Psychiatry · Medical Literature
GH001 Efficacy is Independent of Prior Antidepressant Treatment Failures in Treatment-Resistant Depression: A Post Hoc Analysis of a Phase 2b Randomized Controlled Trial.
Abstract
Approximately 30% of patients treated for major depressive disorder develop treatment-resistant depression (TRD). The STAR*D study demonstrated remission rates decline progressively with each antidepressant failure (37%, 31%, 14%, and 13%, respectively). A single-day individualized dosing regimen of GH001 (synthetic mebufotenin for inhalation) produced rapid, large improvements in depressive symptoms versus placebo in patients with TRD in a Phase 2b trial (least-squares mean difference, -15.5; effect size, -2.0; 57.5% remission at Day 8 versus 0% placebo). The current post hoc analysis examines whether GH001 efficacy varies by number of prior lifetime antidepressant treatment failures. This analysis included all 40 patients who received GH001 in the double-blind part of a Phase 2b trial. Spearman rank correlations between number of prior lifetime antidepressant failures and change from baseline in Montgomery-Åsberg Depression Rating Scale (MADRS) scores were calculated at Day 8 and among 6-month open-label extension completers. Remission rates (MADRS ⩽10) were examined by subgroup (2, 3, 4, or ⩾5 prior lifetime failures). No meaningful correlation was observed between prior lifetime treatment failures and MADRS improvement at Day 8 (r = -0.13; P = 0.44) or 6-month OLE completers (r = -0.10; P = 0.60). Remission rates at Day 8 likewise were similar across subgroups (range, 53.9%-63.6%) and were maintained at end of treatment visit/Month 6 (range, 61.5%-85.7%). Secondary endpoints were not associated with treatment history. The efficacy of GH001 in patients with TRD appears largely independent of number of prior lifetime antidepressant treatments, and further research in patients with extensive treatment histories will be conducted in subsequent development stages.
Clinical relevance
Does the efficacy of GH001 (inhaled synthetic mebufotenin/5-MeO-DMT) in treatment-resistant depression vary according to the number of prior lifetime antidepressant treatment failures? __SNAPSHOT__ {"studyDesign":"Post hoc analysis of the GH001 (active) arm of a Phase 2b randomized, double-blind, placebo-controlled trial, with additional data from a 6-month open-label extension (OLE). Note: this paper analyzes only the 40 patients in the GH001 arm; it is not an independent RCT.","setting":"Multi-center European Phase 2b trial (GH001-TRD-201); specific countries and institution types not reported in the abstract.","sampleSize":"n=40 (all patients who received GH001 in the double-blind phase); OLE completers analyzed separately (exact n not reported in abstract).","population":"Adults with treatment-resistant depression (TRD), defined by ≥2 prior lifetime antidepressant treatment failures. Subgroups stratified by 2, 3, 4, or ≥5 prior lifetime failures.","intervention":"Single-day individualized dosing regimen of GH001 (synthetic mebufotenin for inhalation; a 5-MeO-DMT tryptamine compound) administered via a proprietary inhalation device.","comparator":"No comparator arm in this post hoc analysis. The referenced parent RCT included a placebo arm (n=41), but placebo data were not analyzed in this post hoc subgroup analysis.","primaryOutcome":"Spearman rank correlation between number of prior lifetime antidepressant failures and change from baseline in MADRS score at Day 8; remission rates (MADRS ≤10) by failure-count subgroup at Day 8 and Month 6 (OLE completers).","followUp":"Day 8 (primary timepoint); 6-month open-label extension for OLE completers.","fundingSource":"Industry (GH Research PLC, a clinical-stage biopharmaceutical company; sponsor of GH001 development program)."}
Key findings
No meaningful correlation was observed between the number of prior lifetime antidepressant failures and MADRS improvement at Day 8 (Spearman r = −0.13; P = 0.44) or among 6-month OLE completers (r = −0.10; P = 0.60), indicating no statistically significant linear association. Across subgroups stratified by treatment-failure burden, Day 8 remission rates (MADRS ≤10) ranged from 53.9% to 63.6% for patients with 2 to ≥5 prior lifetime antidepressant failures, with no gradient of decline at higher failure counts. Among 6-month OLE completers, remission rates ranged from 61.5% to 85.7% across the same subgroups. These findings contextualize the parent RCT's primary result: a placebo-adjusted least-squares mean MADRS difference of −15.5 points at Day 8 (effect size −2.0; P < 0.0001) and a 57.5% Day 8 remission rate for GH001 vs. 0% for placebo (P < 0.0001). NNT for remission at Day 8 in the parent RCT: approximately 1.74 (calculated), based on a 57.5% absolute risk difference. Between-subgroup comparisons for this post hoc analysis were not reported as formal statistical tests with p-values; the absence of meaningful correlation does not constitute proof of equivalence across subgroups. No new safety data specific to this post hoc analysis were reported in the abstract.
