The power of the classical twin study

Anxiety disorders and neuroticism: are there genetic factors specific to panic?

Data from 2,903 adult same-sex twin pairs were analysed to investigate whether the genetic determinants of symptoms of panic are different from those underlying the neuroticism personality trait. Our results suggest that much of the genetic variation influencing the physical symptoms associated with panic is of the nonadditive type, perhaps due to dominance or epistasis. In both sexes these nonadditive genetic effects on physical symptoms influence the reporting of "feelings of panic". In males they also account for as much as half the genetic variance in neuroticism. The remainder is additive and also accounts for the balance of genetic variation in "feelings of panic". In females genetic variance in neuroticism is entirely additive but is not an important source of covariation with either panic symptom. Thus, symptoms of panic seem to be shaped in part by unique genetic influences which do not affect other anxiety symptoms. That a substantial part of the genetic variance in neuroticism in males may be due to the nonadditive effects on physical symptoms of panic may help to explain the rather low correlation between the genetic influences found to affect neuroticism in males and their counterparts in females.

Panic disorder: evidence for genetic vulnerability

Many studies have suggested that a genetic predisposition to the development of panic disorder exists. These studies are examined and their limitations discussed. It is suggested that only by the analysis of comprehensive family and twin data, coupled with other measures such as the search for possible single gene association or linkage and study of the children of panic disorder patients, will the mechanism for the 'familiarity' noted in panic disorder patients be elucidated. Delineation of the mode of transmission of panic disorder may allow preventative intervention with those at risk before they develop panic.

Genetic influence on the serum levels of naturally occurring human IgG antibodies to dietary antigens. Quantitative assessment from a twin study

Serum IgG antibodies to ovalbumin (OA) and beta-lactoglobulin (BLG) were quantified by ELISA techniques in 22 monozygotic (MZ) and 24 dizygotic (DZ) healthy twin pairs. Antibody levels were comparable in the MZ and DZ groups both for anti-OA and anti-BLG antibodies. The genetic variance (GWT) was 0.167 for log IgG anti-OA antibodies, and 0.173 for log IgG anti-BLG antibodies, with heritability estimates of 0.44 and 0.37, respectively. No indication was observed of genotype-environmental interaction or differential environmental covariance for the log antibody levels in the MZ and DZ twins. The anti-OA and anti-BLG antibody levels in the same individual correlated only to a low degree. The levels of naturally occurring serum IgG antibodies are significantly influenced by genetic factors.

Prospects for detecting genotype X environment interactions in twins with breast cancer

We consider a study of MZ and DZ twin pairs ascertained because one or both twins have a disease. Genotypes at a major locus are known and putative environmental risk factors have been measured for all individuals. The power of the study to estimate the effect on liability of the measured and residual genetic and environmental effects (Gm, Gr, Em, Er) and all two-way interactions between them (except Gr X Er) is estimated by simulation. If liabilities can be indexed on a continuous scale (eg, blood pressure as an index of liability to hypertension), then a study of 600 MZ and 600 DZ pairs would have sufficient power to detect quite subtle interaction effects, even if ascertainment is greatly biased toward MZ twins. If liabilities cannot be measured and only affection status is known, then the power of the study would be much lower, although not impracticably so. There appears to be no advantage in augmenting the twins with a sample of control individuals who have been drawn at random from the population regardless of disease status, at least for the case we have considered in which the disease threshold on the liability scale is assumed to be known without error. The argument is developed in terms of the utility of the design for research into breast cancer.

Genetic covariation between neuroticism and the symptoms of anxiety and depression

A genetic analysis of the trait of neuroticism and symptoms of anxiety and depression in 3,810 pairs of adult MZ and DZ twins is reported. Differences between people in these measures can be explained simply by differences in their genes and in their individual environmental experiences. There is no evidence that environmental experiences that are shared by cotwins, such as common family environment or social influences, are important. There are differences between the sexes in gene action affecting neuroticism, and genetic effects become more pronounced with age in females. The lack of evidence for dominance variance affecting neuroticism contrasts well with the detection of considerable genetical nonadditivity for extraversion in the same sample and reinforces the view that these two traits are not only statistically, but also genetically, quite independent. An analysis of the causes of covariation between anxiety, depression, and neuroticism shows that additive gene effects are more important causes of covariation than environmental factors. Genetic variation in symptoms of anxiety and depression is largely dependent on the same factors as effect the neuroticism trait. However, there is also evidence for genetic variation specific to depression.

