Diagnostic performance of a 250-ml net ultrafiltration challenge to identify risk of preload-dependence in critically ill patients undergoing continuous renal replacement therapy: a randomized, cross-over trial.

Introduction: During continuous renal replacement therapy (CRRT), preload-independent patients risk of becoming preload-dependent in case of excessive net ultrafiltration (UF_NET). We aimed to evaluate the ability of a UF_NET challenge to identify de novo preload-dependence in preload-independent patients undergoing CRRT.

Materials and methods: We conducted a single-center, randomized, cross-over trial, enrolling adult patients with CRRT, calibrated continuous cardiac index (CCI) monitoring, and preload-independent at time of enrolment. The diagnostic test consisted of 250-ml UF_NET removal over 15 (fast challenge) or 30 min (slow challenge), preceded and followed by a postural maneuver (PM) evaluating preload-dependence using CCI relative variations. Patients underwent both types of challenges, starting with either fast or slow challenges as determined by randomization, separated by a wash-out period of 24 h. We evaluated the performance of UF_NET challenges to diagnose de novo preload-dependence using the area under the receiver operating curve (AUROC) of the relative change in calibrated cardiac index between before and after the challenge (∆CI_UFC), based on the result of the PM performed after the challenge (responder if positive, non-responder if negative). NCT05214729.

Results: We included 20 patients, comprising 36 UF_NET challenges (19 fast and 17 slow challenges). In intention-to-treat (ITT), the rate of preload-dependence after the challenge was 33% (12/36, 95% confidence interval: 19% to 51%). In ITT, the AUROC of ∆CI_UFC to identify de novo preload-dependence was 0.74 (95% confidence interval: 0.58-0.88), with the respective AUROCs of fast and slow challenges not reaching statistical significance. After exclusion of 5 challenges a posteriori identified as being preload-dependent before challenge start (modified intention-to-treat [mITT], N = 31), the AUROC of ∆CI_UFC was 0.83 (0.66-0.99), with ∆CI_UFC not significantly differing between fast and slow challenges. In mITT, CCI variation during the PM preceding the challenge predicted de novo preload-dependence with an AUROC of 0.82 (0.65-0.98), at an optimal threshold of + 5%.

Conclusions: A 250-ml UF_NET challenge had acceptable diagnostic performance to identify preload-independent patients becoming preload-dependent during CRRT, with no detectable difference between fast and slow challenges. A CCI variation ≥ 5% during a PM in preload-independent patients may help identify those at risk of becoming preload-dependent.