Strength training: importance of genetic factors : Medicine & Science in Sports & Exercise

Journal Logo


Strength training: importance of genetic factors


Author Information
Medicine & Science in Sports & Exercise: May 1998 - Volume 30 - Issue 5 - p 724-731
  • Free


The question of whether responses to exercise training are variable in the population and whether the observed heterogeneity in trainability is related to the genotype is referred to as genotype*training interaction. This genotype*training interaction effect has been studied for aerobic(12,23) and anaerobic(7,25) performance in response to exercise training as well as for body composition in response to overfeeding and negative energy balance programs (for review (6)). Only one study reports on the influence of genetic factors in the response to resistance strength training (27). The samples of these studies were small overall (number of pairs between 5 and 14) and included only monozygotic twins. Bouchard et al. (4) apply the experimental strategy of submitting monozygotic twins to the same experimental treatment to induce adaptive responses to the phenotype. A two-way ANOVA is then used to study the response pattern for individuals having the same genotype (within MZ twin pairs) and for subjects with different genetic profiles (between pairs of MZ twins). The interaction term gives the importance of the differences in the response to training as a function of the genotype. Bouchard et al. (5) conclude from these experiments that age and gender as well as prior exercise experience do not significantly contribute to interindividual variation in train-ability of˙VO2max. Pretraining level accounts for about 25% of the observed variance in the response of ˙VO2max. F-ratios of between to within pair variances in training for ˙VO2max after the endurance training response (F value of 6 to 9)(12,23) indicate a six to nine times larger variation in training response between pairs of MZ twins than within pairs of MZ twins having the same genotype, which reveals a strong genotype-dependency of trainability. Short-term anaerobic performance (10-s work output) in response to high-intensity intermittent training is affected little by the genotype (F-ratio is only 2 to 3) (25), while 70% of the variation in training response of long-term anaerobic performance(90-s work output) could be accounted for by genetic factors(25). In a 10-wk knee flexion/extension isokinetic strength training study in five monozygotic twin pairs, significant training effects are observed (mean 24%, SD 12%). However, no significant interaction was found between training and the genotype for the peak torque output(27).

In a behavioral genetics approach, genotype-environment interaction (G*E) is defined as the genetic control of sensitivity to differences in environment(20). Different genotypes react in different ways to the same environment. G*E interaction can explain part of the observed variation and will be part of the environmental variance if not accounted for. When genes responsible for the sensitivity to the environment are not the same as those that control the traits under investigation, G*E interaction is detected by the significant presence of different genes in the phenotype under stressed conditions. One way of testing this is by modeling different sources of genetic variation in fitting the pre- and post-training covariance matrices of monozygotic (MZ) and dizygotic (DZ) twins (20).

This study uses both above mentioned approaches to test genotype*training interaction effects at two points in a 10-wk strength training experiment. There is a general consensus on the sequence of important changes occurring in the muscle and the neurological system because of muscle overload(13). Changes in neurological capacities of the muscle, specific movement and muscle activation patterns, and more general adaptations occur during the first weeks, while muscle hypertrophic factors gain importance in long-term training (17,24). In the present study strength responses to muscle overload are expected to be largely a result of neural adaptations with initiation of muscle morphological changes(muscle hypertrophy) near the end of the program.

The aim of this study is to test the presence of genotype*training interaction in maximal arm muscle strength in males and one of its major determinants, arm muscle cross-sectional area. The null hypothesis could be formulated as the absence of genetic effects in the explanation of responses to strength training, or the lack of evidence for genotype*training interaction as shown in a nonsignifiicant F-ratio of between over within twin pair variation in training responses in the sample of MZ twins. In the modle-fitting approach using data of both MZ and DZ twins, the null hypothesis will be the absence of a specific set of training-specific genetic factors contributing to the variation in post-training phenotype.


