REF-1759 · Main et al., 2026 · European Journal of Applied Physiology, · Paper · PubMed

In 9 well-trained female cyclists, a day of very low carbohydrate (≤1 g/kg) after glycogen-depleting exercise, against ≥9 g/kg, lowered power at the first ventilatory threshold from 152 to 133 W (−19 ± 14 W) and lowered gross efficiency, while power at the lactate threshold and end power in a 3-min all-out test did not differ

A small randomised crossover tested fresh, not after a long ride

Main caveat: Nine riders; one crossover comparison; the abstract gives numbers only for VT1.
limited evidenceMeasured on trained to highly trained athletesRead from the abstract only

Main and colleagues asked whether lowered carbohydrate availability, the state a rider reaches late in a long ride, moves power at the intensity-domain transitions, muscle activation and cycling efficiency in endurance-trained women.

Nine well-trained female cyclists did glycogen-depleting exercise about a day before each test and then ate either at least 9 g/kg of carbohydrate or no more than 1 g/kg, in randomised, counterbalanced order. On the test day an incremental test and a 3-min all-out test followed.

With low carbohydrate, power at the first ventilatory threshold was 19 W lower and gross efficiency in submaximal cycling was lower, while power at the lactate threshold and end-test power in the 3-min test did not differ. Median power frequency of the vastus lateralis and medialis EMG differed between conditions with no difference in EMG amplitude.

The authors propose that glycogen-depleted fibres were compensated for by recruiting higher-threshold motor units, costing efficiency, and conclude that carbohydrate availability is likely important for the durability of the moderate-to-heavy transition.

Participants

n = 9 well-trained female cyclists (endurance-trained).
Not in the abstract: age, VO2max, training volume, competition level, menstrual-cycle phase at testing, hormonal contraceptive use.

Protocol

Randomised, counterbalanced crossover, two conditions.
About 24 h before each test: glycogen-depleting exercise, then carbohydrate intake of ≥9 g/kg (HIGH) or ≤1 g/kg (LOW).
Test day: incremental cycling test (first ventilatory threshold, lactate threshold, gross efficiency, EMG of vastus lateralis and medialis) and a 3-min all-out test (end-test power).
Not in the abstract: the depleting protocol, how long the diet ran and what replaced the carbohydrate energy in LOW, which lactate threshold method was used, whether glycogen was measured, and test-day feeding.

Key Results

OutcomeLOW vs HIGHP
Power at first ventilatory threshold133 ± 24 vs 152 ± 28 W; difference −19 ± 14 W0.011
Power at lactate thresholdno between-trial difference (no values given)—
Gross efficiency, submaximal cyclinglower in LOW (no values given)0.003
EMG median power frequency, vastus lateralishigher in HIGH, as the abstract's results state0.025
EMG median power frequency, vastus medialishigher in HIGH, as the abstract's results state0.007
EMG amplitudeno between-trial difference—
3-min all-out end-test powerno between-trial difference (no values given)—
The title says low carbohydrate "increases" median power frequency, and the authors' explanation (more higher-threshold motor units in LOW) fits that direction, but the abstract's results sentence reports it higher in HIGH; which direction is correct cannot be settled from the abstract.

Practical Application

Starting a session with low carbohydrate stores pulled the first threshold down: in 9 well-trained women, after a day on ≤1 g/kg, power at VT1 was 19 W lower (133 vs 152 W) and cycling efficiency was lower than after ≥9 g/kg, while power at the lactate threshold and 3-min all-out end power did not differ significantly. It fits SYNTH-32's point that durability is partly a fuelling variable and REF-1121's finding that burning more of the body's own carbohydrate deepened the VT1 slide through efficiency; like REF-1121, the CP-like 3-min end power did not move.
What this was not: the women were tested rested after a day of depletion, not at hour three of a ride, and with no carbohydrate during exercise; so it shows what low glycogen availability does to the first threshold, not how much a fuelled rider's threshold falls in a long event.
Inference, not tested here: a session planned at Zone 2 or tempo the day after a hard depleting ride on little carbohydrate will sit relatively higher above the first threshold than the same watts when fuelled; refuelling after hard days, or easing the targets, keeps the session in its intended domain (SYNTH-02 for during-ride carbohydrate).
Not settled by this design: whether the lactate-threshold result really diverges from VT1 or the test lacked power to show it (n = 9), the direction of the EMG change, and whether menstrual phase or contraception played a part (not reported in the abstract).

Limitations

Nine riders; one crossover comparison; the abstract gives numbers only for VT1.
The title and the results sentence of the abstract disagree on the direction of the EMG change.
Glycogen manipulated by diet the day before, tested rested; whether muscle glycogen was measured, and how energy intake was matched between conditions, is not in the abstract, so low carbohydrate and low energy may be mixed.
Menstrual-cycle phase, contraceptive use, age and fitness are not in the abstract.
The lactate-threshold method is not named, so the VT1-but-not-LT result cannot be interpreted further.
Written from the abstract and the PubMed record; full text not read.
How this entry was checked and changed
Why this level

a crossover of 9 (the rubric's B band for crossovers under 15); the power and efficiency outcomes are capacity measures, not markers; the EMG findings are markers and the abstract's EMG direction conflicts with the title