Increased cortical thickness and decreased brain age among special operations veterans with blast TBI after a magnesium-ibogaine protocol

Traumatic brain injury (TBI), particularly blast-related TBI in military veterans, is a major and persistent health problem with few effective treatments. The paper describes blast TBI as a diffuse injury that can produce long-term disability, persistent post-concussion symptoms, psychiatric comorbidity, and possible accelerated brain ageing. The authors note that chronic TBI is often associated with white matter disruption, cortical thinning, and reduced brain volume, but it remains uncertain whether any rapid-acting intervention can reverse or modify these structural changes. Within that context, the paper considers ibogaine, a psychoactive compound of interest for psychiatric sequelae of TBI, including depression, anxiety, and addiction. The researchers state that a prior observational pilot study of a magnesium-ibogaine protocol in special operations veterans had already shown clinical improvements, and they hypothesised that these benefits might be accompanied by measurable changes in cortical thickness, subcortical volume, and predicted brain age on MRI. The study therefore aimed to examine whether structural brain changes could be detected after treatment and whether those changes tracked with the previously reported clinical response. The authors frame this as an early morphometric analysis of MRI data from a clinical treatment cohort, using longitudinal imaging and brain-age modelling to explore possible neurobiological correlates of the protocol.

This was an observational, longitudinal MRI analysis nested within a magnesium-ibogaine treatment protocol for special operations veterans with TBI and multiple comorbidities. Of the 30 treated participants, usable imaging data were available for 25 at baseline and the immediate post-treatment visit, and 22 had usable 1-month follow-up scans. All participants were male; the mean age was 44.5 years, and most had mild TBI, although one had moderate TBI and one had moderately severe TBI. The paper also reports a high lifetime TBI burden and combat exposure, but the extracted text does not clearly restate all inclusion and exclusion criteria. MRI data were acquired on a T1-weighted protocol at approximately the same time of day across visits. The authors processed the scans using the ANTs longitudinal pipeline to derive cortical thickness and log-jacobian determinant maps, and they also ran FreeSurfer longitudinal processing in parallel as a comparison. After pipeline evaluation, they selected ANTs for the formal analyses because it performed better on their dataset. Cortical thickness was quantified in 62 regions of interest, and volumetric change was assessed in 28 subcortical and cerebellar regions from the Mindboggle atlas. The researchers used linear mixed-effects models with study visit as the main fixed effect and covariates including chronological age, combat exposure score, number of TBIs, and estimated total intracranial volume, with a random intercept for participant. They first tested whole-brain average cortical thickness, then regional cortical thickness and volumetric change. Predicted brain age was estimated with the brainageR algorithm from structural MRI. Normative modelling was also used to compare cortical thickness percentile ranks with a normative reference framework. The authors additionally examined associations between imaging changes and clinical or neurocognitive outcomes, including WHODAS-2.0 and other measures, with correction for multiple testing.

The study found a statistically significant main effect of study visit on whole-brain average cortical thickness. Mean cortical thickness increased modestly from 2.08 mm at baseline to 2.13 mm immediately post-treatment and remained at 2.13 mm at 1 month. After multiple-comparison correction, pairwise contrasts were only trend-level for the global measure, but regionally the authors report significant visit effects in 13 of 62 cortical regions. Post-hoc tests showed significant increases from baseline to immediate post-treatment in 11 regions, and additional significant baseline-to-1-month increases in the right entorhinal cortex and right postcentral cortex. There were no significant changes between immediate post-treatment and 1 month, suggesting the increases were sustained rather than continuing to rise further. Normative modelling also showed significant visit-related change. Targeted normative percentile ranks increased by about 7.49 percentile points from baseline to immediate post-treatment and by 8.70 points from baseline to 1 month. For the whole-brain normative analysis, the increases were only trend-level. For subcortical and cerebellar volume, the authors report significant visit effects in 8 regions. Post-hoc analyses identified bilateral volumetric expansion in cerebellar white matter, basal forebrain, and ventral diencephalon, plus the right hippocampus, from baseline to immediate post-treatment. Some of these changes persisted at 1 month, particularly in the ventral diencephalon and left-sided cerebellar white matter and basal forebrain. The left caudate showed a significant contraction over time. No significant baseline-to-1-month increase was seen in the right hippocampus, indicating that the hippocampal expansion was transient. Predicted brain age decreased over time. Mean predicted brain age fell from 39.37 years at baseline to 38.54 years at 1 month, while mean chronological age was about 44.5 years. The mixed-effects model showed a significant visit effect on predicted brain age, and the only significant post-hoc comparison was between baseline and 1 month, corresponding to an estimated mean reduction of 1.3 years. Chronological age and grey matter signal-to-noise ratio were also significant covariates in the model. The exploratory analyses found no significant correlations between imaging changes and WHODAS-2.0 total score after correction. Likewise, most associations between morphometric change and psychiatric or neurocognitive measures did not survive false discovery rate correction. One association between baseline-to-1-month predicted brain age change and 1-month Hamilton Anxiety Rating Scale score normalised to baseline was mentioned, but a subsequent regression did not support a predictive relationship.

The authors interpret the findings as the first group-level evidence that a magnesium-ibogaine protocol may be associated with measurable structural brain changes in humans over one month. They argue that the combination of increased cortical thickness, subcortical volumetric expansion, and reduced predicted brain age may reflect structural neural plasticity and could be linked, at least indirectly, to the clinical improvements reported in their prior work. They also note that many of the cortical changes were right-lateralised, which they consider less likely to be purely artifactual than a more symmetric pattern would be. In relation to earlier research, the authors compare their findings with studies of blast TBI showing cortical thinning, as well as with other rapid-acting interventions such as electroconvulsive therapy and some antidepressant or behavioural interventions that have been associated with regional thickening or volume changes. They suggest that the thickening seen in regions previously reported to thin in military blast TBI may represent a possible reversal-like pattern. They also place the reduction in predicted brain age in the context of prior work showing changes in brain-age estimates after surgery, ibuprofen, and the postpartum period. The authors are cautious about interpretation. They emphasise that machine-learning brain-age measures should not be read literally as the brain becoming younger, but rather as images resembling younger normative patterns. They also note that longitudinal T1 changes may reflect non-structural factors such as tissue water content, so the biological meaning of the observed morphometric changes remains uncertain. For the caudate contraction, they discuss alternative interpretations, including possible changes in synaptic efficiency or network organisation, and argue that it should not automatically be viewed as harmful. Key limitations include the absence of a control group, the small sample size, the lack of biological samples to examine neurotrophic markers, and the inability to verify blast TBI diagnosis with MRI alone. They also note that normative modelling was limited by the absence of healthy controls, and that scanner or processing choices might affect results. The sample is described as highly selected and not broadly generalisable, consisting of physically trained male special operations veterans. The authors conclude that larger randomised controlled studies are needed to determine whether the effects are reproducible and whether they are driven by pharmacology, expectation, the therapeutic setting, or a combination of these factors.