Which of the following might influence the environment of identical twins and therefore complicate the interpretation of twin studies?

Twin Studies

Twin studies of diabetes have made impressive contributions to our understanding of the disease. The study of monozygotic twins of patients with diabetes by Barnett and coworkers74 contributed to the recognition of distinct forms of diabetes, initially termedjuvenile onset andadult onset, laterinsulin dependent andnon–insulin dependent, and now T1DM and T2DM. The concordance rates for monozygotic and dizygotic twins provide important information regarding genetic factors contributing to a given disease, because monozygotic twins share all germline-inherited polymorphisms or mutations, whereas dizygotic twins are similar to nontwin siblings of patients with a disease and have only half of their genes in common. For a locus that contributes to disease in a recessive manner, only one quarter of dizygotic twins would be homozygous to a sibling with diabetes at that locus, but all monozygotic twins would be homozygous for all recessive loci of their diabetic twin. As overall concordance rates of monozygotic twins for T1DM vary across studies, it is likely that T1DM is heterogeneous and that groups of monozygotic twins have different genetic causes for their diabetes. With such genetic heterogeneity, one would expect different concordance rates for different genetic causes.

Redondo and coworkers75 analyzed prospective follow-up data from a large series of initially discordant monozygotic twins from the United Kingdom combined with a series from the United States. Progression to diabetes was identical for both series of twins. There was no length of time of discordance beyond which a monozygotic twin of a patient with T1DM did not have a risk of developing the disease. Nevertheless, the hazard rate for development of diabetes decreased as the period of discordance increased. There was also a marked variation in the risk of diabetes relative to the age at which the disorder developed in the index twin. With long-term follow-up, the overall rate of concordance for monozygotic twins exceeds 50%.72 However, if T1DM developed in the index twin after age 25 years, the concordance rate by life table analysis in the study of Redondo and coworkers75 was less than 10%. If diabetes developed in the index twin before age 5 years, the concordance rate was 70% after 40 years of follow-up. Therefore, environmental factors, random factors, and non–germline-inherited variations (e.g., imprinting, T-cell receptor polymorphisms, somatic mutations) likely contribute to lifetime diabetes risk. Interestingly, studies of dizygotic twins suggest that their risk of diabetes may not differ from that of nontwin siblings or, at most, may be increased by a factor of two as compared with the 10-fold increase for monozygotic twins.76

Genetic factors influence not only the development of diabetes but also the expression of anti-islet autoantibodies. For identical twins, the expression of anti-islet autoantibodies is tightly linked to the eventual progression to overt diabetes, and, true to form, monozygotic twins have a highly concordant prevalence for the presence or absence of anti-islet autoantibodies. Dizygotic twins much less often exhibit concordant positivity for anti-islet autoantibodies, and the prevalence is similar to that of nontwin siblings.77

117.3.4 Twin Studies and Familial PD

Initial twin studies were fairly unambiguous about the lack of a genetic involvement in PD. A more recent twin study in a large cohort from the National Academy of Sciences/National Research Council World War II Veteran Twin Registry advocated the notion that—at least in those in whom PD starts before the sixth decade—genetic factors may play a role (36). Integration of modern imaging techniques identified preclinical loss of dopaminergic input into the striatum in monozygotic twins, initially classified as unaffected with an affected sibling, that are indicative of subclinical PD, demonstrating that twin studies likely underestimate genetic susceptibilities (37). Other recent studies have more closely examined familial aggregation of PD. The odds for any individual to develop PD are increased two to four times in the presence of a first-degree relative with PD (38–40). These odds further increase with increasing numbers of affected relatives and with younger affected relatives. Odds are higher in Caucasians than in African Americans or Hispanics, and they are twice as high for affected male relatives compared with females (39). In general, familial studies and twin studies have supported a much stronger role for genetic factors in individuals with the onset prior to age 50 (e.g. the early-onset PD), but genetic factors could not easily be ruled out in the late-onset disease.

Disadvantages of Twin Studies of Pain Heritability

•

Twin studies may provide estimates of heritability but do not in themselves reveal the genes responsible. In complex disorders involving genes with small effects, one needs association studies, whose optimal design may be different. (However, one may certainly genotype DZ twins and use their data in association studies.)

Twin studies are impractical for the study of phenotypes elicited only by an uncommon disease or traumatic event that is unlikely to occur in both twins, such as post-surgical pain syndromes, post-herpetic neuralgia, or spinal disc herniation with spinal root pain. Consider, for example, the study listed inTable 10-3 for sciatica (Heikkila et al 1989). More than 9000 twin pairs were studied with a survey and examination of hospital records. The 269 pairs in which both subjects reported a medical diagnosis of sciatica at any time in their life was sufficient to compare concordance in MZ and DZ twins and estimate heritability as 20%. However, if the more severe criterion of hospital admission was used, only eight pairs had both suffered sciatica, which is insufficient for a comparison. Furthermore, patient recall of hospital admission for sciatica after 1 or 2 decades was often inaccurate. Forty-one percent of patients 60 years and older who had a hospital discharge summary of sciatica did not recall this at the time of the survey.