Subjects. The sample for this study was derived from within the region of Flemish Brabant, Belgium. Male volunteer twins aged 17-30 yr were included if both members of a twin pair had similar physical activity profiles and did not start or stop performing strength training during the preceding year. Two of 43 twin pairs were excluded because of mental retardation and physical disability. Two subjects had strength training experience but had not trained in the year preceding the study. There were no significant differences in a weighed physical activity score including long-term training history and present physical activities between first- and second-born twins or zygosity groups. Intra-pair correlations were slightly higher (r = 0.70) in MZ twin pairs than in DZ twin pairs (r = 0.54) for these physical activity scores. The remaining 41 twin pairs volunteered to participate. Their mean age was 22.4 yr(SD ± 3.7 yr). Subjects were fully informed of the measurement protocol before giving their written consent. The project was approved by the local medical-ethics committee.

Determination of zygosity was assessed by examination the following genetic markers: ABO, Rhesus (D, C, Cw, c, E, e), MNSs, and Duffy(a,b). The power to detect dizygotic (DZ) twins with this set of genetic markers was 91%. Differences in two genetic markers were used to establish dizygosity. The probability of monozygosity of pairs with the same genetic markers was calculated (31). All monozygotic pairs had a probability of monozygosity of at least 95%. Twenty-five pairs were classified as monozygotic (MZ) and 16 as dizygotic (DZ).

Training protocol. Both members of the twin pair participated in a program in which the elbow flexors were trained. During 10 wk, five sets of biceps curls were performed three times a week on a training apparatus(Kettler Sport type 7408-150, Ense-Parsit, Germany). Every week the load of each set (with a precision of 0.5 kg) was adjusted for each subject to a percentage of the maximum one repetition value (1 RM). This 1 RM was defined as the maximal resistance that could be moved a single time through the full range of motion. Details of the training sessions are outlined inTable 1. A complete repetition cycle lasted 2 s and full relaxation was encouraged between each repetition. The rest period between each set was 2 min. All training sessions were closely supervised by a physical trainer, and each third session of the week the training was monitored objectively by a force and position meter. The training program was followed by a 10-wk detraining period.

Training protocol individually adapted to weekly 1 RM loads(expressed as percentage of the 1 RM obtained at the first set of each week).

Measurement protocol and variables. The total measurement protocol of this study is shown in Figure 1. To test the present hypothesis, only those measurements of strength and muscle size relevant to the hypothesis were tested. The phenotypes under study were: 1 RM, elbow flexion strength: the maximal isometric moment at 110 ° arm flexion, the maximal moments at this angle during maximal concentric and eccentric muscle work at velocities of 30 °·s-1, 60°·s-1, and 120 °·s-1, as well as arm muscle cross-sectional area.

Figure 1-Measurement protocol. A, anthropometric measurements; S, strength evaluation on Promett; PA, Socio-economic and Physical Activity Questionnaire; CT, Computed Tomography scan of the upper arm; B, blood sampling (at week 2: 1, 24 and 48 h after training session 1); U = 24-h urine collection (at week 2: 1-24 h, 24-48 h, and 48-72 h after training session 1).* 1RM,;..... = weekly 7-d recall physical activity questionnaire:

The one repetition maximal value was determined as described above and the score of the 2nd and 10th week were included as pre- and post-training variables in the analyses. Pretraining evaluation of maximal static and dynamic voluntary contractions was done after 1 wk of adaptation to the training apparatus using low training loads (50-70% 1 RM) on an active programmable dynamometer (Promett, Biomechanics Laboratory, Faculty of Physical Education and Physiotherapy, Katholieke Universiteit, Leuven)(30). Post-training evaluations were done after the 10th week. With this system, isometric, concentric, and eccentric contractions were performed at different speeds and amplitudes imposed by the dynamometer. Subjects were seated in a comfortable standardized position strapped to the chair with the right arm resting on the measurement device. They were asked to build up their maximal isometric strength and to hold this maximum for 3 s. The highest registered moment during this contraction was selected as the maximal isometric strength measure expressed in Newton meter (N·m). Measurements were made consecutively at five different angles, covering nearly the whole movement range at the elbow. Test-retest correlations ranged from 0.93 at the extreme angles (170 ° and 50 ° flexion) to 0.97 at the middle angle (110 °, with 180 ° representing full arm extension). The observer was able to evaluate each subject's maximal effort by visualized moment and EMG signals registered from the m. biceps brachii, m. brachioradialis, m. brachialis, and m. triceps brachii. Isometric testing was followed by sets of maximal concentric and eccentric contractions ranging over an amplitude from 170 ° to 50 ° flexion and speeds of 30°·s-1, 60 °·s-1; 120°·s-1; and 240 °·s-1 (only in concentric contractions). Prestretch contractions were further tested at all speeds with eccentric muscle work from 110 ° to 170 ° and concentric work from 170° to 50 ° flexion. All dynamic measures started in full relaxation, were initiated with a small vibrating signal, and subjects were asked to perform maximal flexion moments ending in 2 s static maximal flexion at the end position of the handle until the handle returned to the starting position. Maximal torques at 110 ° arm flexion were selected because of the highest reliability and repeatability of the measurement at that angle and because the highest training loads were in the range of 165 ° to 90 ° of arm flexion. Evaluation of training effects in torques at smaller (80 °, 50°) or larger (170 °) angles would possibly not reflect the maximal training potential. Registered moments smaller than 10 N·m were excluded from the data set (0-7.3% of all cases in dynamic torques).

