Share this article on:

Genetic Predisposition Scores Associate with Muscular Strength, Size, and Trainability


Medicine & Science in Sports & Exercise: August 2013 - Volume 45 - Issue 8 - p 1451–1459
doi: 10.1249/MSS.0b013e31828983f7
Clinical Sciences

Introduction The number of studies trying to identify genetic sequence variation related to muscular phenotypes has increased enormously. The aim of this study was to identify the role of a genetic predisposition score (GPS) based on earlier identified gene variants for different muscular endophenotypes to explain the individual differences in muscular fitness characteristics and the response to training in patients with coronary artery disease.

Methods Two hundred and sixty coronary artery disease patients followed a standard ambulatory, 3-month supervised training program for cardiac patients. Maximal knee extension strength (KES) and rectus femoris diameter were measured at baseline and after rehabilitation. Sixty-five single nucleotide polymorphisms (SNP) in 30 genes were selected based on genotype–phenotype association literature. Backward regression analysis revealed subsets of SNP associated with the different phenotypes. GPS were constructed for all sets of SNP by adding up the strength-increasing alleles. General linear models and multiple stepwise regression analysis were used to test the explained variance of the GPS in baseline and strength responses. Receiver operating characteristic curve analyses were performed to discriminate between high- and low-responder status.

Results GPS were significantly associated with the rectus femoris diameter (P < 0.01) and its response (P < 0.0001), the isometric KES (P < 0.05) and its response (P < 0.01), the isokinetic KES at 60°·s−1 (P < 0.05) and 180°·s−1 (P < 0.001) and their responses to training (P < 0.0001), and the isokinetic KES endurance (P < 0.001) and its change after training (P < 0.0001). The GPS was shown as an independent determinant in baseline and response phenotypes with partial explained variance up to 23%. Receiver operating characteristic analysis showed a significant discriminating accuracy of the models, including the GPS for responses to training, with areas under the curve ranging from 0.62 to 0.85.

Conclusion GPS for muscular phenotypes showed to be associated with baseline KES, muscle diameter, and the response to training in cardiac rehabilitation patients.

1Research Group for Cardiovascular and Respiratory Rehabilitation, Department of Rehabilitation Sciences, KU Leuven, Leuven, BELGIUM; 2Physical Activity, Sports and Health Research Group, Department of Biomedical Kinesiology, KU Leuven, Leuven, BELGIUM; 3Department of Cardiovascular Diseases, University Hospital Leuven, Leuven, BELGIUM; and 4Laboratory of Translational Genetics, Vesalius Research Center, Leuven, BELGIUM

Address for correspondence: Luc Vanhees, Ph.D., Department of Rehabilitation Sciences, KU Leuven, Tervuursevest 101 B15013001, Leuven, Belgium; E-mail:

Submitted for publication June 2012.

Accepted for publication January 2013.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (

Aging is characterized by a decline in functionality due to progressive loss of muscle tissue associated with a decrease in strength and force output. Low skeletal muscle strength has been shown to be an important predictor of all-cause mortality in healthy as well as diseased individuals (19,20). The increasing age of coronary artery disease (CAD) patients accompanied by “fear of moving” and hospitalization in these patients often results in a substantial loss of skeletal muscle mass and muscle strength. It has also been shown that CAD patients suffer from increased muscle fatigability (13). Regular physical activity in cardiac rehabilitation improves aerobic power and skeletal muscle strength and is associated with an increase in survival in these patients (15,18,37). However, individual differences in the response to rehabilitation are large (38), which may be partly related to genetic characteristics (28).

Heritability studies in humans have found a genetic contribution up to 66% to fat-free mass (FFM) (1) and up to 65% to muscle strength (21). Studies in older twins reported that heritability could explain 20% to 52% of the variance in handgrip strength, leg extensor power, and maximal walking speed (2,7,12,21,32–35). Heritability for handgrip strength seems to decrease with increasing age, with a concurrent increase of the relative contribution of environmental effects (7). However, when excluding individuals with age-related chronic diseases, heritability estimates were found to increase (12). Muscle cross-sectional area of upper and lower extremities among young individuals showed to have a heritability up to 85%–95% (17,29,30). In female twins, genetic effects accounted for 52%–84% of the explained variance in lean body mass (26). High heritability estimates have been found in different muscular phenotypes, with multivariate genetic studies showing a shared genetic effect for different muscular characteristics (9,29,31–33). Linkage studies performed on different muscular characteristics have found evidence for shared chromosomal regions (10).

