SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References
  8. Appendices

Objective To determine the relative importance of genetic effects on birthweight, gestational length and small for gestational age.

Design A cohort study, using individual record linkage between the population-based Swedish Twin and Birth Registers to estimate twin similarities in twins with known zygosity.

Population Included were 868 monozygotic and 1141 dizygotic female twin pairs, born in Sweden before 1959, who both delivered single births from 1973–1993.

Methods Quantitative genetic methods, offspring birthweight, gestational length and small for gestational age birth in twin sisters.

Main outcome measures Twin similarities measured as probandwise concordance rates and intra-class correlations for birthweight, gestational length and small for gestational age births.

Results Concordance rates and intra-class correlations for birthweight, gestational length and small for gestational age were consistently higher in monozygotic compared with dizygotic twins. Model fitting suggested heritability estimates in the range from 25% to 40%.

Conclusions This study suggests genetic effects not only for birthweight and fetal growth, but also for gestational length. The mediation of these genetic effects may partly be due to similarities in maternal antropometric measures, lifestyle and medical complications during pregnancy. The study does not distinguish between fetal and maternal genetic effects.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References
  8. Appendices

Genetic factors are thought to influence birthweight primarily through fetal growth1,2. A strong genetic influence on birthweight after controlling for gestational age was shown in a study of the offspring of twin pairs3, and familial effects on fetal growth have been suggested in intergenerational studies4,5. However, contradictory results concluding that genetic factors play a small part in determining fetal growth have also been presented6.

The existence of a genetic or familial tendency to repeat gestational length seems even less certain. Magnus et al.7 found a low correlation coefficient for gestational length across generations, and others have found no increased risk for pre-term birth among mothers themselves born pre-term1,8. However, intergenerational studies may give spurious results due to secular changes in accuracy of estimating gestational length, obstetrical practice or lifestyle.

The population-based Twin and Birth Registers in Sweden provide opportunities to compare twin similarities for reproductive outcomes of twin pairs with known zygosity, which is a highly valid method of evaluating the presence of genetic effects. By comparing twin similarities between twin mothers of different zygosity, we estimate the relative importance of genetic and environmental effects for birthweight, gestational length and birthweight by gestational length, analysed both as dichotomous and continuous variables.

METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References
  8. Appendices

The study base was derived through linkage of the Swedish Twin Register, including data on all twins born in Sweden from 1926–19589, and the Swedish Birth Register, covering more than 99% of all births in Sweden since 197310. Use of the unique personal identification number assigned to each Swedish resident permitted linkage of information about individuals across these registers11. The Twin Register includes 6097 female pairs, born from 1926–1958. Zygosity was determined on the basis of childhood resemblance9 and has been verified with serological data in a subsample including 200 pairs (correctly verified for 99% of the monozygotic and 92% of the dizygotic twins)12.

The Birth Register includes prospectively collected information on demographic characteristics, reproductive history and complications during pregnancy, delivery and the neonatal period.

This study includes female twin pairs with known zygosity, born in Sweden before 1959, who both delivered single births in Sweden during the years 1973–1993. As the focus of the investigation was on genetic effects on birthweight, gestational length and fetal growth among nonanomalous infants, all pregnancies with malformed infants were excluded (International Classification of Diseases, eighth and ninth revision, codes 740–759). A total of 2009 female twin pairs were eligible, of whom 868 were monozygotic and 1141 were dizygotic.

Dependent variables

Birthweight, gestational length and birthweight for gestational length were analysed as dichotomised as well as continuous variables. Low birthweight was defined as a birthweight below 2500 g. Pre-term birth was defined as a birth occurring before 37 completed gestational weeks. Small for gestational age was defined as a weight more than two standard deviations below the mean weight for the gestational age according to the gender specific, ultrasound based Swedish fetal weight curves13. A woman was considered as a low birthweight mother if she had at least one low birthweight infant, and the same applies for pre-term birth and small for gestational age. To estimate gestational length, ultrasound examination was used when available, otherwise the last menstrual period was used. Ultrasound examination no later than 18 completed weeks became increasingly common during the study period14.