Practice implications
This post hoc analysis raises an important hypothesis: that GH001's antidepressant effect may not attenuate with increasing prior treatment failure burden, in contrast to the well-established declining remission pattern seen with successive conventional antidepressants in the STAR*D trial (remission rates of 37%, 31%, 14%, and 13% across steps). If confirmed, this would represent a mechanistically and clinically meaningful departure from standard pharmacotherapy. However, several critical constraints limit interpretation: (1) This is a post hoc, single-arm, exploratory subgroup analysis — it was not pre-specified or powered to detect between-subgroup differences in efficacy. (2) The total sample of 40 patients yields very small subgroup sizes across the four failure-count strata, making all subgroup estimates statistically unstable. (3) The primary correlational approach (Spearman r) tests for a monotonic linear trend but cannot exclude non-linear patterns or adequately characterize heterogeneity across strata. (4) Absence of a statistically significant correlation does not establish treatment-effect equivalence across subgroups. (5) The 6-month OLE is open-label and uncontrolled, and OLE completer analyses are subject to substantial survivorship bias. (6) Only the abstract was available for review, limiting full assessment of methods and confounding. (7) GH001 is investigational and not approved by any regulatory agency. The population studied (European multi-center adults with TRD per study-specific criteria) may not generalize to all TRD populations, including those with highly refractory courses, psychiatric comorbidities, or non-European backgrounds. This analysis does not establish causality or support changing clinical practice at this time. __LIMITATIONS__ Post hoc design: analysis was not pre-specified in the parent trial protocol; results are hypothesis-generating only.,Critically small sample size (n=40 total; further subdivided into 4 subgroups), providing insufficient statistical power to detect meaningful between-subgroup differences in efficacy.,Spearman correlation tests for monotonic linear association only; cannot exclude non-linear or threshold effects across failure-count strata.,Absence of a statistically significant correlation (p=0.44 and p=0.60) does not establish equivalence across subgroups and may reflect type II error given the small n.,Single-arm analysis: placebo arm was excluded, precluding any placebo-adjusted comparison by subgroup.,6-month OLE is uncontrolled and open-label; completer analyses are subject to substantial survivorship bias (early discontinuers excluded).,Exact subgroup n values not reported in abstract, preventing independent verification of remission rate estimates.,Industry funding (GH Research PLC) introduces potential risk of reporting and analysis bias.,Blinding was maintained only in the parent RCT double-blind phase; the OLE was unblinded.,Geographic restriction to European sites may limit generalizability to North American, Asian, and other global TRD populations.,Only the abstract was available; full assessment of statistical methods, confounders, and sensitivity analyses was not possible. __BOTTOM_LINE_CATEGORY__ promising_but_preliminary __BOTTOM_LINE__ This post hoc analysis of 40 patients treated with inhaled GH001 (synthetic mebufotenin/5-MeO-DMT) found no meaningful correlation between number of prior antidepressant failures and MADRS improvement, with Day 8 remission rates of 53.9–63.6% across all failure-burden subgroups — a pattern that contrasts strikingly with the stepwise efficacy decline seen in STAR*D. While clinically intriguing, these findings are hypothesis-generating only, constrained by a post hoc single-arm design, critically small subgroup sizes, and industry sponsorship, and must be prospectively validated in adequately powered Phase 3 trials before informing clinical practice.