🧬 biology

A CGM-Derived Late-Phase Glycemic Response Phenotype During Fasted Moderate-Intensity Running

This study identifies a late-phase glycemic response phenotype (∆G) during fasted moderate-intensity running as a reliable metric only when averaged across repeated trials, revealing that while it correlates with skeletal muscle index in recreational exercisers, this association is confounded by sex and exercise intensity, indicating ∆G is not a direct biomarker of muscle mass.

Original authors: Wooyoung Park, Sanghyeon Ju, Minho Shong, Sukyung Park

Published 2026-07-08

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Wooyoung Park, Sanghyeon Ju, Minho Shong, Sukyung Park

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The Big Idea: Tracking Blood Sugar While Running

Imagine your body is a car, and your blood sugar is the fuel gauge. Usually, we check this gauge when the car is parked (fasting) or right after we fill the tank (eating). But what happens to the fuel gauge while you are driving at a steady speed on a long highway?

This study asked that exact question. Researchers wanted to see if the way your blood sugar behaves during a 60-minute run could tell us something about your muscle mass. They used a special sensor (a Continuous Glucose Monitor, or CGM) that acts like a tiny dashboard camera, recording your fuel levels every few minutes while you run.

The Experiment: The "Steady Cruise" Test

The researchers gathered 30 healthy adults (a mix of men and women with different running backgrounds). They asked them to:

Fast overnight (no food for at least 8 hours) so their "fuel tank" started empty.
Run on a treadmill for 60 minutes.
Run at a "moderate" pace. They didn't just tell people to run "easy." They used heart rate monitors to ensure everyone was working at the same relative effort (like driving at 60% of your maximum speed).

They measured the change in blood sugar during the last 30 minutes of the run. They called this change ΔG (Delta-G). Think of ΔG as the "fuel drift": Did the fuel gauge go up, stay flat, or go down during the second half of the run?

Key Finding 1: One Run Isn't Enough (The "Weather" Analogy)

The researchers found that if you only run once, your blood sugar reading is a bit like checking the weather for a single day. It's noisy and unpredictable.

The Analogy: If you check the temperature once, you might get a weird reading because of a sudden cloud. You need to check it over several days to know the average climate.
The Result: A single run gave a "low reliability" score. However, when they averaged the results from 2 or 3 runs per person, the data became much clearer and more reliable.
Takeaway: To use this blood sugar metric as a personal trait, you can't just do it once; you need to do it a few times and take the average.

Key Finding 2: Muscle Mass vs. Blood Sugar (The "Engine Size" Analogy)

The main question was: Does having bigger muscles (Skeletal Muscle Index) mean your blood sugar behaves differently?

The "Recreational" Runners: For people who run occasionally (the "recreational" group), there was a pattern. People with more muscle tended to have a smaller rise (or even a drop) in blood sugar during the run. It's like a bigger engine (more muscle) is better at burning the fuel efficiently.
The "Trained" Runners: For people who had run half-marathons or longer (the "trained" group), this pattern disappeared. Muscle size didn't seem to predict blood sugar behavior at all.

Why the difference?
The researchers realized that "running at the same heart rate" doesn't mean "running at the same speed."

The Analogy: Imagine two cars driving at 60% of their top speed. A small, old car might be going 30 mph, while a high-performance sports car might be going 80 mph. Even though they are both working at "60% effort," the sports car is doing much more actual work.
The Reality: The trained runners were much faster than the recreational runners, even though their heart rates were similar. Because the trained group was running at a much higher absolute speed, their bodies were reacting differently, and the simple link between "muscle size" and "blood sugar" got lost in the noise.

Key Finding 3: The "Gender" Confusion

The study also found that it was hard to separate the effects of muscle mass from gender.

The Analogy: Imagine trying to figure out if a taller person is faster because of their height or because they are a man. In this group, men tended to be taller and have more muscle, while women tended to be shorter with less muscle.
The Result: When they looked at the "recreational" group, it looked like muscle size mattered. But when they separated men and women, the link disappeared. It turns out the pattern was actually a mix of both muscle size and gender differences in how bodies burn fuel. You can't easily say "muscle size alone caused this."

The Bottom Line

This study proposes a new way to look at blood sugar during exercise, but it comes with a few big "caveats":

It's a "Phenotype," not a simple test: The blood sugar change during a run (ΔG) is a complex pattern, not a simple "good" or "bad" number.
Don't trust a single run: You need to run multiple times and average the results to get a reliable number.
Muscle isn't the whole story: While muscle mass seemed to matter for casual runners, it wasn't a direct, standalone predictor for everyone. Factors like gender, running experience, and how fast you are actually running (not just your heart rate) play huge roles.

In short: The study suggests that tracking blood sugar while running could be a useful way to understand how our muscles work, but we need to be very careful about how we measure it, who we measure, and how many times we measure it before we draw any conclusions.