When continuous variables were used, birthweight was measured in grams, gestational length in days and fetal growth in standard deviations from the gender specific fetal weight curves. The continuous variables were controlled for parity and standardised; values three standard deviations or more above or below the mean values were excluded. If a woman had more than one pregnancy, the mean value of all outcomes was used. In order to control for a possibly confounding effect of different parity within the pair, all analyses were also performed on primiparous twin mothers only. Mothers' birth years ranged from 1928 to 1958 (1953 was most common) and infants' birth year ranged from 1973 to 1993 (1973 was most common). There were no differences between monozygotic and dizygotic twin mothers with regard to mothers' and infants' year of birth. The parity distribution and the proportion of women included once, twice, or more than twice, were also similar among monozygotic and dizygotic twins (data not shown). Thus, the results should not have been confounded by maternal age, parity or number of included births.

Analyses

Twin similarity

Twin similarity was measured as probandwise concordance and tetrachoric correlations for dichotomous variables and as intra-class correlations for the continuous variables. Probandwise concordance is calculated as the number of affected twins (for example with a pre-term birth) in pairs divided by the total number of affected twins with the same zygosity15. Probandwise concordance thus represents the probability of a sister having the gestational outcome in question if her sister had the same outcome. Intra-class correlations measure the degree of within pair similarity (of for example gestational age) as compared with between pair similarity.

Genetic and environmental effects.

The importance of genetic effects is indicated by a higher twin similarity (concordance, correlation) among monozygotic than dizygotic pairs. Shared environmental effects reflect twin similarity that is not explained by genetic effects, for example shared childhood experiences (e.g. diet), whereas nonshared environmental effects are evidenced by within pair differences. The magnitude of heritability and shared and nonshared environmental effects can be estimated and the significance of effects can be formally tested by structural equation modeling16. Model fitting were performed on dichotomised as well as continuous variables. Details of the methods used are described in Appendix 1.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References
  8. Appendices

The rate of mothers with low birthweight infants was 5.1%, while the rates of pre-term birth mothers and small for gestational age mothers were 7.7% and 5.6%, respectively. These rates are thus based on the number of women, not on the number of pregnancies. Corresponding figures in the Swedish Birth Register for the study period are: mothers with a low birthweight infant 5.4%, pre-term birth 6.5% and small for gestational age birth 6.3%.

Table 1 presents the numbers of concordant and discordant pairs for each zygosity group for low birth-weight, pre-term birth and small for gestational age. Among monozygotic twins of whom at least one twin in a pair had a low birthweight infant (n= 90), 18 (9 pairs) also had a sister with a low birthweight infant (probandwise concordance rate = 0.20). Among 114 dizygotic twins giving birth to a low birthweight infant, eight had a twin sister with the same outcome (probandwise concordance rate = 0.07). The correlations of liability for low birthweight were 0.42 and 0.09 for monozygotic and dizygotic pairs, respectively.

Table 1.  Number of concordant and discordant twin pairs, probandwise concordance and correlation for low birthweight, preterm birth and small for gestational age. Female twin pairs born before 1959, who both gave birth in Sweden during 1973–1993.
 No. of pairs  
ZygosityConcordant affectedDiscordantConcordant unaffectedProbandwise concordance rateCorrelation*
  1. *Tetrachoric correlation.

Low birthweight     
Monozygotic9727870.200.42
Dizygotic410610310.070.09
Preterm birth     
Monozygotic14977540.220.40
Dizygotic101649600.110.09
Small for gestational age     
Monozygotic8747830.180.37
Dizygotic612410040.090.11

For pre-term birth, the concordance rates were 0.22 for monozygotic twins and 0.11 for dizygotic twins, while corresponding concordance rates for small for gestational age were 0.18 and 0.09, respectively. Correlations of liability for pre-term birth and small for gestational age were also higher among monozygotic compared with dizygotic twins. Thus, the results indicate a genetic component in all the studied variables.

As the incidence of small for gestational age is higher in pre-term compared with term births17, twin similarity was also calculated for pre-term births of nonsmall for gestational age infants, with similar results (correlations were 0.37 for monozygotic and 0.07 for dizygotic twins).

