The Interplay of Nature, Nurture, and Developmental Influences: Title and subTitle BreakThe Challenge Ahead for Mental Health

The research achievements to date in the fields of psychiatric genetics, psychosocial influences, and development are considerable. The potential for the future is even greater as a result of technical advances in both molecular genetics and functional brain imaging, to give 2 examples. Nevertheless, these3 fields of research have remained regrettably separate (as is obvious from the limited referencing across fields in scientific journals); if the challenges ahead are to be met successfully, better integration will be needed.¹ This article summarizes some of the key findings already available, focusing particularly on interplay issues, and discusses the important challenges ahead.

Numerous studies of twins, adoptees, and families that have used several different quantitative genetic strategies have made it abundantly clear that genetic influences, as they apply to individual differences in the liability to show particular behaviors, are strong and pervasive but rarely determinative.^{2 - 4} Certain assumptions of behavioral genetics need to be questioned more than they have been in the past. For example, there are problems in sample range and bias, particularly with adoptees,⁵ as well as violations of the equal environments assumption.^{6 - 7} Nevertheless, even when these issues have been taken into account, there can be no serious doubt regarding the reality of substantial genetic effects on all forms of human behavior. Some disorders have a heritability in the 60% to 90% range; this probably applies to autism, schizophrenia, bipolar affective disorder, and attention-deficit/hyperactivity disorder. Many more disorders and dimensional traits have a lesser but still important genetic component. Heritability in the 20% to 50% range probably applies to the more common varieties of unipolar depression and to generalized anxiety. It also applies to individual differences in exposure to life circumstances such as poor parenting, divorce, or stressful life events. Effects on life circumstances arise because people's behavior shapes and selects their environment⁸ and influences other people's responses to them (and thereby influences their social environment). Insofar as such behavior is genetically influenced (which applies to all behaviors), it will bring about gene-environment correlations.⁹ It is often supposed that genetic effects plus environmental effects must sum to 100%, but they do not. It is necessary to take into account stochastic effects¹⁰ and developmental influences (possibly including endovariance is accounted for by the joint co-action of genetic and environmental effects. Ordinarily, this variance tends to be wholly attributed to genetic effects, but this is misleading because it involves environmental effects as well.

These and other considerations have implications for the concept of how genes exert their effects. Of course, even with single-gene disorders that involve no environmental liability, the genetic effects on behavior are indirect. Genes affect proteins, and only through the effects of these proteins on the functioning of the brain (or other aspects of the body) are there downstream effects on behavior. For reasons that remain poorly understood, there is often astonishing variation in phenotypic expression, as evident in the fragile X anomaly and tuberous sclerosis. With multifactorial disorders (most psychiatric disorders are of this type), other forms of indirectness need to be recognized. Thus, genetic influences may operate through temperamental dimensions rather than on disorders specifically. This probably applies to neuroticism as a risk factor for depressive and anxiety disorders, as well as sensation-seeking behavior as a risk factor for attention-deficit/hyperactivity disorder and a range of other conditions. Such dimensions may carry both risks and protective effects. For example, behavioral inhibition/emotional hyperreactivity is a risk factor for anxiety disorders but a protective factor against antisocial disorders. A simplistic subdivision of genes that are intrinsically bad and intrinsically good makes little sense.

Indirect genetic causal effects may also occur as a result of influences on individual differences in environmental risk exposure or susceptibility to risk environments (gene-environment correlations and interactions). In such cases, the proximal risk mechanism may involve environmental mediation despite the importance of genetic moderation. Genes affect normal variations in attitudes and behavior in addition to disorders, and such dimensional effects may play a major role in psychopathological risk.

Furthermore, in many cases the development of disorder may depend on the synergistic interaction among particular patterns of genes (epistasis) and not just on the cumulative additive effect of multiple susceptibility genes. This is indicated by the marked falloff in liability in moving from monozygotic to dizygotic co-twins and from first-degree to second-degree relatives, as illustrated by the pattern in autism.¹²

The situation with respect to environmental influences is comparable. The empirical evidence shows that environmental influences are strong and pervasive but rarely determinative.¹³ Geneticists have rightly emphasized that many studies of psychosocial factors misleadingly infer environmental mediation when their designs have failed to test for the possibility that the effects are, at least in part, genetically mediated.^{14 - 15} Developmentalists have reminded researchers that they need to use designs that can separate environmental effects on the person from the person's effects on the environment.^{16 - 17} Nevertheless, an extensive range of research designs can test for environmentally mediated risk effects.⁷ These include examination of the effects of measured environments on monozygotic pairs,^{18 - 20} the effects of the qualities of the adoptive home on within-individual change in late-adopted children,²¹ recovery (partial or complete) following removal from a severely depriving environment,^{22 - 23} dose-response effects with respect to the duration and severity of deprivation and subsequent psychological deficits²⁴ (M.R. and T. O'Connor, PhD, unpublished data, 2002), migration effects on the risk of schizophrenia in Afro-Caribbeans,⁷ and marked secular trends in rates of psychosocial disorder in young people²⁵ and traits such as intelligence^{26 - 27} and height.^{28 - 29} As a result of these and other research strategies, a substantial body of evidence points to environmentally mediated effects on behavior.¹³ Such effects are derived not only from families but also from peer groups,³⁰ communities,³¹ and schools.^{32 - 33}

