REF-726 · Gillen et al., 2019 · Medicine & Science in Sports & Exercise, 2019 Nov;51 · Paper · PubMed
Training with low carbohydrate availability RAISES the protein requirement
The train-low session that buys the adaptation also raises the amino-acid bill
Protocol
n=8 endurance-trained males (27 ± 4 y, 75 ± 10 kg, V̇O₂ 67 ± 10 mL·kg⁻¹·min⁻¹). Two trials matched for energy and macronutrient composition, differing only in how carbohydrate was periodised around a "sleep-low" pattern:
| LOW availability | HIGH availability | |
|---|---|---|
| Before evening HIIT (10 × 5 min at 10-km race pace, 1 min rest) | 7.8 g CHO·kg during the day | 3 g CHO·kg during the day |
| After that session | withheld — 0.2 g·kg | 5 g CHO·kg to resynthesise glycogen |
| Next morning 10 km run (~80% HRmax) | fasted | 1 h after 1.2 g CHO·kg |
Whole-body phenylalanine flux and oxidation measured over 8 h of recovery by oral ¹³C-phenylalanine, while consuming sufficient energy, 7.8 g CHO·kg·d and suboptimal protein (0.93 g·kg·d), as the indicator amino acid oxidation method requires.
Key Findings
Fat oxidation during the morning run was higher in LOW (0.99 ± 0.35 vs 0.60 ± 0.26 g·min⁻¹, p < 0.05) — the train-low protocol did what it is supposed to do.
Phenylalanine oxidation during recovery was higher in LOW (8.8 ± 2.7 vs 7.9 ± 2.4 μmol·kg⁻¹·h⁻¹, p < 0.05). Since oxidation is the reciprocal of synthesis in this method, more amino acid was being burned rather than built with — a greater amino-acid requirement to support the same rate of whole-body protein synthesis.
Phenylalanine flux did not differ between trials. Inference, not tested: the effect is in partitioning, not in overall turnover.
Authors' conclusion, plainly: endurance exercise performed with low carbohydrate availability increases protein requirements.
Practical Application
In 8 endurance-trained men, a sleep-low day (carbohydrate withheld after evening intervals, fasted 10-km run next morning) raised fat oxidation during the run (0.99 vs 0.60 g·min⁻¹) and raised phenylalanine oxidation over 8 h of recovery (8.8 vs 7.9 μmol·kg⁻¹·h⁻¹) versus the same diet with carbohydrate around the sessions — indicating a higher protein requirement. Train-low is not free: it adds to the amino-acid bill.
Inference, not tested: withholding carbohydrate is the intervention; withholding protein as well after the session is not part of it, so protect the post-session protein feeding.
This is a cost, not a veto. The paper's premise is that low-carbohydrate training enhances markers of aerobic adaptation; the performance evidence for train-low is less consistent (REF-543).
Not from this paper; the base's own practice: keep carbohydrate periodised while keeping protein constant and high across the day, rather than periodising both together.
Limitations
n=8, males only, single session, acute 8-hour recovery window. No training-block outcome, no performance measure, no women.
Protein was held suboptimal (0.93 g·kg·d), as the method requires; this means the absolute numbers do not describe an athlete eating adequately.
The finding is a change in amino acid oxidation, an indirect marker — it does not quantify how many extra grams per day are needed, only that the requirement moves up.
One specific sleep-low periodisation pattern; other train-low implementations (fasted morning, twice-a-day, restricted post-session only) were not tested.
Trained males (mean age 27); masters, female and recreational athletes are not represented.
How this entry was checked and changed
Why this level
regraded 2026-09-27 from B+: crossover, n=8
Content correction 2026-09-27
Level B+ -> B: crossover with n=8 (rubric). Sport line narrowed to running (the protocol was running).
"Train-low retains its documented benefits" -> performance evidence is less consistent (REF-543); post-session and energy-deficit points labelled as inference; implementation advice labelled as practice.
Lead rewritten with the abstract's fat- and phenylalanine-oxidation numbers; 1-min rest, 7.8 g CHO·kg·d background diet added.
Byline co-authors were wrong (Parr, Trommelen, Burke, Devlin) -> Gillen, West, Williamson, Fung & Moore per the DOI record.