The continuing value of twin studies in the omics era

The classical twin study has been a powerful heuristic in biomedical, psychiatric and behavioural research for decades. Twin registries worldwide have collected biological material and longitudinal phenotypic data on tens of thousands of twins, providing a valuable resource for studying complex phenotypes and their underlying biology. In this Review, we consider the continuing value of twin studies in the current era of molecular genetic studies. We conclude that classical twin methods combined with novel technologies represent a powerful approach towards identifying and understanding the molecular pathways that underlie complex traits.     

Twin registers across the globe: what's out there in 2002?

Twin research offers the greatest power for the genetic analysis of complex multifactorial traits and diseases in humans. Modern twin analyses extend beyond the classical twin study for estimating the heritability of a trait. The human genome project can fulfil its promises only after functional characterisation of single genes in the context of genetic background and environment. Twin research can make a major contribution in that regard. Twin research is greatly facilitated by the willingness, motivation, cooperation, and generosity of the participants and their families. A second important aspect is the availability of twin registries that serve as a resource for genetic epidemiology. Currently, there is no systematic overview of the twin collections worldwide. This special issue will help to overcome the limited accessibility of this resource by providing basic information on most of the existing twin registers. Furthermore, an additional goal is to facilitate collaboration between registers. Some basic principles, potentials, and problems will be exemplified by my personal experience in the Berlin Twin Study.     

The NAS-NRC Twin Registry of WWII military veteran twins. National Academy of Sciences-National Research Council.

The NAS-NRC Twin Registry is one of the oldest, national population based-twin registries in the United States. It consists of 15,924 white male twin pairs born in the years 1917-1927 (inclusive) both of whom served in the armed forces. The registry, which has been in operation more than 30 years, has collected data from a variety of sources. Records-based, computerized data have come largely from the Department of Veterans Affairs, and there have been three major epidemiologic questionnaires, undertaken roughly every 15 years. Classic twin studies on a variety of medical conditions were the early focus of the registry, which now has a strong focus on chronic disease epidemiology. Work on a DNA specimen bank has been proceeding slowly, but is now a top priority, due to the increasing force of mortality in this twin cohort.     

Similarity of social security numbers among twins: data from the Virginia Twin Registry.

At least two twin registries in the United States have been or are being assembled using the similarity of Social Security Numbers in computerized records to help identify possible twin pairs. While the success of such enterprises depends directly on a high probability of twinness given Social Security Numbers, there are theoretical and practical reasons to study the probability of Social Security Number similarity given twinness. For example, the number of twin pairs with similar Social Security Numbers obviously determines the maximum number of twin pairs that can be discovered by similarity algorithms. To study this issue, we examined the similarity of known Social Security Numbers in twin pairs from the Virginia Twin Registry by age, sex, race, and zygosity of the pair. We found that similarity between the Social Security Numbers of twin pairs varies markedly by age, and MZ twin pairs have significantly more similar Social Security Numbers than DZ pairs at all ages. Among older twins, there are also significant differences by sex and race. For younger twins, algorithms that identify putative twin pairs on the basis of the similarity of their Social Security Numbers hold the promise of being able to identify a large proportion of all true twin pairs. Such algorithms will be substantially less successful, however, in identifying a large proportion of older twin pairs.     

Twin births to mothers who are twins: a registry based study.

OBJECTIVES: To estimate the risk of having twin infants for mothers who are twins; to investigate the genetic influence on twinning. DESIGN: Retrospective study of multiple births in two nationwide registries. SETTING: Sweden. SUBJECTS: Multiple births among 31,586 deliveries between 1973 and 1991 to women who were twins. MAIN OUTCOME MEASURES: Numbers of monozygotic and dizygotic twin births expected and estimated. RESULTS: Women who are dizygotic twins have a moderately increased risk of having twins (relative risk 1.30, 95% confidence interval 1.14 to 1.49) which seems to be completely the result of dizygotic twinning. When a mother is a monozygotic twin, her risk of having twins of the same sex is significantly increased (1.47; 1.10 to 1.97). This is the result of an excess of monozygotic twins (39 pairs estimated, 18 expected). CONCLUSIONS: Women who are twins have an increased risk of giving birth to twins. Genetic components of monozygotic and dizygotic twinning seem to be independent.     

The Vietnam Era Twin Registry.