For the last decade, the number of studies trying to identify genetic sequence variation related to muscular phenotypes has increased enormously. In some candidate genes like ACE, ACTN3, CNTF, and MSTN, specific variants have been repeatedly studied to test for associations in different population groups with different strength measures, albeit with varying success. In more recent years with the introduction of genome-wide association studies (16) and genome-wide linkage studies (11,31,34), novel genes for muscular strength-related phenotypes have emerged. However, to date, most studies have focused on associations between single SNP in single genes and strength phenotypes, with only few studies evaluating haplotype–phenotype associations (reviewed in 6). Most significant associations represented only a small proportion of explained variance in the studied strength phenotypes.

To the best of our knowledge, no study has been executed combining strength-increasing alleles of multiple “strength” genes into a genetic predisposition score (GPS) to search for possible associations with different muscular phenotypes in a nonathlete population. This approach has been performed previously in endurance phenotypes (23–25,39) and in the prediction of elite power-related performance (24).

Therefore, the aim of this study was to identify the role of a GPS based on earlier identified gene variants for different muscular endophenotypes (muscular structure, metabolism, cytokines, growth or differentiation factors, neurotropic factors, and hormones) to explain the individual differences in muscle characteristics and the response to physical training in CAD patients. The backward regression–based selection of phenotype-specific SNP to be included in each GPS was applied in this study, as was introduced by Bouchard et al. (5). A secondary aim of this study was to compare this approach with the construction of a total GPS based on all 54 genotyped SNP (GPS54). We hypothesized a significant contribution for the GPS to explain—at least in part—the variability in strength gain in CAD patients after rehabilitation.

Back to Top | Article Outline


Study sample and training intervention

All patients with CAD (acute myocardial infarction, percutaneous coronary intervention, coronary artery bypass grafting, and/or angina pectoris), who were submitted to the cardiac rehabilitation program at the University Hospital in Leuven, were invited to participate in the Cardiac Rehabilitation and Genetics of Exercise Performance II study. Patients with valve disease, congenital heart disease, major arrhythmia, pacemaker or ICD implantation, heart transplantation, or other cardiac diseases were excluded. Inclusions were held from October 2008 until January 2011. The study protocol was approved by the Ethical Committee of the Faculty of Medicine of the Catholic University of Leuven, and written informed consent was obtained from each participant. In total, 260 CAD patients, who had performed a graded cycle ergometer test with respiratory gas analysis until exhaustion, were included at baseline. Data at peak exercise were collected before and after 3 months of rehabilitation. Patients followed a standard ambulatory supervised cardiac rehabilitation program, three times per week for 3 months with 90 min per session involving cycling, running, arm ergometry, rowing, predominantly isotonic calisthenics, and relaxation. The mean ± SD training frequency (TF) was 2.26 ± 0.03 times per week, and each patient trained on average at an intensity of 80% ± 0.82% (training heart rate/peak heart rate) × 100, where the mean exercise heart rates of the last three exercise sessions and peak heart rates of the exercise testing after training were used. Two hundred and four patients completed the 3-month cardiac rehabilitation program and were included for the analyses of responses to training. The following tests were held before and after 3 months of rehabilitation.

Back to Top | Article Outline

Estimates for body composition

Six skinfolds located at the biceps, triceps, subscapula, suprailiac, mid-thigh, and medial-calf area were taken with a Harpenden caliper. Stature and weight were measured, and percentage of body fat and FFM were estimated using an OMRON handheld body fat monitor (Omron BF 300; OMRON, Matoukasa Co. Ltd., Japan).

Back to Top | Article Outline


Isokinetic testing equipment (BIODEX System 3 Pro; Biodex Medical Systems, Shirley, NY) was used to determine the maximal knee flexion and extension torque and muscular endurance. Isometric knee extension strength (KES) was measured at a 60° knee angle, isokinetic KES was measured at two contraction speeds (60°·s−1 and 180°·s−1), and quadriceps muscle endurance was assessed by the total work delivered during a 25 repetition knee extension–flexion bout at a contraction speed of 180°·s−1. Patients were seated in an upright position with hips and knees flexed 90°. Straps were firmly fastened around the chest, hips, and upper leg to stabilize the trunk and leg. Verbal encouragement was given to achieve maximal effort. Because of a technical problem with the BIODEX testing device, the muscle strength of 14 patients at baseline and 15 patients postrehabilitation could not be tested. This led to a loss of 29 tests for the response in muscular strength to training.

