Cannabis is a commonly used drug, and a substantial
minority of individuals who use it frequently will
develop cannabis use disorder (CUD). Johnson and
colleagues1 identify a high genetic correlation between
liability to frequent cannabis use and CUD. This finding is
in line with previous twin research, which has identified
a high correlation between additive genetic influences
acting on exposure to cannabis, frequency of cannabis
use, and CUD.2 One straightforward interpretation is
that frequent use is on the causal pathway to CUD;
however, Johnson and colleagues1 do not explore
this formally (eg, by using Mendelian randomisation,
which uses genetic variants as proxies for exposures of
interest).
In the present study and the previous twin study, a
moderate proportion of genetic influences acting on
liability to CUD appear to be unique from those acting on
liability to frequency of use. These findings emphasise
the importance of considering the stage-sequential
nature of drug use. CUD cannot occur without exposure
to cannabis and is preceded by escalation from initiation
to frequent use. Research from the USA suggests that
an increase in CUD follows legalisation of recreational
cannabis use.3 It is plausible that this rise in CUD is not
simply a consequence of increased availability, but that
the subsequent increases in population-level cannabis
use allow the expression of genetic liability to CUD
in individuals who were previously unexposed. This
possibility has important implications for public health
in relation to cannabis legalisation, as the perceived
benefits of policy liberalisation should be balanced
against the health risks that might be experienced by a
notable minority of individuals using cannabis.
CUD is characterised by negative health and social
effects, but Johnson and colleagues1 show that polygenic
risk for CUD is also associated with psychiatric
phenotypes. A key question is whether reducing
cannabis use is likely to improve mental health. The
identification of genetic markers for CUD can provide
insights into the causal pathways between CUD and
psychiatric outcomes. Although Johnson and colleagues
explore some causal pathways,1 they do not examine
CUD and schizophrenia. Given ongoing debate about
the causal nature of this relationship, future work on this
area is warranted.

Mendelian randomisation has previously been
applied to lifetime cannabis use and schizophrenia.4
Although lifetime use is not an ideal measure, as it
includes those who might have used cannabis only a
few times, there was stronger evidence that liability
to cannabis use increased the risk of schizophrenia. It
would be interesting to extend this work by looking
at schizophrenia in relation to CUD and cannabis
use frequency (once such genome-wide association
studies become available). Although Johnson and
colleagues1 used Latent Causal Variable analysis to
account for sample overlap, this method does not
allow simultaneous bidirectional effects.5 For future
work, formal methods of Mendelian randomisation
would be more informative, combined with a method
like genomic structural equation modelling to model
directional relationships, which is not biased by sample
overlap.6 Given the high correlation between cannabis
use and tobacco use, and the fact that one of the top
genetic variants associated with CUD is located near a
nicotinic acetylcholine receptor gene, studies on the
association between cannabis use and misuse and
psychiatric outcomes should also account for tobacco
use. This inclusion is particularly important given
evidence for a causal role for tobacco use in psychiatric
outcomes.7,8 Methods, such as multivariable Mendelian
randomisation, would allow the unique causal effects
of tobacco and cannabis use on these outcomes to be
better understood.9 The correlation between cannabis
use and tobacco use also highlights the potential for
pleiotropic effects of genetic variants on these (and
other) outcomes. It will be important to understand
the underlying traits that these variants capture (which
might not necessarily be the measured trait) as they
begin to be used in causal analyses.
Mendelian randomisation is increasingly being used
to investigate causal pathways between modifiable
exposures and a range of outcomes. The work of
Johnson and colleagues1 provides further opportunities
to understand the harms of cannabis use. In particular,
there is now increasing scope to explore the complex
pathways between stages of substance use (eg, from
initiation to frequency of use and dependence), and
the use of multiple substances (eg, the common
use of cannabis and tobacco, which makes their epidemiological dissection challenging). To support strong causal inference, and therefore robust public health policies, findings from Mendelian randomisation
should be considered alongside those from other
methodological approaches that also aim to understand
the harms of cannabis use. This approach is known as
triangulation.10 Understanding these pathways will be
crucial if evidence-based public health policies are to be
implemented.

All authors report contributing to the International Cannabis Consortium.
LAH and JLT contributed equally to the writing of this Comment.
Copyright © 2020 The Author(s). Published by Elsevier Ltd. This is an Open
Access article under the CC BY-ND-ND 4.0 license.
Lindsey A Hines, Jorien L Treur, Hannah J Jones,
Hannah M Sallis, *Marcus R Munafò
marcus.munafo@bristol.ac.uk
Population Health Science, Bristol Medical School (LAH, HJJ, HMS), MRC
Integrative Epidemiology Unit at the University of Bristol (LAH, HJJ, HMS, MRM),
National Institute for Health Research Bristol Biomedical Research Centre,
University Hospitals Bristol NHS Foundation Trust (HJJ, MRM), and School of
Psychological Science (MRM), University of Bristol, Bristol, UK; and Department
of Psychiatry, Amsterdam UMC, and Addiction Development and
Psychopathology (ADAPT) Lab, Department of Psychology (JLT), University of
Amsterdam, Amsterdam, The Netherlands
1 Johnson EC, Demontis D, Thorgeirsson TE, et al. A large-scale genome-wide
association study meta-analysis of cannabis use disorder. Lancet Psychiatry
2020; published online Oct 20. https://doi.org/10.1016/
S2215-0366(20)30339-4.
2 Hines LA, Morley KI, Rijsdijk F, et al. Overlap of heritable influences between
cannabis use disorder, frequency of use and opportunity to use cannabis:
trivariate twin modelling and implications for genetic design. Psychol Med
2018; 48: 2786–93.
3 Cerda M, Mauro C, Hamilton A, et al. Association between recreational
marijuana legalization in the United States and changes in marijuana use
and cannabis use disorder from 2008 to 2016. JAMA Psychiatry 2019;
77: 165–71.
4 Pasman JA, Verweij KJH, Gerring Z, et al. GWAS of lifetime cannabis use
reveals new risk loci, genetic overlap with psychiatric traits, and a causal
influence of schizophrenia. Nat Neurosci 2018; 21: 1161–70.
5 O’Connor LJ, Price AL. Distinguishing genetic correlation from causation
across 52 diseases and complex traits. Nat Genet 2018; 50: 1728–34.
6 Grotzinger AD, Rhemtulla M, de Vlaming R, et al. Genomic structural
equation modelling provides insights into the multivariate genetic
architecture of complex traits. Nat Hum Behav 2019; 3: 513–25.
7 Jones HJ, Gage SH, Heron J, et al. Association of combined patterns of
tobacco and cannabis use in adolescence with psychotic experiences.
JAMA Psychiatry 2018; 75: 240–46.
8 Wootton RE, Richmond RC, Stuijfzand BG, et al. Evidence for causal effects
of lifetime smoking on risk for depression and schizophrenia: a mendelian
randomisation study. Psychol Med 2019: published online Nov 6.
https://doi.org/10.1017/S0033291719002678.
9 Sanderson E, Davey Smith G, Windmeijer F, Bowden J. An examination of
multivariable mendelian randomization in the single-sample and
two-sample summary data settings. Int J Epidemiol 2019; 48: 713–27.
10 Munafo MR, Davey Smith G. Robust research needs many lines of evidence.
Nature 2018; 553: 399–401.