The Vietnam Era Twin (VET) Registry is composed of 7369 middle-aged male-male twin pairs both of whom served in the military during the time of the Vietnam conflict (1965-1975). The Registry is a United States Department of Veterans Affairs resource that was originally constructed from military records; the Registry has been in existence for more than 15 years. It is one of the largest national twin registries in the US and currently has subjects living in all 50 states. Initially formed to address questions about the long-term health effects of service in Vietnam the Registry has evolved into a resource for genetic epidemiologic studies of mental and physical health conditions. The management and administration of the VET Registry is described with particular attention given to the processes involved with database maintenance and study coordination. Several waves of mail and telephone surveys have collected a wealth of health-related information on Registry twins. More recent data collection efforts have focused on specific sets of twin pairs and conducted detailed clinical or laboratory testing. New Registry initiatives for the future include the construction of a web site and the development of a DNA repository.     

Longevity studies in GenomEUtwin.

Previous twin studies have indicated that approximately 25% of the variation in life span can be attributed to genetic factors and recent studies have also suggested a moderate clustering of extreme longevity within families. Here we discuss various definitions of extreme longevity and some analytical approaches with special attention to the challenges due to censored data. Lexis diagrams are provided for the Danish, Dutch, Finnish, Italian, Norwegian, and Swedish Twin registries hereby outlining possibilities for longevity studies within GenomEUtwin. We extend previous analyses of lifespan for the Danish 1870-1900 twin cohorts to include the new 1901-1910 cohorts, which are consistent with the previous findings. The size of the twin cohorts in GenomEUtwin and the existence of population-based, nationwide health and death registers make epidemiological studies of longevity very powerful. The combined GenomEUtwin sample will also allow detailed age-specific heritability analyses of lifespan. Finally, it will provide a resource for identifying unusual sibships (i.e., dizygotic twin pairs) where both survived to extreme ages, as a basis for discovering genetic variants of importance for extreme survival.     

The Italian Twin Project: from the personal identification number to a national twin registry.

The unique opportunity given by the "fiscal code", an alphanumeric identification with demographic information on any single person residing in Italy, introduced in 1976 by the Ministry of Finance, allowed a database of all potential Italian twins to be created. This database contains up to now name, surname, date and place of birth and home address of about 1,300,000 "possible twins". Even though we estimated an excess of 40% of pseudo-twins, this still is the world's largest twin population ever collected. The database of possible twins is currently used in population-based studies on multiple sclerosis, Alzheimer's disease, celiac disease, and type 1 diabetes. A system is currently being developed for linking the database with data from mortality and cancer registries. In 2001, the Italian Government, through the Ministry of Health, financed a broad national research program on twin studies, including the establishment of a national twin registry. Among all the possible twins, a sample of 500,000 individuals are going to be contacted and we expect to enrol around 120,000 real twin pairs in a formal Twin Registry. According to available financial resources, a sub sample of the enrolled population will be asked to donate DNA. A biological bank from twins will be then implemented, guaranteeing information on future etiological questions regarding genetic and modifiable factors for physical impairment and disability, cancers, cardiovascular diseases and other age related chronic illnesses.     

Should all twins be delivered by caesarean section? A preliminary report.

The aim was to analyse the neonatal mortality related to mode of delivery for twins using a population-based registry. In all, 18,125 twins delivered in Sweden between 1991 and 1997, after excluding those with unknown gestational duration, were used to analyse the differences between groups of twins. Results showed the OR for neonatal death, breech vaginal delivery versus caesarean section (all indications) was 1.47 (95% CI 0.99-2.17). The OR at vaginal delivery for neonatal death, twin I in breech versus cephalic presentation was 5.60 (2.62-11.94) and for twin II the corresponding figures were 1.85 (1.03-3.32). Analyses using population-based registries from other countries are needed to confirm or reject the present findings of an increased neonatal mortality for twins in breech presentation delivered vaginally.     

Sex differences in heritability of BMI: a comparative study of results from twin studies in eight countries.