Back to Top | Article Outline

Measurements of rectus femoris diameter

Rectus femoris (RF) diameter was measured by an M-mode ultrasonography wall-tracking ultrasound system (Siemens Vivid 07 GE Siemens, Munich, Germany) with a 12-MHz linear array transducer (12 L transducer GE). The transducer was placed perpendicular to the long axis of the thigh with excessive use of contact gel and minimal pressure to avoid compression of the muscle (3,8). The diameter of the RF was measured at the half point of the length between the epicondylus lateralis and the trochanter major of the femur. Measurements were taken on the patient’s right leg with the patient lying in a supine position with both knees extended but relaxed and toes pointing the ceiling. A set of five pictures was taken and further analyzed offline. Both pre- and postrehabilitation ultrasound measurements were analyzed blind and at random. At baseline, 246 patients were measured and 173 after training. The main causes for missing values were technical in nature (a server crash of the ultrasound system or the inability to visualize the inner wall of the RF with ultrasound). All ultrasound measurements were performed by a single experienced investigator (T.T.), and this method was validated against CT in a similar population (27).

Back to Top | Article Outline


Anonymously coded blood samples were drawn from each patient. Genotyping was performed in a blinded manner using iPLEX technology on a MassARRAY Compact Analyser (Sequenom Inc., San Diego, CA). The selection of SNP was based on recent review articles, genome-wide association studies, and genome-wide linkage studies up to January 2011 in which potential candidate genes, SNP, and regions were identified for aerobic capacity, muscular strength, or muscular endurance as phenotypes (5,6,10,11,25,31,36). Sixty-five SNP in 30 genes were selected for genotyping based on earlier associations with related phenotypes (Table 1, Supplemental Digital Content 1,, Selected candidate genes and SNP). Nine SNP had high linkage disequilibrium with other SNP of the same gene, and two SNP had a genotyping success rate less than 95%. Fifty-four SNP were therefore withheld for further analysis.

Back to Top | Article Outline

Statistical analyses

Data were analyzed using SAS statistical software version 9.2 for Windows (SAS Institute Inc, Cary, NC). Data were reported as mean ± SD for anthropometric measurements, RF diameter, and muscle strength measurements or as number of patients with percentage for dichotomous variables. To test whether the observed genotype frequencies were in the Hardy–Weinberg equilibrium, a χ 2 test with one degree of freedom was used. Because multiple testing induces false-positive or false-negative associations and the correction of P values accordingly will lower the power to identify small genetic effects, “increasing allele” genetic predispositions score (GPS) analysis was performed in which the number of increasing alleles was regressed against the phenotypes of interest. On the basis of our data, backward regression analysis was first applied to detect subsets of SNP to be associated with the different muscular phenotypes and for which the GPS were calculated. Only these significant contributing SNP were included in the GPS. The number of significant contributing SNP and its following GPS were therefore different between the different phenotypes. An additive effect was hypothesized, and equal weights were given for each increasing allele because no well-defined effect sizes were known for the different SNP and weighting of increasing alleles might only have limited effects (14). GPS was calculated for each individual by adding the increasing alleles (0, 1, or 2). The evidence for the GPS influence on the phenotypes under study was analyzed in a general linear model approach (ANCOVA) with age, sex, height, and FFM as covariates for baseline measurements. Age, sex, height, change in FFM, training intensity, and TF were used as covariates for the response to training. Stepwise multiple regression analysis was performed to check for the added explained variance by the GPS for the muscular phenotypes and the response to training.

For all response phenotypes, we performed high- versus low-responder analyses. Patients were categorized in the upper 25% (high-responder group) and lower 25% (low-responder group) using sex-specific cutoff scores. Different logistic regression models were used to test the power of the GPS to discriminate high-responder versus low-responder group status and to produce the receiver operating characteristic (ROC) curve and the area under the ROC curve (AUC). Adjusted odds ratios were calculated to estimate the effect of the GPS as the odds per increasing allele to belong to the high-responder group.

