Challenges to personalized exercise medicine: lack of repeatability and influence of genetics

Bouchard and colleagues’ seminal HERITAGE study revealed heterogeneity in individual responses to exercise training. Since then, many studies have classified participants as ‘low responders’ if they fail to improve a given outcome following exercise training. While the biological mechanisms underlying exercise ‘low response’ remain unclear, recent reports have identified exercise prescription strategies that appear to elicit positive responses in individuals initially identified as low responders. We found that low responders to endurance training exhibit a positive response following sprint interval training (SIT; or vice versa) (Bonafiglia et al. 2016). Our findings, along with other studies comparing responsiveness to different exercise doses, suggest that altering exercise prescription is a viable strategy for rescuing low response. In the current issue of The Journal of Physiology, Marsh and colleagues (Marsh et al. 2020) tested the hypothesis that switching from endurance (END) to resistance (RES) training (or vice versa) will rescue low response. Following a randomized, cross-over design, 34 pairs of twins (n = 68) completed 3 months of END and RES separated by a 3 month washout period. As expected, mean improvements in strength and cardiorespiratory fitness (CRF) were observed following RES and END, respectively. Individual responses to RES and END were not correlated for any variable with ∼50% of participants demonstrating discordant responses: low response to END but a positive response to RES or vice versa. The authors concluded that strength or CRF low response is training mode-specific whereby switching to an alternate mode can elicit a positive response in individuals initially identified as low responders. A major strength of the study by Marsh and colleagues was the use of mono- (MZ; genetically identical) and dizygotic twins (DZ; ∼50% genetically similar) to explore the contribution of genetics toward exercise response variability. As mentioned by the authors, most twin studies only include MZs and interpret significant intraclass correlations (ICCs) between twin pairs as evidence of a genetic contribution. However, because MZ pairs likely have similar environmental factors (e.g. diet, sleep patterns, etc.), it is unclear whether significant ICCs between MZs reflect genetic or environmental contributions. Including DZs in a twin model can discriminate the effects of genetics and environment as DZ pairs also have shared environmental factors but are less genetically similar than MZs. Although Marsh and colleagues found significant ICCs for changes in strength and CRF in MZs but not DZs, the ICCs for MZs were not statistically larger than the ICCs for DZs. These results suggest that environmental factors, and not genetics, explain most of the variance in strength and CRF responses to END and RES. Importantly, and as discussed below, these findings by Marsh et al. add to the growing body of evidence that questions the relative importance of genetics in determining exercise responsiveness. Discordant observed responses to END and RES (Marsh et al. 2020), and to END and SIT (Bonafiglia et al. 2016), highlight potential implications for personalized exercise medicine. Specifically, if an individual presents a low response to a given exercise mode, switching to a different mode may elicit a positive response. However, the conclusion that an individual may respond better to a given mode of exercise training – and the concept that switching exercise modes can rescue low response – is predicated on the assumption that individual responses to exercise training are repeatable. In other words, the conclusions made by Marsh et al. and us (Bonafiglia et al. 2016) assume that an individual's response to a given dose of exercise training is consistent (e.g. a low responder to END will never respond to END or a high responder to SIT will always respond to SIT). To determine whether individuals consistently respond to a particular dose of exercise training, we recently investigated the repeatability of individual CRF responses to exercise training (Del Giudice et al. 2020). Participants completed two identical 4 week periods of high-intensity interval training (HIIT) separated by a 3 month washout period. Despite similar mean CRF improvements following both training periods, individual responses were not repeatable. Previous work also demonstrates a lack of repeatability for exercise performance and skeletal muscle mitochondrial adaptations to single-leg aerobic exercise training (Lindholm et al. 2016). This apparent lack of repeatability in training responses may be explained by poor repeatability in mRNA responses to acute exercise (Islam et al. 2019) – a purported indicator of adaptive potential. Collectively, these findings suggest that an individual's observed response to a given prescription of exercise may not be a stable, repeatable trait. A lack of repeatability may question the notion that individuals preferentially respond to a given mode of exercise training (Bonafiglia et al. 2016; Marsh et al. 2020). To help illustrate this point, we recreated Fig. 3 from the Marsh et al. study using CRF data from our repeatability study (Del Giudice et al. 2020). As opposed to zero, our response thresholds and corresponding shaded areas were built using the typical error of measurement and a smallest worthwhile change (see Fig. 1 caption for details). Despite completing two identical training periods, re-exposure (period 2) appeared to rescue CRF low response