We wish to test the hypothesis that alleles leading to lower transcription rates of the serotonin transporter gene are associated with significantly higher levels of emotionality in early life. Our objective is to rigorously test this hypothesis by exploiting the size and representativeness of the unique ALSPAC sample and by using a research design that includes stringent control measures.

Identifying the genetic and environmental determinants of variation in human traits is seen as one of the central components of what is described as translational research in the biomedical sciences. Following this research direction, we propose using a large birth cohort in order to probe the relationship between a polymorphism in the promoter region of the serotonin transporter gene and a well-established early risk factor for psychiatric problems.

A paradigmatic case of studying the genetic underpinnings of emotions and behaviour has been that of establishing the association between human emotionality and a functional polymorphism at the promoter region of the serotonin transporter gene, designated 5-HTTLPR. Emotionality, commonly referred to as neuroticism in the literature, is measured as a continuous personality trait and has been shown to be an important risk factor for depressive and anxiety disorders. The short allele (S) at the 5-HTTLPR leads to lower transcriptional activity and , thus, reduced serotonin re-uptake at the synapse compared to the long (L) allele. A high-profile publication in the mid '90s reported that possession of the short allele (s) of the 5-HTTLPR is significantly associated with higher scores on neuroticism measures compared to possession of the long allele. Indeed, a series of replications and extensions have lent further support to this important finding; although there have also been notable non-replications as well as ambiguous meta-analytic results. Given the long-known links between neuroticism and depression/anxiety, and also the central role ascribed to serotonin metabolism in depression/anxiety, this finding has boosted the hopes for arriving at a molecular explanation of mental disorder.

One of the most important questions arising in view of the potential importance of these findings is how early in life does their relationship between 5-HTTLPR and emotionality become apparent? In other words, how early does 5-HTTLPR start operating as a putative pathophysiologic mechanism and what would its potential utility be as an early biomarker? Considering that about 70% of mental illness has its origins in pre-adulthood, the potential importance of an answer to this question becomes particularly pressing. Some existing studies indicating that there is an early link between 5-HTTLPR and emotionality, other studies do not find a link. However, the small sample size of most of these studies does not allow for inferences to be drawn with a great degree of confidence. Here we propose to use a large, representative and well-characterized birth cohort, the ALSPAC dataset to study the link between 5-HTTLPR and emotionality in early life. Moreover, as we will detail in the research design that will help us overcome some of the typical challenges facing research into early emotionality, such as specificity of the findings and influence of reporter effects.

In summary, we propose to study a question that is central to current behavioural research-the genetic underpinnings of early human emotionality-and do so by using adequately-powered and rigorously controlled research methodology.

Work undertaken so far

1) Preliminary research within the ALSPAC cohort.

We have studied the instrument that we intend to use for measuring early emotionality, namely the well established Emotionality Activity Sociability (EAS) measure. As shown in Table 1, emotionality forms a distinct factor in the ALPAC population; it also shows good internal reliability. Moreover, as shown in Figure 1, we have demonstrated that emotionality measured at 38 months is an important predictor of internalising and some, but, crucially, not all forms of psychopathology in middle childhood (98 months).

2) Power analysis using the ALSPAC cohort.

Please see under "plan of investigation" for the relevant power analyses.

3) Published results

Our research efforts so far have focused on establishing the cross-sectional, three-year-, and 20-year follow-up associations of an important aspect of youth emotionality, irritability, in community samples.

In addition, we have established the prevalence and morbid associations of emotional instability (mood lability) in the general population.

* Brief plan of investigation, including methodology and design

Sample: The Avon Longitudinal Study of Parents and Children is a well-described birth cohort. DNA has been extracted and is available for 7000 mothers and 10000 children in the cohort.

Genotyping: Given previous reports about the 5-HTTLPR being functionally tri-allelic because of a substitution within the L allele, we will genotype the variable number tandem repeat (VNTR; short and long allele) as well as the single nucleotide polymorphism (SNP).

While the 5-HTTLPR has been partly genotyped before in ALSPAC, this was only done for a fraction of the children in the sample and the error rates were higher than what is commonly encountered in the literature. We have received advice on this matter and have established that the genotyping can be done at the Molecular Genetics Lab in the Institute of Psychiatry with a very high (less than 1%) precision rate (Professor Ian Craig, personal communication).

Phenotype: We will use the well-established (see above) EAS scale to measure emotionality and activity.

Main Analysis:

To test the main hypothesis that the low-expression genotype, an ANOVA with genotype as the categorical 3-level variable (low-, medium-, and high-expression) and human emotionality as the independent variable with planned post-hoc analysis to compare the low-expression genotype with the other two genotypes.

Controls:

1) Specificity of the finding: We will compare the results of emotionality with those of one of the other factors in the EAS for which we have established that it is a predictor of illness. This is in order to ensure that any the connection between 5-HTTLPR and emotionality is specific rather than a diffuse marker of pathology.

2) Impact of maternal genotype: It is well conceivable that maternal genotype influences ratings of emotionality either due to bias in reporting (mothers with a genotype linked to high emotionality may be biased towards scoring their children higher, for example) or an interaction between mothers and their children (mothers with a genotype linked to high emotionality may, for example, create an enviroment that is conducive to developing higher rates of emotionality). For this reason, we will conduct subanalyses stratifying for maternal genotype. For similar reasons, we will conduct subsidiary analyses stratifying for maternal history of mental illness.

Power Analysis:

We have used G-Power Version 3 (University of Dusseldorf) to calculate the power afforded by our sample to answer the main hypothesis and subsidiary questions of this study.

Assuming a sample size of 7000 children, an alpha level of 0.05 we would have a power of 100% to estimate an effect size as small as 0.1 in an F-test with a three-level categorical predictor variable. After stratifying for three maternal genotypes, essentially creating a 9-level variable, our power to detect a significant difference at 0.05 for an effect size as small as 0.1 would be 100%.

We will seek to confirm these estimates in our meeting with Dr Daniel Stahl, Department of Biostatistics, in a week's time.