Precise Use of Psychedelics in Treating Addiction

Despite decades of neuroscientific investigation and billions invested in brain imaging, clinical outcomes in addiction remain staggeringly poor. A recent WHO report estimates that only 1.8% of individuals with substance use disorders (SUDs) globally receive effective treatment (1). This figure is particularly sobering when set against the backdrop of nearly 40 years of neuroimaging research in addiction, much of which has yet to yield actionable tools for clinical use (2).

To date, the field has produced 479 imaging paradigms across 409 distinct protocols, the vast majority relying on human functional magnetic resonance imaging (fMRI) (2). Notably, over 70% of human addiction brain biomarkers studies have been aimed at diagnostic classification, with task-based cue-reactivity emerging as the lead candidate for a regulatory-approved biomarker of addiction (3). Yet the diagnostic yield of this work has been minimal. Few, if any, biomarkers have been translated into routine clinical practice or drug development (2). Efforts to improve translation from preclinical to clinical biomarkers are however underway and opportunities to improve this translational bridge are emerging and are discussed more in depth here but are beyond the scope of this review (4).

This disproportionate focus on diagnostic biomarkers has created a stagnation in the field of addiction neuroscience. What is urgently needed now is a redirection toward developing translational and theragnostic biomarkers (for definitions, see Table 1) to advance therapeutic development and a move away from basic science experiments that do not directly serve to advance treatments. Shifting the aim of addiction neuroimaging biomarker research from predominantly diagnostic classification to theragnostic development can provide clinical value in four key ways: (1) catalyse the discovery of novel therapeutics; (2) inform biotype-specific treatment assignment; (3) move beyond DSM-defined diagnoses towards biologically meaningful patient stratification and biotyping; and (4) provide predictive or prognostic indicators to optimise treatment outcomes. This shift would align addiction medicine with a broader trend across psychiatry and indeed across the whole of medicine towards precision-guided, mechanism-based approaches (5, 6).

Encouragingly, early work in this space suggests that neuroimaging biomarkers may, in some cases, outperform clinician judgement in predicting treatment response, and some studies have begun to assess how neuroimaging data can influence or moderate treatment outcome, with some replication studies beginning to validate this approach (7, 8). However, a major recent systematic review on neuroimaging biomarkers in addiction suggests less than 5% of studies adopt the strategy of trying to understand the predictive or prognostic potential of these markers, and no validated theragnostic tools currently exist for addiction therapeutic development (2).

This perspective calls for a reorientation of addiction neuroscience research over the next decade. We feel it is timely to refocus efforts from diagnostic classification to a translational, precision-medicine framework guided by biomarker discovery and validation. Particularly, this perspective is the first to our knowledge to integrate multimodal imaging into a theragnostic framework for biomarker-guided psychedelic addiction medicine.

To anchor these arguments in practical application, we highlight ongoing psychedelic therapy (PT) research at Imperial College London, specifically focused on psilocybin as an investigational therapeutic for addiction. We show how this work integrates neuroimaging biomarkers within experimental and translational medicine frameworks to accelerate both drug development and mechanistic understanding that could build the foundations for precision psychedelic medicine. It is our proposal that this convergence of neuroscience, biomarkers, and these novel and highly promising therapeutics holds the greatest potential for transformative change in addiction treatment and validated biomarker development over the next decade.

It is the hope that this perspective combines mechanistic insights into both psychedelic research and biomarker discovery in addiction, bridging these as yet disparate domains and providing a translational roadmap for mechanism-based, biomarker-informed psychedelic research.

The process of drug development traditionally begins with the identification of a biological target implicated in a given disease process. This is followed by high-throughput screening of candidate molecules, evaluating their binding affinity, bioavailability, and pharmacological profiles in vitro. A refined subset is then progressed into preclinical animal models, where safety, toxicity, pharmacokinetics (PK), pharmacodynamics (PD), and target engagement are assessed in vivo. Lead compounds that meet these criteria advance to early-stage human trials (9).

