Anti-anhedonic effect of ketamine and its neural correlates in treatment-resistant bipolar depression

Lally and colleagues situate the study in the context of a large unmet need: anhedonia—reduced pleasure or interest in previously rewarding activities—is highly prevalent in mood disorders, predicts poorer prognosis and is not specifically treated by current approved pharmacotherapies. Prior work suggests that anhedonia in depression is predominantly a deficit in motivational or anticipatory reward processes rather than consummatory pleasure, and that dopaminergic and glutamatergic systems are implicated in these aspects of reward. Conventional antidepressants have limited or inconsistent effects on anhedonia, and there is interest in rapid-acting agents such as the NMDA receptor antagonist ketamine, but it was unknown whether ketamine specifically reduces anhedonic symptoms in depressed patients. This randomised, double-blind, placebo-controlled, crossover study therefore tested whether a single subanesthetic ketamine infusion produces rapid anti-anhedonic effects in treatment-resistant bipolar depression and sought to identify neural correlates of any such effects using [18F]FDG positron emission tomography (FDG-PET). The investigators emphasised measuring anticipatory anhedonia with the Snaith–Hamilton Pleasure Scale (SHAPS) and relating clinical change to regional cerebral glucose metabolism (rCMRGlu), an indirect marker of glutamatergic neurotransmission.

Participants were 36 inpatients aged 18–65 with bipolar I or II disorder currently experiencing a major depressive episode and meeting criteria for treatment resistance; all had Montgomery–Åsberg Depression Rating Scale (MADRS) scores ≥20 at screening. Inclusion required failure to respond to at least one adequate antidepressant trial and a prospective open-label mood stabiliser trial (lithium or valproic acid) prior to study entry. Exclusion criteria included pregnancy or nursing, serious suicidal ideation, substance abuse/dependence within the past 3 months, and prior ketamine exposure. Participants continued their mood stabiliser throughout the trial. The clinical design was a randomised, double-blind, placebo-controlled, crossover trial. Each subject received two intravenous infusions separated by 2 weeks: a single subanesthetic dose of ketamine hydrochloride (0.5 mg kg-1) and placebo (0.9% saline), each administered over 40 minutes. Clinical ratings were obtained 60 minutes before each infusion and repeatedly thereafter (40, 80, 120, 230 minutes and days 1, 2, 3, 7, 10 and 14). Anticipatory anhedonia was measured with the SHAPS (14–56 scoring used here, higher = greater anhedonia); overall depressive symptoms were assessed with the MADRS. A priori the 230-minute time point was selected for primary imaging-related analyses because prior work had shown it sensitive to ketamine’s antidepressant effects and it precedes PET acquisition. FDG-PET imaging was performed in a subset of 21 patients. Resting-state FDG-PET scans began about 2 hours after infusion and comprised a brain emission scan of ~1.5 hours. Images were acquired following an intravenous bolus of 4.5 mCi [18F]FDG and rCMRGlu values were computed using a cardiac input function. Structural MRI was used for anatomical localisation. PET analyses combined region-of-interest (ROI) approaches (ventral striatum—nucleus accumbens and olfactory tubercle—and orbitofrontal cortex) with whole-brain voxelwise analyses in SPM5; images were normalised by global mean and smoothed, and difference images (post-ketamine minus post-placebo) were created. ROIs were adjusted per subject and normalized by total grey matter. Clinical statistical analyses used linear mixed models (heterogeneous first-order autoregressive covariance, restricted maximum likelihood) with fixed effects of time and drug and a random subject effect; baseline SHAPS at each infusion day was included as a covariate. To test whether anti-anhedonic effects were independent of overall mood improvement, models were repeated with MADRS total score (minus item 8, ‘‘inability to feel’’) as an additional covariate. Post-hoc Bonferroni-corrected comparisons examined time points. PET–behaviour relationships used Pearson correlations between percent SHAPS improvement at 230 minutes and change in ROI rCMRGlu, plus multiple regression to separate variance attributable to SHAPS versus MADRS change. Whole-brain regressions examined percent SHAPS improvement on difference images, and analyses were repeated with SHAPS change orthogonalised to MADRS change (removal of shared variance) to test specificity. Cluster-level correction for multiple comparisons was applied (Gaussian random field theory), with some thresholds tightened in the presence of very large clusters.