As an alternative approach, we created a total GPS (GPS54) based on the total set of 54 SNP for isokinetic KES at 180°·s−1 to compare the amount of explained variance of an overall approach (GPS54) with the GPS based on a significant subgroup of SNP after backward regression. Isokinetic KES at 180°·s−1 was used because this is a measurement that has a widespread use within muscle testing and because the training regimen in cardiac rehabilitation was predominantly dynamic in nature. ANCOVA analysis and multiple regression were used to test the association and percentage explained variance of GPS54 in baseline and response to training values. The power of the total GPS54 to discriminate between high and low responders was analyzed using ROC and AUC, and adjusted odds were calculated. All statistical tests were performed two-sided at a significance level of 5%.

Back to Top | Article Outline


Descriptive baseline patient characteristics, cardiac history, medication use, and response to 3 months of cardiac rehabilitation are described in Table 1. Baseline characteristics did not differ between the patients who dropped out of the study and the patients who participated for three months. Peak V˙O2 increased by 21.6% ± 15.9% (P < 0.001) and peak heart rate by 8.2% ± 11.9% after 3 months of training, and isometric KES increased by 11.5% ± 16.0% (P < 0.0001), isokinetic KES (60°·s−1) by 17.0% ± 23.1% (P < 0.0001), isokinetic KES (180°·s−1) by 16.5% ± 20.2% (P < 0.0001), and isokinetic quadriceps muscle endurance by 18.8% ± 23.3% (P < 0.0001) (Table 2). RF diameter was increased by 5.4% ± 11.2% (P < 0.0001). Body weight remained similar, whereas the relative and absolute values of body fat decreased significantly by approximately 3% (P < 0.001) (Table 2).





Backward regression analysis revealed subsets of SNP to be significantly associated with the respective muscular phenotypes at baseline and the response to training (strength and RF diameter). In particular, we identified 2 SNP for baseline isometric and isokinetic KES (60°·s−1) and up to 11 SNP for the response of RF diameter and knee extensor endurance strength. Table 2 (Supplemental Digital Content 2, gives an overview of the SNP included in the GPS for each baseline and response phenotype. Only these significantly contributing SNP were included in the GPS. To avoid the possibility of false-positive or false-negative results and the lack of statistical power by the smaller groups at the two tails—small number of patients with either a small or high number of increasing alleles—we combined the two lower and two upper GPS groups, respectively. Results of ANCOVA analyses of the GPS influence on baseline and response variables, with age, sex, height, and FFM as covariates for baseline measurements and age, sex, height, change in FFM, training intensity, and TF for response measurements, are shown in Figure 1. The procedure general linear model showed that each GPS was significantly associated with RF diameter (P < 0.01) and RF diameter response (P < 0.0001), isometric KES (P < 0.05) and change in isometric KES (P < 0.01), isokinetic KES at 60°·s−1 (P < 0.05) and 180°·s−1 (P < 0.001) and their respective response to training (P < 0.0001), and isokinetic knee extensor muscle endurance (P < 0.001) and its change after 3 months of training (P < 0.0001).



Stepwise multiple regression analysis showed a total explained variance between 36% and 57% for the baseline muscle phenotypes (Table 3). GPS was found as an independent determinant in all baseline muscle phenotypes, except for baseline isometric KES with partial r between 0.16 and 0.30 (Table 3). The significant b coefficients in Table 3 indicate that each increasing allele in the GPS results in an additional 6.65 N·m, 3.47 N·m, 124.62 J, and 0.04 cm in baseline isokinetic KES (at 60°·s−1 and 180°·s−1), baseline muscle endurance, and baseline RF diameter, respectively. GPS54 was not significantly related to baseline isokinetic KES at 180°·s−1. Table 4 shows the multiple stepwise regression analysis for the training response parameters. Total explained variance ranged between 6% and 26%, with GPS as a significant independent determinant in all response phenotypes (r between 0.25 and 0.48). Each additional increasing allele adds 6.92% to the gain in isometric KES, 1.38% in isokinetic KES at 60°·s−1, 6.60% in isokinetic KES at 180°·s−1, 7.60% in knee extension endurance, and 3.38% to the gain in RF diameter. Analyses of the high- versus low-responder groups were performed for the GPS of all phenotypes after training. For all response phenotypes, GPS was the only statistically significant contributing variable, except for the response in isokinetic KES at 180°·s−1 for which change in FFM was an additional significant predictor. Results per phenotype are shown in Table 5.







