Genetic Pathways Link Cannabis Use Disorder to Psychosis Risk

A genetics study from researchers at King’s College London has identified, for the first time, three distinct groups of genetic variants likely involved in the pathway from heavy cannabis use to psychosis — findings that researchers say could eventually support personalized risk assessment for people who develop psychosis in the context of cannabis use.

The paper, published May 13, 2026 in Biological Psychiatry Global Open Science, is a meta-analysis of genome-wide association study (GWAS) data from people diagnosed with schizophrenia and bipolar I disorder. Led by first author Dr. Isabelle Austin-Zimmerman and senior author Professor Marta Di Forti at King’s Institute of Psychiatry, Psychology & Neuroscience (IoPPN), the team combined both diagnostic groups into a single broader psychosis category — a deliberate methodological choice. Cannabis-related psychosis in epidemiological research doesn’t map cleanly onto either diagnosis alone, and prior analyses restricted to schizophrenia have likely underrepresented the full genetic overlap.

What the meta-analysis found

The analysis turned up more than 500 genetic loci associated with psychosis, including 122 not previously reported in this context. The more analytically significant result came from the pathway analysis: biological pathways shared between psychosis and cannabis use disorder were more numerous than would be expected by chance alone. That’s not proof of a deterministic chain from genotype to clinical outcome, but it does put molecular specificity on an epidemiological association that has long resisted a clean biological explanation.

The study also adds scale to what’s already known about directionality. Prior research has established that the relationship between cannabis use and psychosis runs both ways — psychosis appears to increase cannabis use, and cannabis use appears to increase psychosis risk. What this analysis contributes is a comparison of magnitude: the causal association running from cannabis use disorder to psychosis is substantially larger than the reverse.

The researchers identified three distinct groups of genetic variants likely involved in the cannabis-use-disorder-to-psychosis causal pathway — variants associated with genes relevant to neurodevelopment, neuronal signaling, and other biological processes. Only one comparable group of variants was detected running the other direction. Di Forti described that asymmetry as consistent with clinical observation: there appear to be more biological routes through which cannabis use contributes to psychosis than routes through which psychosis leads to heavy cannabis use. The mechanisms aren’t simply the same pathway operating in two directions — they differ depending on which condition is the starting point.

The glutamate connection

Among the specific systems the pathway analysis flagged, glutamate-related genes stand out as candidates for predicting psychosis risk in cannabis users. Glutamate is a neurotransmitter whose signaling THC — cannabis’s primary psychoactive compound — is known to disrupt. Identifying glutamate-pathway gene variants as potentially predictive of risk is mechanistically coherent: it connects established pharmacology to a genetic signal that might eventually be measurable at the individual level.

Di Forti described the glutamate findings as potential “groundwork for prediction models that identify those most at risk, as well as informing the development of new targeted treatments.” That framing is appropriately cautious. The distance between identifying a candidate genetic system and deploying a validated clinical risk score is considerable — and the authors don’t compress it.

For anyone tracking cannabis and neurodevelopmental risk, the involvement of neurodevelopment-related variants in the cannabis-use-to-psychosis pathway adds molecular context to what epidemiology has already suggested. Whether the developmental stage interacts with these genetic signals is a question the current analysis wasn’t designed to answer.

What the evidence can and can’t support

A GWAS meta-analysis identifies associations between genetic variants and a trait — here, psychosis is defined broadly to include schizophrenia and bipolar I. The Mendelian randomization approach used to assess causal direction is a validated method for inferring causality from genetic data, but it carries assumptions that can strain under complex polygenic conditions like psychosis. These are signals, not deterministic markers.

The populations in most GWAS datasets remain predominantly European in ancestry, limiting how well any derived risk patterns generalize across other groups. The loci identified here explain only a portion of psychosis heritability. Identifying genes linked to neurodevelopment and neuronal signaling doesn’t specify which cellular mechanisms are most relevant — that work requires follow-up experimental studies.

Austin-Zimmerman framed the findings with appropriate precision: “a step towards understanding the potentially different etiology in psychosis with and without cannabis use.” The distinction matters clinically. If specific genetic profiles can eventually separate cannabis users at elevated psychosis risk from those who aren’t, then risk counseling and targeted intervention can be matched to the underlying biology — rather than applied uniformly across a population with very different individual risk landscapes.