Mean muscle cross-sectional area was determined by Computed Tomography(28). Starting from the mid-humerus position, three scans were done at 3-cm intervals in the direction of the hand. A fourth scan was taken at the second position with the relaxed arm in 150 ° flexion. Tissue within the limits of -50 Houndsfield U (HU) and +200 HU was defined as muscle. Technical error of measurement for muscle area was 0.16 cm2 with a reliability of 0.99. The average muscle cross-sectional area of all four scans was used in the final analyses. To minimize radiation, only pre- and post-training values were taken.

Genetic analyses. Genotype*training interaction effects were tested by two approaches: 1) a two-way ANOVA for repeated measurements on the treatment effect in the MZ group and the intra-pair similarity in relative training responses (4), 2) the testing of alternative biometrical models on observed covariance matrices of pre- and post-training measurements in MZ and DZ twins. In the latter approach, phenotypic variation is a function of variation in genotype and in environment.

Sources of variation are considered additive genetic variation (A), which reflects the sum of the average effects of the individual alleles at all loci, dominance genetic effects (D), the environmental variation that is unique to an individual (E), and common environmental variation (C) which is shared by family members. Univariate model-fitting applied to data on MZ and DZ twins reared together can test for the significance of additive genetic variance, specific environmental factors, and either common environmental or dominance effects. Several alternative models were tested previously on pretraining phenotypes (29) indicating little evidence for common environmental influences on maximal strength. Although low DZ intrapair correlations could indicate genetic dominance factors, no evidence was found for a significant dominance contribution to the observed variance. In post-training phenotypes, dominant genetic factors seemed to be important in isometric torque at 110 ° flexion, and common environmental effects in eccentric strength at 120 °·s-1 (unpublished results). The latent factors in the bivariate analysis were, however, restricted to additive genetic factors (A) and unique environmental factors (E). Before model-fitting was applied, normality of the distribution as well as birth order effects and equality in means and total variance of both pre- and post-training phenotypes between MZ and DZ twins were tested. Two of the three alternative hypotheses in the bivariate analysis of pre- and post-training phenotypes are given inFigure 2. In panel A, both genetic and environmental factors are decomposed by the Cholesky factorization(14,20), in which a first factor (Ac and Ec) influences both occasions, while an independent second factor(As and Es) only influences the post-training phenotype. Evidence for genotype*training interaction is found when this post-training specific set of genetic factors (path coefficient as) is significant. A submodel (model Abis,) from model A tested if the loadings of acpr and acpo could be equated to each other. For the null hypothesis (model B), as shown in panel B of Figure 2, it was hypothesized that the specific genetic factor was not significant and that the loadings on the common genetic factor could be equated (acpr = acpo). This meant that both pre- and post-training phenotypes were influenced by the same genes in the same proportions without evidence for genotype*training interaction. The most parsimonious model was selected by the lowest Akaike's Information Criterion (26), a measure that evaluates simultaneously the goodness-of-fit of the model and its simplicity. Model parameters were estimated by maximum likelihood procedures in Mx (21). The contribution of each path to the variation in pre- and post-training phenotypes is given by its variance component. These genetic and environmental variance components are calculated as squared standardized path-coefficients, and all variance components of all paths pointing to a phenotype add up to one. The genetic correlation between both pre- and post-training measurements indicates the importance of the same set of genes in pre-training and trained conditions, or pleiotropic genetic action at both pre- and post-training phenotypes, while the environmental correlation indicates factors from the environment for each individual that are shared at both occasions (e.g., diet components, physical activity during work).