In addition, we performed an analysis on the GPS54 score based on the 54 selected SNP, for isokinetic KES at 180°·s−1 and the response after training. Although there was a theoretical spread of GPS54 score between 0 and 108, the GPS54 ranged from 42 to 60 for baseline and between 40 and 65 for the sample with response data. The groups at the two tails were combined, and the GPS54 was analyzed with a spread from 46 to 57 for baseline and 47 to 58 for the response. When GPS54 was analyzed without covariates for isokinetic KES (180°·s−1) at baseline, a trend could be observed that a higher GPS54 results in a higher baseline KES (P = 0.06). When age, sex, height, and FFM were added as covariates the model was significant (P < 0.001), however, GPS54 had no significant contribution (P = 0.50). For the response in isokinetic KES (180°·s−1), GPS54 was a significant (P < 0.001) independent variable when analyzed without covariates and the model with covariates was also significant (P < 0.01). In the latter procedure, GPS was the only significant contributor to the model (partial P = 0.0029). When GPS54 was entered into a multiple stepwise regression analysis with all covariates to explain response in isokinetic KES at 180°·s−1, GPS54 had a partial R 2 of 9.5% and age had a partial R 2 of 2.6%. GPS54 contributed significantly to the distinction between high- and low-responder groups (AUC = 0.68, 95% confidence interval [CI] = 0.56–0.80; adjusted odds ratio = 1.23, 95% CI = 1.06–1.42) (Fig. 1, Supplemental Digital Content 3,, Overlay of four different models to discriminate high versus low responder in isokinetic KES [180°·s−1] after training). Change in FFM also contributed significantly to the high response status in isokinetic KES at 180°·s−1 (AUC = 0.64, 95% CI = 0.51–0.76, P < 0.05). When adding change in FFM to the model with GPS54, the AUC increased to 0.72 (95% CI = 0.61–0.84), but the model did not differ significantly from the model with GPS54 alone (P = 0.28).

Back to Top | Article Outline


Because most previous genetic association analysis studies mainly focus on one phenotype and one gene (variant), the first aim of this study was to identify combinations of gene variants that were associated with different muscular phenotypes and to quantify the degree of explained variances of these GPS for the variability in strength and strength gains in CAD patients. In this study, we found an increase in muscle strength between 11% and 17% and in RF diameter of 5% after 3 months of rehabilitation in CAD patients with a predominantly aerobic exercise-training program. However, a large interindividual variability could be observed.

Previous studies applying the candidate gene approach have found significant associations of SNP with different muscular phenotypes, but with only limited explained variance by a single SNP. Recent genome-wide studies have identified novel SNP and regions of interest for associations with muscular phenotypes, providing additional potential genetic information. However, muscular strength phenotypes are multifactorial and polygenic traits. To the best of our knowledge, this was the first study to search for combinations of SNP in different muscular (endophenotype)-related genes. By means of backward regression analysis, we were able to identify 10 sets of SNP to be associated with different muscular phenotypes. rs17602729 in the AMPD1 gene and rs1016732 and rs2854248 in the ATP1A2 gene showed a large overlap between the different strength phenotypes. Rs17602729 of the AMPD1 gene has previously been shown to be associated with a diminished aerobic capacity and cardiorespiratory response to exercise (22,28) and a decrease in exercise duration for 20 years (25). Likewise, markers from ATP1A2 were associated with a decrease in exercise duration (25). The majority of SNP included in the GPS for these muscular phenotypes were located in genes that were functionally categorized into muscle metabolism or muscle growth and differentiation.

ANCOVA analyses showed that all created GPS were significantly associated with their respective phenotype under study. At baseline, CAD patients with a higher GPS showed higher baseline isometric and isokinetic knee extensor strength, with every additional increasing allele accounting for a surplus in strength. Subjects with a higher GPS also showed a higher muscle strength or muscular endurance response to training. Moreover, we were able to show that GPS contributes significant to the discrimination between high- and low-responder status for the phenotypes under study. GPS had a significant AUC between 0.62 and 0.85 and adjusted odds ratio between 1.90 and 2.84 for the different muscular phenotypes. A patient with a high GPS has a higher probability to end up in the group with the 25% highest response after training. For all phenotypes under study, GPS was the best independent variable associated with a high-responder status; and except for isokinetic KES at 180°·s−1, it was the only contributor to high response values.

