Lack of association between I/D dimorphism in the АСЕ gene and success in a chosen sport

Full Text

BACKGROUND

In recent decades, numerous attempts have been made to identify genes that determine various morphofunctional and psychophysiological traits associated with outstanding athletic performance [1, 2]. Montgomery et al. [3] and Rivera et al. [4] were the first to study the genetic determinants of athletic performance. In 1997, Bouchard et al. [5] published the first monograph on the genetic basis of physical activity— Genetics of Fitness and Physical Performance. Case-control association studies are based on comparing the frequency of alleles/genotypes in specific genes or extensive genomic regions between athletes and non-athletes. The results of these studies were used to identify specific gene variants associated with an athlete’s qualification or their physiological characteristics. Several editions of genetic maps have been published, clearly demonstrating the association or linkage with the phenotype of athletes [6]. Some of the most studied aspects in sports genetics are the insertion-deletion dimorphism (indel) in the angiotensin-I converting enzyme gene (ACE I/D) and the C/T polymorphism in the alpha-actinin-3 gene (ACTN3, rs1815739). Earlier, we conducted a similar test for the C/T polymorphism of the ACTN3 gene [7].

The very concept of an elite athlete does not have a clear or unambiguous definition, because of this, it is difficult to compare various case-control studies, comparing the elite athlete group to the controls [8]. The lack of clear phenotypic (anthropometric, physiological, ethnic, etc.) characteristics of the examined groups of athletes is one of the bottlenecks in sports genetics. The absence of a clearly defined phenotype inherent to a top-level athlete warrants the development of phenomics, a new field of science at the intersection of sports physiology, psychology, anthropology, and genetics. Phenomics aims to accumulate and analyze multidimensional data on various characteristics of athletes at the organism level [9].

Reasonable criticism is also caused by the inclusion of individuals who do not have any significant sports achievements at the time of the study in the control group of non-athletes [10]. This group often consists of volunteers who have never been engaged in any professional sport and led a sedentary lifestyle. That is why it is impossible to realistically assess their predisposition to high athletic performance. It would probably be advisable to form a control group consisting of individuals who were engaged in the specific sport, but without any significant results, such as being awarded the title of Candidate Master of Sports.

Most of the studies assume that there are alleles (genotypes) that enhance an individual’s speed-strength qualities, and opposing genetic variants that enhance aerobic qualities. Conventionally speaking, the existence of the genotype of an outstanding sprinter and the opposite genotype of an outstanding long-distance runner is suggested. Sports that require an individual to simultaneously exhibit high-speed strength and aerobic qualities (all-around events, complex coordination sports, martial arts) are not suitable for this approach. Due to the great diversity of sports, sports disciplines, and sports roles, it is necessary to assess the potential of sports genomics in creating sets of genetic markers that increase an individual’s chances of achieving high sports performance in a chosen sport.

AIM. The aim of the study was to evaluate the utility of the ACE gene I/D dimorphism (rs1799752) as a predictive marker of exceptional athletic performance, based on the analysis of the available scientific sources.

MATERIALS AND METHODS

The search was conducted using the principles of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) [11]. Publications for the analysis were retrieved using the keywords: ACE, sport genetics, athletes, SNP, sport selection in the PubMed, Google Scholar, and eLIBRARY databases The analysis included full-text articles that met the study objectives. As a result of the search, 60 studies were selected, and 47 studies were excluded from the analysis due to insufficient data on control groups. The final sample included 13,776 individuals (3536 athletes and 10,240 controls) [12–35].

For each analyzed study, the correspondence of genotype frequencies with the Hardy–Weinberg equilibrium (HWE) was checked. Mid--p-values, i.e., the exact p-values adjusted for the conservativeness of exact tests [36], were calculated using an online program (https://www.cog-genomics.org/software/stats). It is known that p-values reflect neither the probability of the absence of an effect (the probability of the null hypothesis), nor the sign of the effect, or its size. Therefore, interval estimation of the size of the effect is more informative and has long become a mandatory procedure in statistical analysis. One of the main indicators of observed genotype frequency deviation from Hardy–Weinberg equilibrium is the fixation index F_IS (inbreeding coefficient). Thus, to check the overall correspondence of genotype frequencies with Hardy–Weinberg equilibrium for F_IS, 95% confidence intervals (CI) were calculated and checked to see whether they cover the equilibrium value of F_IS=0 or not. To verify the concordance of the frequencies of each genotype with those expected from HWE, 95% CIs were calculated and checked to see whether they cover the expected values or not. To assess the equality of genotype or allele frequencies in the compared groups, a 95% CIs for the difference in D frequencies were calculated and checked whether they cover the indifferent value Δ=0 or not. To calculate these parameters, we used the original FixIndAll program, which the authors will provide to interested researchers upon request. Unlike other similar programs, in FixIndAll analysis of genotype and allele frequencies, their comparisons, and accordance with Hardy–Weinberg equilibrium is based on the calculation of Bayesian intervals of statistically permissible values (credible intervals) for them, for their differences, and for F_IS.

