Longitudinal muscle ultrasound changes and exploratory early risk assessment for ICU-acquired weakness: a prospective exploratory study

ICU-acquired weakness (ICU-AW) is a frequent complication of critical illness, associated with prolonged mechanical ventilation and poor functional outcomes. Early detection remains challenging because conventional strength testing requires patient cooperation, whereas electrophysiological assessment, the diagnostic reference standard in sedated patients, is technically demanding. This study investigated longitudinal changes in multimodal quadriceps ultrasound parameters and their preliminary value for identifying and assessing the risk of ICU-AW. In this prospective observational study, 188 ICU patients were screened; 100 met eligibility and provided consent, and 43 completed serial ultrasound examinations on Days 1, 5, and 10 together with an evaluable Day-10 electrophysiological assessment. Parameters included rectus femoris cross-sectional area (RF-CSA) and thickness, vastus intermedius pennation angle (VI-PA), subcutaneous fat thickness, shear wave velocity, and elastic modulus. ICU-AW was determined on Day 10 using prespecified electrophysiological criteria. Longitudinal trajectories were compared between ICU-AW and non-ICU-AW groups. ROC analyses evaluated the concurrent discriminatory performance of Day 1–10 changes and the exploratory early risk-assessment value of Day-5 measurements. Nested logistic regression models assessed whether early ultrasound added information beyond age and APACHE II. Given the small sample size, internal validation was performed using 1000 bootstrap resamples, but findings remain hypothesis-generating and require external validation. Among 43 completers, 23 (53.5%) developed ICU-AW. These patients were older and had higher APACHE II scores, though baseline ultrasound parameters did not differ significantly. Patients with ICU-AW showed greater declines in RF-CSA and VI-PA, with significant time-by-group interactions in side-specific analyses. The relative reduction in mean RF-CSA from Day 1 to Day 10 showed strong discriminatory performance in this exploratory cohort (AUC 0.913, 95% CI 0.807–0.987), with a preliminary cutoff of 24.09%. For VI-PA, the AUC was 0.850 (95% CI 0.722–0.954), with a cutoff of 26.82%. Day-5 VI-PA alone yielded an AUC of 0.788. Adding Day-5 VI-PA to age and APACHE II increased the apparent AUC from 0.854 to 0.896 and improved model fit (likelihood-ratio χ² = 5.079, P = 0.024), although the pairwise AUC comparison was not statistically significant (DeLong P = 0.269), whereas adding Day-5 RF-CSA did not (χ² = 0.029, P = 0.865). These AUC estimates were derived and evaluated in the same cohort, and the wide 95% CI approaching 1.0 (0.807–0.987) indicates statistical instability due to the small sample size; they should therefore not be interpreted as expected performance in new populations. Serial quadriceps ultrasound revealed distinct muscle deterioration patterns in patients with electrophysiologically confirmed ICU-AW. RF-CSA and VI-PA changes over 10 days were associated with concurrent ICU-AW, while Day-5 VI-PA improved model fit but did not significantly improve discrimination beyond clinical variables (likelihood-ratio χ² = 5.079, P = 0.024; DeLong P = 0.269). Because patients who died or left the ICU before Day 10 were necessarily excluded, the findings apply to a selected intermediate-risk spectrum and should not be extrapolated to all critically ill patients. These preliminary, hypothesis-generating findings require validation in larger independent cohorts before clinical translation.