Table 2 presents the intra-class correlations for monozygotic and dizygotic twins when continuous variables were used. Correlations for birthweight, gestational length and fetal growth rate are presented separately for each zygosity group. To evaluate possible effects of parity, estimates are presented for all women as well as for primiparas separately. The correlation for birthweight in offspring of monozygotic twins was 0.46, while the corresponding figure for dizygotic twins was 0.23. If only primiparas were considered, the corresponding correlations were 0.41 and 0.16, respectively. For gestational length and fetal growth rate the higher correlations among monozygotic compared with dizygotic twins also indicate genetic effects.

Table 2.  Number of monozygotic and dizygotic twin pairs and correlation by zygosity
 MonozygoticDizygotic
MeasuresNo. of pairsCorrelation*No. of pairsCorrelation*
  1. *Intra-class correlation.

Birthweight8540.4611300.23
Primiparae5250.416610.16
Gestational age8500.3112250.13
Primiparae5150.256440.11
Fetal growth8520.4411250.25
Primiparae5270.316650.18

In the quantitative genetic analyses presented in Table 3, the results for both dichotomous and continuous variables are presented. For low birthweight, the estimates of heritability is 37%. When birthweight was analysed as a continuous variable, estimates did not change noticeably, but since there is more power in analyses of continuous than in dichotomous variables the heritability estimate is significant and confidence intervals are more narrow. The estimates also remain in the same range if only primiparous twin pairs are considered. For pre-term birth, heritability is estimated at 36%. When gestational length is used as a continuous variable, the heritability estimate remained essentially unchanged. Fit-of-model for pre-term birth is adequate, whereas the models using gestational length as a continuous variable only has a modest fit. For small for gestational age, the heritability estimate was 34%. If fetal growth rate is considered as a continuous variable, the estimates are again very similar, and the heritability estimate is significant.

Table 3.  Parameter estimates and 95% confidence intervals (CI) from quantitative genetic analyses of genetic and environmental effects for different gestational outcomes, measured as both dichotomous and continuous variables. SD = standard deviation.
 Parameter estimate (95% CI)Fit of model
MeasuresHeritabilityShared environmentNonshared environmentχ2dfP
  1. *Measured as continuous variables.

Low birthweight (< 2500 g)0.37 (0.0–0.56)0 (0.0–0.32)0.63 (0.44–0.83)3.6730.30
Birthweight*0.42 (0.27–0.49)0.03 (0.0–0.14)0.56 (0.51–0.61)4.4430.22
Primiparae0.39 (0.27–0.45)0 (0.0–0.09)0.61 (0.55–0.68)2.5730.46
Preterm birth0.36 (0.03–0.51)0 (0.0–0.23)0.64 (0.49–0.82)3.3030.35
(< 37 gestational weeks)      
Gestational age*0.31 (0.2–0.36)0 (0.0–0.07)0.69 (0.64–0.75)9.3330.02
Primiparae0.25 (0.09–0.32)0 (0.0–0.12)0.75 (0.68–0.83)8.6530.03
Small for gestational age (< 2 SD)0.34 (0.0–0.53)0 (0.0–0.31)0.66 (0.47–0.87)2.9530.40
Fetal growth rate*0.42 (0.28–0.50)0.03 (0.0–0.14)0.55 (0.50–0.60)4.3730.22
Primiparae0.32 (0.11–0.40)0.01 (0.0–0.17)0.67 (0.60–0.75)3.3630.34

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References
  8. Appendices

Genetic effects explain, according to this population-based study, 25–40% of the variations in birthweight, fetal growth rate and gestational duration. Similar estimates of genetic effects were obtained when dichotomous and continuous variables were used, which implies that the heritabilities are of the same magnitude whether variation for extreme groups or for the whole normal distribution are analysed. Thus, the results give no evidence for specific genes operating at the end of the distribution, but rather that many genes probably account for the individual differences across the distributions.

The present study is based on births within a national cohort of female twin pairs with known zygosity. Provided that the present results are valid also for genetic effects on pregnancy outcome in singleborn mothers, the population-based study design should ensure that the results are possible to generalise. As it is based on differences within twin pairs, it reduces the need of controlling for confounding factors. Information on pregnancy outcome was collected independent of zygosity, and recall bias is therefore inconceivable. Distributions of maternal age, parity and year of birth were similar among monozygotic and dizygotic twins, and confounding by these factors are therefore unlikely to affect the results.