When considering the preclinical addiction drug development approaches, a variety of in vivo animal models exist to assay mechanisms and the potential of candidate compounds (10). However, capturing the nuanced behavioural and environmental complexities of addiction in animal models has proved difficult, and translating observations from preclinical animal models to humans has had limited, albeit some, success (11, 12).

Conversely, preclinical psychedelic drug development has benefitted from back-translation of human neuroimaging to determine occupancy of psychedelics in rodent models, which has benefitted a more optimised and targeted development of compounds, including novel second and third generation psychedelics and non-psychedelic analogues (13). Indeed, harmonising approaches across preclinical and clinical neuropsychopharmacology will provide a deeper understanding of the molecular and functional mechanisms of drug action, and this approach presents a unique opportunity for addiction psychedelic drug development, of which some companies (Alvarius Pharmaceuticals) are now spearheading. A recent review highlights how the adoption of such translational assays is critical for the interpretation of animal studies of psychedelic effects on cognition, affect and reward and how these studies have the potential to help unravel the mechanisms which contribute to their therapeutic effects, but only if they involve relevant doses (14).

In parallel to aligning preclinical and clinical mechanistic research on interventions, ongoing human multimodal research into the development of validated, reliable and scalable biomarkers in addiction remains essential. Importantly, though this review focuses primarily on neuroimaging biomarkers, it is imperative to test and consider the association between CNS and non-CNS markers relevant to addiction too, as research demonstrates the important role of the involvement of peripheral markers. Despite recent advances in identifying candidate biomarkers of addiction across neural, plasma-based, genetic and epigenetic domains, no single measure has yet demonstrated sufficient clinical or translational utility when used in isolation. While some show promise in substance use and behavioural addictions, the complexity and heterogeneity of addictive disorders demand a more integrated approach. Multimodal research that combines multiple biological systems, psychological testing and behavioural markers offers a more realistic path forward, not only for improving biomarker precision but also for individualised risk profiling, for guiding treatment selection and monitoring therapeutic response (15). The future of addiction medicine lies in embracing this complexity, rather than oversimplifying it.

To date, neuroimaging biomarkers have been the most studied biomarker in addiction drug development. Studies that clarify developmental trajectories, neurocircuit dysfunction, and molecular targets in addiction are crucial to the rational design of new therapeutics (16). This, along with advances in large-scale genetic consortia (e.g., the Psychiatric Genomics Consortium), digital phenotyping, and biomarker technologies, are converging to define a new era of precision psychiatry (17) of which addiction medicine could greatly benefit. These tools are not only shaping therapeutic design but are increasingly integrated into clinical trials to support stratification, enrich for likely responders, and guide outcome prediction as reviewed here (18).

Increasingly, insights from genetics, digital phenotyping, and multimodal biomarkers are converging to define a new era of precision psychiatry. These tools are not only shaping the design of more targeted compounds but are also supporting clinical trial enrichment and mechanism-based therapeutic discovery with multistakeholder precision psychiatry roadmaps now being adopted across the EU and North America for mental health conditions more broadly (19, 20).

At each clinical trial phase (Figure 1), neuroimaging-informed, brain-based biomarkers [e.g., electroencephalography (EEG), functional magnetic resonance imaging (fMRI), and positron emission tomography (PET)] can play a critical role in assessing pharmacodynamic responses, establishing target engagement, and discriminating treatment responders from non-responders (21). When integrated into early-phase study design, these markers can also inform ‘go/no-go’ decisions and reduce the high failure rates that characterise psychiatric drug development. This is particularly relevant in addiction and psychiatry, where nearly three-quarters of investigational compounds fail to progress beyond Phase 2 (22, 23). Poor mechanistic understanding, clinical heterogeneity, and inadequate biomarkers all contribute to these failures. Efforts to incorporate experimental medicine frameworks with embedded mechanistic endpoints are gaining traction across psychiatric trials, including in initiatives such as the NIH’s Experimental Therapeutics Approach and the European Medicines Agency’s Biomarker Qualification Programme (24, 25).