Education policy and the heritability of educational attainment

Many workers assume that genetically determined differences in intellectual ability will be influenced little by changes in educational policy or other environmental interventions. Others, however, have suggested that increasing equality of educational opportunity will lead to an increase in the heritability of educational attainment. The resolution of this issue has been delayed until now because of the extremely large sample sizes which would be required. Education data on twins and their parents, from the Norwegian twin panel, provide a unique opportunity to determine the impact on the heritability of educational attainment of the more liberal social and educational policies introduced in Norway after the Second World War. As reported here, for individuals born before 1940 there is a strong effect of family background on educational attainment, accounting for 47% of the variance, though genetic factors account for an additional 41% of the variance. For females born after 1940 and before 1961, the relative importance of genetic (38-45%) and familial environmental (41-50%) differences changes very little. For males born during the same period, the broad heritability of educational attainment has increased substantially (67-74%), and the environmental impact of family background has correspondingly decreased (8-10%). For males, at least, having well-educated parents no longer predicts educational success, as measured by duration of education, independent of the individual's own innate abilities.

Causes of variation in drinking habits in a large twin sample

A genetic analysis of alcohol consumption in 3810 pairs of adult twins is reported. When no correction was made for age, individual environmental variance, including non-repeatable errors of reporting, accounted for approximately 44% of variation in both sexes. In females, there was no evidence of shared environmental effects and 56% of the variance was genetic in origin. In males, only 36% of the variance was genetic and common environmental effects accounted for the remaining 20% of individual differences. For females, the results for younger (30 years and under) and older (over 30) twins were similar. For males, however, the effect of age was striking. In younger male twins over 60% of the variance was genetic in origin, with the remaining variance due to environmental influences unique to the individual. In older twins genetic differences do not appear to be important, with approximately 50% of the total variance due to individual environmental differences and the remaining 50% due to the effect of the common family environment. Our results suggest that both age and sex need to be considered when analysing the causes of variation in alcohol consumption.

Genetic and environmental causes of variation in renal tubular handling of sodium and potassium: a twin study

We have conducted a study of renal sodium and potassium reabsorption in 205 pairs of twins on freely chosen diets; 89 of the subjects were studied on more than one occasion. Renal tubular sodium and potassium handling, as measured by the fractional excretions FENa and FEK, show repeatable differences between individuals. Siblings (in this case monozygotic and dizygotic pairs of twins) are more alike in this respect than unrelated individuals. Comparison of monozygotic and dizygotic twin pairs indicates that genetic, rather than shared environmental, factors are probably responsible for this similarity, with heritability estimates of 0.5 for sodium and 0.6 for potassium. There are indications of sex differences in the sizes of the genetic and environmental effects for both variables and indications that the genetic effects may be qualitatively different for FEK. Such findings need further investigation.

The effects of inheritance on constituents of plasma: a twin study on some biochemical variables

Heritability and within-person repeatability of thirteen constituents of plasma were assessed in a study of 206 pairs of male and female twins. Repeat measurements were available on 44 pairs. For bilirubin, calcium, creatinine, phosphate and potassium, the individuality of these characteristics was genetic in origin. Total protein, albumin and globulin showed significant heritability but considerable variation between occasions, while bicarbonate, chloride, iron, sodium and urea showed mainly environmental effects.

Plasma lipids in twins. Environmental and genetic influences

A study on 205 pairs of male and females twins, aged from 18 to 34 years. showed significant heritabilities for total and high density lipoprotein cholesterol and for triglycerides. Significant effects of shared environment were also found for total and HDL cholesterol, possibly to a greater extent in women than in men. Triglycerides showed greater variance in men but a model specifying different sized environmental and genetic parameters in the two sexes gave a good fit and indicated that the factors influencing plasma triglycerides are the same in men and women although the effects they produce are scaled differently.

Genetic and environmental influences on body measurements of Belgian twins

A study of 100 MZ and 67 DZ twin pairs aged 18 to 25 years has shown a highly significant genetic contribution to the following measurements: standing and sitting height; height on tragus; arm length; biacromial, biiliac, and bitrochanteric diameter; weight; circumference of thigh and of upper arm relaxed and contracted; and head length and breadth. The twins were brought up together and of the same socioeconomical, geographical, ethnical, and cultural origin. Zygosity diagnosis was based on 22 to 26 blood groups. The means of the measurements were smaller in MZ than in DZ twins, some of them significantly so. The intraclass correlation coefficients of the MZ cotwins were all significant and greater than those of the DZ cotwins. In families of same geographical origin, sib-sib correlations were somewhat smaller than those of the DZ cotwins, but for the three diameters of the body the order was reversed.

Evidence for directional non-additivity in the genetics of finger ridge counts

A genetical analysis of variation in finger ridge counts of 221 pairs of twins and 80 pairs of opposite sex siblings has been carried out. Negative regression of DZ and sibling pair variances on pair means suggests the action of non-additive genes or unequal gene frequencies tending to increase finger ridge counts. Negative skewness of the distributions supports this view. While models including dominance or epistasis are not a significant improvement over purely additive genetic models, it is regarded as important that large and positive values of non-additive genetic variance are estimated. The evolutionary importance of dominance and epistasis for greater finger ridge counts is discussed.

