Acute kidney injury (AKI) is a frequently encountered complication in the intensive care unit (ICU), affecting on average 25 to 35% of patients. It is associated with an increased mortality, proportional to AKI severity. RRT induces important shifts of water and electrolytes. Thus, significant amount of chloride might unintentionally be transferred to patients.

Chloride is the main anion of the organism. It is involved in the regulation of numerous physiological processes. Thus, significant and rapid modification of chloride amount contained in the organism (as might be induced by renal replacement therapy) may be responsible for important, and potentially deleterious, consequences to critically ill patients.

Studies have shown that the administration of high amounts of chloride rich solutions (such as sodium chloride (NaCl) 0,9%) was associated with the development of hyperchloremic acidosis in a dose-dependent manner. This hyperchloremic acidosis could also be theoretically associated with deleterious physiological effects. However, the true clinical consequences of administration of high amounts of chloride rich solutions remains unclear. Their effect on mortality remains a matter of debate, the results of studies being very conflicting in that respect. Nevertheless, hyperchloremia itself and/or the rise of chloremia in the intensive care unit seems to be associated with increased mortality. Moreover, the impact of those chloride rich solutions on the development of acute kidney injury is also a subject of controversy, data from the literature being here again very conflicting.

A recent study already showed that continuous RRT (CRRT) techniques induce a significant transfer of sodium to patients benefiting from those techniques. In that study, the amount of sodium transferred depended mainly on the difference between patient's natremia and sodium concentration in dialysate and/or replacement fluid (usually higher than patient's natremia) used.

By analogy, it is likely that an occult transfer of chloride also happens during RRT, given the high chloride concentration of dialysate fluids (in continuous veno-venous dialysis, CVVD) and replacement fluids (in continuous veno-venous hemofiltration, CVVH), or when these 2 modalities are combined (continuous veno-venous hemodiafiltration, CVVHDF). Finally, the investigators suspect, although it remains undemonstrated so far, that the RRT technique (convective vs. diffusive) may influence this transfer, to an unknown extent. Nevertheless, this transfer and its potential determinants have never been studied yet.

If chloride overload (and its potential clinical consequences) induced by the administration of solutions such as NaCl 0,9% is being extensively studied, no study has ever focused on chloride transfer that may result from the use of renal replacement therapy. However, as mentioned above, it is very likely that such a chloride transfer to patients happens, and that its magnitude depends on different parameters such as RRT modality, RRT fluids characteristics, or patient's chloremia at the start of RRT.

The investigators conduct the present study to describe and compare the intensity of chloride transfer during the first 24 hours of renal replacement therapy by continuous veno-venous hemofiltration (CVVH), continuous veno-venous hemodialysis (CVVD),or continuous veno-venous hemodiafiltration (CVVHDF), and to determine if that transfer is more important with one or the other of those two techniques, in ICU patients affected with severe AKI requiring RRT. Secondary aims are to describe and compare the effects of chloride transfer under 3 RRT modalities (CVVD, CVVH and CVVHDF) on patient's outcome, organ failures, electrolyte and acid-base balance, fluid balance and hemodynamics. Finally, the investigators aim to develop a pharmacokinetic compartment model of chloride transfer during different modalities of RRT.

Inclusion Criteria:

• Adult patients (age > 18 years) affiliated to a social security regimen.
• Presenting with stage III AKI according to Kidney Disease: Improving Global Outcome (KDIGO) classification28.
• Treated with continuous RRT (CVVH or CVVD or CVVHDF) for less than 24 hours, whatever the method used.
• Mechanically ventilated at the time of inclusion in the study.

Exclusion Criteria:

• Patients on chronic hemodialysis / peritoneal dialysis.
• Patients on extracorporeal membrane oxygenation
• Patients on intermittent hemodialysis (acute or chronic)
• RRT initiated for a different reason than stage III AKI according to Kidney Disease: Improving Global Outcome (KDIGO) classification
• Withholding of life sustaining treatment concerning RRT, mechanical ventilation or cardiopulmonary resuscitation.
• Patients whose life expectancy is lower than 24 hours
• Patients under guardianship or another juridical protection
• Patient's or next of kin opposition to participate
• Patients previously included in the study

We will enroll all consecutive adult ICU patients, with stage III AKI (according to Kidney Disease: Improving Global Outcome (KDIGO) classification), and requiring continuous RRT (CVVH or CVVD or CVVHDF). Moreover, patients included will all be mechanically ventilated because of their initial clinical condition. This allows the selection of the most severe typology of ICU patients, in which the evaluation of chloride transfer may be the most relevant.