British Journal of Psychiatry, (1989), 155 (suppl. 6), 18-26 The Genetics of Depression: Current Approaches P. McGUFFIN and R. KATZ It is widely accepted that depression shows a tendency to run in families. This is partly due to genetic influences - and the subject has been covered in recent reviews. This article, therefore, does not aim to recapitulate the genetic evidence but rather aims to elucidate the mode of transmission, to achieve an understanding of relevant gene-environment co-actions and interactions, and to describe attempts to identify and localise major genes. That depression shows a tendency to run in families is a widely accepted and uncontroversial fact. There is good evidence that the tendency to familial clustering is partly due to genetic influences, and this has been the subject of a number of recent reviews (e.g. Reich et al, 1982; McGuffin & Katz, 1986, 1989). The purpose of this article, therefore, is not to recapitulate in detail the evidence in favour of a genetic contribution to depression, but rather to focus on recent studies which assume that the case for a genetic predisposition is proven. The main endeavours then are to elucidate the mode of transmission, to achieve an understanding of relevant gene-environment co-actions and interactions, and to identify and localise the major genes. The evidence from genetic studies suggests that both genes and environment are important in the pathogenesis of depression. Environmental influences are probably most important in the less severe, more broadly defined depressive disorders, including those where a neurotic pattern of symptoms is found, but even in typical bipolar affective illness, the existence of discordant identical twins points to some degree of environmental determination. In Table I, data from recent twin studies are used to estimate the comparative contributions to the variance in liability to depression of genes (h 2 ), common environment (c 2 ), and non-shared environment (s ) i.e. those environmental influences which are specific to the individual and which do not impinge on his/her relatives. These calculations assume a multifactorial liability-threshold model, in which a variable termed 'liability to develop the disorder' is normally distributed in the population, such that only those individuals whose liability at some stage exceeds a certain threshold manifest the disorder (Falconer, 1965; Reich et al, 1972). These estimates have large standard errors because the sample sizes, although currently the best available, are fairly small (McGuffin & Katz, 1986, 1989). Nevertheless, the evidence provides a strong impression that Kraepelinian-type manic-depressive illness, as studied by Bertelsen et al (1977), is a predominantly genetic condition, while neurotic depression is mainly of environmental origin and clusters in families because of common family 18 environmental effects (Torgersen, 1986). Major affective disorder, as judged from the preliminary results of a recent Maudsley Hospital Register-based study (McGuffin & Katz, 1989), appears to occupy an intermediate position. We view the intermediate position of major affective disorder as scarcely surprising, since clinical experience suggests that the DSM - III or DSM - III-R category (American Psychiatric Association, 1980, 1987) is a mixture of what in Britain would be classified as either ICD-9 manic-depressive illness (296) or depressive neurosis (300.4). It is important to stress that concepts such as heritability have serious limitations, and that estimations of h 2 and c 2 should not be seen as ends in themselves. Nevertheless, the results summarised in Table I provide more than a hint that if our aim is to search for major genes, we should concentrate on the more severe forms of manic-depressive disorder, while if our main aim is to investigate the interplay between genes and environment, we should probably focus on the more broadly defined milder forms of depression. The search for major genes for severe affective disorders When dealing with a trait which does not lend itself to a clear all-or-none dichotomy, or where typical Mendelian segregation patterns are not observed, three complementary strategies can be used to detect major gene effects. The first is to analyse the distribution of the trait in the population. If some continuous measure can be applied and a mixture of two or more distributions can be demonstrated, then this can be taken as suggestive evidence of a major gene (particularly if there are three distributions in the proportions predicted by the Hardy- Weinberg Law). This approach has been successfully used for traits such as platelet monoamine oxidase activity, which are of potential psychiatric interest (Rice et al, 1982). Unfortunately, there is as yet no satis- factory continuous trait measure which is strongly associated with liability to become depressed and is not