For meta-analysis, the MetaGenyo online program (https://metagenyo.genyo.es/) was used [37]. In addition to the usual assessment of the pooled effect (OR, odds ratio), 95% prediction intervals (PI) were calculated using the Meta-Essentials package (https://www.erim.eur.nl/research-support/meta-essentials/) [38] and/or CMA Prediction Intervals (https://meta-analysis-workshops.com/pages/predictionintervals).

RESULTS

A statistically significant deviation from Hardy–Weinberg equilibrium was detected in nine cases in the athlete subgroups and in six controls (mid-p < 0.05). Of them, five and three were with mid-p < 0.01, respectively. In 56 subgroups, the fixation index (F_IS) significantly deviated from zero, indicating either inbreeding, outbreeding, and/or an excessively wide 95% confidence interval, suggesting probable genotyping errors. In three cases [12, 17, 33], statistically significant differences in allele frequencies (Δ) were found between the groups of athletes and the control group. However, these studies demonstrated a significant deviation of F_IS from zero, and in the study by Varillas-Delgado et al. [33], there was a deviation from Hardy–Weinberg equilibrium in the control group. In all other cases, the differences between genotype frequencies were not statistically significant.

The MetaGenyo program provides meta-analysis for three inheritance models: dominant, codominant, and recessive. The most significant results were obtained for the dominant model, which is represented in Fig. 1 and in Table 1. But even in this case, all 95% CIs for OR (except for one) cover the indifferent value of OR=1 or have a very wide CI. No sport or athletic specialization was identified in which the ACE gene I/D dimorphism could serve as a reliable marker for predicting individual predisposition to high athletic performance.

Fig. 1. Forest-plot for the dominant model (DD+ID vs II) meta-analysis.

Table 1. Results of the meta-analysis for the dominant model (DD+ID vs II)

Note. CI, confidence interval; OR, odds ratio; PI, prediction interval.

DISCUSSION

The obtained results reinforce doubts that adaptively and evolutionarily neutral genetic polymorphism can serve as a tool for selection or prediction in sports [10, 39]. Currently, most of the identified associations have not proven their practical value.

The assumption that athletic talent is a fixed trait that can be identified at an early stage, the belief in the exceptional influence of talent on the development of athletic qualities, various levels of risk in decision-making with regard to athlete selection, biases in approaches to athlete selection, inadequacy of modern statistical approaches, issues with using current results to predict future outcomes, short-term benefits and competition between different sports for promising athletes dramatically decrease the overall effectiveness of sports development systems [40]. One of the most significant issues in predicting high athletic performance based on presumed individual genetic predisposition is uncertainty, measured using PI [41]. Most meta-analyses so far are limited to reporting confidence intervals only. However, this is insufficient. CI is an indicator of the accuracy of effect measurement, but it does not indicate the degree of variation in effect size. Variation, the dispersion of the effect, is reflected by the confidence interval, which is mandatorily calculated in meta-analyses [41, 42]. In this meta-analysis, the 95% CI ranges from 0.58 to 1.16. It covers the indifferent value of OR=1, meaning it is not statistically significant. This means that in 95% of subsequent studies comparable to those presented in this analysis, the actual effect size will fall within this interval and will generally be practically useless.

The practical infeasibility of using individual genes to predict sports talent at the individual level is due to the following limitations: genotyped variants are not functionally significant and show incomplete linkage with other significant gene variants; low statistical power of studies, lack of population stratification, heterogeneity of the phenotypes and loci studied. It was previously shown that the effectiveness of using a genetic marker for testing a binary trait (healthy/sick, athlete/non-athlete) depends on the frequency of occurrence of the given genotype (allele, haplotype) and the frequency of manifestation of the studied phenotype [43]. If the OR is less than 2.2, then at any frequency of marker occurrence, it does not have any diagnostic and/or prognostic value. Only with an OR greater than 5.4 and a population frequency above 0.3 can be recognized as a marker suitable for mass screenings and professional selection [43]. But in sports genetics, such markers have not been identified and are unlikely to ever be discovered. There are such markers for certain cancers.

Contradictory research results, as well as the lack of data on their real practical value for the search, selection, and profiling of young athletes and the choice of a training program, caused leading scientists in the field of sports genetics to release a joint statement in 2016. The outline of the statement is as follows: there is currently no scientifically proven basis to consider that the studied molecular-genetic markers have prognostic value for selecting talented athletes or for individualizing the training process; test systems based on the results of these studies are misleading and should not be used for the stated purposes [44]. Undoubtedly, the complex of morphofunctional and psychophysiological traits inherent in high-level athletes is based on a multitude of genes. However, the mechanisms that determine these connections are still virtually unexplored [45].