Figure 2-Bivariate genetic models for pre- and post-training phenotypes. In circles: A and E are latent additive genetic (A) and unique environmental factors (E). Observed pre- and post-training phenotypes are in squares. Small letters are path coefficients. MODEL A.: A:
c and E c represent genetic and environmental factors that are common to both pre- and post-training phenotypes. A c also serves as the unique source of genetic variation in the pretraining phenotype (E c as the only environmental). A s and E s represent genetic and environmental factors that contribute uniquely to the variation in the post-training phenotype. Significant inclusion of A s is indicative for genotype*training interaction(training-specific genes). MODEL B.: Paths of A c are equated for both phenotypes and represent complete genetic pleiotropy between measures. A c and A s are correlated 1.0 between both MZ twin members and 0.5 between both DZ twin members.


ANOVA on monozygotic twins. Since the Shapiro-Wilk test showed significant deviation from normality for 1 RM scores, ANOVA and models were therefore fitted to the logarithmic transformed values. The increase in the training-specific strength measure (1 RM) as well as in the maximal isometric strength at 110 ° flexion and all concentric torques at 110 ° flexion were highly significant (Table 2). Strength increases were on average 2.7 to 8.5 N·m (20% of the pretraining value), for 1 RM the average increase was 10.5 kg (45.8%). Training effects in maximal eccentric moments were not significant for the slowest contraction at 30°·s-1, and significantly smaller moments were observed after training for eccentric contractions at 60 °·s-1 and 120°·s-1. There was an average increase of 2.2 cm2(4.4%) in arm muscle area. Coefficients of variation in responses to training ranged from 34% for 1 RM to 504% in the eccentric strength at 30°·s-1 indicating a large variability in training responses.

Effects of training on strength and arm muscle indicators in 25 MZ twins: G*E interaction F-ratio test and twin resemblance in training response.

The F-ratio of the within pairs over between pairs sum of squares indicated 3.5 times more variation in 1 RM increase between MZ twin pairs than within pairs of identical genotype (P < 0.01). The response to training in maximal isometric contraction at 110 ° flexion is also significantly linked to the genotype as indicated by the significantF- ratio test (F = 1.83, P < 0.05). No genotype*training effect was found for any of the other dynamic strength tests or muscle hypertrophy. The MZ intra-pair resemblance in training responses showed a moderate correlation in 1 RM (0.46) and isometric strength (0.30) increases in MZ twins, while nonsignificant low (0.07 to 0.30) and negative correlations (-0.04 to -0.40) were found for isometric strength, dynamic strength, and muscle cross-sectional area.

Bivariate genetic model fitting. Means and variances in pre- and post-training phenotypes were equal in MZ and DZ twins, and zygosity effects were absent in the responses to training except for a larger increase in muscle cross-sectional area in DZ twins than in MZ twins (P < 0.05). Table 3 shows the goodness-of-fit indices(χ2, Akaike's Information Criterion) of three alternative models. A model excluding path a2cpo (not shown in Table 3) resulted in very high χ2 values, indicating that a path representing genetic covariance between pre- and post-training data should be included. The observed data were well described by the most parsimonious model except for the concentric strength measure at 60 °·s-1. The most parsimonious model indicated the presence of As, as a genotype*training interaction factor, for the training-specific 1 RM measure, isometric strength, and concentric strength at 120 °·s-1. There was no evidence for genotype*training interaction for concentric and eccentric strength at other velocities or for muscle cross-sectional area. The variance components contributing to the variation in pretraining phenotypes indicated a high contribution of genetic factors (heritability h2pre) for 1 RM (0.77), isometric strength (0.69), eccentric strength (0.65-0.77), and muscle cross-sectional area (0.85). Environmental factors accounted for the rest of the variation and their importance increased with velocity in concentric strength measures (0.50 at 30°·s-1 to 0.69 at 120 °·s-1). The component a2cpo, reflected the proportion of variation in post-training performance and muscle cross-sectional area that could be explained by the same genes that caused variation in the pretraining phenotype. The heritability h2post indicated the total contribution of genetic influences on the post-training phenotype and is calculated as the sum of a2s and a2cpo. The genetic correlation rg reflected the amount of genes with shared action on both pre- and post-training phenotypes. In the variables without genotypes*training interaction, the genetic correlation was, as defined by the model, 1.0. In these cases, the heritability of post-training phenotypes equated the contribution of a2cpo. For 1 RM, isometric strength at 110 ° and concentric strength at 120 °·s-1, genetic correlations were high (0.75-0.88), which was consistent with the low contribution of training-specific factors in the post-training phenotypes (a2s = 0.19-0.23). The heritability of 1 RM and concentric strength at 120°·s-1 increased over training by 4% and 23%, respectively, remained unchanged over training for muscle cross-sectional area and concentric strength at 30 °·s-1, and decreased over training for all other measurements. Environmental factors were to a smaller extent shared in pre- and post-training dynamic strength measurements (re = 0.03-0.38). For 1 RM, isometric strength and muscle cross-sectional area, moderate environmental correlations were found (re = 0.41-0.66).

