Journal Logo

SPECIAL COMMUNICATIONS: Contrasting Perspectives

Limits to the Evidence that DNA Sequence Differences Contribute to Variability in Fitness and Trainability


Author Information
Medicine & Science in Sports & Exercise: August 2019 - Volume 51 - Issue 8 - p 1786-1789
doi: 10.1249/MSS.0000000000001977
  • Free


In this contrasting perspective, I argue that the evidence a genetic explanation accounts for the variability seen in human cardiorespiratory fitness and its trainability is limited. To make my argument, I first outline key steps in the oxygen transport cascade central to cardiorespiratory fitness. I then define genetic causation and ask how much variation in DNA sequence influences key traits relevant to oxygen transport. Finally, I highlight ongoing issues relevant to the current narrative about genetic causation and complex human phenotypes.

The Oxygen Transport Cascade and Cardiorespiratory Fitness

The oxygen transport cascade describes the steps involved in oxygen transport from the air to the tissues. In the case of endurance exercise, the tissue of interest is skeletal muscle because that is where the vast majority of oxygen consumption occurs. The major steps in the cascade include: 1) pulmonary ventilation, 2) diffusion of O2 across the alveolar capillary membrane, 3) delivery of O2 to peripheral tissues via a combination of cardiac output and hemoglobin in red blood cells, 4) distribution of a high fraction of cardiac output (e.g., blood flow) to the contracting muscles, and 5) diffusion of O2 from the microcirculation to the tissues and oxidative metabolism in the mitochondria.

Importantly, acquired or rare inborn defects in one or more of these steps can result in reduced cardiorespiratory fitness and blunted responses to endurance exercise training (1,2). However, in large populations of healthy humans (including elite endurance athletes), a combination of peak cardiac output and red cell mass is the deterministic element of the oxygen transport cascade that dominate cardiorespiratory fitness (3). This is shown via the tight linkage between V˙O2max values ranging from 2 to 8 L·min−1 and cardiac output shown in Figure 1. Thus, any “causal” explanations for cardiorespiratory fitness at the level of DNA variation must explain the range in cardiac output and related physiological parameters.

V˙O2max is linearly related to cardiac output () across a wide range of values in humans. The vast majority of this relationship is explained by differences in stroke volume. To date, there is no clear genetic explanation for this variability in stroke volume or the wide range of increases in stroke volume that can occur in response to endurance exercise training. Data courtesy of C. Lundby, Ph.D.

Genetic Causation Defined

With modern DNA sequencing technology, it is possible to search for variation at the level of DNA sequence and link any variation with phenotypic variation and make a causal argument that variation in gene X explains some of the variation in phenotypic trait Y. This approach has superseded statistical techniques relating phenotypes of interest with assumptions about the magnitude of shared genetic influences in monozygotic or dizygotic twins or other forms of family or population studies. Concurrently, there has been a shift in the definition of “what is a gene” from a phenotypically derived unit of inheritance to something embedded in DNA sequence (4). The epistemology behind this shift and its implications for biological theory are beyond the scope of this essay, but excellent resources are available for those interested. (5)

Along these lines, five key criteria for claims about genetic causation at the level of DNA variation are required to explain complex phenotypes like cardiorespiratory fitness:

  1. identification of potentially causal variants,
  2. linkage of these variants to deterministic physiological mechanisms,
  3. explanation by these variants of more than a small fraction of the phenotypic trait of interest,
  4. demonstration that the gene or pathway is obligatory for an adaptive response,
  5. applicability of criteria 1 to 4 when a “maximal” adaptive stimulus has been applied.

Using this general framework, I now ask how much variation in key traits relevant to oxygen transport can be accounted for by variation in DNA sequence.

DNA Sequence and Key Traits Relevant to Oxygen Transport

Before deconstructing the relationship between the oxygen transport cascade and DNA variation, it is important to realize that with the advent of the Human Genome Project and related initiatives, it was widely believed that a few gene variants (~10) would explain the vast majority of phenotypic variation. For noncommunicable diseases, this was known as the Common Disease/Common Variant hypothesis. It was summarized by Francis Collins for Type 2 diabetes in a 2006 Wall Street Journal interview (6):

“I expect there are about 12 genes involved, and that all of them will be discovered in the next two years.”

That having been said, it now appears that tens if not hundreds of gene variants with tiny effect sizes contribute to human phenotypic variability. Of note, many of these effect sizes are typically reported in increments (like a fraction of a millimeter of mercury of blood pressure) that are not measurable in a laboratory (7). As a result of these limitations, polygenic scores that incorporate many small effect size variants have now emerged and phenotypes are compared in groups with high (top quintile or quartile) and low (bottom quintile or quartile) gene scores (8).