In recent years, the number of commercial companies providing direct-to-consumer (DTC) services for predicting individual predisposition to a specific type of sport/group of sports and injury risk has significantly increased [39]. In several countries, DTC tests are widely used as one of the stages of selecting gifted athletes at the initial stages [44]. This approach and the scale that DTC tests have acquired in sports genetics raise reasonable concerns in the professional scientific community both in the Russian Federation [10] and abroad [46]. Such services are provided based on research of a limited number of loci (from 1 to 37 depending on the company). Some companies do not provide the results of genotyping claimed in gene testing, and the criteria for conclusions to be based on are not universally accepted in the global community. There is no single diagnostic procedure that has been approved by official regulatory and supervisory authorities and has proven effectiveness for assessing the claimed qualities in children, adolescents, and/or adults [45–47]. Most test systems have not undergone randomized blind trials. In fact, none of these companies provide information on the quantitative (statistical) characteristics of the predictive ability of their test systems (specific values of PPV, NPV, likelihood ratios, and the limits of their interval estimates) [45].

Let us assume that we have identified a certain number of independent (unlinked) genetic markers that actually have a significant impact on the development and manifestation of traits necessary for achieving high athletic performance. Then the question arises regarding the probability of detecting a carrier of all these variants (G⁺) in the population, and the predictive ability of a diagnostic test based on the analysis of these variants/variant Table 2 represents quantitative assessments of these parameters, indicating that theoretically, if an individual has 9–10 predisposing variants, a high predictive probability (0.91–0.95) can be achieved. However, the probability of existence of such an individual in the population is negligible (10⁻⁹–10⁻¹⁰): less than one person in the entire population of the Earth. And this is despite the fact that these calculations do not consider the inevitable epistasis and pleiotropy, where different genotypes can suppress effects of each other.

As we already mentioned, the phenotype of an elite athlete is complex, with each of its individual components being polygenic in nature, meaning it is controlled by several genes. Obviously, there is a need to simultaneously consider multiple genes (a polygenic profile) for a quantitative assessment of their cumulative impact on the phenotype and the chances of a particular individual achieving outstanding athletic results. The next stage in the development of sports genetics was the introduction of a method for calculating the total genotype score (TGS) to assess the polygenic profile of athletes [48]. The first model for assessing the polygenic profile and its associations with athletic achievements was the additive model proposed by Williams and Folland in 2008 [48]. It involves assignment of scores to alternative genotypes of each polymorphic system based on the results of association studies, then the results for all studied systems are summed and divided by the total number of alleles. This approach is also known as the polygenic or genetic predisposition score [49].

Table 2. Correlation of predictabilities with the number of independent predisposing genetic markers combined in the genotype of an individual

Note: PPV, probability of the presence of the studied phenotype in a carrier of a given genotype (positive predictive value). It is simplistically assumed that the prevalence of each genotype is the same (0.1).

The assessment of a polygenic profile, including 7 genes, in a group of elite long-distance runners led to the conclusion that the average genetic score of the athletes is higher than that of the control group [25]. This result was confirmed in a group of athletes highly qualified in endurance sports (using 6 genes), with the average score being higher than that of the controls and long-distance runners [50]. The presence of all six genotypes associated with endurance sports was found in 9% of athletes, and no long-distance runner had the “ideal” polygenic profile. The distribution spectra of TGS in the control group and the groups of endurance and aerobic athletes significantly overlap. This means that many individuals in the control group had a TGS equal to or exceeding that of highly qualified athletes [50]. The analysis of TGS for a larger number of genes (22 and 23) showed significant similarity between groups of athletes and non-athletes, which also does not allow using this approach to identify individuals with a greater or lesser predisposition to a particular sport/group of sports [51].

The inadequacy of calculating TGS for assessing a polygenic profile to identify gifted individuals at the individual level, is illustrated by one of the recent studies [52]. The control group included 503 non-athletes, while the experimental group consisted of five international-level track and field athletes, including one Olympic champion. Two TGS values were calculated based on different polygenic profiles: 68 loci for stayers and 48 loci for sprinters. It turned out that both TGS values in sprinter athletes were higher than those in elite long-distance runners. Moreover, in 70 representatives of the control group, the “sprint” TGS values were higher than those of elite sprinters. A similar result was obtained in a Japanese study on the polygenic profile for marathon runners (n = 211), which included 21 genes. In the control group, the TGS (n = 649) was 49.0% ± 7.6%, which was higher than that in professional stayers at the international (48.2% ± 7.0%), national (49.1% ± 5.7%), and regional (47.3% ± 7.6%) levels [1].