The GPS was an independent determinant for all, except isometric extensor strength, baseline phenotypes with a partial R 2 ranging between 3% and 9% (Table 3). For the response to training, only the change in FFM and age were previously found as determining variables for some response phenotypes (Table 4). The partial explained variance by adding the GPS for the response to training had an R 2 between 6% and 23%. The higher explained variance by the GPS of the difference in response to training, compared with the baseline measurement, might be explained by the higher number of other determinants at baseline. At baseline, the largest part of the variability in muscle strength and diameter could be explained by covariates such as age, sex, height, and FFM, which are already under genetic influence. Furthermore, effect sizes of gene variants might be larger when the muscular system is challenged to be active in repair and metabolic optimization in response to regular training compared with a stable baseline condition.

Related to the secondary aim of this study, we compared two strategies to determine the GPS. The option to determine each GPS phenotype-specific based on significantly contributing SNP is less practical in further applications of this approach as the set of SNP differs according to different phenotypes. This approach might also inflate the probability of finding significant GPS–phenotype associations. An overall GPS based on all 54 SNP (GPS54) was calculated and compared with the GPS based on the significant subsets of SNP for isokinetic KES at 180°·s−1 at baseline and the response after training. Indeed, the explained variance in response to training was higher when the GPS was based on the significant subgroup of SNP (13.7%) when compared with GPS54 (9.5%). Both approaches proved GPS to discriminate significantly (P < 0.01) between high and low responders to training with a similar AUC of 0.70 (95% CI = 0.58–0.81) and 0.68 (95% CI = 0.56–0.80) (Supplemental Digital Content, Fig. 1, Overlay of four different models to discriminate high versus low responder in isokinetic KES [180°·s−1] after training) for subgroup and total group of SNP respectively. However, the adjusted odds ratio is clearly in favor of the smaller group of SNP, 1.90 (95% CI = 1.24–2.93), versus GPS54, 1.23 (95% CI = 1.06–1.42). Constructing a GPS based on 54 SNP induces background “noise,” meaning that subjects might carry a small or larger amount of “increasing” alleles, whereas part of those alleles are only having very limited effect or no effect on the phenotype. For the GPS54 score groups, a nonlinear GPS–phenotype curve is observed (plateau in response scores for GPS54 scores 51 up to 56, results not shown), which also induces a lower degree of explained variance (R 2). By first using a backward regression analysis on the total group of SNP, we and others who applied this similar approach (5) were able to filter out the SNP that had very small contributions or no influence on the phenotype under study. Inherent to the chosen construction of the GPS (whether as a total set of SNP or subset of SNP), we assumed allelic effects to be codominant and each SNP to have an equal additional effect. Also by using GPS, information on which combination of increasing alleles is responsible for the specific effect is lost, as individuals with four increasing alleles might possess those from various combinations of increasing alleles of different gene variants. Finally, we are aware that the selection of genes in the initial gene list might have left out other SNP that might be associated with muscular phenotypes.

There are some limitations to this study. This study was performed in a predominantly Caucasian, older, male CAD population. According to required sample sizes in genetic epidemiological studies, the sample size is small (4). This approach therefore needs further replication, and results might be expected to be different in other populations or in response to other exercise regimens. The cardiac rehabilitation training was mostly focused on aerobic training with a smaller part of dynamic resistance training (3 sets of 15 reps) of the lower extremities with own body weight as resistance. Unfortunately, some measuring devices had technical errors; so in some individuals, not all tested parameters were available. In addition, there is a reasonable dropout rate in patients after cardiac rehabilitation due to work resume or long distance to the rehabilitation center, which led to a lower number of response measurements. However, subanalyses of dropout versus nondropout groups did not reveal significant differences between both groups at the start of the rehabilitation.

Back to Top | Article Outline


Constructions of multiple SNP into GPS for different muscular phenotypes showed to be associated with the baseline muscle strength, the muscle diameter, and the response of these parameters to training in cardiac rehabilitation patients. The GPS could explain up to 23% of the variance in these muscular phenotypes and was able to discriminate high- versus low-responder status on different muscular phenotypes. Phenotype-specific GPS selected by backward regression show higher GPS–phenotype associations compared with the application of a GPS based on all listed gene variants.

This study was supported by the Fund for Scientific Research–Flanders “Fonds voor Wetenschappelijk Onderzoek–Vlaanderen,” Belgium (F.W.O. grant nos. G.0624.08 and G.0124.02). The authors declare no conflict of interest. Results of the present study do not constitute endorsement by the American College of Sports Medicine.