Technical Summary: A CGM-Derived Late-Phase Glycemic Response Phenotype During Fasted Moderate-Intensity Running

Problem Statement
While skeletal muscle is the primary site of glucose disposal during exercise, it remains unclear whether glycemic responses measured by Continuous Glucose Monitoring (CGM) during exercise accurately reflect muscle-related characteristics (such as muscle mass) in healthy adults. Existing literature largely focuses on basal or static metabolic challenges (e.g., fasting glucose, HOMA-IR), leaving the dynamic relationship between exercise-induced glucose fluctuations and muscle mass underexplored. Furthermore, CGM-derived glycemic responses during exercise exhibit high day-to-day variability and diverse temporal phenotypes (increases, stability, or decreases), making it difficult to define a stable, individual-level metric. The study addresses the challenge of standardizing these complex time-series data into a reliable phenotype and determining its association with the Skeletal Muscle Index (SMI).

Methodology
This exploratory, repeated-measures study involved 30 healthy adults (19 men, 11 women) with varying running experiences. Participants underwent fasted moderate-intensity running on a treadmill for 60 minutes, with intensity controlled within 40–60% of Heart Rate Reserve (HRR).

Data Collection: Participants wore Dexcom G7 CGM sensors. Measurements were taken on 2–3 separate occasions per participant (76 valid trials total) to assess test-retest reliability.
Metrics Defined:
- $\Delta G$ (Glucose Increase): The change in mean glucose during the final 30 minutes of exercise relative to the fasting baseline ( $G_0$ ). This was intended to capture the stable late-phase response.
- $\Delta G_{early}$ : The change in mean glucose during the first 30 minutes relative to baseline.
Analysis:
- Unsupervised Learning: $k$ -shape clustering and Principal Component Analysis (PCA) were applied to CGM time-series data to identify dominant temporal patterns and validate whether $\Delta G$ captures the primary variance.
- Reliability: Intraclass Correlation Coefficients (ICCs) were calculated for single trials and averaged repeated measurements.
- Associations: Linear and multiple regression models were used to test the association between $\Delta G$ and SMI, stratified by sex and running experience (classified post hoc as "recreational" vs. "trained" based on half-marathon completion history).

Key Contributions

Definition of a Task-Specific Phenotype: The study proposes $\Delta G$ (late-phase glucose increase) as a candidate CGM-derived glycemic response phenotype. PCA results indicated that $\Delta G$ closely corresponds to the dominant axis of time-series variation (67% of total variance, $R^2 = 0.92$ ), validating it as a summary metric for the late-phase response.
Reliability Characterization: The study demonstrates that single-trial $\Delta G$ has low reliability (ICC = 0.424) due to day-to-day variability. However, averaging across 2–3 repeated measurements improves reliability to a moderate level (ICC = 0.642), suggesting that $\Delta G$ should be interpreted as a repeated-measurement average rather than a single-session value.
Contextualized Muscle-Response Relationship: The research clarifies that the relationship between SMI and glycemic response is not a simple univariate association. It highlights the confounding effects of sex and the limitations of HRR-based intensity prescription when comparing groups with different training statuses.

Results

Temporal Patterns: CGM time series exhibited diverse phenotypes (increasing, stable, decreasing). However, the late-phase response ( $\Delta G$ ) was the primary driver of inter-individual variation.
SMI Association:
- In the full sample, the association between SMI and $\Delta G$ was negative but not statistically significant ( $p = 0.084$ ).
- In the recreational group ( $N=21$ ), a significant negative association was observed ( $R^2 = 0.24, p = 0.024$ ), where higher SMI correlated with lower $\Delta G$ . However, this association was not independent of sex; when analyzed separately by sex or adjusted for sex in a multiple regression, the significance disappeared. This suggests the finding reflects the joint distribution of SMI, sex, and $\Delta G$ rather than an independent effect of muscle mass.
- In the trained group ( $N=9$ ), no significant association was found.
Intensity Discrepancy: Despite similar HRR ranges (40–60%) between recreational and trained groups, the trained group ran at significantly higher absolute speeds. This indicates that HRR-based prescription did not guarantee matched absolute exercise intensity, potentially obscuring the SMI- $\Delta G$ relationship in the trained group due to differences in metabolic load and training adaptations.

Significance and Claims
The paper concludes that $\Delta G$ is not a direct biomarker of muscle mass but serves as a candidate CGM-derived glycemic response phenotype. Its interpretation requires:

Repeated Measurements: To overcome inherent day-to-day variability and achieve moderate reliability.
Contextual Consideration: Careful accounting for sex, running experience, and the distinction between relative (HRR) and absolute exercise intensity.

The study does not claim to elucidate the specific physiological mechanisms (e.g., separating hepatic glucose production from muscle uptake) but establishes the experimental conditions necessary to interpret CGM-derived exercise responses as individual-level metrics. It suggests that future research must address sample stratification (sex-balanced groups) and use more precise intensity controls (e.g., VO2max or lactate thresholds) to isolate the effects of muscle mass on exercise glycemia.