In this study, heritability for birthweight was estimated to be 42%. It should, however, be noted that our design does not permit a separation of maternal genetic effects from fetal genetic effects, because the genetic relationship for both the twins and their offspring are dependent on the zygosity of the woman (the offspring of monozygotic mothers are halfsibs, whereas the offspring of dizygotic mothers are ordinary cousins). A previous study18 has tried to distinguish between maternal and fetal genetic effects with various results. Magnus reports heritabilities for birthweight between 50% and 70%, and most of these effects are due to fetal genes. On the other hand, in another study of offspring of twins, Nance et al.19 report maternal genetic effects with more modest heritabilities (40%). In a recent study of birthweight in children born after ovum donation, there was a correlation between the infant's birthweight and the ovum recipient's weight20. However, donor birthweight and weight of the donor's own children were not significantly correlated with infant's birthweight. Another recent study concludes that both the father's physical stature and his own birthweight at term influence the birthweight of the offspring21, but others have found that both maternal and paternal birthweight are poor predictors of outcome, and explain 5.6% of the variance in birthweight at term22. Thus, the accumulated evidence from this and other studies seems to agree that genetic effects account for less than 50% of the individual differences in birthweight, even though the issue of maternal versus fetal genetic effects is still unresolved.

For gestational length the role of genetic influence is less clear. In the present investigation we found a significant genetic contribution not only to low birthweight and small for gestational age but also to pre-term birth. Previous studies, which have found a weak or negligible genetic influence on pre-term birth are intergenerational1,7,8,23. In the previously mentioned twin study3 variance in birthweight was reduced when gestational length was included in the model, suggesting that part of the genetic effects on birthweight are mediated through gestational length. However, because our measures of fetal growth were also influenced by genetic effects, there must be at least some sets of genes that are independent for the two outcomes. Previous intergenerational studies also suffer the disadvantage of being influenced by the effects of time. Whereas birthweight is often assiduously recorded, gestational length is less reliable and the accuracy in reporting has changed over time14. Results from intergenerational studies may also be influenced by changes in sociode-mographic factors, lifestyle, or differences in obstetrical practices, such as induction of pre-term delivery.

Analyses of continuous variables gave higher precision in the estimates, and the genetic parameters were all significant. For the dichotomised variables, only pre-term birth reached statistical significance, although all heritability estimates were of the same magnitude regardless whether continuous or dichotomised variables were used. Thus, genetic effects seem to account for the individual differences across the whole distribution.

The mediation of genetic effects on birthweight and gestational length are largely unknown. Antropometric and lifestyle factors, such as maternal height24, smoking, and pregnancy complications, such as pre-eclampsia and gestational hypertension, are well known risk factors for fetal growth retardation and pre-term birth25. Genetic or familial factors also influence both height and smoking habits26,27, as well as risk of pre-eclampsia and gestational hypertension28,29. A recent study30 indicates that gene expression can be modified by environmental factors, such as smoking. Probably, genes influencing birthweight and gestational length operate through many channels, among others anthropometric characteristics and susceptibility to diseases during pregnancy. Thus the maternal genetic influence on fetal growth can differ between populations.

The present study, in agreement with previous studies, finds a significant genetic contribution to both low birthweight and small for gestational age birth, but also finds a stronger genetic contribution to gestational length than previously described. One reason for conflicting results between studies can be that gene expression depends on environmental factors. Due to differences in basic health status between study populations the genetic effects will be more important in a well-fed and well-nourished population, whereas in a more heterogeneous population with a higher level of maternal morbidity, environmental causes of pre-term birth and small for gestational age births will probably be more important.

Acknowledgements

The Swedish Twin Registry is supported by grants from the Swedish Council for Planning and Coordination of Research (FRN) and the John D. and Catherine T. MacArthur Foundation. The study was supported by grants from the Swedish Medical Research Council (grant number 27X-12276).

