A GWAS of 184,765 individuals identified 8 significant genetic variants for lifetime cannabis use in 6 genomic regions, with the strongest finding at CADM2 (linked to risk-taking), 11% SNP heritability, and Mendelian randomization evidence that schizophrenia risk causally increases cannabis use.
Genetics researchers; psychiatrists studying cannabis-psychosis relationships; public health researchers modeling bidirectional risk.
Schizophrenia risk causally increases cannabis use (Mendelian randomization)
The Backstory
When scientists ask why some people use cannabis and others don't, the usual answers involve environment: peer groups, availability, cultural norms, stress, opportunity. Jacqueline Pasman's 2018 genome-wide association study asked a different question: how much of the variation in cannabis use is written into your DNA?
The answer, derived from the genomes of 184,765 people, was both illuminating and unsettling. Eight genetic variants across six regions of the genome were significantly associated with lifetime cannabis use. And the genetic architecture of cannabis use overlapped substantially with the genetic architecture of schizophrenia — a finding that complicates the cannabis-psychosis debate in ways that neither side of that argument may want to hear.
The Largest Genetic Study of Cannabis Use
Pasman's team, a massive international collaboration published in Nature Neuroscience, conducted a SNP-based genome-wide association study meta-analysis across multiple cohorts. The scale was unprecedented for cannabis genetics: 184,765 individuals and nearly 12 million genetic variants tested.
11%
of the variance in lifetime cannabis use was explained by measured genetic variants — a substantial proportion for a complex behavioral trait, comparable to the genetic contribution to traits like educational attainment or body mass index.
Gene-based tests identified 35 significant genes across 16 genomic regions. S-PrediXcan analyses found 21 genes with different expression levels in cannabis users versus non-users. The strongest signal was CADM2 — a gene previously associated with substance use, risk-taking, and cognitive function.
Pasman et al. (2018), Nat Neurosci 21(9):1161-1170
Eleven percent may not sound like a lot, but for a behavior influenced by dozens of environmental factors — where you grew up, who your friends were, whether cannabis was available, cultural attitudes, legal status — explaining 11% from genetics alone is substantial. It means your DNA has a measurable influence on whether you'll try cannabis, even after accounting for everything else in your environment.
The Genes That Emerged
The eight genome-wide significant variants pointed to specific biological pathways:
CADM2 — the strongest hit — is particularly interesting. It encodes a cell adhesion molecule involved in synapse formation, and it's been associated with a cluster of traits that paint a behavioral portrait: substance use, risk-taking, lower cognitive performance, and higher BMI. The gene doesn't code for a "cannabis receptor" or a "drug-seeking pathway." It appears to influence a broader behavioral tendency toward novelty-seeking and risk tolerance, of which cannabis use is one expression.
The Schizophrenia Bombshell
Genetics
The Cannabis-Schizophrenia Genetic Overlap
Finding
Significant positive genetic correlation between cannabis use and schizophrenia — the same genes that predispose you to psychosis also predispose you to using cannabis
Mendelian randomization
Analysis suggested a causal direction: genetic liability for schizophrenia increases the probability of cannabis use — not the other way around
What this means
Some of the apparent 'cannabis causes psychosis' signal may be reverse causation — people genetically predisposed to psychosis are more likely to use cannabis, perhaps as self-medication for prodromal symptoms
What it doesn't mean
This does not prove cannabis is harmless for psychosis-vulnerable people. The Di Forti (2012) AKT1 data and the EU-GEI population data show that cannabis exposure genuinely amplifies psychosis risk in susceptible individuals. Both things can be true.
Pasman et al. (2018); Di Forti et al. (2012, 2019)
This is the finding that keeps psychiatric geneticists up at night. The Mendelian randomization analysis — which uses genetic variants as natural experiments to infer causal direction — provided evidence that schizophrenia liability causally increases the likelihood of cannabis use. In other words: the arrow may point in both directions. Cannabis may increase psychosis risk (as Di Forti's AKT1 study and EU-GEI data suggest), AND people genetically predisposed to psychosis may be more likely to use cannabis in the first place.
