Bioimpedance and Multiparametric Ultrasound-Guided Fluid Management After Kidney Transplantation

RATIONALE Fluid management after kidney transplantation is a two-sided problem. The newly implanted graft is an encapsulated, denervated organ with interrupted lymphatic drainage and endothelial injury from ischaemia-reperfusion. Venous hypertension and interstitial expansion raise intracapsular pressure, reduce the net filtration gradient and peritubular capillary flow, and may prolong recovery from acute tubular necrosis. At the same time, the post-ischaemic graft has lost autoregulation, so its perfusion becomes pressure-passive and intravascular depletion is equally harmful. A useful monitoring strategy must therefore detect both overload and depletion, and must distinguish between the compartments involved.

Bioimpedance quantifies extracellular overhydration, the venous excess ultrasound score (VExUS) grades systemic venous congestion, and lung ultrasound detects extravascular lung water. In this trial, the intrarenal component of VExUS is obtained from the graft itself, and graft venous pulsatility is added as a confirmatory marker of congestion.

PILOT DATA

A prospective single-centre cohort of 80 kidney transplant recipients (Hospital Universitario Ramón y Cajal, 2023-2024) showed that:

• On day 1, bioimpedance identified overhydration in 80% of patients, whereas physical examination and VExUS classified most of them as non-congested. Extracellular overload frequently coexisted with preserved central venous haemodynamics, so the methods are complementary rather than interchangeable.
• Graft venous pulsatility of 30% or more was highly specific for VExUS-defined congestion (specificity 88%) but insensitive (sensitivity 42%).
• Ultrasound information prompted a change in management in 64% of patients on day 1, mostly a reduction in intravenous fluids, without a predefined algorithm.
• No method discriminated delayed graft function. Among deceased-donor recipients, dialysis-defined delayed graft function occurred in 17% and functional delayed graft function in 44%; half of early dialysis sessions were indicated for hyperkalaemia and did not reflect the severity of graft injury.

The limitations of the pilot (single centre, a single operator for most examinations, no treatment algorithm, post hoc exclusion of patients with severe outcomes and no concurrent control group) determine the requirements of the present design.

DESIGN RATIONALE Because the intervention is a strategy rather than a test, the trial randomises only the disclosure of assessment results and their translation into a treatment algorithm. Keeping the assessment schedule identical in both arms equalises surveillance intensity, removes the surveillance bias of the pilot study, allows a paired diagnostic comparison with physical examination in every participant and makes contamination of the control arm measurable. Individual randomisation was preferred to cluster randomisation because the small number of centres would make a cluster design inefficient and unstable; contamination is accepted, mitigated (recommendations are delivered only by the ultrasound team) and accounted for in the sample size. The treatment algorithm will be finalised by Delphi consensus among the participating centres and frozen before the first participant is enrolled.

The hierarchical primary endpoint integrates survival, early need for dialysis and graft function, and handles death and graft loss as outcomes rather than as missing data, avoiding the survivorship bias of the pilot study. A single dialysis session for isolated hyperkalaemia in the first 24 hours is excluded from the dialysis level because, in the pilot data, it reflected potassium management rather than graft injury.

MEASUREMENT QUALITY Operators are certified against a central imaging core laboratory before enrolment. Two blinded central readers from different centres read all VExUS and graft venous Doppler examinations and a random 20% of lung ultrasound examinations; the local reading guides management and the central reading is used for analysis. A reliability substudy separates acquisition and interpretation variability, with weighted kappa, Gwet AC1/AC2, intraclass correlation, Bland-Altman analysis and minimal detectable change. Operator drift is monitored with CUSUM charts, with recertification every 6 months.

SAMPLE SIZE The sample size was estimated by Monte Carlo simulation (5,000 iterations) bootstrapping individual data from the 78 deceased-donor recipients of the pilot cohort. Assumptions: no effect on death or graft loss; relative risk of 0.75 for sustained dialysis; an improvement of 5 mL/min/1.73 m2 in eGFR at 1 month; 10% contamination of the control arm; and 5% loss to follow-up. The expected win ratio is 1.33, giving 80% power with a two-sided alpha of 0.05. A blinded sample size re-estimation will be performed at 50% of recruitment, with a predefined cap.

STATISTICAL ANALYSIS The primary analysis follows the intention-to-treat principle using a win ratio stratified by centre and donor type, with win odds, net benefit and the contribution of each hierarchical level. Prespecified sensitivity analyses vary the eGFR tie margin and the dialysis definition, use the per-protocol population, exclude the largest centre and use local instead of central readings. Key secondary outcomes are tested hierarchically to control type I error. Missing 1-month eGFR in participants alive with a functioning graft is handled by multiple imputation.

OVERSIGHT AND ETHICS A blinded adjudication committee reviews every dialysis indication, acute pulmonary oedema, graft loss and the diagnostic reference standard. An independent Data Monitoring Committee reviews safety data every 6 months and performs a non-binding futility analysis at 50% of the information. Concealment of results in the control arm is justified by clinical equipoise: there is no evidence that acting on these measurements improves transplant outcomes, and in haemodialysis a lung ultrasound-guided strategy changed management without improving clinical outcomes. Consent is sought while candidates are on the waiting list, so that the decision is not made under the time pressure of a deceased-donor offer.