Science and sensibility

Last month media from New Zealand and the world reported the latest results of a study based on people from my home town of Dunedin. The results suggested that some people may carry a genetic predisposition to psychotic illness that is triggered by smoking marijuana as a teenager. The story caused a little bit of a fuss here with cannabis law reform proponents decrying the study as being mere statistical fudging. At the time the story hit the news wires the actual paper hadn't been published so I refrained from adding anything until the whole story came to light. So now, with the paper published in the Journal of Biological Psychiatry, let's see what it actually says and just how strongly it can say it.

The Dunedin Multidisciplinary Health and Development Study started in 1972. It has followed 1037 children born in a local maternity hospital in that year with interviews and medical examinations at roughly three year intervals since that time. This webpage describes the things that the study has investigated. In 2002 a team of researchers form King's college in London used the Dunedin study's data to investigate the hypothesis that smoking cannabis as a teenager increased ones risk of developing psychotic symptoms in later life. That analysis showed that people who used cannabis by age 15 were four times more likely to show psychotic symptoms by age 26 than those who did not. However, when those people that showed symptoms indicative of psychosis before they where 11 where removed from the analysis the distinction between the two groups lost statistical significance.

The team at King's thought that cannabis could still be playing a role in some people developing psychotic symptoms. They set out to find another factor that some people in the study might have which, when combined with exposure to cannabis, could rise the person's risk of becoming psychotic. A prime contender was a gene that makes the enzyme catechol-O-methyltransferase (COMT.) COMT is a good candidate because its gene is based in a section of chromosome 22 that has been associated with schizophrenia in QTL studies like the height study I summarised last year. Additionally, COMT's function is to break down the neurotransmitter dopamine in the brain. Dopamine regulation is thought to play an important role in schizophrenia.

The other interesting thing about the COMT gene is that it is polymorphic in our population. That is, there are two distinct versions (alleles) of COMT out there which code for very slightly different enzymes. The two enzymes differ at one amino acid with variants having either valine (val) or methionine (met). Since we each have two copies of COMT (one from each parent) we can have either have two copies of the methionine version, two copies of the valine version or one of each version. These states are called 'genotypes' and can be shortened to met|met, val|val and met|val respectively Roughyl 25% of people are met|met, 50% met|val and the remaining 25% are val|val.

The valine containing version of COMT has been shown to be less active in degrading dopamine which led the King's College team to hypothesise that people with the val|val genotype that are exposed to cannabis as teenagers might be at greater risk of becoming psychotic. To investigate this hypothesis they took another look at the Dunedin data, this time placing people in groups based on both their exposure to cannabis and their genotype for the COMT gene. Their results are summarised in the table below:

The first few rows are to show us that some other things that have been suggested to be good markers for psychosis have no statistical relevance in this study. And to show one more thing IQ tests are no good for. It also backs up the earlier study by showing that people that experience symptoms of psychosis as a child are slightly more likely to experiment with marijuana as adolescents. Probably the most obvious fact made by these data is that COMT genotype has little effect on any of it - there is much more variation between the early dope smokers and their more conservative counterparts than with either group.

The interesting results are to be found in the lower rows. For instance look at how adolescent cannabis use affects each COMT genotype's chances of developing the psychotic illness schizophreniform disorder. For people with the met|met COMT genotype early cannabis use seems to have no bearing on the development of schizophreniform. The story gets a little different for people with met|val; in the study people with this genotype double their chances of developing schizophreniform disorder by smoking cannabis before they turn 18. It's even worse for people with the val|val genotype; for them the risk of developing schizophreniform is increased by a factor of 9. These data are summarised in this graph:

Grey boxes represent non adolescent cannabis smokers. Black boxes represent adolescent canabis users.

All the other measure of psychosis used in the study show similar results. At the time the (unpublished) research hit the newswires a lot of pro-cannabis voices expressed doubts in the validity of the study. They suggested there simply wouldn't be enough people in the adolescent onset smokers group to draw meaningful conclusions and anything published would be statistical extrapolation. In fact, of the 803 people whose data was considered 193 had smoked cannabis before the age of 18. In absolute numbers 54 of the respondents had smoked dope before they turned 18 and had the val|val genotype. 7 of these developed schizophreniform disorders. The group with the same genotype but no experience of marijuana before their 18th birthday comprised 148 members of whom only 2 experienced schizophreniform. The paper uses the precise and arcane measures of statistical analysis to show that these differences are very much significant. It also shows that allowing for increased use of other drugs by early onset cannabis users, childhood psychotic symptoms and underling effects of the valine version of COMT as confounding factors didn't remove this significance.

The results provide an intersting counterpoint to the "gene for x" reporting we tend to here in the media. You know articles that start with the line "researchers have decoded the gene for cancer/highIQ/homosexuality/telepathy" then suggests treatments are least ten years away. COMT is not a gene for psychosis, it's a gene for regulating dopamine in the brain which has one allele that can, under the right circumstances, raise ones rick of developing psychosis. Similarly there are no genes 'for' cancer. There are genes that regulate cell division or cell death or DNA repair or many other processes that are designed to stop cancers developing. Some of those genes are going to be polymorphic and some alleles are going to be better at their job than others - there will be genes that raise the owners risk of developing cancer. But that's a very long way from the deterministic little bodies that media seems to be talking about that, independent of anything else, assign their owner to a life of illness or stupidity or a love of musical theatre or whatever the flavour of the day is.

OK, so what does the study have to say about cannabis law? It seems pretty clear that the valine version of COMT interacts with something that results from smoking cannabis and that interaction increases ones chances of developing psychosis later in life, if the interaction occurs during ones teens. If the results of this study can be reproduced then there will clearly be ramifications in debate surrounding the legal status of marijuana. 75% of the population have at least one copy of the valine version of COMT and 25% have the particularly bad val|val genotype. How should we react to this information? The role of science in matters of ethics and society is not to provide the answer but to give the facts that the debate should revolve around. If it can be shown to be true that early exposure to cannabis carries a major risk to a significant proportion of the population then it is clearly very important that cannabis is kept out of the hands of young people. It's up to society and politicians to decide if that is to be done through restrictive laws or decriminalisation and subsequent regulation along the lines of cigarettes and alcohol.

Reference: Avshalom Caspi et al. (2005) Moderation of the Effect of Adolescent-Onset Cannabis Use on Adult Psychosis by a Functional Polymorphism in the Catechol-O-Methyltransferase Gene: Longitudinal Evidence of a Gene X Environment Interaction. Biological Psychiatry, 57:10