However, studies of environmental effects have universally shown that there are huge individual differences in response, with some individuals severely affected and others with few sequelae. This has given rise to the concept of resilience: relative resistance to the ill effects of psychosocial adversity.^{34 - 35} This multifaceted phenomenon reflects a range of mechanisms operating before, during, and after risk exposure. Although the concept has been little explored up to now, it is clear that genetic influences are often implicated.³⁶ Just as environmental effects are often most apparent in genetically vulnerable individuals,⁹ resistance to environmental hazards may result from genetically influenced protective effects.

The third area of research, developmental influences, also has substantial empirical achievements to its credit, although much remains unknown about the underlying biological processes. Different phenomena reflect a variety of developmental mechanisms. First, there are age differences in response to some risk factors. This is well demonstrated with respect to the effects of unilateral brain injury on language and visuospatial functions.³⁷ Age effects are also probable in children's responses to hospital admission and other forms of stressful separation.³⁸ For reasons that are poorly understood, children and adolescents with major depressive disorders are less likely to show the beneficial response to tricyclic medication seen in most adults.³⁹ Second, there are marked age trends in the onset of particular disorders. For example, although depressive disorders have a low incidence and equal sex distribution before puberty, during adolescence they become much more common in girls, with an increasing female preponderance.⁴⁰ Evidence suggests that this is a result of genetic effects on both the liability and susceptibility to negative life events.^{41 - 43} It also demonstrates a psychopathological progression from anxiety to depressive disorders.⁴³

Other heterotypic psychopathological progressions, the third developmental feature, include those of hyperactivity to oppositional defiant disorder or conduct disorder, which reflects a substantial degree of shared genetic liability,^{44 - 46} and substance use/abuse to depression, which involves both environmentally mediated mechanisms and shared liability.⁴⁷ Severe developmental disorders of receptive language are frequently followed in adult life by serious impairments in social relationships⁴⁸ ; the mechanisms are unclear, but the pattern suggests that they reflect the basic underlying liability. Although evidence has accumulated on the frequency with which early neurodevelopmental impairment precedes the onset of schizophrenic psychoses in late adolescence and early adult life, controversy continues regarding whether it reflects genetic liability or some form of environmental insult.⁴⁹ The frequent development of epileptic seizures in late adolescence in individuals with autism,^{50 - 51} an atypical age of onset in both the general population and in people with mental retardation, remains a puzzle. This occurrence is not usually accompanied by any marked change in clinical conditions, but it must reflect underlying neuropathological and/or neurophysiological processes.

Fourth, several long-term longitudinal studies have shown a strong tendency for conduct disorders in childhood (and to a lesser extent, emotional disturbances) to be followed by an increased rate of psychosocial stresses and adversities in adult life ^{52 - 53} and an increased incidence of teenage parenthood.⁵⁴ In other words, adult environments are substantially predicted by childhood behavior. Nevertheless, evidence suggests that environments brought about by people's own actions can still have a major influence on their subsequent behavior.^{54 - 57}

Finally, studies of severely deprived children from institutions in Romania who have been adopted into well-functioning English families demonstrate that some show persistent cognitive deficits associated with reduced head circumference, implying an effect of deprivation on brain growth²² (and M.R. and T. O'Connor, PhD, unpublished data, 2002).

It has been argued that we will soon be awash with susceptibility genes for disorders.⁵⁸ One would hope that this is the case, but it seems unlikely. It has proved difficult to identify susceptibility genes for multifactorial disorders because of small effects, genetic heterogeneity, and unknown patterns of interplay with the environment, apart from the difficulties in phenotypic definition. It has become clear that the following guidelines are necessary: very large sample sizes (implying the need for collaborations and meta-analyses), high-quality standardized measures (both categorical and dimensional), focused hypotheses on possible genetic heterogeneity, multiple complementary strategies (linkage, association, and quantitative trait loci), and the study of causal hypotheses from dimensional risk to categorical disorder.^{1 ,59 - 60} However, despite its difficulty thus far, identifying susceptibility genes may turn out to be the(relatively) easy step. The great challenge is to discover the effects of genes on proteins and the causal pathways leading from such protein products to the clinical phenotype. Research methods should include animal models, focused biological (integrative physiological) studies, molecular epidemiological studies, and proteomics.⁵⁹

