Hypothesis

Circadian‑adjusted postprandial glucose recovery slope (PRGS) combined with concurrent ketone body concentration predicts clamp‑measured metabolic flexibility (ΔRQ) in healthy adults.

Rationale

CGM captures glucose excursions but lacks validation against gold‑standard ΔRQ measures[[1]][[2]]. Postprandial glucose decline reflects hepatic insulin sensitivity and mitochondrial capacity to switch from glucose to fat oxidation. Ketone bodies rise when fat oxidation increases, providing a direct read‑out of lipid utilization. Circadian timing modulates both hepatic glucose output and ketone production via clock genes such as REV‑ERBα[[3]][[4]]. Therefore, a metric that integrates glucose recovery speed, ketone level, and time of day should more closely mirror the substrate‑shift index ΔRQ obtained during hyperinsulinemic‑euglycemic clamps[[5]][[6]].

Predictions

1. PRGS (negative slope of glucose concentration from peak to baseline) will correlate positively with ΔRQ (higher slope = greater flexibility) only when adjusted for circadian phase (morning vs evening).
2. Adding postprandial ketone concentration to the model will significantly improve prediction accuracy (increase in R² > 0.15) compared with PRGS alone.
3. Traditional CGM metrics such as postprandial AUC, glycemic variability, or recovery time will show weaker or non‑significant associations with ΔRQ after controlling for circadian phase and ketones.
4. The relationship will hold across a mixed‑meal challenge containing 50 g carbohydrate, 30 g fat, 20 g protein.

Experimental Design

• Recruit 60 non‑diabetic adults aged 20‑35, balanced for sex.
• Equip participants with a blinded CGM and a continuous ketone monitor.
• Conduct two standardized mixed‑meal challenges: one at 08:00 h and another at 19:00 h, separated by at least 48 h.
• Compute PRGS for each challenge: fit a linear regression to glucose values from peak (max) to return to fasting baseline; slope (mg/dL min⁻¹) is taken as negative value.
• Record average ketone concentration during the same window.
• Within one week, perform a hyperinsulinemic‑euglycemic clamp with indirect calorimetry to determine ΔRQ (difference between basal and insulin‑stimulated respiratory quotient).
• Perform hierarchical regression: Model 1 = PRGS + circadian dummy; Model 2 = Model 1 + postprandial ketone; Model 3 = Model 2 + traditional CGM metrics.
• Primary outcome: change in R² and standardized beta coefficients.

Potential Outcomes

If the hypothesis is correct, Model 2 will show a significant increase in explanatory power over Model 1 (p < 0.01) and PRGS + ketone will retain significance in Model 3, whereas traditional CGM metrics will not. Failure to observe improved prediction would falsify the claim that ketone‑adjusted glucose recovery reflects metabolic flexibility, suggesting that additional factors (e.g., mitochondrial enzyme activity) are required for a valid CGM‑based proxy.