Body mass index (BMI), a simple anthropometric measure, is the most frequently used measure of adiposity and has been instrumental in documenting the worldwide increase in the prevalence of obesity witnessed during the last decades. Although this increase in overweight and obesity is thought to be mainly due to environmental changes, i.e., sedentary lifestyles and high caloric diets, consistent evidence from twin studies demonstrates high heritability and the importance of genetic differences for normal variation in BMI. We analysed self-reported data on BMI from approximately 37,000 complete twin pairs (including opposite sex pairs) aged 20-29 and 30-39 from eight different twin registries participating in the GenomEUtwin project. Quantitative genetic analyses were conducted and sex differences were explored. Variation in BMI was greater for women than for men, and in both sexes was primarily explained by additive genetic variance in all countries. Sex differences in the variance components were consistently significant. Results from analyses of opposite sex pairs also showed evidence of sex-specific genetic effects suggesting there may be some differences between men and women in the genetic factors that influence variation in BMI. These results encourage the continued search for genes of importance to the body composition and the development of obesity. Furthermore, they suggest that strategies to identify predisposing genes may benefit from taking into account potential sex specific effects.     

The cholinesterases: analysis by pharmacogenomics in man

We have undertaken a study on variations in cholinesterase (ChE) genes in relation to cardiovascular (CV) function and the metabolic syndrome. Peripheral and central nervous system control of cardiovascular (CV) function mediated through cholinergic pathways is critical in homeostatic maintenance of blood pressure and responsiveness to stress. For acetylcholinesterase (AChE; EC 3.1.1.7) our focus is to identify single nucleotide polymorphisms (SNPs) in the gene that are linked to cardiovascular function. For butyrylcholinesterase (BChE; EC 3.1.1.8) we examined whether BChE activity correlated with parameters of the metabolic syndrome and cardiovascular function. ChE can be found in whole blood enabling a characterization of biochemical phenotype in addition to correlating genotype with phenotypic physiologic responses. Analysis of enzymatic activity was determined spectrophotometrically in blood samples from twin and other subject registries. Correlation analysis revealed significant relationships between enzyme activity and certain CV endpoints. Linkage analysis with data from a dizygotic (DZ) twin set showed a suggestive linkage at the BChE locus, and statistical analysis revealed a high correlation between BChE activity and variables associated with cardiovascular risk and the metabolic syndrome. Pattern of within-pair twin correlations by zygosity and the ACE model-fitting findings suggest the major source of this variation (65%) is attributable to an additive genetic component. To date 19 SNPs have been identified by the re-sequencing of AChE including four nonsynonymous coding SNPs (cSNPs).     

Using the tools of genetic epidemiology to understand sex differences in neuropsychiatric disorders

Many neuropsychiatric disorders exhibit differences in prevalence, age of onset, symptoms or course of illness between males and females. For the most part, the origins of these differences are not well understood. In this article, we provide an overview of sex differences in psychiatric disorders including autism spectrum disorder (ASD), attention deficit/hyperactivity disorder (ADHD), anxiety, depression, alcohol and substance abuse, schizophrenia, eating disorders and risk of suicide. We discuss both genetic and nongenetic mechanisms that have been hypothesized to underlie these differences, including ascertainment bias, environmental stressors, X‐ or Y‐linked risk loci, and differential liability thresholds in males and females. We then review the use of twin, family and genome‐wide association approaches to study potential genetic mechanisms of sex differences and the extent to which these designs have been employed in studies of psychiatric disorders. We describe the utility of genetic epidemiologic study designs, including classical twin and family studies, large‐scale studies of population registries, derived recurrence risks, and molecular genetic analyses of genome‐wide variation that may enhance our understanding sex differences in neuropsychiatric disorders.     

Utilizing Twins as Controls for Non-Twin Case-Materials in Genome Wide Association Studies

Twin registries around the globe have collected DNA samples from large numbers of monozygotic and dizygotic twins. The twin sample collections are frequently used as controls in disease-specific studies together with non-twins. This approach is unbiased under the hypothesis that twins and singletons are comparable in terms of allele frequencies; i.e. there are no genetic variants associated with being a twin per se. To test this hypothesis we performed a genome-wide association study comparing the allele frequency of 572,352 single nucleotide polymorphisms (SNPs) in 1,413 monozygotic (MZ) and 5,451 dizygotic (DZ) twins with 3,720 healthy singletons. Twins and singletons have been genotyped using the same platform. SNPs showing association with being a twin at P-value < 1 × 10^-5 were selected for replication analysis in 1,492 twins (463 MZ and 1,029 DZ) and 1,880 singletons from Finland. No SNPs reached genome-wide significance (P-value < 5 × 10^-8) in the main analysis combining MZ and DZ twins. In a secondary analysis including only DZ twins two SNPs (rs2033541 close to ADAMTSL1 and rs4149283 close to ABCA1) were genome-wide significant after meta-analysis with the Finnish population. The estimated proportion of variance on the liability scale explained by all SNPs was 0.08 (P-value=0.003) when MZ and DZ were considered together and smaller for MZ (0.06, P-value=0.10) compared to DZ (0.09, P-value=0.003) when analyzed separately. In conclusion, twins and singletons can be used in genetic studies together with general population samples without introducing large bias. Further research is needed to explore genetic variances associated with DZ twinning.     

