PP343 - DETERMINATION OF INSULIN RESISTANCE IN ICU, A PHYSIOLOGICAL CLINICAL STUDY, THE DIRICU STUDY

PP343

DETERMINATION OF INSULIN RESISTANCE IN ICU, A PHYSIOLOGICAL CLINICAL STUDY, THE DIRICU STUDY

C. Dupuis^1,*, V. Alves de Araújo², C. Annicchiarico³, G. Macedo³, C. Pinto Teixeira³, V. Uyttendaele⁴, T. Desaive⁴, F. Taccone³, C. Miriam⁵, T. Carton⁵, M. Seret⁴, J.-C. Preiser²

¹Intensive care medicine, Clermont Ferrand University hospital, Clermont Ferrand, France, ²intensive care, ³Intensive care medicine, Hopital Universitaire libre de Bruxelles, Bruxelles, ⁴Université de Liège, Liège, ⁵Hopital Universitaire libre de Bruxelles, Bruxelles, Belgium

Rationale: Insulin resistance (IR) is a key feature of critical illness, with dynamic changes across metabolic phases. However, its bedside assessment remains challenging. Continuous glucose monitoring (CGM)–derived metrics may offer a pragmatic alternative, but their validity as surrogates of intrinsic insulin sensitivity (SI) is unclear. This study aimed to evaluate CGM-derived indices against SI measured by intravenous glucose tolerance testing (SI_IVGTT).

Methods: In this prospective single-center physiological study, 20 non-diabetic ICU patients with stress hyperglycemia were included within 48 hours of admission. CGM (Dexcom G7) was initiated, and IVGTT was performed on days 2 and 4. SI_IVGTT was estimated using the minimal model. CGM-derived indices (CGM-KG, CGM-T50, CGM-AUC0–180), HOMA-IR, and SI derived from the Intensive Control Insulin-Nutrition-Glucose model (ICING) (SI_ICING) were compared with SI_IVGTT.

Results: No significant correlation was observed between CGM-KG and SI_IVGTT at day 2 (ρ=0.07, p=0.76). At day 4, moderate associations emerged (ρ=0.59, p=0.016) but were not significant after correction (FDR-adjusted p=0.069). CGM-derived indices overall showed limited performance. In contrast, SI_ICING showed the strongest association with SI_IVGTT at day 4 (ρ=0.64, p<0.001, FDR-adjusted p=0.048), with good discrimination (AUC 0.87). HOMA-IR was not informative. CGM indices were more relevant in patients without insulin therapy.

Conclusion: CGM-derived metrics do not reliably reflect intrinsic insulin sensitivity in ICU patients, especially under insulin therapy. Treatment-integrated indices such as SI_ICING may better capture metabolic responsiveness. Further studies are needed to improve metabolic monitoring strategies in critical illness.

Disclosure of Interest: C. Dupuis Other: Dexcom supplied the CGM devices but had no role in the study design, data collection, analysis, or manuscript preparation., V. Alves de Araújo: None declared, C. Annicchiarico: None declared, G. Macedo: None declared, C. Pinto Teixeira: None declared, V. Uyttendaele: None declared, T. Desaive: None declared, F. Taccone: None declared, C. Miriam: None declared, T. Carton: None declared, M. Seret: None declared, J.-C. Preiser: None declared