The use of a large number of markers leads to the probability of detecting a carrier with an ideal or near-ideal polygenic profile being negligible.

In the overwhelming majority of cases, commercial entities providing genetic predisposition services to the general public, do not consider themselves obligated to adhere to international standards for the use and protection of the data they obtain. Such structures often transfer the obtained data to third parties (research groups or other organizations) and also use the accumulated data for purposes not specified in the informed consent [44, 46].

Regarding minors, the following issues should be addressed: whether sport clubs, sections, and state institutions may require trainees to provide data on individual genetic characteristics; whether it is permissible to deny a young athlete the right to engage in a specific sport based on genetic data; who can have access to data on the genetic parameters of a minor; what is the mechanism for protecting a child from discrimination based on genetic reasons; what consequences may arise for an athlete if they refuse to undergo genetic testing [47]. It should also be recognized that in the practical use of such genetic testing, there is an inevitable high risk of obtaining both false-positive and false-negative results and conclusions based on them [10, 43]. This approach may be acceptable at the population level, but cannot be used for individual assessment. Currently, the predictive ability of sports genetics is zero. There is no direct evidence that genetic indicators of athlete success exist. An athlete’s performance primarily depends on socioeconomic, cultural, and environmental factors. So, the stopwatch predicts a runner’s athletic achievements much better than all the genetics.¹

Despite numerous attempts to identify genetic variants associated with success in high-performance sports, progress in this area remains insignificant and disappointing. The advantage of genetic testing over standard pedagogical and anthropometric testing also remains open. We should keep in mind that a specific phenotype can be a product of completely different genotypes and even genomes. This is strongly evidenced by the phenomenon of doppelgängers, where unrelated individuals, sometimes living on different continents, look surprisingly alike. This example illustrates the complexity of the task of guessing or predicting the phenotypic manifestation of a given specific genotype. The phenotype prediction algorithm should take into account the frequency of phenotypic expression of a specific genotype (penetrance), even if it concerns rare alleles that have a pronounced effect on the phenotype. For example, carrying rare high-penetrance pathogenic alleles that cause monogenic diseases in children does not always lead to the development of the disease. The study of more than half a million genomes revealed 13 adults who were carriers of eight rare pathogenic variants, yet the disease did not manifest in them [53].

CONCLUSION

The development of modern technologies in the field of genomics (high-throughput sequencing, formal analysis of big data, use of artificial intelligence, genome editing) should contribute to the emergence of tools for personalized medicine and gene therapy as part of everyday practice. However, the emerging opportunities raise some ethical, moral, social, and personal questions for society. The field of genomics in physical activity is also influenced by emerging genomic

technologies, warrants an urgent need to develop common principles and approaches to the procedures of genetic testing for athletes. Modern technologies for obtaining genetic data and the rate of their accumulation significantly outpace our current ability to interpret and correctly apply them.

ADDITIONAL INFORMATION

Author сontributions: N.N. Khromov-Borisov: data collection and analysis, formal analysis, writing—original draft, writing—review & editing; E.A. Bondareva: sources review, data collection and analysis, writing—original draft, writing—review & editing. All authors confirm that their authorship meets the ICMJE criteria (all authors made substantial contributions to the conceptualization, investigation, and manuscript preparation, and reviewed and approved the final version prior to publication).

Funding sources: No funding.

Disclosure of interests: The authors have no relationships, activities, or interests for the last three years related to for-profit or not-for-profit third parties whose interests may be affected by the content of the article.

Statement of originality: No previously published material (text, images, or data) was used in this work.

Data availability statement: The editorial policy regarding data sharing does not apply to this work, as no new data was collected or created.

Generative AI: No generative artificial intelligence technologies were used to prepare this article.

Provenance and peer review: This paper was submitted unsolicited and reviewed following the standard procedure. The peer review process involved two external reviewers, a member of the editorial board, and the in-house scientific editor.

¹ The reasons why Kenyans always win marathons lie in one region. Available at: http://news.menshealth.com/why-kenyans-keep-winning-marathons/2011/06/03/

About the authors

Nikita N. Khromov-Borisov

Email: nikita.khromovborisov@gmail.com
ORCID iD: 0000-0001-6435-7218
SPIN-code: 1086-2105

Cand. Sci. (Biology)

Elvira A. Bondareva

Author for correspondence.
Email: bondareva.e@gmail.com
ORCID iD: 0000-0003-3321-7575
SPIN-code: 6732-2072

Cand. Sci. (Biology)

References

Supplementary files

Supplementary Files

Action

1. JATS XML

Download

2. Fig. 1. Forest-plot for the dominant model (DD+ID vs II) meta-analysis.

Download (720KB)

Indexing metadata