Genetic and environmental effects on age at menarche,and its relationship with reproductive health in twins

INTRODUCTION:

Menarche or first menstrual period is a landmark in reproductive life span and it is the most prominent change of puberty. The timing of menarche can be under the influence of genes as well as individual environmental factors interacting with genetic factors.

OBJECTIVE:

Our study objectives were (a) to investigate the heritability of age of menarche in twins, (b) to obtain the association between age of menarche and childhood factors, and reproductive events/behavior, (c) to examine whether or not having a male co-twin affects early/late menarche.

METHODOLOGY:

A group of female-female identical (n = 108, 54 pairs), non-identical twins (n = 68, 34 pairs) and 17 females from opposite-sex twin sets were identified from twin registries of Malaysia and Iran. Genetic analysis was performed via two methods of Falconers’ formula and maximum likelihood.

RESULTS:

Heritability was found to be 66% using Falconers’ formula and 15% using univariate twin analysis. Model analysis revealed that shared environmental factors have a major contribution in determining the age of menarche (82%) followed by non-shared environment (18%).

DISCUSSION:

Result of this study is consistent with that of the literature. Timing of menarche could be under the influence of shared and non-shared environmental effects. Hirsutism was found to have a higher frequency among subjects with late menarche. There was no significant difference in age of menarche between females of opposite-sex twins and females of same-sex twins.

CONCLUSION:

It is concluded that twin models provide a powerful means of examining the total genetic contribution to age of menarche. Longitudinal studies of twins may clarify the type of environmental effects that determine the age of menarche.     

Genetic susceptibility to prostate, breast, and colorectal cancer among Nordic twins

To investigate the role of genetics in the development of cancer, we developed a new approach to analyze data on prostate, breast, and colorectal cancer from the Swedish, Danish, and Finnish twin registries on monozygotic (MZ) and same-sex dizygotic (DZ) twins. In the spirit of a sensitivity analysis, we modeled genetic inheritance as either an autosomal recessive or dominant cancer susceptibility (CS) genotype that involves either a single gene, many genes with equal allele frequencies, or three genes with a ninefold range of allele frequencies. We also modeled the joint probability of cancer incidence among five age categories, conditional on the presence or absence of the CS genotype. The main assumptions are: (1) The joint distribution of unobserved environmental effects in a twin pair conditional on the presence or absence of the CS genotype is the same for MZ and DZ twins, (2) the probability of cancer conditional on the presence or absence of the CS genotype and the unobserved environmental effects (i.e., the gene-environment interaction) is the same for MZ and DZ twins, and (3) the probability of cancer is independent between twins with the CS genotype. Estimation was maximum likelihood via a search over allele frequency and two levels of EM algorithms. Models had acceptable or good fits. Variability was estimated using a bootstrap approach, but only 50 replications were feasible. The 94th percentile of bootstrap replications for the estimated fraction of cancers with the CS genotype ranged, over the various genetic models, from 0.16 to 0.45 for prostate cancer, 0.12 to 0.30 for breast cancer, and 0.08 to 0.27 for colorectal cancer. We conclude that genetic susceptibility makes only a small to moderate contribution to the incidence of prostate, breast, and colorectal cancer.     

Selection strategies for linkage studies using twins.

Genetic linkage analysis for complex diseases offers a major challenge to geneticists. In these complex diseases multiple genetic loci are responsible for the disease and they may vary in the size of their contribution; the effect of any single one of them is likely to be small. In many situations, like in extensive twin registries, trait values have been recorded for a large number of individuals, and preliminary studies have revealed summary measures for those traits, like mean, variance and components of variance, including heritability. Given the small effect size, a random sample of twins will require a prohibitively large sample size. It is well known that selective sampling is far more efficient in terms of genotyping effort. In this paper we derive easy expressions for the information contributed by sib pairs for the detection of linkage to a quantitative trait locus (QTL). We consider random samples as well as samples of sib pairs selected on the basis of their trait values. These expressions can be rapidly computed and do not involve simulation. We extend our results for quantitative traits to dichotomous traits using the concept of a liability threshold model. We present tables with required sample sizes for height, insulin levels and migraine, three of the traits studied in the GenomEUtwin project.     