When considering these general findings in the context of the first criteria for genetic causation outlined above, it appears that clearly causal variants have not been identified. Instead, probabilistic assertions about many small effect size genes are now in vogue. In the case of oxygen transport both individual DNA variants and gene scores for lung volume and maximum heart rate—physiological variables of clear relevance to cardiorespiratory fitness—can explain only a tiny fraction of the physiology of oxygen transport, and they do not rise above any reasonable physiological signal to noise criteria (9,10). Additionally, there is essentially no information on the effects sizes or number of variants that might contribute to pulmonary diffusing capacity, cardiac output or stroke volume, and red cell mass or key indices of microcirculatory function. In the case of skeletal muscle oxidative capacity, it has been known since the work of John Holloszy in the 1960s that mitochondrial content is extremely plastic and can increase dramatically with exercise training in almost all humans (11). Because DNA variation has not been linked to deterministic physiological mechanisms, this means that the second of the five criteria for genetic causation is also lacking. Likewise that a probabilistic collection of variants that might under some circumstances explain no more than a small fraction of the variance of the trait of interest means that positive evidence relevant to the third criteria for causation noted above is also lacking.

This lack of linkage to physiological mechanisms is highlighted by two key studies. First, in the HERITAGE study, although the increases in V˙O2max observed in response to a standardized 20-wk program of endurance exercise training was heritable in the pre-DNA variant statistical context, no DNA variants associated with any key elements of oxygen transport cascade were identified (12). Second, in a large survey of elite endurance athletes no common genetic signature associated with very high V˙O2max values (2–3× normal) was found. In both cases, there was no DNA variant–related signal that could be linked to the physiology of cardiac output or red cell mass (13).

When the “is it obligatory” question related to causation is asked about acute and chronic responses to exercise, it is clear that multiple pathways thought to be central to exercise are in fact not obligatory. Many key pathways thought to be “obligatory” can be knocked out at the level of the genome, inhibited by drugs, or altered by surgical interventions like cardiac denervation with only limited impact on phenotypic adaptation (14–17). If whole pathways can be knocked out or inhibited via other means with limited impact on phenotype, how is it possible that small effect size DNA variants alone, or in combination, will have the sorts of deterministic effects once thought to be certain?

In terms of the cardiorespiratory fitness responses to exercise training, the fifth question is what happens when a maximal physiological stimulus is applied? A number of small N studies that have used very high-intensity interval training suggest that all humans exhibit at least a modest level of trainability (18,19). Whether this sort of training—especially if applied for years—would merely shift that range of training responses to higher values with little impact on range, or whether the range might be compressed around a higher average value is not known. Of note, there are anecdotal reports of individuals who, as a result of years of endurance training, have undergone a complete shift from a sprinter to endurance phenotypes (20). Such extreme phenotypic plasticity was recently confirmed in a pair of middle-age monozygotic twins highly divergent for exercise over decades. Among other notable observations, the endurance trained twin had 90% slow myosin, whereas the untrained twin had only 40% (21). Thus, application of a maximal physiological training stimulus over time can clearly disrupt conventional concepts about phenotype and heritability.

Further Limitations to the Genotype/Phenotype Narrative

Proponents of a more deterministic relationship between genotype and phenotype will frequently point to the clear genetic signatures that have emerged in strains (or breeds) of inbred animals. As a result of artificial selection, dog breeds for example show huge phenotypic variation with a limited number of traits that can be mapped to a few areas of the genome (22). By contrast, in humans, there are only a few examples like lactase persistence that can clearly be ascribed to natural selection (23). There is also simply no evidence that remarkable cardiorespiratory fitness is required for activities like persistence hunting and thus might have been subject to selection pressure in our forbearers.

In the case of inbred models of hypertension, chromosome substitution experiments show that when a single chromosome from a normotensive rat strain is switched into genome of a hypertensive strain, blood pressure responses in the offspring can be surprisingly normal. (24) Additionally, experimental evolution studies and convergent evolution paradigms show that while selection pressure typically generates phenotypic convergence, the DNA based components of this convergence can be variable (25). Finally, a case study on the genetics of height is perhaps instructive. Former professional basketball player Shawn Bradley has a height of 2.29 m which is about 50 cm greater than average (26). He also has a remarkably high gene score for height. However, his gene score only predicts that his height will be approximately 1 cm above average.