Sample and procedural notes: the clinical sample comprised 36 participants (21 female). Of the 36, 21 underwent FDG-PET; one PET subject was excluded for failure to obtain the cardiac input function and another lacked SHAPS data at the 230-minute time point for both sessions, yielding smaller PET analysis samples reported in the text (correlation degrees of freedom indicate n ≈ 20). Clinical (behavioural) outcomes: linear mixed models revealed a significant main effect of drug (ketamine vs placebo) on SHAPS scores (F(1,110)=24.71, P<0.001) and a main effect of time (F(9,194)=2.69, P=0.006), with a trend-level drug-by-time interaction (F(9,218)=1.90, P=0.053). Bonferroni-corrected post-hoc tests showed that ketamine produced significantly greater reductions in anhedonia than placebo at multiple post-infusion time points across the 14-day follow-up. SHAPS and MADRS total scores correlated at baseline (r=0.53, P<0.001). Crucially, when MADRS (minus item 8) was entered as a covariate, the main effect of drug on SHAPS remained significant (F(1,123)=7.713, P=0.006), indicating an anti-anhedonic effect above and beyond general mood improvement. Post-hoc tests in this covaried model found significant ketamine-related reductions in SHAPS at days 1, 3, 7 and 14. Adding mood stabiliser as a covariate produced a trend suggesting greater anti-anhedonic response with lithium than valproate (F(1,57)=3.57, P=0.06), though there was no significant drug-by-stabiliser interaction. A positive family history of alcohol use disorder was associated with an enhanced anti-anhedonic response to ketamine (drug-by-family history interaction F(1,109)=4.12, P=0.045); the ketamine effect was strong in those with such a family history (F(1,116)=11.13, P=0.001) but not in those without (F(1,111)=1.22, P=0.27). Personal history of substance misuse did not significantly influence response. PET ROI results: percent improvement in SHAPS at 230 minutes correlated with ketamine-induced change in ventral striatum rCMRGlu (r(19) = -0.52, P = 0.02), interpreted as larger increases in ventral striatum metabolism being associated with greater anti-anhedonic response. Orbitofrontal cortex rCMRGlu did not show a significant relationship (r(19) = -0.37, P = 0.12). However, in a multiple regression including both SHAPS and MADRS change, change in MADRS (t = -2.18, P = 0.045) but not SHAPS (t = -0.05, P = 0.96) predicted change in ventral striatum metabolism, implying that ventral striatal metabolism related more to overall depressive improvement than to anhedonia specifically. Absolute SHAPS scores post-placebo and post-ketamine were not correlated with ROI rCMRGlu. PET whole-brain results: whole-brain analyses showed that greater percent reduction in SHAPS scores was significantly associated with increased glucose metabolism in the dorsal anterior cingulate cortex (dACC; peak [x = -6, y = 40, z = 43], t(17)=4.39, P_corr=0.016). A cerebellar increase also survived correction, and there was a trend-level increase in striatal metabolism (cluster extending from caudate to putamen into ventral striatum; peak [x = 14, y = 4, z = 10], t(17)=4.28, P_corr=0.051). When SHAPS percent change was orthogonalised to MADRS change (removing shared variance with overall depression), the relationship specific to anhedonia remained significant in the dACC ([x = -8, y = 40, z = 28]; t(17)=4.89, P_corr=0.027) extending into pregenual cingulate/callosal regions and right dorsolateral prefrontal cortex; trends were noted in the fusiform gyrus extending into the claustrum and putamen but not the ventral striatum. There was no significant relationship between change in dACC metabolism and SHAPS change at day 14 (r(18)=0.11, P=0.66).

Lally and colleagues interpret their findings as evidence that a single subanesthetic ketamine infusion produces rapid reductions in anticipatory anhedonia in treatment-resistant bipolar depression, with effects observable within 40 minutes and persisting up to 14 days for some patients. They emphasise that the anti-anhedonic effect remained significant after statistically controlling for changes in overall depressive symptoms, suggesting a degree of specificity beyond ketamine’s general antidepressant action. Neurobiologically, the investigators found that increases in glucose metabolism in the dorsal anterior cingulate cortex (dACC) and, tentatively, the putamen were associated with the magnitude of anhedonia improvement. They note that these regions have been implicated in reward anticipation, motivation and decision-making in prior human and animal studies, and suggest that increased dACC and putamen metabolism may reflect improved capacity to anticipate or be motivated by pleasurable experiences. By contrast, ventral striatal metabolism appeared more closely tied to overall MADRS-measured depression change than to SHAPS-assessed anhedonia after accounting for shared variance, a finding the authors discuss in terms of clinical heterogeneity and the distinct constructs captured by the two scales. Mechanistically, the authors propose that ketamine’s effects may arise from modulation of glutamatergic circuits and downstream dopaminergic signalling; ketamine blocks NMDA receptors and can alter glutamate cycling, and it has been reported to increase striatal dopamine. They also report a suggestive interaction with concurrent mood stabiliser use: lithium was associated with a trend towards greater anti-anhedonic response than valproate, citing preclinical work that lithium may potentiate ketamine’s plasticity-related effects. The paper acknowledges several limitations that temper interpretation. First, the crossover design lacked a baseline PET scan, so carryover effects cannot be excluded. Second, the subjective psychoactive effects of ketamine likely compromised blinding. Third, all participants remained on lithium or valproate, which may have masked or enhanced ketamine’s effects; the generalisability to unmedicated patients is therefore uncertain. Fourth, applicability to major depressive disorder remains unknown. The authors recommend future studies include baseline and multiple imaging time points, examine medication-free samples, extend investigations to MDD, and probe reward processing more directly with cognitive testing. Finally, they situate the findings in a public-health frame: given the absence of approved therapies for anhedonia, the rapid anti-anhedonic effects observed here and their neural correlates support further investigation of NMDA receptor antagonists for this symptom domain.