In the twin studies of clinical pain disorders listed inTable 10-3, one cannot distinguish whether the heritability component reflects genetic effects on development of the structural disease causing pain, on processing of pain from a given lesion, or both. This can be sorted out only if one can measure the magnitude of structural injury at the time that the pain is occurring, which is particularly difficult if the disorder strikes members of the twin pair at different times. It may be possible to address these issues in twin registries that include access to hospital records and imaging studies of osteoarthritis, spinal degeneration, or other disorders routinely assessed with high-resolution imaging methods.

Critics of the classic MZ–DZ design point out a flaw in the model’s assumption that MZ twins and DZ twins have similar variance in their environment. Because MZ twins look so much like each other, people are likely to treat them similarly in many ways; for example, parents often dress them identically. Such environmentally driven inflation of the VDZ term will spuriously increase the value ofh2 in the equation presented earlier.

Twin studies

Twin studies are studies that typically compare identical (monozygotic) to fraternal (dizygotic) twins. Identical twins are the only human beings alive who are genetically identical. Fraternal twins are genetically siblings only; they carry similar genes, as siblings do, but not identical genes as identical twins do. Therefore, if a behavior is found more common in both identical twins and less commonly in both fraternal twins, then one conclusion is that that behavior may be genetically heritable. This is called ‘concordance.’ Hudziak and his colleagues, who studied 492 twin pairs (comparing identical to fraternal twins), completed one classic twin study on aggression. The researchers surveyed the twins' parents about a broad range of behaviors, including aggression, and then cross-checked to see if aggressive behavior was more commonly found in both identical twins then in both fraternal twins. Indeed, that was precisely what they found, and the researchers estimated that, in their sample, genetics accounted for 70–77% of the variance in aggression. Another study of twins and triplets found an astonishing 96% heritability among 9- and 10-year olds for antisocial and aggressive behaviors (based on behavior ratings from the subjects themselves, their teachers, and their caregivers). Such a finding suggests a strong genetic influence on aggression.

Another twin study compared 183 monozygotic (identical) with 64 dizygotic (fraternal) twins. They found not only that individual measures of aggression were heritable – that is, more commonly found in both identical twins – but also that the same genotype might produce a variety of aggressive behaviors. Twin studies are compelling, but they suffer from a common flaw: not only do identical twins share genes, but also they often share an environment that treats them similarly. Because of this, identical twins are not the perfect genetics–environment experiment. Another method, which seeks more fully to separate genetics from psychological environment, is the method known as adoption studies.

Twin Studies

Comparison of monozygotic to dizygotic twins and/or plain siblings reared in virtually identical environments have been informative.617,618 Goldfarb and coworkers sampled dizygotic and monozygotic twins from the Vietnam Era Twin Registry and found a concordance rate of 32% in monozygotic twins compared with 17% in dizygotic twins, an effect that cannot be explained by the documentable dietary information.619 Other twin studies obtained urinary chemistries and found the hereditability of urinary calcium excretion rate to be around 50%.617,618

8.1 Overview

Twin studies are helping to shed some light on the correlations of DNA methylation variants with disease phenotypes that have been reported. They also help to improve our understanding of the nature of DNA methylation variation in general. Thus, twin studies represent an important subset of human DNA methylation studies as they provide experimental design advantages otherwise impossible to achieve with human subjects. Monozygotic twins in particular are an ideal system to study interindividual variation in DNA methylation for several reasons. First, there are no “genetic background effects,” as the two individual twins share the same DNA sequence. Second, twin pairs also share other factors that usually must be controlled in DNA methylation studies, such as age and sex. Third, most twins are raised in similar environments, thus minimizing early life environmental effects and allowing us to potentially identify late life environmental differences that might explain variability among twins [121,122]. Given these characteristics, monozygotic twins present the best opportunity for well-controlled studies of DNA methylation in humans.

Twin Studies and the Epigenetics of Cardiovascular Disease

Twin studies show a heritability of 40–60% for risk of coronary heart disease mortality [41,42]. A number of environmental factors, such as smoking, obesity, lack of exercise, and alcohol abuse, have been identified to increase risk of coronary heart disease. However, the genetic factors that increase the risk of coronary heart disease remain poorly understood.

Cardiovascular disease risk factors such as nutrition, smoking, pollution, and stress have been correlated with modification of epigenetic marks [43]. Low birth weight is a well-recognized risk factor for cardiovascular disease [44]. A possible contribution of epigenetic memory to this risk factor is suggested by twin studies that found up to 60% discordance in methylation profiles between the heaviest and lightest cotwins [16]. Significantly, the genes whose methylation was highly correlated with birth weight showed enrichment for functions and pathways associated with growth, metabolism, and cardiovascular disease.

Taken together, these findings support a role for environmental factors that act through epigenetics to influence cardiovascular disease. Further detailed studies will be required to determine which epigenetic markers are most relevant to the disease state, both as biomarkers and as potential targets for therapeutic intervention.