For the reasons highlighted above, it is time to move beyond the DNA-based “gene” or “genes for” narrative as it applies to both the inherent and acquired biological determinants of cardiorespiratory fitness and its response to training.


Professor Bouchard has succinctly and comprehensively outlined the data supporting the concept that there is a genetic component to variability in fitness and trainability (27). However, all of the support for his overall position comes from correlational observations in humans and data from inbred animal models. These data have been interpreted to suggest that DNA sequence variation is responsible. However, to date, there is no clear evidence that DNA sequence variation in deterministic physiological pathways (e.g., maximal cardiac output and red cell mass) is responsible for the variation in fitness and trainability seen in humans (28).

Of note, Professor Bouchard also touches on the potential role of physiological redundancy in generating complex adaptive traits and issues related to gene and protein expression. These factors or mechanisms could clearly generate marked differences in the appearance and performance of animals with the same DNA sequence. They also explain a number of the observations in my competing perspective, including that adaptations to exercise training are possible in knockout animals missing genes deemed as critical to exercise responses. If responses can be seen when whole genes are absent, it seems reasonable to question the idea that any modest changes in gene or subsequent protein function associated with small effect size single nucleotide polymorphisms will make a major difference in the physiological responses to exercise. Likewise, observations from experimental evolution and convergent evolution studies show that there are many potential DNA sequence pathways associated with a given trait.

As I read and considered the Bouchard piece further, I was pleased to see the many limitations and caveats he acknowledged to the DNA sequence centric world view for complex biological traits. These limitations and caveats are consistent with the challenges to the explanatory power of molecular reductionism that numerous investigators raised many years ago (29–31). That they are being publically discussed by a geneticist and thought leader as prominent as Professor Bouchard is heartening to those of us who have challenged the DNA centric view of biology that has emerged over the last 30 yr. However, they are nothing new.

For example, in their 2000 paper Weiss and Terwilliger (29) argued for an approach to genetics that interfaces more intimately with biology. They also noted that complex traits (like cardiorespiratory fitness) likely emerged via noise-tolerant evolutionary mechanisms—meaning that an acceptable complex trait could emerge in the face of variability at lower levels of biological integration—and were thus unlikely to be genetic in the traditional deterministic sense of the concept. Additionally, they warned about genetic studies that attempt to mimic Mendel’s approach that focus on a few states or interventions. Finally, they noted that anyone who challenged the emerging obsession with molecular reductionism risked being categorized as a naysayer.

In the context of the above comments, it seems to me that now is the time to abandon the reductionist obsession with ever larger sample sizes that generate an ever larger list of small effect size gene variants that are weekly correlated with a trait of interest. Focus should return to the multiscale physiological responses that generate different adaptations and traits across the biological world (32). Thus, it seems fair to close by asking how much farther along the field might be if concerns raised by the skeptics had been considered all along. These concerns were clearly hiding in plain sight.