This doesn't invalidate the cannabis-psychosis link — it makes it more complicated. People in the prodromal phase of schizophrenia (before full onset) often experience anxiety, sleep problems, and social withdrawal. Cannabis temporarily alleviates those symptoms. If genetic predisposition drives some people toward cannabis as self-medication for emerging psychiatric symptoms, then observational studies linking cannabis to later psychosis will overestimate the causal effect of cannabis itself.
The honest conclusion: both directions are probably real. Cannabis use is partly genetically determined, schizophrenia and cannabis use share genetic risk factors, cannabis genuinely increases psychosis risk in vulnerable individuals, and genetically vulnerable individuals are more likely to use cannabis. Untangling these overlapping causal pathways is one of the central challenges in psychiatric genetics.
Beyond Cannabis: The Common Liability Model
Pasman's study found significant genetic correlations with 14 of 25 tested traits: smoking initiation, cigarettes per day, alcohol consumption, schizophrenia, ADHD, major depression, risk-taking, and others. This pattern supports what addiction researchers call the "common liability" model — the idea that there isn't a specific "cannabis gene" or "addiction gene," but rather a constellation of genetic variants that influence broad traits like impulse control, reward sensitivity, and risk tolerance, which in turn affect the probability of using any number of substances.
This is the same model Vanyukov et al. (2012) invoked to challenge the gateway hypothesis: the sequential pattern of substance use (cannabis before other drugs) may reflect shared genetic and environmental vulnerability rather than a pharmacological escalation pathway.
Does this mean cannabis use is genetic and I can't control it?
No. An 11% genetic contribution means 89% of the variance is environmental. Genes influence probability, not destiny. Many people with high genetic risk never use cannabis, and many people with low genetic risk do. The genetic contribution affects the threshold at which environmental factors (availability, peer influence, stress) tip you toward use — but your choices still matter enormously.
If schizophrenia genes drive cannabis use, does that mean cannabis doesn't cause psychosis?
It means the relationship is more complex than a simple causal chain. Both directions appear to be real: genetic predisposition increases cannabis use, AND cannabis exposure increases psychosis risk in genetically vulnerable individuals. The practical advice doesn't change — people with family histories of psychosis should be especially cautious with cannabis — but the scientific explanation for why they should be cautious now involves bidirectional causation rather than a single arrow.
Could these genetic findings be used for personalized cannabis advice?
Not yet, but potentially in the future. Polygenic risk scores (combining information across many genetic variants) could theoretically estimate an individual's genetic predisposition to cannabis use disorder or cannabis-related psychosis. This type of personalized risk assessment is being developed for other conditions but is not yet clinically validated for cannabis.
GWAS of Lifetime Cannabis Use Reveals New Risk Loci, Genetic Overlap with Psychiatric Traits, and a Causal Influence of Schizophrenia
Pasman JA, Verweij KJH, Gerring Z, Stringer S, Sanchez-Roige S, Treur JL, Abdellaoui A, Nivard MG, Baselmans BML, Ong JS, et al. (2018) · Nature Neuroscience
What the researchers found
In the largest GWAS of lifetime cannabis use to date, researchers analyzed 184,765 individuals and identified eight genome-wide significant SNPs in six genomic regions.
All measured genetic variants combined explained 11% of the variance in cannabis use.
Gene-based tests revealed 35 significant genes in 16 regions. The strongest finding was CADM2, previously associated with substance use and risk-taking behavior.
S-PrediXcan analyses showed 21 genes had different expression levels between cannabis users and non-users.
Significant genetic correlations were found with 14 of 25 tested traits, including smoking, alcohol use, schizophrenia, and risk-taking.
Mendelian randomization analysis provided evidence for a causal positive influence of schizophrenia risk on cannabis use. This means genetic liability for schizophrenia increases the probability of using cannabis, not just the reverse.
Why it matters
The Mendelian randomization finding is a game-changer for the cannabis-schizophrenia debate. While epidemiology has long shown that cannabis users have higher psychosis risk, this genetic analysis shows the arrow can also point the other direction: people genetically predisposed to schizophrenia are more likely to use cannabis, possibly as self-medication.