A twin study of skin reflectance

Skin colour has been measured by reflectance spectrophotometry on 134 pairs of twins at three sites, forehead, forearm and upper arm, each at three wavelengths, 425, 545 and 685 nm. Tanning is measured most reliably at 685 nm and at this wavelength the heritability is high at the least exposed upper arm site, intermediate on the forearm, while on the forehead variation is entirely environmentally determined. The same gradient is observed, through less strikingly at 545 nm, but at 425 nm, where haemoglobin is reflecting most of the light, the degree of genetic determination is the same at all sites.

The advantages of analysing human variation using twins and twin half-sibs and cousins

The "twin family" design is a new strategy for studying quantitative characters in man which overcomes most limitations of earlier designs and which is readily accessible to existing twin research units as it uses only adult twins (both identical and non-identical) and their spouses and offspring and juvenile twins (both identical and non-identical) and their parents. The design yields all information inherent in ordinary twin studies but also permits the simultaneous estimation of more components than any other design. Tests for most genetical and environmental components of variation and the assumptions of the design are provided. Particular advantages are the unambiguous separation of sex-linkage and maternal inheritance, the analysis of the mechanism of assortative mating and the specification of more realistic environmental models. Although several components are confounded the biases are not seriously misleading. However, it would be necessary to include adoption data to resolve the effects of cultural transmission which are otherwise confounded with the family environment. Nevertheless, this design provides a wealth of data on a diversity of relationships and promises to be a valuable tool for the analysis of individual differences in man.

Intrapair similarity in frequency of disfluency in monozygotic and dizygotic twin pairs containing stutterers

It is known that concordance for stuttering is higher in monozygotic (MZ) than in dizygotic (DZ) twins. Evidence for genetic contribution to the frequency of speech disruption in stutterers was examined in 17 pairs of MZ and 13 pairs of DZ twins, each pair containing at least one stutterer. Intraclass correlations and analyses of variance indicated positive evidence for a genetic contribution to the overall frequency of disfluency in speech and to the frequency of certain types of speech disruption (blocked and prolonged sounds). There was little evidence for genetic influence on other types of speech disruption (various types of repetitions and interjections) In those disfluency types for which genetic predictions were confirmed, evidence of excessive dissimilarity in DZ cotwins suggests that a simple, additive gene-environment model is unlikely to be appropriate for disfluency frequency data.

The monozygotic twin half-sib method for analysing maternal effects and sex-linkage in humans

The influence of maternal effects and sex-linked loci upon the statistics derived from an analysis of the families of monozygotic twins (the so called MZ half-sib design of Nance and Corey, 1976) is examined. If, as in the published analyses, the progeny sexes are considered jointly, maternal effects and sex-linkage may be confused. In the absence of sex-linkage a comparison of maternal and paternal half-sib covariances will often be a more powerful test for maternal effects than the conventional comparison of maternal and paternal parent-offspring covariances. If the progeny sexes are considered separately the analysis of sex-linkage presents no fundamental problems.

Sex differences in the inheritance of some anthropometric characters in twins

Biometrical genetical techniques have been applied to the analysis of certain anthropometric characters measured in 134 pairs of adult twins. After allowing for assortative mating it appears that there is a family environment (E2) component for variation in height larger than previously reported. "Fatness" traits - weight, ponderal index, and skinfold thickness - all show higher heritabilities in males and substantial E2 components in females, and reasons for this are discussed. The same is true for cephalic index and forearm length but the reason for these differences is not so obvious. Head length shows a much higher heritability than head breadth. A larger sample of DZ opposite-sex pairs would allow more powerful discrimination, but the variety of patterns of variation revealed by the model-fitting approach used here justify its use over more traditional techniques.

Model-fitting approaches to the analysis of human behaviour

Model-fitting methods are now prominent in the analysis of human behavioural variation. Various ways of specifying models have been proposed. These are identical in their simplest form but differ in the emphasis given to more subtle sources of variation. The biometrical genetical approach allows flexibility in the specification of non-additive factors. Given additivity, the approach of path analysis may be used to specify several environmental models in the presence of assortative mating. In many cases the methods should yield identical conclusions. Several statistical methods have been proposed for parameter estimation and hypothesis testing. The most suitable rely on the method of maximum likelihood for the estimation of variance and covariance components. Any multifactorial model can be formulated in these terms. The choice of method will depend chiefly on the design of the experiment and the ease with which a data summary can be obtained without significant loss of information. Examples are given in which the causes of variation show different degrees of detectable complexity. A variety of experimental designs yield behavioural data which illustrate the contribution of additive and non-additive genetical effects, the mating system, sibling and cultural effects, the interaction of genetical effects with age and sex. The discrimination between alternative hypotheses is often difficult. The extension of the approach to the analysis of multiple measurements and discontinuous traits is considered.