Adjusting for reporting delay in cancer incidence when combining different sets of cancer registries

Cancer registries collect cancer incidence data that can be used to calculate incidence rates in a population and track changes over time. For incidence rates to be accurate, it is critical that diagnosed cases be reported in a timely manner. Registries typically allow a fixed amount of time (e.g. two years) for diagnosed cases to be reported before releasing the initial case counts for a particular diagnosis year. Inevitably, however, additional cases are reported after the initial counts are released; these extra cases are included in subsequent releases that become more complete over time, while incidence rates based on earlier releases will underestimate the true rates. Statistical methods have been developed to estimate the distribution of reporting delay (the amount of time until a diagnosed case is reported) and to correct incidence rates for underestimation due to reporting delay. Since the observed reporting delays must be less than the length of time the registry has been collecting data, most methods estimate a truncated delay distribution. These methods can be applied to a group of registries that began collecting data in the same diagnosis year. In this paper, we extend the methods to two groups of registries that began collecting data in two different diagnosis years (so that the delay distributions are truncated at different times). We apply the proposed method to data from the National Cancer Institute's Surveillance Epidemiology and End Results (SEER) program, a consortium of U.S. cancer registries that includes nine registries with data collection beginning in 1981 and four registries with data collection beginning in 1992. We use the method to obtain delay‐adjusted incidence rates for melanoma, liver cancer, and Hodgkin lymphoma.     

The role of hospital-based cancer registries in low and middle income countries-The Nigerian Case Study

Background: The incidence of cancer continues to rise all over the world and current projections show that there will be 1.27 million new cases and almost 1 million deaths by 2030. In view of the rising incidence of cancer in sub-Saharan Africa, urgent steps are needed to guide appropriate policy, health sector investment and resource allocation. We posit that hospital based cancer registries (HBCR) are fundamental sources of information on the frequent cancer sites in limited resource regions where population level data is often unavailable. In regions where population based cancer registries are not in existence, HBCR are beneficial for policy and planning. Materials and methods: Nineteen of twenty-one cancer registries in Nigeria met the definition of HBCR, and from these registries, we requested data on cancer cases recorded from January 2009 to December 2010. 16 of the 19 registries (84%) responded. Data on year hospital was established; year cancer registry was established, no. of pathologists and types of oncology services available in each tertiary health facility were shown. Analysis of relative frequency of cancers in each HBCR, the basis of diagnosis recorded in the HBCR and the total number of cases recorded by gender was carried out. Results: The total number of cancers registered in these 11 hospital based cancer registries in 2009 and 2010 was 6484. The number of new cancer cases recorded annually in these hospital based cancer registries on average was 117 cases in males and I77 cases in females. Breast and cervical cancer were the most common cancers seen in women while prostate cancer was the commonest among men seen in these tertiary hospitals. Conclusion: Information provided by HBCR is beneficial and can be utilized for the improvement of cancer care delivery systems in low and middle income countries where there are no population based cancer registries.     

Paper 3: EUROCAT data quality indicators for population-based registries of congenital anomalies

The European Surveillance of Congenital Anomalies (EUROCAT) network of population-based congenital anomaly registries is an important source of epidemiologic information on congenital anomalies in Europe covering live births, fetal deaths from 20 weeks gestation, and terminations of pregnancy for fetal anomaly. EUROCAT's policy is to strive for high-quality data, while ensuring consistency and transparency across all member registries. A set of 30 data quality indicators (DQIs) was developed to assess five key elements of data quality: completeness of case ascertainment, accuracy of diagnosis, completeness of information on EUROCAT variables, timeliness of data transmission, and availability of population denominator information. This article describes each of the individual DQIs and presents the output for each registry as well as the EUROCAT (unweighted) average, for 29 full member registries for 2004-2008. This information is also available on the EUROCAT website for previous years. The EUROCAT DQIs allow registries to evaluate their performance in relation to other registries and allows appropriate interpretations to be made of the data collected. The DQIs provide direction for improving data collection and ascertainment, and they allow annual assessment for monitoring continuous improvement. The DQI are constantly reviewed and refined to best document registry procedures and processes regarding data collection, to ensure appropriateness of DQI, and to ensure transparency so that the data collected can make a substantial and useful contribution to epidemiologic research on congenital anomalies.