1. Dowman L, Hill CJ, Holland AE. Pulmonary rehabilitation for interstitial lung disease. Cochrane Database Syst Rev. 2014(10).
2. Haller RG, Lewis SF, Estabrook RW, DiMauro S, Servidei S, Foster DW. Exercise intolerance, lactic acidosis, and abnormal cardiopulmonary regulation in exercise associated with adult skeletal muscle cytochrome c oxidase deficiency. J Clin Invest. 1989;84(1):155–61.
3. Lundby C, Montero D, Joyner M. Biology of VO2 max: looking under the physiology lamp. Acta Physiol (Oxford). 2017;220(2):218–28.
4. Noble D. Genes and causation. Philos Trans A Math Phys Eng Sci. 2008;366(1878):3001–15.
5. Rheinberger H, Müller-Wille S, Meunier R. “Gene”, The Stanford Encyclopedia of Philosophy (Spring 2015 Edition), Edward N. Zalta (ed.), URL = <
6. Regaldo A. New genetic tools may reveal roots of everyday ills-rapid DNA tests can search many variations at once; probing obesity, memory. Wall Street J. 2006;5:2018.
7. International Consortium for Blood Pressure Genome-Wide Association Studies, Ehret GB, Munroe PB, et al. Genetic variants in novel pathways influence blood pressure and cardiovascular disease risk. Nature. 2011;478(7367):103–9.
8. Khera AV, Chaffin M, Haas ME, et al. Genome-wide polygenic scores for common diseases identify individuals with risk equivalent to monogenic mutations. Nat Genet. 2018;50(9):1219–24.
9. van der Plaat DA, de Jong K, Lahousse L, et al. Genome-wide association study on the FEV1/FVC ratio in never-smokers identifies HHIP and FAM13A. J Allergy Clin Immunol. 2017;139(2):533–40.
10. Ramirez J, Duijvenboden SV, Ntalla I, et al. Thirty loci identified for heart rate response to exercise and recovery implicate autonomic nervous system. Nat Commun. 2018;9(1):1947.
11. Holloszy JO, Coyle EF. Adaptations of skeletal muscle to endurance exercise and their metabolic consequences. J Appl Physiol Respir Environ Exerc Physiol. 1984;56(4):831–8.
12. Bouchard C, Sarzynski MA, Rice TK, et al. Genomic predictors of the maximal O(2) uptake response to standardized exercise training programs. J Appl Physiol (1985). 2011;110(5):1160–70.
13. Rankinen T, Fuku N, Wolfarth B, et al. No evidence of a common DNA variant profile specific to world class endurance athletes. PLoS One. 2016;11(1):e0147330.
14. Leick L, Wojtaszewski JF, Johansen ST, et al. PGC-1alpha is not mandatory for exercise- and training-induced adaptive gene responses in mouse skeletal muscle. Am J Physiol Endocrinol Metab. 2008;294:E463–74.
15. Fentz J, Kjobsted R, Kristensen CM, et al. AMPKalpha is essential for acute exercise-induced gene responses but not for exercise training-induced adaptations in mouse skeletal muscle. Am J Physiol Endocrinol Metab. 2015;309(11):E900–14.
16. Lloyd PG, Prior BM, Li H, Yang HT, Terjung RL. VEGF receptor antagonism blocks arteriogenesis, but only partially inhibits angiogenesis, in skeletal muscle of exercise-trained rats. Am J Physiol Heart Circ Physiol. 2005;288(2):H759–68.
17. Donald DE, Milburn SE, Shepherd JT. Effect of cardiac denervation on the maximal capacity for exercise in the racing greyhound. J Appl Physiol. 1964;19:849–52.
18. Hickson RC, Bomze HA, Holloszy JO. Linear increase in aerobic power induced by a strenuous program of endurance exercise. J Appl Physiol Respir Environ Exerc Physiol. 1977;42(3):372–6.
19. Bacon AP, Carter RE, Ogle EA, Joyner MJ. V˙O2max trainability and high intensity interval training in humans: a meta-analysis. PLoS One. 2013;8(9):e73182.
20. Joyner M. Human limits—exploring performance and health with Michael J. Joyner, M.D. 2014. Available from:
21. Bathgate KE, Bagley JR, Jo E, et al. Muscle health and performance in monozygotic twins with 30 years of discordant exercise habits. Eur J Appl Physiol. 2018;118(10):2097–110.
22. Rimbault M, Ostrander EA. So many doggone traits: mapping genetics of multiple phenotypes in the domestic dog. Hum Mol Genet. 2012;21(R1):R52–7.
23. Segurel L, Bon C. On the evolution of lactase persistence in humans. Annu Rev Genomics Hum Genet. 2017;18:297–319.
24. Cowley AW Jr., Roman RJ, Kaldunski ML, et al. Brown Norway chromosome 13 confers protection from high salt to consomic Dahl S rat. Hypertension. 2001;37(2 Pt 2):456–61.
25. Natarajan C, Hoffmann FG, Weber RE, Fago A, Witt CC, Storz JF. Predictable convergence in hemoglobin function has unpredictable molecular underpinnings. Science. 2016;354(6310):336–9.
26. Sexton CE, Ebbert MTW, Miller RH, et al. Common DNA variants accurately rank an individual of extreme height. Int J Genomics. 2018;2018:https//
27. Bouchard CB. DNA sequence variations contribute to variability in fitness and trainability. Med Sci Sports Exerc. 2019;51(8):1781–5.
28. Joyner MJ, Lundby C. Concepts about VO2max and trainability are context dependent. Exerc Sport Sci Rev. 2018;46(3):138–43.
29. Weiss KM, Terwilliger JD. How many diseases does it take to map a gene with SNPs? Nat Genet. 2000;26(2):151–7.
30. Weiss KM. Tilting at quixotic trait loci (QTL): an evolutionary perspective on genetic causation. Genetics. 2008;179(4):1741–56.
31. Joyner MJ. Giant sucking sound: can physiology fill the intellectual void left by the reductionists? J Appl Physiol (1985). 2011;111(2):335–42.
32. Noble D, Jablonka E, Joyner MJ, Muller GB, Omholt SW. Evolution evolves: physiology returns to centre stage. J Physiol. 2014;592(11):2237–44.
Copyright © 2019 by the American College of Sports Medicine