Hippocampal volume loss in individuals with a history of non‐fatal opioid overdose

Incidence of opioid‐related overdoses in the United States has increased dramatically over the past two decades. Despite public emphasis on overdose fatalities, most overdose cases are not fatal. Although there are case reports of amnestic syndromes and acute injury to the hippocampus following non‐fatal opioid overdose, the effects of such overdoses on brain structure are poorly understood. Here, we investigated the neuroanatomical correlates of non‐fatal opioid overdoses by comparing hippocampal volume in opioid use disorder (OUD) patients who had experienced an opioid overdose (OD; N = 17) with those who had not (NOD; N = 32). Voxel‐based morphometry showed lower hippocampal volume in the OD group than in the NOD group, which on post hoc analysis was evident in the left but not the right hippocampus. These findings strengthen the evidence that hippocampal injury is associated with non‐fatal opioid overdose, which is hypothesized to underlie overdose‐related amnestic syndrome.


| INTRODUCTION
Opioid use disorder (OUD) is a chronic, relapsing disease with high mortality and morbidity.OUD disproportionally affects younger adults, with opioid-related fatalities accounting for up to 20% of deaths among individuals 24-35 years of age.In the United States, the number of opioid overdoses (ODs) continues to rise and was estimated at nearly 80 000 in 2021. 1 Overall, 73% of OD fatalities were among men, 2 with a disproportionately high risk among American Indian and Alaska Native men. 3 Naloxone, an opioid receptor antagonist, is effective in reversing OD and has prevented thousands of deaths. 4[7] Given the high incidence of non-fatal ODs, their neurobiological mechanisms and sequelae need to be better understood.
Opioids cause respiratory suppression through agonism of brainstem μ-opioid receptors and, at sufficiently high doses, can cause respiratory arrest, leading to loss of consciousness and death. 8,9In preclinical studies, opioids reduced respiratory rate and diminished oxygen entry into the brain. 8,10,116][17][18] Neuroimaging, neuropsychological and postmortem brain studies identified several neurocognitive sequelae of acute opioid ODs that align with hippocampal injury. 7For example, leukoencephalopathy, a demyelinating syndrome associated with cognitive and memory impairments was observed following episodes of ODinduced hypoxia. 19,20Additionally, over 20 cases of an amnestic syndrome attributable to opioid-induced acute hippocampal injury have been reported in the United States [21][22][23][24][25] and France. 26Thus, respiratory depression following non-fatal opioid OD may elicit hypoxic injury to susceptible brain regions such as the hippocampus and lead to chronic neurocognitive deficits in survivors. 7,9However, few studies to date have systematically compared the neuroanatomical correlates of non-fatal opioid OD in a well-characterized sample of patients with OUD with those without a reported history of OD.
Here, we examined the effects of non-fatal opioid ODs on hippocampal volume in individuals with OUD using structural magnetic resonance imaging (MRI) with voxel-based morphometry.We hypothesized that individuals with a previous OD will display lower hippocampal volume than those with no prior OD (NOD).

| Participants
Study participants comprised two cohorts of individuals with OUD that underwent a structural MRI as part of participation in other imaging research.The first cohort (N = 35) participated in a neuroimaging study at the University of Pennsylvania (UPenn) with the following inclusion criteria [27][28][29][30] : 18 to 60 years old; DSM-IV-TR diagnosis of opioid dependence; active opioid use confirmed by urine toxicology screen and self-reported daily heroin or prescription opioid use for more than 2 weeks in the past 3 months; good physical health; free of chronic psychiatric disorders and not receiving treatments that may affect the cerebrovascular system; free of contraindications for MRI; negative urine toxicology result for opioids after detoxification; free of contraindications for treatment with extended-release naltrexone and for female participants, not pregnant or breastfeeding.The second cohort (N = 14) participated in a study on dopaminergic signalling in OUD at the National Institutes of Health (NIH) (clinical trial NCT03190954), which had the following inclusion criteria: 18 to 80 years old; ability to provide informed consent; a DSM-5 diagnosis of a lifetime moderate or severe OUD; no current major medical problems that could impact brain function (e.g., seizures, stroke and traumatic brain injury), no head trauma with loss of consciousness; not pregnant or currently breastfeeding and no contraindications to MRI.
All participants gave written informed consent to participate in the study protocols approved by the UPenn or NIH Institutional Review Boards.See the Supplementary Material (Data S1) for the estimation of statistical power.