Goodness-of-fit indices (χ2, Akaike's Information Criterion, Probability) of three alternative models testing for genotype training interaction. Maximum likelihood estimates of the proportions of explained variance in pre-training (a2 cpr, e2 cpr) and post-training (a2 cpo, a2 s, e2 po, e2 s) phenotypes and genetic (rg) and environmental(re) correlations of the most parsimonious model (underlined) (25 MZ, 16 DZ twins).


Training effects on 1 RM loads, isometric and concentric elbow torques, and muscle cross-sectional area were significant and showed a large interindividual variability. A large number of subjects did not improve their eccentric strength with this training program, whereas isometric and concentric strength, however, did improve considerably by about 20%. The lack of improvement in eccentric strength could result from lack of motivation or an increased pain sensation in performing eccentric muscle work at the end of the training program compared with that at the pretraining evaluation. Also, eccentric strength increases might be expected to be smaller than isometric and concentric strength increases, following a resistance-program of predominantly concentric muscle action. However, some part of the movement in lowering the load did include eccentric muscle work, which would be in favor of increases in eccentric muscle strength. Our results showed only weak evidence for contraction-type specificity of training, and adds to the contradictory results of previous studies on contraction-type specificity of different training modes, as reviewed by Morrissey et al.(19). Higher increases were found in 1 RM, compared to more objectively and standardized measurements of isometric and dynamic strength recorded on the programmable dynamometer system. Increases in strength were considerably larger than the reported muscle hypertrophic effects, this observation being consistent with most high-resistance training studies (13,24).

Both analytical approaches identified a genotype*training interaction for one repetition maximum performance and isometric strength. However, using model-fitting there was also evidence for a set of specific genetic factors for concentric elbow flexion at 120 °·s-1. There was no evidence for genotype-dependent training responses in the other concentric and eccentric torques or in muscle cross-sectional area. Although the present results seem to confirm each other, different information is obtained by both methods. Considering the repeated measurement ANOVA in MZ twins, Bouchard et al. (4) argue that adding DZ twins to the protocol would not contribute to the study design as a way of getting information or testing for common environmental effects as in the classical twin study design, because both groups of twins are exposed to similar environmental conditions during the duration of the experiment. However, if all twins are exposed to the same environmental stress, namely, the load that is relative to their own maximum 1 RM loads, in our opinion, this relative load would not be interpreted as a common environmental effect but rather as a generally constant stress factor that does not differ between twin pairs. In the biometrical approach, a common environmental factor is only detected when environmental factors shared by twin members cause differences between pairs of twins from different families. With the available data from monozygotic and dizygotic twins, Figure 3 shows a plot of increases in 1 RM for twin 1 against twin 2. The intra-pair correlation for strength gain in 1 RM load in DZ twins is about half that of the MZ twins, which is an indication for a significant but moderate contribution of additive genetic factors. However, for training effects for all other strength and muscle characteristics, DZ correlations were as high or higher as the MZ correlations, which would be consistent with a decomposition of the variation into unique and common environmental effects.

