Adipose tissue thickness and wearable muscle oxygen saturation during graded treadmill running: a site-specific mixed-effects analysis

Wearable continuous-wave near-infrared spectroscopy (CW-NIRS) provides a device-derived estimate of local muscle oxygen saturation (SmO₂), rather than a direct or absolute measure of microvascular oxygen saturation. Subcutaneous adipose tissue thickness (ATT) may influence the interpretation and reliability of wearable SmO₂, particularly in between-subject comparisons during graded exercise. This study examined the site-specific associations of ATT with the SmO₂–intensity relationship during graded treadmill running and generated model-based predictions at a common ATT reference. A total of 157 healthy adults, including 79 habitually physically active and 78 recreationally untrained participants, completed a graded treadmill exercise test with continuous SmO₂ monitoring at the bilateral vastus lateralis and right gastrocnemius lateralis. Local ATT was estimated from skinfold thickness at each monitoring site. Stage-wise SmO₂ responses were analysed using site-specific linear mixed-effects models including relative exercise intensity, group, sex, mean-centred ATT, group × intensity, and ATT × intensity, with a participant-specific random intercept. Adjusted marginal predictions were subsequently generated at the site-specific mean ATT. SmO₂ declined progressively with increasing exercise intensity at all monitored sites, with significant negative associations at VL-R (β = −66.18, p < 0.001), VL-L (β = −64.26, p < 0.001), and GL-R (β = −35.39, p < 0.001). Habitually physically active participants exhibited lower exercise SmO₂ than recreationally untrained participants at the vastus lateralis, with significant group effects at VL-R (β = 11.57, p < 0.05) and VL-L (β = 12.47, p < 0.05). ATT effects were site-specific, with a significant ATT × intensity interaction at VL-L (β = 2.60, p < 0.05) and a significant positive ATT main effect at GL-R (β = 3.20, p < 0.05). Predictions generated at a common ATT reference showed reduced ATT-related dependence while retaining the observed between-group patterns. ATT was associated with the between-subject variation in wearable CW-NIRS-derived SmO₂ during graded treadmill running in a site-specific manner. Site-specific mixed-effects adjustment can provide marginal SmO₂ predictions at a common ATT reference and reduce ATT-related dependence in model-derived estimates. This statistical procedure does not correct the underlying optical signal or establish improved measurement validity; external validation against an independent criterion is required.