Jamie Timmons
Loughborough University, UK

https://doi.org/10.36866/pn.90.22

Most readers of this article would support the belief that our ‘genes’ interact with a variety of environmental factors, such as nutrients and physical activity, to shape our health and risk of chronic disease. Indeed, this basic premise dominates the research fields of ageing, cancer, cardiovascular and metabolic disease. Yet it is very rare indeed that a direct causal connection between a particular ‘behaviour’ and a health outcome has been established. Equally, the number of established links between a specific DNA sequence and any of the major multifactorial chronic disease states is limited.

There are numerous unknowns when it comes to the validity and application of high throughput genomics being used to help us understand how our genes impact on our disease risk. We need to ask the question, is it possible to genuinely embrace modern evidence-based medicine when scientists with no multidisciplinary training manage the multidisciplinary science? Technical limitations within data-sets are unappreciated, and no common minimum acceptable level of ‘evidence’ to enable evidence-based medical practice is established.

This article addresses some of the barriers to advancing the contribution genomic and epigenomic technologies can make to our understanding of human health and physiology. What I want to know in the end, is, if I take up jogging today, and do my 150 minutes a week, every week, like a good citizen, will I live a healthier and longer life? Or will it just feel longer?

Lifestyle modification has been shown to improve the group incidence of type II diabetes in intervention trials, but no robust exercise-only interventions have been carried out. So it would be easy to focus on risk-factor modification and not give due care and attention to the fact that these behaviours are not necessarily causal for future disease risk. In fact, the available intervention data demonstrates that lifestyle modification involving exercise training does not reduce the incidence of major cardiovascular events (such as myocardial infarction) at all, though it does reduce the incidence of diabetes and modify other important risk factors (Wing, 2010). The significant healthcare costs and levels of morbidity associated with cardiovascular events are presumably what justified the costs of the studies in question.

If our genes haven’t changed it must be the environment, right?

In a recent meta-analysis, Lee and colleagues conclude, “the association of lung cancer with smoking is strong, evident for all lung cancer types, dose-related and insensitive to covariate-adjustment. This emphasises the causal nature of the relationship” (Lee et al. 2012). While it is rather insensitive to label the discipline of epidemiology (addressing any topic) as ‘largely’ correlational, what we can reflect on is that the relative risk of developing lung cancer from smoking ranges from <2 up to >40-fold increase. The risk also appears to vary from continent to continent, perhaps reflecting differing genetics (but also data recording methodologies). Put bluntly, one of the most accepted truisms in evidence-based medicine interacts with other strong factors, including genetic predisposition, to dramatically determine the risk to the individual.

Is the successful identification of a strong link between cigarette-smoking and lung cancer proof that epidemiological methodology is a valid synonym for ‘evidence-based medical research’, or is it an exception? A number of scientists have recently made the rather perplexing statement that low physical activity is the ‘new smoking’. The American College of Sports and Medicine has launched an initiative called ‘Exercise is Medicine’, trade-marking the name in the process! Meanwhile, numerous academics have extensively documented the claim that ‘exercise’ offers a cure to almost all known chronic disease states (Pedersen & Saltin, 2006), a mantra unquestionably repeated by others (Booth & Laye, 2010). If true, where is there room for genetic or familial influences – influences that have been established in supervised intervention studies in humans (Bouchard et al. 1986; Oppert et al. 1995)?

The truth about exercise is that there are very few examples of properly controlled human intervention trials that have measured a reduction in disease, a topic I explore in greater detail online (Timmons, 2012). What we do have is extensive evidence that exercise, as part of an overall lifestyle modification strategy, modifies ‘risk factors’ for poor health and chronic disease. To address the question “Are we in control of our own health?”, we must progress with caution through the minefield of epidemiological evidence, and ask honestly whether any existing evidence reaches a level of confidence sufficient to interfere with an individual’s behaviour choices. To draw analogy, the basis for the strongly held views on the causal benefits of increased physical activity would be like basing the conclusion that cigarette-smoking causes lung cancer because both are associated with ‘coughing’. Nevertheless, smoking will almost certainly get you some other way, e.g. chronic obstructive pulmonary disease.

Variation is a physiologist’s new best friend

Firstly, most interventions that I am aware of (except prolonged anoxia!), result in a variable outcome in outbred mammals, with marked inter-subject and inter-study heterogeneity. This is true in drug trials and true in studies of the impacts of exercise or lifestyle modification. Such variation has been unintentionally overlooked through a focus on group average responses (or actively denied by some too attached to their pre-existing beliefs). This biological variation is also considered the ‘enemy’ of the research scientist, creating concern over whether a study may be inadequately powered to reveal a ‘significant’ effect from the intervention. I would argue that, in many cases, the mean effect of an intervention is meaningless to the individual, and thus so are those misguided grant-review questions on power calculations. The key feature of a quality human physiology study, moving forward, will be independent replication of the observation and not power to detect a mean effect.