Figure 3-Intra-pair resemblance in monozygotic (128) and dizygotic(143) twins for strength training increase in 1RM load after 10 wk of strength training (r:
MZ = 0.49, P < 0.01; r DZ = 0.22, n.s.).

Using both pre- and post-training phenotypes in a model-fitting approach to explain observed covariations in MZ and DZ twin data has the advantage of 1) testing different hypotheses about genetic and environmental sources of variation, 2) evaluating the goodness-of-fit of the hypothesized model, and 3) quantifying the importance of genetic and environmental factors acting in prestressed phenotypes and in phenotypes under stress and their genetic and environmental correlations. Training-specific genetic factors were significant for the post-training measures and were estimated to explain about 20% of the variation in the achieved 1 RM and isometric strength measures.

Heritability estimates of arm strength and arm muscle cross-sectional area can only be compared with estimates derived from untrained relatives. Arm strength is often based on hand grip or arm pull, and arm circumferences have been used as an indicator for arm musculature. Heritability coefficients for arm strength reported herein were comparable with estimates found in twin studies using the same methodology. In a sample of 10-yr-old twins, 72% of the variation in arm pull could be attributed to additive genetic factors(16). When data of their parents were included, similar results were found for the broad heritability of arm pull; however, there was evidence for dominance genetic variation or for a reduced genetic transmission between both generations (16). Our heritability estimates were, however, higher compared with the genetic transmissibility component of 0.30 in isometric knee extension torque in a family study of Pérusse et al. (22), and the zero transmissibility estimate of hand grip strength by Devor and Crawford(9). The high heritability of arm muscle cross-sectional area is in concordance with the heritability estimates of arm circumferences from a longitudinal study of twins aged 10 to 14 yr(15).

To our knowledge only one study including only five monozygotic twins of a similar age range (17-26 yr) investigated genotype*training interaction in muscle strength (27). In this study twins were trained for 10 wk using an isokinetic knee flexion/extension protocol. Despite significant strength increases and considerable variation in training responses (24% ± 12%), the authors report evidence for genotype-dependent changes in muscle oxoglutarate dehydrogenase concentrations but not for strength increases in peak torque output, creatine kinase, muscle kexokinase, malate dehydrogenase, or 3-hydroxyacyl CoA dehydrogenase, as tested by the two-way ANOVA also used herein. The results of the present study, for dynamic strength measurements, are consistent with these observations. There is no clear explanation for the fact that maximal dynamic torques did not express genotype*environment interaction (except for concentric contraction at 120 °·s-1), while 1 RM and static strength did. Firstly, the lack of training effects in the mean value of a trait (like eccentric strength) does not necessarily reflect the absence of a significant MZ intra-pair correlation or a set of training-specific genetic factors. Secondly, training-specificity effects could not completely account for these results, although 1 RM and static strength measures could be seen as the most trained trait and the easiest contraction type, respectively. The higher complexity of dynamic muscle contractions versus static contractions could also induce more important neuromotor adaptations. These learning effects can themselves be determined be genetic and environmental factors and could lower the intra-pair similarity in strength gains over a period of training. Also a higher intra-individual variability in dynamic strength performances compared with static strength performances at both pre- and post-training evaluation increases the specific environmental variation component in an absolute or relative gain score. At present, it is difficult to give an exact interpretation of the underlying mechanisms of differences observed in static versus dynamic training effects or to assign possible different pathways of neural and muscular adaptations that could be linked to the presence or absence of these genes, or switching on of genes, specific to the different muscle contraction types.

In contrast to the findings on trainability of strength, responses in˙VO2max, 90-min, and 90-s work output after aerobic and anaerobic training protocols, respectively, depended largely on the genotype(7,12,23,25). Studies on endurance were therefore directed to the identification of polymorphisms or variants in gene products of the oxidative pathway in human muscle(3,11). Trends were found toward an improved aerobic and anaerobic trainability in muscle creatine kinase variants and adenylate kinase variants, respectively (3); however, no charge variants were detected in nine enzymes of the tricarboxylic acid cycle and 11 enzymes of the glycolytic pathway (2,17). There was also evidence for a polymorphism in mitochondrial DNA (subunit 5 of NADH gene), that resulted in smaller training responses in ˙VO2max(11).