Although real progress has been made in the testing of environmental risk mediation hypotheses, much more is necessary to move from the identification of risk indicators to an understanding of causal risk processes. Genetically sensitive designs that are environmentally informative and a broader range of "natural experiments" are required.⁷ Distal risks, meaning those that merely foster other risks, must be differentiated from proximal risk processes, meaning those that directly and immediately cause psychopathological characteristics.³⁰ The identification of the causes of individual differences in response to psychosocial risk (identified as a key question 30 years ago⁶¹ )will have to be tackled directly rather than dismissed as the result of an unexplained constitutional factor. It has long been observed in both human and animal studies that stressful experiences can have either sensitizing or steeling effects (ie, increasing or decreasing susceptibility to later stresses). Why does this occur, and what mediating mechanisms are involved? Finally, there must be direct investigation of the genetic moderation of environmental susceptibility.

Although it is apparent that gene-environment correlations and interactions are pervasive and important, little is known about the underlying causal mechanisms by which they occur, and even less is known about their functional consequences for psychopathological effects. In addition, there is negligible understanding(with respect to mental disorder) of the role of epistatic effects, which cause the actions of one gene to be influenced by the operation of other genes. The pattern of family findings for some disorders (such as autism and schizophrenia) strongly suggests synergistic effects, but so far it has proved difficult to determine the specifics. The molecular genetic identification of individual susceptibility genes should help greatly. The role of epigenetic effects (the genetic processes involved in the expression of particular genes in individual cells¹¹ ) also has yet to be adequately understood.

Early childhood experiences can sometimes have psychopathological effects that extend into adult life.^{62 - 63} However, surprisingly little attention has been paid to the need to identify the factors within or outside the organism that mediate such long-term consequences. The possibilities include effects on cognitive sets or models, effects on interpersonal interactions, changes in the neuroendocrine system, and structural damage to the brain.⁶⁴ Kindling effects, by which the induction of psychopathological characteristics brings about internal changes that predispose the individual to persistence or recurrence of mental disorder, must also be considered.^{65 - 66} Early studies of visual function emphasized the need to consider experience-expectant biological programming, or the process in which normal somatic development depends on environmental input that may be expected in all ordinary circumstances.⁶⁷ In addition, however, it has become apparent that normal development adapts to the environmental circumstances prevailing during the sensitive period when the relevant somatic structures are being created: so-called experience-adaptive programming.^{1 ,25 ,68 - 69} Psychosocial research will have to become better integrated into biological studies to elucidate these or other mediating mechanisms.

Almost all causal research, both genetic and psychosocial, has focused on individual differences. Although this research involves an important set of causal questions, they are not the only ones. During the past half century, there has been a major rise in the rates of psychosocial disorders in young people, including suicide, drug use/abuse, and crime.²⁵ There have been similar dramatic changes in mean IQ,^{26 - 27} height,^{28 - 29} and age of menarche⁷⁰ as well as infant mortality and life expectancy rates. It is by no means self-evident that the causal factors responsible for these changes will be the same as those that underlie individual differences.

Similarly, there are well-established sex differences in psychopathological characteristics.⁵⁴ In general, there is a male preponderance for early neurodevelopmental disorders (such as autism, attention-deficit/hyperactivity disorder, and developmental disorders of language) but a female preponderance for adolescent-onset emotional conditions (such as depression and eating disorders).

In both cases, there is a need to test a range of competing hypotheses. These include the possibilities that the group differences are artifacts reflecting differences in pattern, ascertainment, or diagnosis; that they reflect differences in social context; that they reflect differences in susceptibility to risk factors; that the groups experience different types of risk factors; and that there are group-specific protective factors.

Reference has already been made to the increased rate of depression in adolescence⁴¹ and to the progression from anxiety to depressive disorders.⁴² In addition, there is a frequent progression from antisocial behavior to drug use/abuse to depressive disorder⁷¹ as well as the usual fall in rate of antisocial behavior in early adult life.³⁰ Only recently has research begun to investigate the various possible causal processes behind these changes.⁴⁷

There is a need to tackle causal questions in new ways; a focus on the interplay among nature, nurture, and development, with a real appreciation of multifactorial etiology, provides the means to do that. It will require the integration of basic and clinical neurosciences along with the use of epidemiological studies to test causal hypotheses.

Submitted for publication January 15, 2002; accepted May 2, 2002.

Presented in part at the Academy of Medical Sciences, London, England, November 20, 2001, to mark the award (shared with Solomon Snyder, MD) of the2001 Bernard and Rhoda Sarnat International Prize in Mental Health from the US Institute of Medicine, Washington, DC.

Corresponding author: Michael Rutter, MD, FRS, Box P080, SGDP Research Centre, Institute of Psychiatry, De Crespigny Park, Denmark Hill, London SE58AF, England (e-mail: j.wickham@iop.kcl.ac.uk).