Translational neuropsychopharmacology leverages human neuroimaging to bridge the gap between basic neuroscience and clinical treatment development. In addiction research, neuroimaging has been pivotal in advancing our understanding of the pathophysiology underlying key features of addiction (such as disrupted reward, salience, executive, and affective processing), and identifying potential biomarkers that may serve as targets or readouts for novel therapeutics (2, 26).

Integrating imaging tools into early- and mid-phase clinical trials facilitates the evaluation of the mechanistic effects of new interventions, including psychedelic compounds. Linking brain-based biomarkers with clinical and behavioural outcomes offers a richer and more precise account of how interventions act on the neural substrates of addiction. Such approaches have been successful in identifying putative predictive and prognostic psychedelic biomarkers in depression (27, 28). There are many ways to implement this approach in addiction medicine, such as integrating functional, structural, and molecular neuroimaging biomarkers with peripheral metrics, psychometric, behavioural and neurocognitive assessments (29).

Importantly, neuroimaging biomarkers can also support regulatory decisions by offering mechanistically grounded evidence of treatment effects, as was seen in the case of the FDA approval of Alzheimer’s drug aducanumab (30). As the field begins to shift towards mechanism-of-action-driven trial design, particularly in early-phase drug development, neuroimaging provides a means to test whether a candidate agent is engaging its intended target, altering disease-relevant circuitry, and producing measurable change in brain function.

These approaches are not hypothetical. Numerous working groups, including those supported by the National Institute on Drug Abuse (NIDA), the European College of Neuropsychopharmacology (ECNP), and international consortia such as ENIGMA-Addiction, have called for the systematic development of neuroimaging biomarkers to guide precision medicine in addiction and psychiatry at large. A white paper co-authored by 30 global leaders in addiction science laid out a translational roadmap to incorporate neuroimaging in future treatment trials, specifically advocating for their use in drug development and regulatory qualification processes (31, 32).

Despite these advances, industry uptake has lagged. To date, few neuroimaging biomarkers have been formally qualified by regulators for psychiatric indications, and even fewer are embedded in addiction drug development programmes. Strengthening academic-industry partnerships, improving standardisation across imaging pipelines, and demonstrating clinical utility through prospective trials are critical next steps.

In this context, psychedelic research offers a promising testing ground for embedding neuroimaging into translational pipelines (27). These compounds act rapidly, target specific receptor systems (e.g., 5-HT2A), and induce measurable changes in brain function and subjective experience, making them ideal for the development of pharmacodynamic biomarkers (33). In the following sections, we examine how fMRI and EEG have been used to support drug development in addiction and how these modalities can be deployed in experimental medicine studies to guide the advancement of psychedelic therapies.

Neuroimaging has played a foundational role in contemporary psychedelic research. A wide range of acute and longitudinal studies have investigated how serotonergic psychedelics alter brain dynamics, modulate neural circuits, and drive therapeutic outcomes. These approaches have helped define the psychedelic brain state, which is characterised by distinct patterns of functional reorganisation, desynchronisation, increased signal complexity, and altered connectivity (33).

To date, such measures have been sparsely used in psychedelic drug development outside of academia. However, there has been a strong case made for the utility of imaging to help guide drug development and precision and personalised psychedelic medicine (27, 69).

In this section, we highlight emerging biomarkers from psychedelic EEG, fMRI, PET, and magnetic resonance spectroscopy (MRS) studies and discuss how they can inform drug development pipelines by indexing target engagement, brain exposure, and treatment response more broadly. We also illustrate how these tools may enable stratification and biotype-specific treatment selection, which are key priorities for theragnostic implementation (Figure 2).

While the theragnostic rationale for biomarker-guided psychedelic research is compelling, several conceptual and methodological limitations currently constrain its translation. First, state-dependence presents a fundamental challenge: neural signatures identified under the acute or sub-acute psychedelic condition are not readily assessed in the same patients in follow-up states in clinical studies. This complicates interpretation of whether observed changes due to the transient pharmacological effects relate to durable therapeutic mechanisms. Longitudinal designs are required to dissociate drug-related from trait-related biomarker shifts.