We found no evidence for genotype*training interaction for muscle cross-sectional area, although small hypertrophic effects were present. It is possible that the strength training needs to be extended in time to reach more substantial increases in muscle tissue to detect gene*environment interaction. Studies on rodents, however, stress the importance of exercise-induced factors, such as insulin-like growth factors and thyroid hormone(10), that intervene in the transcriptional and translational processes that control contractile protein synthesis and degradation (1,8,17), resulting in muscle fiber hypertrophy and changes in myosin heavy chain isoform expressions. These exercise-induced factors should be characterized by genetic polymorphisms or by a specific set of combination of alleles (haplotypes of multiple genes) that are switched on by the exercise stimulus, to be detected in a separate set of genetic factors (As in model A).

One line of research to decipher the genetic basis of muscle strength and the evidence for the contribution of some training-specific genetic factors in strength training could therefore involve the study of genetic variation in gene products (enzymes, contractile proteins, etc.) and the study for polymorphisms in promoter regions that could influence gene expression or, more generally, the study of Quantitative Trait Loci.

Results from this study should be interpreted with the following limitations. It is possible that, although the sample in the present study is larger than previous studies on environmentally challenged MZ twins, we were not able to detect the presence of smaller contributions of training-specific genetic factors by the model-fitting procedures used. The present results are also specific to the characteristics of the training program, the elbow flexor muscles and to postpubertal, sedentary, or moderate active males. Further research is needed to confirm these findings in similar and other training programs.


