Continuous glucose monitoring maps a late-race metabolic shift in marathon runners
A marathon does not fail at a single point. In this cohort of 26 elite amateur runners, the breakdown unfolded gradually, with continuous glucose monitoring capturing a rise in interstitial glucose during the race, followed by a decline in the final 11 kilometers as speed also dropped in the latter stages. The study, conducted during the 2023 Wuxi Marathon, followed 24 men and 2 women across 10-kilometer segments and linked glucose dynamics to pacing patterns rather than to a single static number. That design matters because endurance performance is not governed by one fuel source alone; glycogen depletion, fluid balance, heat strain, neuromuscular fatigue, and pacing strategy all press on the finish time at once.
Mean interstitial glucose concentration during the race was significantly higher than baseline. Even so, the relationship between glucose and speed was not straightforward. Generalized linear mixed models showed significant differences in interstitial glucose across segments, with declines in the 31–40 km and 41–42 km portions compared with the first 10 km. Speed followed its own trajectory, with significant slowing in the 21–30 km, 31–40 km, and 41–42 km segments versus the opening 10 km. The two curves moved in the same race, but not in a simple lockstep pattern.
Late-race slowing appeared before glucose alone could explain it
The final third of a marathon is where peripheral fatigue, central drive, and substrate availability begin to collide. Muscle glycogen depletion can reduce the capacity for sustained high-intensity effort, while the body may raise circulating glucose through counterregulatory hormones in an attempt to preserve fuel delivery. A rise in interstitial glucose during competition can therefore reflect stress physiology as much as nutritional adequacy. The drop seen late in the course is compatible with increased muscular extraction and a waning ability to maintain stable metabolic supply, but the data do not support a one-variable explanation for the decline in speed.
No significant linear relationship emerged between interstitial glucose concentration and speed, either across the whole race or within individual 10-kilometer segments. That finding is clinically and physiologically important for sports medicine because it argues against treating CGM values as a direct surrogate for pacing quality. Runners can slow while glucose remains relatively elevated, and glucose can fall without a proportional immediate collapse in speed. Marathon fatigue is therefore distributed across multiple systems, not concentrated in one biomarker.
Pre-race carbohydrate intake also failed to show a significant association with speed or interstitial glucose during the first 10 kilometers. In practical terms, the opening segment did not reveal a measurable performance edge tied to carbohydrate consumed before the start. The absence of a signal in that early race window does not mean carbohydrate loading has no value; rather, it suggests that within this specific cohort and race context, early pacing and pre-race intake were not enough to separate runners on the basis of the measured outcomes.
In-race carbohydrate intake tracked longer finish times, not better glucose control
A higher mean carbohydrate intake rate during the race was associated with longer completion time. The direction of that association runs counter to what athletes often hope for, but it is not difficult to explain on physiologic grounds. Slower runners spend more time on course, which gives them more opportunity to ingest carbohydrate and inflate the average intake rate recorded over the event. The data do not show that carbohydrate intake caused slower performance; they show that intake rate and race duration moved together in this observational setting. No significant correlation was found between mean carbohydrate intake rate and mean interstitial glucose concentration.
This is where observational sports physiology has to resist overinterpretation. A CGM trace can describe the fuel environment, but it cannot determine whether a runner slowed because the glucose curve changed, because the legs faded, or because the cardiovascular and thermoregulatory burden mounted beyond tolerance. The analysis used generalized linear mixed models and linear mixed models with runner as a random intercept, a sensible approach for repeated measures across segments, yet even strong statistical handling cannot convert association into mechanism. The study instead points to a more restrained conclusion: glucose is part of the marathon response, not the whole response.
The segment-level findings make that point more concrete. Glucose held up or even rose above baseline during much of the event, then declined when fatigue became most visible at the back end of the course. Speed worsened earlier and more broadly in the second half. Those two trajectories overlap, but they are not identical. A runner can retain measurable glucose availability while still being unable to convert it into the force, cadence, and economy needed to sustain pace.
What this marathon study can and cannot tell coaches and clinicians
The sample was small, consisting of 26 elite amateur runners, and the sex distribution was heavily skewed toward men. The work was observational, which limits causal interpretation, and the report does not justify extending these findings to recreational runners, professional marathoners, or athletes racing under different environmental conditions. Even so, the study contributes a useful physiologic snapshot: marathon performance should not be interpreted through glucose monitoring alone, because late-race slowing reflects a broader exhaustion state that CGM cannot fully capture.
For clinicians and sports scientists, the practical message is narrow but useful. CGM can reveal how glucose behaves across the race course, yet it should be paired with pacing data, carbohydrate strategy, and a broader assessment of fatigue physiology. A single biomarker cannot explain the collapse that often appears in the final 10 kilometers. The finish line is where metabolic strain becomes visible, but the cause is usually spread across the whole race.