Twin Studies

Twin and adoption studies are the two principal approaches for distinguishing genetic and environmental contributions to MDD. The twin study requires affected probands from both pairs of monozygotic (MD) and dizygotic twins, and the presence (concordance) or absence (discordance) of the disorder in the cotwins of these affected probands is assessed. Concordance and discordance proportions are used to determine the relative contributions of genetic and environmental factors in the development of illness, which are conceptualized as genes, shared environment (that is, environmental factors that make twins more alike—socioeconomic status, for example), and unique environment (that is, nonshared environmental factors that makes twins dissimilar).

In an adult study, heritability computed from data from five twin studies (three studies conducted in the community, one study done in a clinical setting, and one study using population-based samples both in the community and a clinical setting) was 37% (95% confidence interval (CI) = 31–42%) (Sullivan et al., 2000). On the contrary, heritability of depressive symptoms in youth varies across studies (Rice et al., 2002a; Thapar & McGuffin, 1994), depending on who rated the symptoms and what measurements were selected. In the review by Rice et al. (2002a), heritability estimates for parent-rated depressive symptoms varied from 30% to 80%, while lower heritability estimates were found for self-reported depressive symptoms. Despite these inconsistent results, the only consistent finding from several twin studies is that the genetic factors becoming more important from childhood to adolescence (Rice, Harold, & Thapar, 2002b; Thapar & McGuffin, 1994), which might be due to uncovered gene–environment correlation, gene–environment interaction, or additive genetic influence in adolescence (Eley & Stevenson, 1999). Additionally, two studies from the national Swedish Twin Registry suggested higher heritability in women than in men (40% versus 30%: PMID11352363 and 42% versus 29%; Kendler, Gatz, Gardner, & Pedersen, 2006), and clear evidence was found for sex-specific genetic effects with genetic correlations estimated at +0.55 and +0.63.

14.10.3 Nonshared Environment and Chronic Disease

Twins studies have made a tremendous contribution to the understanding of the causes of chronic disorders, from cancers to neurodevelopmental conditions such as schizophrenia, allergies, and cardiometabolic disorders such as cardiovascular disease [7,173,204,226,227]. Such disorders originate in very early life [228–230] and like the epigenetic state itself, are mainly influenced by the nonshared environment [231]. This finding was initially surprising because maternal diet and lifestyle had been assumed to be influential on the developing fetus; however, multiple nonshared factors have now been shown to influence risk for chronic disease [116,173,232–234]. Most of these factors are associated with the “fetoplacental unit” [235]—the placenta, cords, and fetus. Such factors include uterine implantation site, the placental location at which the umbilical cord is inserted, and the physical characteristics of the umbilical cord such as length, width, and torsion. All these factors have the potential to affect the growth rate of individual twins via the transplacental transport of oxygen, nutrients, and teratogens.

MZ twins can be discordant for inflammatory state [236–240], which is a major contributor to chronic disease risk [241,242]. Twin studies have shown that the nonshared environment dominates as the largest component of variance of immune factors in adults [243]. Prenatal inflammation can occur in the umbilical cord or placenta of a single MZ twin [237,244] and soluble inflammatory factors can pass to the associated fetus [245]. More longitudinal twin birth cohorts are required to improve our understanding of the way the nonshared environment influences health and disease and its epigenetic mediators.

Twin studies

Twin studies are conducted on MZ (identical) and DZ (fraternal) twins. They aim to reveal the importance of environmental and genetic influences for traits, phenotypes, and disorders—if done with care and caution. The major premises of twin studies are: (1) MZ twins share 100% of their genes, while DZ twins share only about 50% (on average) of their genes, and (2) MZ and DZ twins share 100% of their environment. Hence, any difference in disease status between MZ and DZ twins would be likely to be attributable to the difference in shared genes.

While twin studies have been used extensively in human behavioral studies to estimate the contributions of genetic factors to phenotypic variability, the methodology that allows the estimation entails several crucial assumptions. The most critical is that MZ and DZ twins are equally similar environmentally [42,43]. Although MZ twins are genetically more similar than DZ twins, they are often environmentally more similar. It has been shown that, even in the complete absence of any genetic factor and of any biases, the greater environmental similarity alone in MZ twins can result in a higher concordance rate in MZ twins than in DZ twins [44]. This is especially true when there are multiple environmental factors, which may have multiple exposure levels and/or interact strongly, although each of them may be of low risk [44].

In two twin studies that reported the heritability for endometriosis as being in the range of 47%–51% [45,46], which have been widely cited [8,47], both assumed that MZ and DZ twins are equally similar environmentally. In addition, additive genetic action was also assumed [45,46].

This review is not to discredit the published twin studies in endometriosis. Rather, by enumerating explicitly all assumptions and limitations of the statistical methodology, we can better understand what the heritability truly means and what its limitations are. All in all, we cannot rely uncritically on twin studies without checking the validity of the assumptions underlying the methodology.