It is also known that the impact of exercise training on health biomarkers, after supervised intervention, varies dramatically and, for most people, an adverse change in a health biomarker will also occur (Bouchard et al. 2012). Thus, if you have elevated blood pressure and you have an adverse blood pressure response to a one-size-fits-all exercise regimen, then it is very clear Exercise Cannot Be Medicine (™!). If you have elevated blood glucose levels and you become more insulin resistant with exercise training, then physical activity is not “a drug for people with type 2 diabetes” (exerciseismedicine.org). Generalised advice on exercise prescription (in the absence of a valid personalised plan for the individual) is not only unscientific, it could in some circumstances be irresponsible.

Can we use genomics technologies to solve this public health challenge? Well, certainly not if we propose that everyone does 5 hours of power walking (Karstoft et al. 2013), rather than 3–10 minutes of sprint interval training per week (Metcalfe et al. 2012; Gillen et al. 2012) as the majority of humans don’t value perpetual motion nor have we undergone such behaviour at any time in our history.

Which flavour of ‘omics’ will deliver progress?

Assuming common sense is applied, then it will be the integration of new technologies to study human variation – requiring multidisciplinary groups to come together and effectively communicate the limitations of all methods being employed – that will make progress. There is little point perfecting the determination of gene sequence or DNA methylation, with high fidelity, only to correlate it with unreliable plasma glucose samples, or not include the appropriate time control in your study (Barrès et al. 2012). To understand why humans respond to exercise training to a highly variable degree, human physiology experts have come together with genomics and systems biology teams (Timmons et al. 2012). Currently, the physiological journals do a rather poor job when it comes to the evaluation of studies utilising systems biology tools, preferring to question the basis of such research (Timmons, 2011).

To progress evidence based medicine, we need to train individuals in multidisciplinary research. It is not sufficient to put together some esteemed colleagues from distinct fields (Turan et al. 2011) – without a comprehensive appreciation of obvious biological or clinical covariates by all parties no progress can be made. Modelling suboptimal ‘omic’ data can do much damage, as the lack of quality of the clinical data (Mootha et al. 2003) is often put to one side, especially in so-called high-impact journals. Thus, the first breakthrough required is not a technology platform, but a more sustained effort to train numerically-confident physiologists with an interest in computer science, genomics and big data, and put to one side academic debates about physiology versus systems biology.

Will nucleic acid sequence-based analysis provide the genomic tools for future physiological studies? To date, it has only been possible to unveil a small number of variations in the human genome that determine the outcome of physical activity. These include stable sequence related variations in nucleic acid content that predict the gains in aerobic fitness gained from exercise training (Bouchard et al. 2011) and variations in the abundance of a set of RNA molecules which do not vary by muscle activity (Timmons et al. 2010), but rather are set to a pre-exercise level by mechanisms not fully established (some link to DNA variants while others may be epigenetically regulated).

This was the first example in human physiology where a mathematical classification model (www.medicalprognosis. com) was used to mine RNA transcriptomic data sets to then select genomic loci for targeted genotyping to yield a new diagnostic (www.XRgenomics.co.uk). I feel this is an important example where human physiology and ‘big science’ methods have been combined to produce progress beyond that which has been achieved with genome-wide DNA screening epidemiology. In the genome-wide genotyping studies, limited clinical phenotyping almost certainly rate-limits progress. By using data integration methods, analysis can be applied to hundreds, rather than thousands, of samples, making intervention studies affordable. What is now required is larger scale cooperation across the human physiological community to create ‘pools’ of high-quality human physiological data sufficient in size to carry out this type of molecular epidemiology.

The EU FP7 project, Metapredict, is one example, where more than 10 teams have come together to share existing data and generate biological materials and physiological data on over 1500 humans (www.metapredict.eu). This study will integrate a variety of ‘omic’ technology to produce novel diagnostics for blood pressure, aerobic fitness and glucose tolerance responses to supervised time-efficient exercise training (<10 mins per week). Based on these integrated studies we should be able to utilise DNA, RNA and metabolite profiles to predict which health biomarkers are going to be positively modified in the individual, using a blood sample, DNA swab or micro-tissue biopsy. We are applying these strategies to a variety of interventions, including an intervention that overcomes societal limitations of current public health advice (i.e. a perceived lack of time). Indeed, the laboratories of Kraus and Kujala have made some excellent progress, discovering metabolite signatures that co-vary with clinical status (Shah et al. 2009; Kujala et al. 2013). The next step is to prove they are predictive of future events and integrate such data into multi-’omic’ models.