The numbers in context
N = 184,765. 8 genome-wide significant SNPs in 6 regions. 35 significant genes via gene-based tests. 11% variance explained by all SNPs. 21 genes with differential expression. 14 significant genetic correlations with other traits. CADM2 was the strongest finding.
How the study worked
Genome-wide association meta-analysis. N = 184,765. Gene-based tests (MAGMA). S-PrediXcan transcriptomic analyses. LD score regression for genetic correlations. Mendelian randomization for causal inference.
What this study cannot tell us
European-ancestry cohorts only. Lifetime cannabis use is a binary measure that does not capture dose, frequency, or recency. Mendelian randomization assumptions may be violated. 11% variance explained means most genetic variance is still unaccounted for.
How to read the evidence
Strong. Published in Nature Neuroscience with the largest sample to date, rigorous statistical methods, and novel causal inference approach.
When this study was published
Published in 2018 in Nature Neuroscience. Subsequent GWAS with even larger samples have continued to build on these findings.
The bigger picture
This study reframes the cannabis-psychosis relationship as bidirectional: cannabis may increase psychosis risk AND psychosis risk increases cannabis use. Understanding both directions is essential for developing effective prevention and treatment strategies.
Questions still open
- How do these genetic variants affect brain function to influence cannabis use? Can polygenic risk scores identify individuals at highest risk for both cannabis use and psychosis? Would genetic counseling about cannabis risk be feasible or ethical?
Common questions
Does this mean schizophrenia causes cannabis use?
What is CADM2?
Read the original research
GWAS of lifetime cannabis use reveals new risk loci, genetic overlap with psychiatric traits, and a causal influence of schizophrenia.
Nature neuroscience, 21(9), 1161-1170
Citation
Pasman, Joëlle A; Verweij, Karin J H; Gerring, Zachary; Stringer, Sven; Sanchez-Roige, Sandra; Treur, Jorien L; Abdellaoui, Abdel; Nivard, Michel G; Baselmans, Bart M L; Ong, Jue-Sheng; Ip, Hill F; van der Zee, Matthijs D; Bartels, Meike; Day, Felix R; Fontanillas, Pierre; Elson, Sarah L; de Wit, Harriet; Davis, Lea K; MacKillop, James; Derringer, Jaime L; Branje, Susan J T; Hartman, Catharina A; Heath, Andrew C; van Lier, Pol A C; Madden, Pamela A F; Mägi, Reedik; Meeus, Wim; Montgomery, Grant W; Oldehinkel, A J; Pausova, Zdenka; Ramos-Quiroga, Josep A; Paus, Tomas; Ribases, Marta; Kaprio, Jaakko; Boks, Marco P M; Bell, Jordana T; Spector, Tim D; Gelernter, Joel; Boomsma, Dorret I; Martin, Nicholas G; MacGregor, Stuart; Perry, John R B; Palmer, Abraham A; Posthuma, Danielle; Munafò, Marcus R; Gillespie, Nathan A; Derks, Eske M; Vink, Jacqueline M. (2018). GWAS of lifetime cannabis use reveals new risk loci, genetic overlap with psychiatric traits, and a causal influence of schizophrenia.. Nature neuroscience, 21(9), 1161-1170. https://doi.org/10.1038/s41593-018-0206-1
Explore the wider topic
- Cannabis and Your Cardiovascular System: Heart Risk, Stroke, and What the Research Shows
- Cannabis and Your Heart: What the Cardiovascular Research Shows
- Coughing Up Black Stuff After Quitting Weed: What It Means
- Lung Recovery After Quitting Weed: The Complete Timeline
- Weed and Your Lungs: Recovery Timeline After Quitting
- Quitting Weed as a Woman: Hormones, Cycles, and Recovery
- Cannabis, Weight, and Diet: What Quitting Does to Your Appetite
- Sex After Quitting Weed: What Changes and What Doesn't
- Cannabis and Driving: DUI Laws, Impairment Science, and What You Need to Know
- Weed and Acne: Can Quitting Cannabis Clear Your Skin?
- Weed and Fertility: What Couples Should Know
- Weed and Your Gut: How Cannabis Affects Digestion
- Weed and Your Heart: Cardiovascular Effects and Recovery
- Weed, Testosterone, and Male Health: What the Research Actually Shows