References

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References
  8. Appendices
  • 1
    Klebanoff MA, Meirik O, Berendes HW. Second generation consequences of small-for-dates birth. Pediatrics 1989; 84: 343347.
  • 2
    Emanuel I, Filakti H, Alberman E, Evans SJW. Intergenerational studies of human birthweight from the 1958 birth cohort. 1. Evidence for a multigenerational effect. Br J Obstet Gynaecol 1992; 99: 6774.
  • 3
    Magnus P. Causes of variation in birthweight: a study of offspring of twins. Clin Genet 1984; 25: 1524.
  • 4
    Ounsted M, Scott A. Transmission through the female line of a mechanism constraining human fetal growth. Ann Hum Biol 1986; 13: 143151.
  • 5
    Klebanoff MA, Graubard BI, Kessel SS, Berendes HW. Low birthweight across generations. JAMA 1984; 252: 24232427.
  • 6
    Carr-Hill R, Campbell DM, Hall MH, Meredith A. Is birthweight determined genetically BMJ 1987; 295: 687689.
  • 7
    Magnus P, Bakketeig LS, Skjaerven R. Correlations of birthweight and gestational age across generations. Ann Hum Biol 1993; 20: 231238.
  • 8
    Alberman E, Emanuel I, Filakti H, Evans SJW. The contrasting effects of parental birthweight and gestational age on the birthweight of offspring. Paediatr Perinat Epidemiol 1992; 6: 134144.
  • 9
    Medlund P, Cederlöf R, Floderus-Myrhed B, Friberg L, Sörensen S. A new Swedish twin registry. A new Swedish twin registry 1976; 600 (Suppl): 1111.
  • 10
    Cnattingius S, Ericson A, Gunnarskog J, Källén B. A quality study of a medical birth registry. Scand J Soc Med 1990; 18: 143148.
  • 11
    Lunde AS, Lundeborg S, Lettenström GS, Thugesen L, Huebner J. The person number systems of Sweden, Norway, Denmark and Israel. The person number systems of Sweden, Norway, Denmark and Israel, series 2, (DHHS publications no (PHS) 80–1358). Hyattsville, MD: National Center of Health Statistics, 1980; 84: 511.
  • 12
    Cederlöf R, Friberg L, Jonsson E, Kaij L. Studies on similarity diagnosis in twins with the aid of mailed questionnaires. Acta Genet Med Gemellol 1961; 11: 338362.
  • 13
    Marsàl K, Persson P-H, Larsen T, Lilja H, Selbing A, Sultan B. Intrauterine growth curves based on ultrasonically estimated foetal weights. Acta Paediatr 1996; 85: 843848.
  • 14
    Högberg U, Larsson N. Early dating by ultrasound and perinatal outcome: a cohort study. Acta Obstet Gynecol Scand 1997; 76: 907912.
  • 15
    McGue M. When assessing twin concordance, use the probandwise not the pairwise rate. Schizophr Bull 1992; 18: 171176.
  • 16
    Neale MC, Cardon LR. Methodology for Genetic Studies of Twins and Families. Dordrecht , The Netherlands : Kluwer Academic Publishing, 1992.
  • 17
    Clausson B, Cnattingius S, Axelsson O. Pre-term and term births of small for gestational age infants: a population based study of risk factors among nulliparous women. Br J Obstet Gynaecol 1998; 105: 10111017.
  • 18
    Magnus P. Further evidence for a significant effect of fetal genes on variation in birth weight. Clin Genet 1984; 26: 289296.
  • 19
    Nance WE, Kramer AA, Corey LA, Winter PM, Eaves LJ. A causal analysis of birth weight in the offspring of monozygotic twins. Am J Hum Genet 1983; 35: 12111223.
  • 20
    Brooks AA, Johnson MR, Steer PJ, Pawson ME, Abdalla HI. Birth weight: nature or nurture Early Hum Dev 1995; 42: 2935.
  • 21
    Klebanoff MA, Mednick BR, Schulsinger C, Secher NJ, Shiono PH. Father's effect on infant birthweight. Father's effect on infant birthweight. 1998; 178: 10221026.
  • 22
    Langhoff Roos J, Lindmark G, Gustavsson KH, GebreMedhin M, Meirik O. Relative effect of parental birthweight on infant birthweight at term. Clin Genet 1987; 32: 240248.
  • 23
    Klebanoff MA, Yip R. Influence of maternal birthweight on rate of fetal growth and duration of gestation. J Pediatrics 1987; 111: 287292.
  • 24
    Klein J, Stein Z, Susser M. Conception to birth: epidemiology of prenatal development. Monographs in Epidemiology and Biostatistics. Oxford : Oxford University Press, 1989: 231240.
  • 25
    Kramer MS. Intrauterine growth and gestational duration determinants. Paediatrics 1987; 80: 502511.
  • 26
    Heath AC, Madden PAF. Genetic influences on smoking behaviour. In: TurnerJR, CardonLR, HewittJK, editors. Behaviour Genetic Approaches In Behavioral Medicine. New York : Plenum Press, 1995: 4566.
  • 27
    Stunkard AJ, Harris JR, Pedersen NL, McClearn GE. The body mass index of twins who have been reared apart. N Engl J Med 1990; 322: 14831487.
  • 28
    Sutherland A, Cooper DW, Howie PW, Liston WA, MacGillivray I. The incidence of severe pre-eclampsia amongst mothers and mothers-in-law of pre-eclamptics and controls. Br J Obstet Gynaecol 1981; 88: 785790.
  • 29
    Arngrimsson R, Bjömsson S, Geirssson RT, Björnsson H, Walker JJ, Snaedal G. Genetic and familial predisposition to eclampsia and pre-eclampsia in a defined population. Br J Obstet Gynaecol 1990; 97: 762769.
  • 30
    van Baal CGM, Boomsma DI. Etiology of individual differences in birthweight of twins as a function of maternal smoking during pregnancy. Twin Res 1998; 1: 123130.
  • 31
    Neale MC MX. Statistical Modeling. Richmond , Virginia : Medical College of Virginia, 1994.

