Fuel for the Work Required: A Theoretical Framework for Carbohydrate Periodization and the Glycogen Threshold Hypothesis

On the basis that reduced pre- [15] and post-exercise [16] muscle glycogen availability augments expression of genes regulating substrate utilization and mitochondrial biogenesis, initial training studies adopted a ‘training twice every second day versus once daily’ research design. In this approach, subjects complete a morning training session to reduce muscle glycogen followed by several hours of reduced CHO intake so that the second training session of the day is commenced with reduced muscle glycogen. Using this model, 3–10 weeks of ‘train low’ increases oxidative enzyme activity [9, 17–19], whole body [17, 19] and intramuscular lipid utilization [19] and improves exercise capacity [9] and performance [20]. It is difficult to ascertain if the enhanced training response is mediated by the upregulation of transcriptional responses induced by CHO restriction in recovery from the morning exercise session and/or the enhanced cell signalling responses [21–24] associated with commencing the afternoon training session with reduced muscle glycogen. In addition to low glycogen availability, there may also be compounding effects of stacking training sessions in close temporal proximity where altering the mechanical, metabolic and hormonal environment may also modulate the enhanced training response. Notwithstanding a potential reduction in absolute training intensity in the afternoon session [17], the twice per day model provides a practical framework whereby the accumulative total time with reduced muscle glycogen is increased. Depending on the length of the interval between the first and second session (i.e. recover low) and the actual duration of the second training session (i.e. train low), the accumulated low glycogen period could range from 3 to 8 h.

Exercising fasted represents a simpler model of ‘train low’ whereby breakfast is consumed after a morning training session. In this model, pre-exercise muscle glycogen is not different between fasted or fed conditions, but liver glycogen and circulating glucose is higher during fed conditions. In contrast, increased free fatty acid (FFA) availability and lipid oxidation occur in fasted conditions when exercise is matched for intensity, duration and work performed [25]. Depending on the timing of CHO feeding in relation to the commencement of exercise (e.g. > 60 min before exercise versus < 10 min before and/or during exercise), such differences in FFA availability may manifest from the beginning of exercise [25] or not until after 30–40 min of exercise, respectively [26]. Exercising fasted increases AMP-activated protein kinase (AMPK) activity [27] and post-exercise gene expression [28, 29], while several weeks of fasted training increases oxidative enzyme activity [30], lipid transport protein content [31] and resting glycogen storage [32]. However, it is noteworthy that such adaptations are likely regulated via CHO restriction as opposed to true fasted conditions. Indeed, consuming approximately 20 g of whey protein before and during CHO-restricted training sessions still permits mobilization of FFAs [33] and activation of the AMPK signalling axis [34], while also improving net muscle protein balance [35].

In the ‘sleep low, train low’ model, participants perform an evening training session, restrict CHO during overnight recovery, and then complete a fasted training session the following morning. The accumulative total time with reduced muscle glycogen could therefore extend to 12–14 h depending on the timing and duration of the training sessions and sleep period. Acute models of ‘sleep low, train low’ (where morning exercise is commenced with glycogen < 200 mmol/kg dw) enhance the activation of AMPK, p38 mitogen-activated protein kinase (p38) and tumour protein p53 (p53) signalling [36–38], although responses are attenuated if trained cyclists complete a morning steady-state session (thus suggesting that both intensity and training status are important) where pre- and post-exercise glycogen content is 350 and 250 mmol/kg dw, respectively [39]. Using a sleep-low model similar to that used by Lane and colleagues [39], Marquet et al. [40, 41] observed that 1–3 weeks of sleep-low training in elite triathletes and cyclists improves cycling efficiency (3.1%), 20 km cycling time-trial performance (3.2%) and 10 km running performance (2.9%) compared with traditional train-high approaches. At present, the extent of CHO restriction required to elicit conditions considered best representative of ‘sleep low’ are not well-defined. However, it is noteworthy that studies demonstrating beneficial effects on cell signalling [38] and performance adaptations [40, 41] have completely restricted CHO after the evening training session and after subjects were fed a protein-only meal or beverage. Nonetheless, similar to other train-low models, the sleep-low paradigm is subject to the obvious limitations of lack of a placebo controlled, double-blinded design. Such limitations are particularly relevant where performance improvements have been observed with only three sessions of ‘sleep low, train low’ [41].

In addition to manipulating CHO availability, it is possible that associated elevations in circulating FFA availability also regulate relevant cell signalling pathways. Indeed, the exercise-induced activation of p38MAPK is suppressed with pharmacological ablation of FFA availability [42]. Nonetheless, while we acknowledge the potential role of acute FFA-mediated signalling (as occurring secondary to the primary intervention of CHO manipulation), it is unlikely that CHO restriction in combination with chronic high-fat feeding is beneficial for training adaptations. Indeed, 1–5 days of high-fat feeding reduces the expression [43] and activity of the pyruvate dehydrogenase (PDH) complex [44], ultimately impairing CHO oxidation and high-intensity performance. Furthermore, Burke et al. [45] observed that exercise economy and performance were negatively impacted in elite race walkers following 3 weeks of a high-fat diet when compared with periodized and high CHO availability. Emerging data also suggest that high-fat feeding may impair muscle protein synthesis [46], potentially via reduced activation of mammalian target of rapamycin (mTOR) and ribosomal protein S6 kinase (p70S6K) signalling [43].

In the real-world environments of elite endurance athletes, it is likely that athletes practice an amalgamation of the aforementioned train-low paradigms (either through default of their current training structure or via coach and sport scientist-led practices), as opposed to undertaking one strategy in isolation. Additionally, elite athletes may also undertake 20–30 h of training per week, whereas many of the study designs reviewed thus far have utilized training programmes of < 10 h per week. The complexity of practical train-low models is exacerbated by observations that endurance athletes practice day-to-day or longer-term periods of energy periodization (as opposed to CHO per se) in an attempt to reduce body and fat mass in preparation for competition [10, 47] (Morton, unpublished observations). Indeed, the performance improvements observed by Marquet et al. [40] were also associated with a 1 kg reduction in fat mass induced by the sleep-low model. Notwithstanding reductions in body mass, it is indeed possible that many of the skeletal muscle adaptations associated with ‘train low’ are mediated by repeated and transient periods of energy restriction as opposed to CHO restriction per se. We observed that the post-exercise expression of peroxisome proliferator-activated receptor gamma coactivator 1-α (PGC-1α), p53, mitochondrial transcription factor A (Tfam) and peroxisome proliferator-activated receptor (PPAR) messenger RNA (mRNA) were elevated with similar magnitude and time-course when a low CHO and high-fat diet was consumed versus an isoenergetic high-CHO feeding strategy [43]. Such data conflict with previous observations from our laboratory [38], where post-exercise CHO and calorie restriction augments the expression of many of the aforementioned genes. Given the similarities in metabolic adaptation to both CHO and calorie restriction, such data raise the question as to whether the enhanced mitochondrial responses observed when training low are actually due to transient periods of calorie restriction (as mediated by a reduction in CHO intake) as opposed to CHO restriction per se. This point is especially relevant given that many endurance athletes present daily with transient periods of both CHO and calorie restriction due to multiple training sessions per day, as well as longer-term periods of suboptimal energy availability [47].