A very trendy term is emerging in the literature to explain what can’t be explained by sequence-based inheritance. That term is ‘epigenetics’. If you don’t understand something, then it’s definitely epigenetic modification in action! The mysterious little bioactive RNA molecules (Gallagher, 2010; Davidsen et al. 2011) called microRNA’s, are highly abundant products of longer RNAs and, in turn, ‘genes’, which are produced by a process analogous to any protein. Bizarrely, they are frequently listed as mediators of epigenetic action, highlighting the general level of mystery around both concepts – they are just ordinary genes, doing an ordinary job, at the RNA level. In the past year, epigenetic analysis of human muscle tissue has yielded claims of both importance and surprising plasticity in the ‘stable’ DNA modification that one can detect in human cells (Barrès et al. 2012). It is rather surprising that a single bout of exercise can ‘modify’ your DNA (as the press release suggested). The truth is that the genes profiled switch on and off with circadian pattern in the at-rest human (Vissing et al. 2005) and this, more than anything, highlights the extreme dangers of poor peer-review and candidate-based analysis. If a global analysis of transcriptional events had been coupled with a global and detailed map of methylation, the results would not have been nearly as newsworthy. There remains to be a technically valid and scientifically robust study of epigenetic regulation in human muscle and the impact of exercise and lifestyle.

To pursue this correctly, the theme mentioned above is critical. You need to assemble a multi-disciplinary team with a genuine understanding of physiology, bioinformatics, statistics and DNA biochemistry (for example, I’ve never had a clear answer to the question, how do you isolate DNA from muscle tissue and ensure epigenetic events are captured?). Never has physiology been so important, and yet never has fashion-driven technology so dominated biomedical science. Fifteen years after the emergence of gene-chip technology for global RNA profiling, my laboratory still works with the technology and we are still discovering better ways to use the technology and data. Grant reviewers still attempt to force us to move to next-generation sequencing of RNA, a technology that does not yet work reliably (Hansen et al. 2011).

Conclusions

Why have the epidemiological predictions failed us on the impact of lifestyle modification and the link between diabetes and cardiovascular disease? The truth is, they haven’t. There was abundant data that hyperglycaemia is not a strong predictor of macrovascular disease in the UKPDS analysis, while beneficial effects on microvascular disease were predicted. Sadly, the details are being ignored by many in favour of a good simple headline – “Exercise prevents cardiovascular disease by preventing type 2 diabetes”. Politically-correct thinking is driving a public health strategy and that now needs revision. Other researchers are still attempting to evaluate high-volume organised physical activity (Beck-Nielsen et al. 2012), a behaviour that modern humans dislike and a behaviour that humans probably never exhibited, given the energy–transfer efficiency of food into locomotion. Think about it; why would we move around all day ‘chasing after’ food, as surely there is no faster route to the eventual extinction of a species and its habitat, through excessive consumption. It’s also an effective strategy for increasing the carbon footprint of humans on this planet!

So how do we combine the latest ‘omic’ developments to advance this area of public health? Firstly, what we need is exercise-only intervention studies, using modes of exercise that are likely to overcome real-world barriers and that are genomic-driven, tailored to the individual. Also, as most studies apply one-size-fits-all exercise intervention plans, we need strategies to overcome the limitations of such studies using individualised diagnostics. In that way the individual’s outcome, for each major health parameter, can be re-evaluated on the basis, for example, of their ‘omic’ responder status. Only then will we edge closer to the truth about exercise.

References

American College of Sports and Medicine, The. Exercise is Medicine resources page [Online]. Available: http://exerciseismedicine.org/resources.htm [18 Feb 2013].

Barrès R, Yan J, Egan B, Treebak JT, Rasmussen M, Fritz T, Caidahl K, Krook A, O’Gorman DJ & Zierath JR (2012). Acute exercise remodels promoter methylation in human skeletal muscle. Cell Metab 15, 405–411.

Beck-Nielsen H, Solomon TP, Lauridsen J, Karstoft K, Pedersen BK, Johnsen SP, Nielsen JS, Kryger TB, Sortsø C & Vaag A (2012). The Danish Centre for Strategic Research in Type 2 Diabetes (DD2) study: expected outcome from the DD2 project and two intervention studies. Clin Epidemiol 4, 21–26.

Booth FW & Laye MJ (2010). The future: genes, physical activity and health. Acta Physiol (Oxf) 199, 549–556.

