Brief History
The Portuguese territory comprises a mainland located in the most western part of Europe, and two archipelagos both in the Atlantic Ocean: Azores (nine islands), and Madeira (two islands). The Portuguese twin research started in 1999 in the mainland by the first author (JARM), and then expanded to the Azores in 2001. In 2004, together with DLF, a new research branch started in Madeira. The focus of our research is on human physical activity and related traits, including physical growth, biological maturation, health perceptions, physical fitness, gross motor coordination, neuromotor development, body composition, bone, and metabolic syndrome (MetS). Our research started without funding. It was only in 2002 that the Portuguese Foundation for Science (FCT) gave a grant to JARM to study twins’ physical activity, physical fitness, and body composition. Later on, in 2004, another grant from FCT allowed a joint venture with DLF, the late Gaston Beunen and Martine Thomis from the Katholieke Universiteit Leuven, Belgium, and colleagues from the University of Madeira Genetics Department. This ‘research adventure’, named genetic and environmental influences in physical activity, physical fitness and health (GEAFAS) (Madeira Family Study), was broad in its sampling schema (three-generation families) and scope (from quantitative to molecular genetics probing the value of several candidate genes). As no published twin research was available in the Portuguese language, we produced several papers regarding twin methodology and books/reports for Portuguese-speaking sport and human movement sciences researchers from Portugal, Brazil, Mozambique, and Angola (Maia, Reference Maia2001; Maia et al., Reference Maia, Garganta, Seabra and Lopes2004, Reference Maia, Seabra, Silva, Freitas, Gouveia and Basso2011; Sapage & Maia, Reference Sapage and Maia2007). Furthermore, over the years, and in conjunction with the Portuguese Twin Group Register, four twin meetings were organized around the theme of active and healthy lifestyles.
Recruitment Procedures
As the main focus of our twin research is physical activity and related traits during the childhood and adolescent growth period, our target twin population was school-aged children aged 4–18 years. Twin recruitment was made in several ways: first, through school lists and information from physical education teachers; second, through announcements in local newspapers and TV; third, through advertisements with local governmental bodies that facilitated all our twin meetings; fourth, through the Portuguese Twin Group.
Zygosity Assignment
Two procedures were used to assess zygosity: (1) by the Peeters et al. (Reference Peeters, Van Gestel, Vlietinck, Derom and Derom1998) questionnaire, that was later cross-validated in Portugal using DNA data (Maia et al., Reference Maia, Fernandes, Amorim, Alves, Gusmão and Pereira2007); (2) by DNA extraction with a method based on the use of Chelex resin. Genotyping was performed on an ABI 310 Genetic Analyzer (AB Applied Biosystems, Carlsbad, California, USA), according to the manufacturer's instructions. In all DNA samples, the analysis of 17 short tandem repeats autosomal (CSF1PO, D2S1338, D3S1358, D5S818, D7S820, D8S1179, D13S317, D16S539, D18S51, D19S433, D21S11, FGA, PD, PE, TH01, TPO, and VWA) and the Amelogenin locus (sex determination) was performed. Allele frequencies of the different genetic markers in the north and center of Portugal were used in the computation of probabilities of being monozygotic (MZ) twins.
Measures
Table 1 presents research domains, tests used, phenotypes, and equipment used in different twin studies. Twin research in Portugal has been initiated from scientists within the field of sport and movement sciences, with a special focus on children and adolescent populations. This ‘initiation site’ is reflected in the set of phenotypes under study that are mainly related to human growth, physical fitness, physical (in-)activity, neuromotor development, bone, and metabolic health.
Domains, Test Batteries, Phenotypes, and Equipment
Main Findings
Findings for distinctive research domains are summarized in Tables 2, 3, 4–5. Studies within the area of growth, physical fitness, physical activity, motor development, bone, and metabolic health require the measurement of young children and adolescents in a school or laboratory setting, with often multiple days of assessment. Therefore, sample sizes are often small in the different studies. Pearson correlation (r), intra-class correlation coefficients (t), or structural equation modeling were used to analyze twin covariation. In most studies, age effects are not modeled as a confounding factor within the structural equation-modeling framework. Given the small sample sizes in most studies, the analytical option was to first regress age and sex from each phenotype, and then compute heritabilities. Had our samples been larger, not only more precise genetic and environmental effects would be computed, but also more firm conclusions would be drawn.