Tom Thomaes and Martine Thomis are equally contributing first authors.

Back to Top | Article Outline


1. Abney M, McPeek MS, Ober C. Broad and narrow heritabilities of quantitative traits in a founder population. Am J Hum Genet. 2001; 68 (5): 1302–7.
2. Arden NK, Spector TD. Genetic influences on muscle strength, lean body mass, and bone mineral density: a twin study. J Bone Miner Res. 1997; 12 (12): 2076–81.
3. Arts IM, Pillen S, Schelhaas HJ, Overeem S, Zwarts MJ. Normal values for quantitative muscle ultrasonography in adults. Muscle Nerve. 2010; 41: 32–41.
4. Bouchard C. Overcoming barriers to progress in exercise genomics. Exerc Sport Sci Rev. 2011; 39 (4): 212–7.
5. Bouchard C, Sarzynski MA, Rice TK, et al. Genomic predictors of the maximal O2 uptake response to standardized exercise training programs. J Appl Physiol. 2011; 110 (5): 1160–70.
6. Bray MS, Hagberg JM, Perusse L, et al. The human gene map for performance and health-related fitness phenotypes: the 2006–2007 update. Med Sci Sports Exerc. 2009; 41 (1): 35–73.
7. Carmelli D, Reed T. Stability and change in genetic and environmental influences on hand-grip strength in older male twins. J Appl Physiol. 2000; 89 (5): 1879–83.
8. Delaney S, Worsley P, Warner M, Taylor M, Stokes M. Assessing contractile ability of the quadriceps muscle using ultrasound imaging. Muscle Nerve. 2010; 42: 530–8.
9. De Mars G, Thomis MA, Windelinckx A, et al. Covariance of isometric and dynamic arm contractions: multivariate genetic analysis. Twin Res Hum Genet. 2007; 10 (1): 180–90.
10. De Mars G, Windelinckx A, Huygens W, et al. Genome-wide linkage scan for maximum and length-dependent knee muscle strength in young men: significant evidence for linkage at chromosome 14q24.3. J Med Genet. 2008; 45 (5): 275–83.
11. De Mars G, Windelinckx A, Huygens W, et al. Genome-wide linkage scan for contraction velocity characteristics of knee musculature in the Leuven Genes for Muscular Strength study. Physiol Genomics. 2008; 35 (1): 36–44.
12. Frederiksen H, Gaist D, Petersen HC, et al. Hand grip strength: a phenotype suitable for identifying genetic variants affecting mid- and late-life physical functioning. Genet Epidemiol. 2002; 23 (2): 110–22.
13. Ghroubi S, Chaari M, Elleuch H, et al. The isokinetic assessment of peripheral muscle function in patients with coronary artery disease: correlations with cardiorespiratory capacity. Ann Readapt Med Phys. 2007; 50: 295–301.
14. Janssens AC, Moonesinghe R, Yang Q, Steyerberg EW, van Duijn CM, Khoury MJ. The impact of genotype frequencies on the clinical validity of genomic profiling for predicting common chronic diseases. Genet Med. 2007; 9: 528–35.
15. Kavanagh T, Mertens DJ, Hamm LF, et al. Prediction of long-term prognosis in 12 169 men referred for cardiac rehabilitation. Circ. 2002; 106: 666–71.
16. Liu XG, Tan LJ, Lei SF, et al. Genome wide association and replication studies identified TRHR as an important gene for lean body mass. Am J Hum Genet. 2009; 84, 418–23.
17. Loos R, Thomis M, Maes HH, et al. Gender-specific regional changes in genetic structure of muscularity in early adolescence. J Appl Physiol. 1997; 82 (6): 1802–10.
18. Myers J, Prakash M, Froelicher V, Do D, Partington S, Atwood JE. Exercise capacity and mortality among men referred for exercise testing. N Engl J Med. 2002, 346: 793–801.
19. Rantanen T. Muscle strength, disability and mortality. Scand J Med Sci Sports. 2003; 13: 3–8.
20. Rantanen T, Harris T, Leveille SG, et al. Muscle strength and body mass index as long-term predictors of mortality in initially healthy men. J Gerontol A Biol Sci Med Sci. 2000; 55: M168–73.
21. Reed T, Fabsitz RR, Selby JV, Carmelli D. Genetic influences and grip strength norms in the NHLBI twin study males aged 59–69. Ann Hum Biol. 1991; 18 (5): 425–32.