Bouchard C, Blair SN, Church TS, Earnest CP, Hagberg JM, Häkkinen K, Jenkins NT, Karavirta L, Kraus WE, Leon AS, Rao DC, Sarzynski MA, Skinner JS, Slentz CA & Rankinen T (2012). Adverse metabolic response to regular exercise: is it a rare or common occurrence? PloS one 7, e37887.

Bouchard C, Lesage R, Lortie G, Simoneau JA, Hamel P, Boulay MR, Perusse L, Theriault G & Leblanc C (1986). Aerobic performance in brothers, dizygotic and monozygotic twins. Med Sci Sports Exerc 18, 639–646.

Bouchard C, Rankinen T & Timmons JA (2011). Genomics and Genetics in the Biology of Adaptation to Exercise ed. Terjung R. Compr Physiol 1603–1648.

Davidsen PK, Gallagher IJ, Hartman JW, Tarnopolsky MA, Dela F, Helge JW, Timmons JA & Phillips SM (2011). High responders to resistance exercise training demonstrate differential regulation of skeletal muscle microRNA expression. J Appl Physiol 110, 309–317.

Gallagher I (2010). Integration of microRNA changes in vivo identifies novel molecular features of muscle insulin resistance in Type 2 Diabetes. Genome Med 2, 9.

Gillen JB, Little JP, Punthakee Z, Tarnopolsky MA, Riddell MC & Gibala MJ (2012). Acute high-intensity interval exercise reduces the postprandial glucose response and prevalence of hyperglycaemia in patients with type 2 diabetes. Diabetes Obes Metab 14, 575–577.

Hansen KD, Wu Z,Irizarry RA & Leek JT (2011). Sequencing technology does not eliminate biological variability. Nat Biotechnol 29(7), 572–573.

Karstoft K, Winding K, Knudsen SH, Nielsen JS, Thomsen C, Pedersen BK & Solomon TPJ (2013). The Effects of Free-Living Interval-Walking Training on Glycemic Control, Body Composition, and Physical Fitness in Type 2 Diabetic Patients: A randomized, controlled trial. Diabetes care 36, 228–236.

Kujala UM et al. (2013). Long-term Leisure-time Physical Activity and Serum Metabolome. Circulation 127, 340–348.

Lee PN, Forey BA & Coombs KJ (2012). Systematic review with meta-analysis of the epidemiological evidence in the 1900s relating smoking to lung cancer. BMC cancer 12, 385.

Metcalfe RS, Babraj JA, Fawkner SG & Vollaard NBJ (2012). Towards the minimal amount of exercise for improving metabolic health: beneficial effects of reduced-exertion high-intensity interval training. Eur J Appl Physiol 112, 2767–2775.

Mootha VK et al. (2003). PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat Genet 34, 267–273.

Oppert JM, Nadeau A, Tremblay A, Després JP, Thériault G, Dériaz O & Bouchard C (1995). Plasma glucose, insulin, and glucagon before and after long-term overfeeding in identical twins. Metabolism 44, 96–105.

Pedersen BK & Saltin B (2006). Evidence for prescribing exercise as therapy in chronic disease. Scand J Med Sci Sports 16 Suppl 1, 3–63.

Shah SH, Hauser ER, Bain JR, Muehlbauer MJ, Haynes C, Stevens RD, Wenner BR, Dowdy ZE, Granger CB, Ginsburg GS, Newgard CB & Kraus WE (2009). High heritability of metabolomic profiles in families burdened with premature cardiovascular disease. Mol Syst Biol 5, 258.

Timmons JA (2011). What happens if you pose the wrong questions? J Physiol 589, 4799–4801.

Timmons JA et al. (2010). Using molecular classification to predict gains in maximal aerobic capacity following endurance exercise training in humans. J Appl Physiol 108, 1487–1496.

Timmons JA (2012). The Truth About Exercise and Public Health [Online]. Available: http://vimeo.com/channels/176195/57838149 [18 Feb 2013].

Timmons JA, Baar K, Davidsen PK & Atherton PJ (2012). Is irisin a human exercise gene? Nature 488, E9–E10.

Turan N, Kalko S, Stincone A, Clarke K, Sabah A, Howlett K, Curnow SJ, Rodriguez DA, Cascante M, O’Neill L, Egginton S, Roca J & Falciani F (2011). A systems biology approach identifies molecular networks defining skeletal muscle abnormalities in chronic obstructive pulmonary disease. PLoS comput biol 7, e1002129.

Vissing K, Andersen JL & Schjerling P (2005). Are exercise-induced genes induced by exercise? Faseb J 19, 94–96.

Wing RR (2010). Long-term effects of a lifestyle intervention on weight and cardiovascular risk factors in individuals with type 2 diabetes mellitus: four-year results of the Look AHEAD trial. Arch Intern Med 170, 1566–1575.