Physical Activity
m = male; f = female; MZ = monozygotic twin; DZ = dizygotic twin; A = additive genetic effects; D = nonadditive genetic effects; C = shared environmental effects; E = nonshared environmental effects including measurement error.
Health-Related Physical Fitness
MZ = monozygotic twin; DZ = dizygotic twin; A = additive genetic effects; D = nonadditive genetic effects; C = shared environmental effects; E = nonshared environmental effects including measurement error.
Gross Motor Coordination, and Neuromotor Development
MZ = monozygotic twin; DZ = dizygotic twin; A = additive genetic effects, D = nonadditive genetic effects, C = shared environmental effects, E = nonshared environmental effects including measurement error.
Metabolic Syndrome
MZ = monozygotic twin; DZ = dizygotic twin; A = additive genetic effects; D = nonadditive genetic effects; C = shared environmental effects; E = nonshared environmental effects including measurement error.
Physical Activity
When using time-consuming interviews, pedometers and accelerometers to assess physical activity measures, our sample sizes are small. In children, genetic factors are small (a 2 from 24% to 34%) for the Godin and Shephard (Reference Godin and Shephard1985) questionnaire phenotypes (low, moderate, and high intensity, as well as weekly total activity), and in line with the accelerometer data. These results were consistent in mainland and Azorean subjects. In adolescents, questionnaire data (Baecke) showed sex-specific estimates for sport participation, sports index (males: a 2 = 63% and c 2 = 37%; females: a 2 = 32%, c 2 = 38%, e 2 = 30%), and leisure time physical activity index (a 2 = 40%, c 2 = 28%, e 2 = 32%), which were coherent with accelerometer low and very vigorous physical activities — in males the genetic additive factors were always greater, and their magnitude was moderate-to-high (a 2 from 42% to 79%). When studying physical activity patterns using approximate entropy as a suitable descriptor or a regular pattern of daily activities intensity, genetic factors account for about 50% of the total variation.
Motor Performance
Genetic factors are low to moderate in children's and adolescents’ health-related physical fitness (see Table 3). Endurance running and muscular strength phenotypes had higher a 2 estimates. Gross motor coordination tests presented low genetic influence in children's performance (see Table 4). Twins’ MZ and dizygotic (DZ) intra-class correlation coefficients suggest that neuromotor development from childhood to adolescence may be under genetic control, although environmental forces associated with daily living histories and twins’ lifestyle may play an important role.
MetS
Within the Madeira Family Study (GEAFAS), DNA was collected for zygosity determination and future association studies with an array of candidate genes. From a vast set of data collected from all three-generation families, MetS information (systolic blood pressure, waist circumference, triglycerides, HDL-cholesterol, and glucose) is available from all subjects. Here, we only report twin data (see Table 5). In all five MetS indicators, MZ had higher intra-class correlations (from 0.56 to 0.81) than DZ twins (from 0.26 to 0.53). The prevalence of MetS (having, at least, any combination of three indicators) was very low: MZ = 2.4%, DZ = 5.3%. A moderate-to-high additive genetic component was found in four of the five MetS indicators (a 2 from 55% to 80%; and e 2 from 20% to 45%), but except HDL-cholesterol (a 2 = 34%, c 2 = 44%, e 2 = 22%).
Future Directions
There is a strong need to merge all of our data sets to address the following issues: (1) to study in more detail physical activity patterns using accelerometer data; (2) to link physical activity levels with health-related physical fitness; (3) to study body composition and its relationship with physical activity levels; (4) to have a careful look at bone phenotypes; (5) to associate MetS indicators with physical activity and physical fitness; (6) to investigate the role of health perceptions in metabolic health and body composition; (7) to study the links of physical self-concept, body composition, and health-related physical fitness. The inter-relations among the phenotypes we examined suggest shared genetic and environmental factors for these phenotypes. In future studies, we plan to apply multivariate twin modeling to the covariations of these phenotypes. The Madeira Islands study includes three-generation families with phenotypes on human physical growth, body composition, biological maturation, physical activity, physical fitness, bone, health-related quality of life, and MetS and has also DNA data collected from 1,200 subjects. As such, an international collaboration is needed to extract the most from this unique data set. Furthermore, a call for education of Portuguese sport and human movement sciences researchers in the field of twin and family genetic methodologies is needed.
Acknowledgment
The authors would like to express their gratitude to Yoon-Mi Hur for her contributions to the development of this manuscript.