1. Baldwin, K.M. Effects of altered loading states on muscle plasticity: what have we learned from rodents? Med. Sci. Sports Exerc. 28:S101-S106, 1996.
2. Bouchard, C., M. Chagnon, M. C. Thibault, M. R. Boulay, M. Marcotte, and J. A. Simoneau. Absence of charge variants in human skeletal enzymes of the glycolytic pathway. Hum. Genet. 78:100, 1988.
3. Bouchard, C., M. Chagnon, M. C. Thibault, M. R. Boulay, et. al. Muscle genetic variants and relationship with performance and trainability. Med. Sci. Sports Exerc. 21:71-77, 1989.
4. Bouchard, C., L. Pérusse, and C. Leblanc. Using MZ twins in experimental research to test for the presence of a genotype-environment interaction effect. Acta Gen. Medic. Gemell. 39:85-89. 1990.
5. Bouchard, C., F. T. Dionne, J-A. Simoneau, and M. R. Boulay. Genetics of aerobic and anaerobic performances. Exerc. Sport Sci. Rev. 20:27-58, 1992.
6. Bouchard, C. and L. Pérusse. Genetics of Obesity.Annu. Rev. Nutr. 13:337-354, 1993.
7. Boulay, M. R., G. Lortie, J.-A. Simoneau, and C. Bouchard. Sensitivity of maximal aerobic power and capacity to anaerobic training is partly genotype dependent. In: Sport and Human Genetics, Ch 11. R. M. Malina and C. Bouchard (Eds.). Champaign, IL: Human Kinetics, 1986, pp. 173-182.
8. Chambers R. L. and J. C. McDermott. Molecular basis of skeletal muscle regeneration. Can. J. Appl. Physiol. 21:155-184, 1996.
9. Devor, E. J. and M. H. Crawford. Family resemblance for neuromuscular performance in a Kansas Mennonite Community. Am. J. Physiol. Anthropol. 64:289-296, 1984.
10. Devor, S. T. and T. P. White. Myosin heavy chain phenotype in regenerating skeletal muscle is affected by thyroid hormone.Med. Sci. Sports Exerc. 27:674-681, 1995.
11. Dionne, F.T., L. Turcotte, M. C. Thibault, M. R. Boulay, J. S. Skinner, and C. Bouchard. Mitochondrial DNA sequence polymorphism. ˙VO2max, and response to endurance training.Med. Sci. Sports. Exerc. 23:177-185, 1991.
12. Hamel, P., J. A. Simoneau, G. Lortie, M. R. Boulay, and C. Bouchard. Heredity and muscle adaptation to endurance training. Med. Sci. Sports. Exerc. 18:690-696, 1986.
13. Kraemer, W. J., S. J. Fleck, and W. J. Evans. Strength and power training: physiological mechanisms of adaptation. Exerc. Sport Sci. Rev. 24:363-397, 1996.
14. Loehlin, J. C. The Cholesky Approach: a cautionary note. Behav. Genet. 26:65-69, 1996.
15. Loos, R., M. Thomis, H. H. M. Maes, et al. Genetic analysis of muscularity in early adolescence: sex-specific regional changes of the genetic structure with age. J. Appl. Physiol. 82:1802-1810, 1997.
16. Maes, H. H. M., G. P. Beunen, R. F. Vlietinck, et. al. Inheritance of physical fitness in 10-yr-old twins and their parents.Med. Sci. Sports Exerc. 28:1479-1491, 1996.
17. Marcotte, M., M. Chagnon, C. Coté, M. C. Thibault, M. R. Boulay, and C. Bouchard. Lack of genetic polymorphism in human skeletal muscle enzymes of the tricarboxylic acid cycle. Hum. Genet. 77:200, 1988.
18. McComas, A. J. Human neuromuscular adaptations that accompany changes in activity. Med. Sci. Sports Exerc. 26:1498-1509, 1994.
19. Morrissey, M. C., E. A. Harman, and M. J. Johnson. Resistance training modes: specificity and effectiveness. Med. Sci. Sports Exerc. 27:648-660, 1995.
20. Neale, M. C. and L. R. Cardon. Methodology for genetic studies of twins and families. Dordrecht: Kluwer, 1992 pp. 1-496.
21. Neale, M.C. Mx: Statistical Modeling, 3rd Edition, Richmond, VA: Department of Psychiatry, Virginia Commonwealth University, 1995, pp. 1-142.
22. Pérusse, L., G. Lortie, C. Leblanc, A. Tremblay, G. Thériault, and C. Bouchard. Genetic and environmental sources of variation in physical fitness. Ann. Hum. Biol. 14:425-434, 1987.
23. Prud'Homme, D., C. Bouchard, C. Leblanc, F. Landry, and E. Fontaine. Sensitivity of maximal aerobic power to training is genotype dependent. Med. Sci. Sports Exerc. 16:489-493, 1984.
24. Sale, D. G. Neural adaptations to strength training. In: Strength and Power in Sports. The Encyclopaedia of Sports Medicine. P. Komi (Ed.). Oxford, UK: Blackwell, 1992, pp. 249-265.
25. Simoneau, J. A., G. Lortie, M. R. Boulay, M. Marcotte, M. C. Thibault, and C. Bouchard. Inheritance of human skeletal muscle and anaerobic capacity adaptation to high-intensity intermittent training.Int. J. Sports Med. 7:167-171, 1986.
26. Tanaka, J.S. Multifaceted conceptions of fit in structural equation models. In: Testing Structural Equation Models. K. A. Bollen and J. S. Long (Eds.). Newbury Park, CA: Sage Publications, pp. 10-39. 1993.
27. Thibault, M. C., J. A. Simoneau, C. Côté, et al Inheritance of human muscle enzyme adaptation to isokinetic strength training. Hum. Hered. 36:341-347, 1986.
28. Thomis, M., A. L. Claessens, R. Vlietinck, G. Marchal, and G. Beunen. Accuracy of anthropometric estimation of muscle-crosssectional area of the arm in males. Am. J. Hum. Biol. 9:73-86, 1997.
29. Thomis, M., M. Van Leemputte, M. H. Maes, et al. Multivariate genetic analysis of maximal isometric muscle force at different elbow angles. J. Appl. Physiol. 82:959-967, 1997.
30. Van Leemputte, M. and E.J. Willems. EMG quantification and its application to the analysis of human movements. Med. Sports. Sci. 25:177-194, 1987.
31. Vlietinck, R. F. Determination of the Zygosity in Twins. Doctoral Dissertation. Leuven, Belgium: Katholieke Universiteit, 1986 pp. 1-124.


© Williams & Wilkins 1998. All Rights Reserved.