–phenotype relationships. The review finds no robust evidence that the OPRM1 A118G variant modifies antidepressant or antisuicidal effects in the studies assessed, although debate remains in the literature about whether opioid system activation contributes to ketamine’s effects. GRIN gene family variants were implicated in some genome‑wide analyses with links to suicidality and treatment resistance, but single‑study results for EP were null. NET polymorphism (rs28386840) was associated with acute hypertensive responses in a healthy volunteer study, leaving open whether cardiovascular effects stem from peripheral noradrenergic mechanisms or central neurotransmitter modulation. Key limitations acknowledged by the authors include small sample sizes and low frequencies of some genotypes, heterogeneity in study design, populations, dosing and outcome measures, and generally short follow‑up durations (many ketamine/esketamine trials were 28 days or shorter). Because of these factors, quantitative pooling was not feasible and the authors caution that current evidence is preliminary. They recommend larger, prospectively designed studies that account for clinical covariates (for example inflammatory markers, metabolic comorbidities), examine dose–genotype interactions, and formally evaluate the clinical utility of genotyping for improving patient outcomes before pharmacogenomic testing for ketamine can be recommended in practice.

The authors conclude that pharmacogenomics offers a potentially useful avenue to individualise ketamine therapy, but current evidence is preliminary. As testing becomes cheaper and more accessible, pharmacogenomic information may increasingly inform clinical decision‑making for ketamine; however, confirmatory prospective studies are required to validate reported genotype–treatment associations, discover new relevant variants, and establish whether genetic testing improves clinical outcomes and justifies routine clinical implementation.