| Assessments
Both UPenn and NIH participants were assessed using the Addiction Severity Index (ASI), 5th edition.Drug use severity was indexed by the ASI drug composite score. 31We used the following ASI question to differentiate OUD patients with a history of opioid overdose (OD) from patients without any prior opioid OD (NOD): Question D18: 'How many times have you overdosed on drugs?' ODs were defined as opioid ODs that required intervention by someone to recover, not simply sleeping it off, and included intentional suicide attempts.The OD group was defined as individuals who reported one or more ODs, whereas individuals in the NOD group reported none.All participants completed the Wechsler Abbreviated Scale of Intelligence (WASI-II) subtests Matrix Reasoning and Vocabulary as a proxy for general intelligence. 32For each individual, the mean bilateral hippocampal volume was calculated as the average of all voxels within the bilateral hippocampal ROI.

| MRI data analysis
A general linear model (GLM) implemented in R (www.R-project.org)evaluated the difference between OD and NOD individuals in mean hippocampal volume while controlling for TIV, age, sex and cohort/ scanner (UPenn vs. NIH).We also performed exploratory post hoc analysis on the left and right hippocampal ROIs separately.The second analysis used a voxel-wise approach to identify the subregion(s) within the hippocampus that showed volume differences between OD and NOD individuals.We conducted a whole-brain GLM analysis using the bilateral hippocampal ROI as the search volume (#voxels = 9400) while controlling for TIV, age, sex and cohort/scanner.Significant regions were detected using the threshold-free cluster-enhancement algorithm with 5000 iterations at cluster-level p < 0.05 corrected for familywise error. 33| RESULTS

| Demographics and clinical characteristics
The demographic characteristics of the participants are summarized in Table 1.No significant differences were observed between the OD    We investigated the structural brain correlates of prior non-fatal opioid OD using volumetric analysis of MRI data from individuals with OUD.The group with a history of opioid OD had lower bilateral hippocampal volume than the group without such a history.Previous studies comparing hippocampal volume in OUD and healthy individuals showed no significant differences, although they did not specify whether individuals had a prior OD. 13,27,34Age-related hippocampal volume loss (À1.55% ± 1.38% per year) has been described in healthy aging 35 ; however, we found no difference in the effect of age on hippocampal volume between the OD and NOD groups (Data S1).

| Hippocampal volume
These results suggest that the observed differences in hippocampal volume could be attributed to a prior OD.These findings are consistent with previous case reports/series that illustrated an association between fentanyl OD and acute hippocampal injury with diffusionweighted imaging. 22,24,36though our study was not designed to unravel the mechanisms of hippocampal volume loss in OD, its results could support several non-mutually exclusive hypotheses.7][38] Anatomically, the hippocampus The left hippocampus showed lower volume in the overdose (OD) group than the no prior OD (NOD) group (Montreal Neurological Institute space coordinates: x/y/z = À24/À15/À16).
is located deep within the medial temporal lobe and receives its vascular supply from collateral branches of the posterior cerebral and anterior choroidal arteries.Its location in a vascular border zone makes it especially vulnerable to hypoperfusion and hypoxic-ischemic injury during episodes of respiratory depression. 18,39Hypoxic stress may be exacerbated by the hippocampus's greater reliance than other neural tissues on oxidative metabolism to meet its energy demands. 18,37,40,41Clinical evidence also supports hippocampal susceptibility to hypoxic injuries.Individuals who experienced an anoxic episode had lower grey matter volume bilaterally in the hippocampus than healthy controls, a difference not seen in other brain regions. 157][48][49] Tau accumulation is shown to be associated with brain tissue atrophy in both preclinical models and humans. 50,51Consequently, elevated hippocampal tau levels following OD-associated hypoxia could additionally mediate brain structural abnormalities.
Future studies are needed to assess the effects of non-fatal opioidrelated OD on hippocampal tau levels in vivo using positron emission tomography.
3][54][55][56][57] Here, we found a higher prevalence of current alcohol use disorder in the OD cohort.It is consistent with recent reports that 15%-30% of opioid OD deaths involved alcohol, 53,54 suggesting that the combination of alcohol and opioids likely increases the risk of opioid OD.We also found a lower prevalence of current cannabis use disorder in the OD cohort.Although there is currently a paucity of data examining this relationship, it will be important to investigate it in future studies given the recent trends in medical cannabis legalization and harm reduction efforts.Despite the different prevalence of alcohol and cannabis use disorders in the OD group, controlling for these differences did not alter our main findings of lower hippocampal volume in the OD compared with the NOD group.
Our study has several limitations.First, it lacks details about the number, timing and clinical severity of past ODs.[24][25]36,58 Second, the lack of data on the participants' neuropsychological performance limited our interpretation of the observed effect of OD on hippocampal volume.WASI IQ scores did not differ statistically between the OD and NOD groups, making it unlikely that the OD individuals had severe global intellectual deficits.Future work is needed to associate VBM variables with domain-specific neuropsychological measures (e.g., episodic memory) to fully understand the behavioural and clinical consequences of nonfatal opioid ODs.Third, our sample is relatively small for a VBM analysis.Thus, we focused our primary analysis on the hippocampus rather than on a conventional whole-brain voxel-wise analysis that would
Structural and voxel-based analyses of MRI data were performed using Statistical Parametric Mapping (SPM) version 12 (www.fil.ion.ucl.ac.uk/spm) and Computational Anatomy Toolbox (CAT) version 12 (www.neuro.uni-jena.de/cat).A priori hippocampal regions of interest (ROIs) were anatomically defined based on the Neuromorphometrics atlas (www.neuromorphometrics.com; Figure 1).Total intracranial volume (TIV) was estimated by CAT12 for each individual.For voxelbased morphometry analyses, the grey matter probability maps were spatially normalized to the Montreal Neurological Institute (MNI) stereotaxic space with 1-mm isotropic voxels using diffeomorphic anatomical registration through an exponentiated lie (DARTEL) algebra.The images were further modulated by the Jacobian determinants and smoothed by a Gaussian filter with full width at half-maximum set to 8 mm.We performed two complementary statistical analyses to evaluate the association with a history of OD.The first analysis aimed to investigate the overall hippocampal volume in OD vs NOD individuals.