22. Rico-Sanz J, Rankinen T, Joanisse DR, et al. Associations between cardiorespiratory responses to exercise and the C34T AMPD1 gene polymorphism in the HERITAGE Family study. Physiol Genomics. 2003; 14: 161–6.
23. Ruiz JR, Gómez-Gallego F, Santiago C, et al. Is there an optimum endurance polygenic profile? J Physiol. 2009; 587: 1527–34.
24. Santiago C, Ruiz JR, Muniesa CA, González-Freire M, Gómez-Gallego F, Lucia A. Does the polygenic profile determine the potential for becoming a world-class athlete? Insights from the sport of rowing. Scand J Med Sci Sports. 2009; 20: e188–94.
25. Sarzynski MA, Rankinen T, Sternfeld B, et al. Association of single-nucleotide polymorphisms from 17 candidate genes with baseline symptom-limited exercise test duration and decrease in duration over 20 years: the Coronary Artery Risk Development in Young Adults (CARDIA) fitness study. Circ Cardiovasc Genet. 2010; 3 (6): 531–8.
26. Seeman E, Hopper JL, Young NR, Formica C, Goss P, Tsalamandris C. Do genetic factors explain associations between muscle strength, lean mass, and bone density? A twin study. Am J Physiol. 1996; 270 (2 Pt 1): E320–7.
27. Thomaes T, Thomis MA, Onkelinx S, Coudyzer W, Cornelissen V, Vanhees L. Reliability and validity of the ultrasound technique to measure the rectus femoris muscle diameter in older CAD-patients. BMC Med Imaging. 2012; 12 (1): 7.
28. Thomaes T, Thomis MA, Onkelinx S, et al. A genetic predisposition score for muscular endophenotypes predicts the increase in aerobic power after training: the CAREGENE study. BMC Genet. 2011; 12: 84.
29. Thomis MA, Van Leemputte M, Maes HH, et al. Multivariate genetic analysis of maximal isometric muscle force at different elbow angles. J Appl Physiol. 1997; 82 (3): 959–67.
30. Thomis MA, Beunen GP, Maes HH, et al. Strength training: importance of genetic factors. Med Sci Sports Exerc. 1998; 30 (5): 724–31.
31. Thomis MA, De Mars G, Windelinckx A, et al. Genome-wide linkage scan for resistance to muscle fatigue. Scand J Med Sci Sports. 2011; 21 (4): 580–8.
32. Tiainen K, Sipila S, Alen M, et al. Heritability of maximal isometric muscle strength in older female twins. J Appl Physiol. 2004; 96 (1): 173–80.
33. Tiainen K, Sipila S, Alen M, et al. Shared genetic and environmental effects on strength and power in older female twins. Med Sci Sports Exerc. 2005; 37 (1): 72–8.
34. Tiainen KM, Perola M, Kovanen VM, et al. Genetics of maximal walking speed and skeletal muscle characteristics in older women. Twin Res Hum Genet. 2008; 11 (3): 321–34.
35. Tiainen K, Sipila S, Kauppinen M, Kaprio J, Rantanen T. Genetic and environmental effects on isometric muscle strength and leg extensor power followed up for three years among older female twins. J Appl Physiol. 2009; 106 (5): 1604–10.
36. Timmons JA, Knudsen S, Rankinen T, et al. Using molecular classification to predict gains in maximal aerobic capacity following endurance exercise training in humans. J Appl Physiol. 2010; 108 (6): 1487–96.
37. Vanhees L, Fagard R, Thijs L, Staessen J, Amery A. Prognostic significance of peak exercise capacity in patients with coronary artery disease. J Am Coll Cardio. 1994; 23: 358–63.
38. Vanhees L, Stevens A, Schepers D, Defoor J, Rademakers F, Fagard R. Determinants of the effects of physical training and of the complications requiring resuscitation during exercise in patients with cardiovascular disease. Eur J Cardiovasc Prev Rehabil. 2004; 11 (4): 304–12.
39. Williams AG, Folland JP. Similarity of polygenic profiles limits the potential for elite human physical performance. J Physiol. 2008; 586: 113–21.


© 2013 American College of Sports Medicine