to the first training period (participants in the top-left yellow shaded area). This apparent inconsistency in individual responses to the same training protocol may question the interpretation of discordant CRF responses to two different training modes as evidence that individuals preferentially respond in CRF to one training mode over another. However, it is important to note that our repeatability study had a small sample size (n = 14), used short training periods (4 weeks each), and only examined CRF responses to HIIT. Thus, future repeatability studies with larger sample sizes, longer training periods, and different outcomes and modes of training – including changes in strength following resistance training – are warranted to confirm our findings. Response thresholds were calculated based on the typical error of measurement and a 50% confidence interval constructed around a smallest worthwhile change (0.2 times baseline standard deviation). The vertical dashed lines represent response thresholds for classifying individuals as CRF ‘high responders’ (above threshold) greater than threshold (above 0) or not responders (below threshold). Thus, participants in the green or red areas were ‘high responders’ or ‘low responders’ to both conditions (concordance), respectively, whereas participants in the yellow areas ‘responded’ to only one condition (discordance). The lack of repeatability of observed responses (Lindholm et al. 2016; Islam et al. 2019; Del Giudice et al. 2020) coupled with the results from Marsh et al. challenge the concept that genetics is a major determinant of exercise response variability. Importantly, the possibility that environmental factors may be stronger determinants than genetics (Marsh et al. 2020) supports the notion that adaptations to exercise training are complex phenotypes arising from intricate gene–environment interactions. Consequently, ascribing phenotypic variation in complex traits (e.g. CRF responses) to specific DNA sequences or ‘candidate’ genes is of little use for personalized exercise medicine. Several expert physiologists, most recently Dr Michael Joyner (Joyner, 2019), have deconstructed the idea that genetics is a major determinant of exercise response variability. We present some of their key arguments below to expand upon Marsh and colleagues’ discussion on genetics and exercise responses. Importantly, although only Dr Joyner's points are highlighted below, equally compelling arguments were presented by Dr Claude Bouchard in support of the role of genetics in determining exercise responses. We therefore encourage the reader to refer to the point–counterpoint between Drs Joyner and Bouchard for a more detailed discussion on this topic (see Joyner, 2019). There is a large degree of redundancy in the molecular pathways that underpin complex adaptive phenotypes. This redundancy likely explains why many purported ‘master regulators’ of muscle metabolic capacity are dispensable for skeletal muscle adaptations to training (Joyner, 2019). Relatedly, studies have failed to identify DNA sequence variants that can independently account for a substantial proportion of the explained variance in training adaptation (see Joyner (2019) for examples). The inability of single genetic variants to predict training adaptations supports the ‘omnigenic hypothesis’: adaptations to exercise training are complex traits that arise from vast gene regulatory networks as opposed to a single or small collection of genes. Additionally, the above-mentioned repeatability issue (Fig. 1) may question the meaningfulness of genetic predictors of exercise response variability. The idea that DNA sequence variations will dictate cellular abundance of the encoded protein, and thus cell function, assumes that the gene will yield a functional protein. In reality, each step in the cascade of events from gene to protein (e.g. transcription, translation, post-translational modifications, etc.) is highly complex, independently regulated, and can be impacted by environmental factors. This complexity likely explains why cellular mRNA levels account for (at most) ∼40% of the variance in tissue protein abundance in humans (de Sousa Abreu et al. 2009). Since mRNA expression lies downstream of DNA sequence variation in the sequence of events leading from gene to protein, DNA sequence variation likely accounts for even less of the explained variance in a given phenotype. Thus, the biological complexity linking DNA sequence to phenotype supports the lack of a genetic influence on training adaptations reported by March and colleagues. The influence of environmental factors on strength and CRF adaptations may instead be mediated by lifestyle-induced epigenetic, transcriptomic and/or proteomic changes occurring over the course of the intervention. Overall, the findings by Marsh and colleagues provide important insight toward individual responsiveness to exercise training and the lack of a contribution of genetics. Future work is needed to continue exploring the repeatability of individual responses to exercise training and to elucidate the mechanisms that contribute to response variability. None. All authors were involved in the conception, writing, editing, and final approval of the manuscript. All authors agree to be accountable for all aspects of the work. All persons designated as authors quality for authorship, and all those who quality for authorship are listed. None. We thank our supervisors Dr Brendon J. Gurd and Dr Chris McGlory for their careful evaluation of our article and constructive feedback.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.