Appendices

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. References
  8. Appendices

Appendix 1.

Probandwise concordance is calculated in the following manner: the number of affected twins in concordant pairs (a pair where both sisters have a low birthweight infant, a pre-term birth or an small for gestational age birth) is divided by the total number of affected twins with the same zygosity. Probandwise concordance thus represents the probability of a sister having the gestational outcome in question if her sister had the same outcome.

Tetrachoric correlations (correlations of liability) were computed for the dichotomised variables. Here an underlying normal distribution of liability is assumed, with a critical threshold value exceeded by affected subjects. The tetrachoric correlation is not a correlation between a pair of score values. Rather it is an estimate of the correlation between two latent variables η and ζ underlying x and y, where η and ζ are assumed to have a bivariate normal distribution.

For continuous variables intra-class correlations were calculated. The intra-class correlation could be considered as a measure of the degree of homogeneity of the classes (twin pairs) relative to the total variability. The formula for calculating intra-class correlation (ri) among twin pairs is:

  • image

where Vb and Vw are the between and within estimates respectively.

Model fitting

In quantitative genetic analyses, the phenotype is assumed to be the sum of effects of the genotype and the environment16. By comparing observed pheno-typic covariance between twins of different zygosity, the proportion of variance of genetic (heritability), shared environmental and nonshared environmental effects can be estimated. Monozygotic twins share identical genotypes, whereas dizygotic twins share on average 50% of their segregating genes. Hence, genetic effects are indicated if monozygotic twins are more similar than dizygotic twins. Shared environmental effects reflect twin similarity that is not explained by genetic effects, whereas nonshared environmental effects are manifest as within-pair differences. The phenotype of one of the twins in a pair can be symbolised as:

  • image

where P1, G1, EC1, and En1 are the phenotype, genotype, shared environment, and nonshared environment of the first twin in the pair. A similar equation can be written for the second twin and the model can be represented as the path model in Fig. 1. The correlations between the latent factors (i.e. environmental and genetic effects) are set to fixed values according to the theoretical expectations, and the model can be represented by the following three equations:

  • image
image

Figure 1. Univariate genetic analysis with twin data. The model estimates genetic (G), shared environmental (Es) and nonshared environmental (En) influences on the measured variables. MZ = monozygotic; DZ = dyzygotic.

Download figure to PowerPoint

Variance component estimates for genetic and environmental effects were obtained by entering zygosity specific contingency tables for dichotomised variables and covariance matrices for continuous variables into a structural equation model-fitting program31.

By comparing observed phenotypic covariance between twins of different zygosity, the proportion of variance of genetic (heritability), shared environmental and nonshared environmental effects can be estimated. Monozygotic twins share identical genotypes, whereas dizygotic twins share on average 50% of their segregating genes. Hence, genetic effects are indicated if monozygotic twins are more similar than dizygotic twins. Shared environmental effects reflect twin similarity that is not explained by genetic effects, whereas non-shared environmental effects are manifest as within-pair differences.