T A B L E 1
Demographic and clinical characteristics of the OD and NOD cohorts.
require a correction for multiple comparisons and thus lead to a loss of statistical power.Future VBM studies with larger samples could shed light on OD-induced structural changes in other brain regions such as the prefrontal cortex.Fourth, we did not investigate the interaction of opioid OD with polysubstance use and the neurotoxic effects of other substances or drug contaminants that could have contributed to the observed differences in hippocampal volume.It will be important for future research with larger samples to elucidate how other substance use modulates the neurological effects of non-fatal OD.Finally, as in all similar studies, the observed relationships are correlational, as it would not be safe or ethical to conduct a causal study of opioid OD effects in humans.Worth noting, the current hippocampal findings were detected in OUD cohorts recruited prior to the nearly complete replacement of heroin by fentanyl in the regional and national drug supply.Given the potency and cytotoxicity of fentanyl, it is likely that our report may be conservative: more pronounced neuroanatomical and neurocognitive effects of OD in cohorts who use fentanyl would be expected.In conclusion, we found an association between non-fatal opioid-related ODs and lower hippocampal volumes in individuals with OUD.These findings underscore the need for further studies of the neurobiological mechanisms underlying hippocampal injury in non-fatal opioid-related OD, which could aid in preventing and treating the neuropsychological consequences of OD and thereby enhance relapse prevention efforts in individuals with OUD.AUTHOR CONTRIBUTIONS Study conception and design: Zhenhao Shi, Corinde E. Wiers.Acquisition of data: Nora D. Volkow, Peter Manza, Daniel D. Langleben, Anna Rose Childress, Zhenhao Shi, Corinde E. Wiers.Analysis and interpretation of data: Dustin R. Todaro, Zhenhao Shi, Corinde E. Wiers.Drafting of manuscript: Dustin R. Todaro, Xinyi Li, Zhenhao Shi, Corinde E. Wiers.Critical revisions of the manuscript: Xinyi Li, Laís S. Pereira-Rufino, Peter Manza, Ilya M. Nasrallah, Sandhitsu Das, Anna Rose Childress, Henry R. Kranzler, Nora D. Volkow, Daniel D. Langleben.ACKNOWLEDGMENTS This work was supported by the Commonwealth Fund of Pennsylvania CURE grant SAP#4100055577 (Childress); the NARSAD Young Investigator Grant from the Brain & Behavior Research Foundation (#30780, Shi); the following National Institutes of Health grants: DA036028 (Langleben), DA051709 (Shi), DA028874 (Childress, Li), DA046345 (Kranzler, Childress, Wiers), AA026892 (Wiers) and MH119043 (Todaro) and intramural support from the National Institute on Alcohol Abuse and Alcoholism (Y1AA-3009, Volkow).CONFLICT OF INTEREST STATEMENT Dr. Kranzler is a member of advisory boards for Dicerna Pharmaceuticals, Sophrosyne Pharmaceuticals, Clearmind Medicine and Enthion Pharmaceuticals; a consultant to Sobrera Pharmaceuticals; the recipient of research funding and medication supplies for an investigatorinitiated study from Alkermes; a member of the American Society of Clinical Psychopharmacology's Alcohol Clinical Trials Initiative, which was supported in the last 3 years by Alkermes, Dicerna, Ethypharm, Lundbeck, Mitsubishi, Otsuka and Pear Therapeutics; and a holder of US patent 10,900,082 titled: 'Genotype-guided dosing of opioid agonists', issued 26 January 2021.The other authors report no conflicts of interest.