Second, blinding and expectancy effects remain significant obstacles. Given the distinctive subjective experiences induced by psychedelics, full participant and investigator blinding is rarely achievable, introducing potential expectancy-related confounds (140). Active placebos or sub-perceptual comparator designs may partially address this issue or multiple dose studies.

Third, physiological confounds, including alterations in cerebral blood flow, heart rate, and respiration can influence BOLD-fMRI signals. Incorporating concurrent physiological monitoring, e.g., through peripheral wearables or utilising complementary modalities (e.g., PET, EEG, ASL-fMRI) can enhance interpretability.

Beyond methodological issues, multimodal imaging pipelines remain costly and resource-intensive, limiting scalability, though some companies are now beginning to explore this space by utilising imaging to guide drug development and patient stratification, such as Alto Neuroscience Inc. Studies involving PET are particularly resource-intensive, with availability of particular PET ligands (e.g., [11C]Cimbi-36) also sharply limited to a small number of sites worldwide. Standardisation of methods across sites and use of open-source analysis frameworks is also a current challenge, but will be critical for reproducibility and inclusion into existing frameworks such as ENIGMA and following BEST procedures (141) will be necessary to validate such biomarkers for clinical adoption.

Together, these considerations underscore that biomarker integration into psychedelic drug development remains an evolving endeavour. Addressing these challenges will be essential for moving from correlational insight toward clinically actionable, biologically grounded theragnostic tools.

To transform psychedelic therapy into a precision treatment for addiction, biomarkers must do more than explain; they must predict, stratify, and scale. Neuroimaging markers that track changes in craving, control, and reward circuits can help identify responders, optimise dosing, and de-risk development. Large-scale data across consortium is needed to effectively validate these biomarkers utilising platforms created by organisations such as ENIGMA, the ISAM the ECNP precision psychiatry working group. Validated across multi-site and multi-centre clinical trials, these tools could support drug development, enabling the kind of targeted, biomarker-guided treatments that have revolutionised oncology, alzheimer’s and gene therapy.

Realising the vision of theragnostic psychedelic addiction medicine will require more than compelling science; it will demand collaboration across sectors. Academia must work hand-in-hand with industry, regulatory bodies, healthcare providers, funders, and patient communities to co-develop validated, scalable biomarker platforms. This includes partnerships with pharmaceutical and biotech companies developing current and next-generation psychedelics; regulatory engagement with agencies such as the MHRA, EMA, and FDA to align on biomarker qualification pathways; and integration with public and private healthcare systems to support real-world implementation.

Crucially, this vision must be properly financed. Just as oncology and neurology benefitted from multi-million-dollar precision medicine investments, building a biomarker-driven infrastructure for addiction will require dedicated funding from national research councils, philanthropic foundations, venture capital, and governmental innovation missions. The tools, data, and frameworks now exist to move beyond trial-and-error in addiction and psychiatric medicine more broadly. What’s needed next is the collective will to build it.

The author(s) declared that financial support was received for this work and/or its publication. Dr Rayyan Zafar has received funding from Wellcome Accelerator Award to support writing this research “Translational biomarker development for precision psychedelic medicine in addiction”.

RZ: Conceptualization, Project administration, Supervision, Writing – original draft, Writing – review & editing. PK: Writing – original draft, Writing – review & editing. DK: Writing – review & editing. MW: Writing – original draft, Writing – review & editing. DE: Conceptualization, Supervision, Writing – original draft, Writing – review & editing.

RZ reports to have received scientific consultancy fees from Drug Science Consulting, RELM, Beautiful Spaces and has share options in ITER. DE reports to have received scientific consultancy fees from Aya Biosciences, Lophora Aps, Clerkenwell Health, and Mindstate Design Lab. MW’s primary employer is Perceptive Inc., a contract research organisation that provides research services to the pharmaceutical and biotechnology industries.

The remaining author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer DZD declared a shared affiliation with the author(s) DE to the handling editor at time of review.

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.