Intermittent hemodialysis (IHD) and continuous RRT (CRRT) provided as continuous hemofiltration or hemodiafiltration are the main RRT modalities in ICU. Randomized controlled trials (RCTs) comparing IHD and CRRT for AKI have not shown an indisputable benefit of one technique over the other. However, these studies were conducted more than 15 years ago. In addition, several recent RCTs on RRT initiation strategies have completely modified both knowledge and practice of RRT initiation.

The main objective is to evaluate whether IHD is not inferior to CRRT with regard to overall incidence of a composite outcome of death, persistent renal dysfunction and dialysis dependency at day 90 in critically ill patients with severe AKI (Major Kidney Event 90, MAKE 90). The primary endpoint will be the proportion of patients who will meet one or more criteria for a major adverse kidney event 90 days after randomization (MAKE90). The MAKE will be the composite of death, renal replacement therapy dependence and/or an increase in serum creatinine above 25% of its basal value.

This is a non-inferiority multicenter open-label randomized controlled trial with two parallel groups. Randomization will take place 1:1 to 2 groups: a group receiving IHD and a group receiving CRRT. Randomization will be stratified according to center, dose of vasopressor and cumulative fluid balance from ICU admission. Treatment will be initiated and monitored by the physician responsible for patient. Whatever the group, investigators will follow recommendations to achieve optimal metabolic control and hemodynamic stability. The investigators plan to include 1000 patients.

Intermittent hemodialysis (IHD) and continuous RRT (CRRT) provided as continuous hemofiltration or hemodiafiltration are the main RRT modalities in ICU. Randomized controlled trials (RCTs) comparing IHD and CRRT for AKI have not shown an indisputable benefit of one technique over the other. However, these studies were conducted more than 15 years ago. Major progresses were eventually made leading to much lower mortality rates in recent studies of RRT for AKI. In addition, an important pitfall of previous studies on RRT modalities is that the indications for initiating RRT were disputable in the light of present knowledge. Indeed, these studies included a noticeable proportion of patients who could have recovered without ever receiving RRT. Three recent RCTs on the timing of RRT for AKI have shown that, in the absence of immediate life-threatening complication, RRT initiation can be safely delayed, allowing up to 49% of severe AKI patients to escape RRT. Many such patients who did not actually require RRT were included in previous studies on modalities. This justifies a reevaluation of this issue by the performance of a new RCT that takes recent knowledge on RRT initiation into account.

The main objective is to evaluate whether IHD is not inferior to CRRT with regard to overall incidence of a composite outcome of death, persistent renal dysfunction and dialysis dependency at day 90 in critically ill patients with severe AKI. The primary endpoint will be the proportion of patients who will meet one or more criteria for a major adverse kidney event 90 days after randomization (MAKE90). The MAKE will be the composite of death, renal replacement therapy dependence and/or an increase in serum creatinine above 25% of its basal value.

The secondary objectives will address the effect of IHD versus CRRT on:

1. Survival (at different time points: D28, D60, D90)
2. Persistence of renal dysfunction (at different time points: D28, D60, D90)
3. Dialysis dependency (at different time points: D28, D60, D90)
4. ICU and hospital length of stay
5. Duration of RRT
6. Number of organ failure-free days (catecholamine-free days, ventilator-free days, RRT-free days) at day 28
7. Estimated renal function at hospital discharge
8. Occurrence of adverse events (hypotension, bleeding, thrombocytopenia, hypoglycaemia, hypophosphataemia, hypothermia, cardiac rhythm disorders, air embolism, bloodstream infection)
9. Medico-economic aspects

Another secondary objective will be to assess the ability of plasma and urinary AKI biomarkers collected at D0 an ICU discharge to predict renal function recovery.

The secondary endpoints will be:

1. Time to death; day-28; day-60 and day-90 mortality

2. Number of patients with more than a 25% increase in serum creatinine from baseline at different time points: D28, D60, D90

3. Number of patients with dialysis dependency at different time points: D28, D60, D90

4. ICU and hospital length of stay

5. Time until cessation of RRT

6. Catecholamine-free days, ventilator-free days and RRT free days through day 28.

7. Estimated glomerular filtration (eGFR) rate at hospital discharge

8. The number of episodes of adverse events from inclusion until day 28 (or ICU discharge):

   • Hypotension: systolic arterial pressure < 80 mmHg or a fall > 50 mmHg from baseline value
   • Hemorrhage requiring red blood cell transfusion or surgical procedure
   • Thrombocytopenia < 100 000 platelets/mm3
   • Hypoglycaemia: blood glucose concentrations<3.0 mmol/L
   • Hypophosphataemia: serum phosphate concentration<0.6 mmol/l
   • Hypothermia: core body temperature <34°C
   • Cardiac rhythm disorders: ventricular tachycardia or ventricular fibrillation or torsade de pointes or new episode of atrial fibrillation requiring medical treatment or external electric counter shock
   • Air embolism
   • Catheter-related (or unknown origin) bloodstream infection

9. Incremental cost-effectiveness ratio in € per MAKE-90 averted

Another secondary endpoint will be the association between the concentration of several plasma and urinary AKI biomarkers (KIM-1, CCL-14, PENKID, NGAL) collected at D0 and at ICU discharge and the MAKE-90.

This is a non-inferiority multicenter open-label randomized controlled trial with two parallel groups. Randomization will take place 1:1 to 2 groups: a group receiving IHD and a group receiving CRRT. Randomization will be stratified according to center, dose of vasopressor and cumulative fluid balance from ICU admission. Treatment will be initiated and monitored by the physician responsible for patient. Whatever the group, investigators will follow recommendations to achieve optimal metabolic control and hemodynamic stability.

IHD group: a central venous access, a biocompatible membrane and bicarbonate dialysate will be used. In accordance with the management recommendations, the investigators will plan at least 3 sessions of 4 to 6 hours per week each with blood flow > 200ml/min, dialysate flow>500ml/min, high sodium concentration (>145 mmol/L) and low temperature (35°C) in the dialysate. The investigators will recommend urea reduction ratio > 65% for each session.

CRRT group: a central venous access and a biocompatible membrane will be used. The investigators will plan continuous treatment with a change of membrane every 72 hours (unless clotting occurs before). Choice between continuous veno-venous hemodialysis (CVVHD), continuous veno-venous hemofiltration (CVVHF), or continuous veno-venous hemodiafiltration (CVVHDF) will be left at physician discretion. The investigators will recommend a minimum delivered dose of dialysis of 20-25 ml/Kg/h of effluent by filtration and/or diffusion.

For both groups, anticoagulation will be implemented according to center practice (regional citrate will be strongly recommended for CVVHDF). The randomly assigned treatment modality will be unchanged for at least 4 consecutive days unless renal recovery occurs before and mandates RRT discontinuation. A shift for the other modality will be allowed after this delay according to clinician decision.

Discontinuation of RRT will be contemplated when spontaneous diuresis is >500 ml/24h, and highly recommended if diuresis >1000 ml/24h without diuretic or >2000 ml/24h in patients receiving diuretics, as in our previous studies (AKIKI and AKIKI2).

Vital status, need for dialysis and serum creatinine will be collected at day 28, day 60 and day 90. Patients or their surrogates or the general practitioner (GP) will be contacted by phone. Prescriptions for plasma creatinine measurements (at D60 and D90) will be provided to the patient at discharge from the hospital.

Inclusion criteria

1. Adults (> or= 18 years old) in ICU
2. Receiving (or who have received) invasive mechanical ventilation and/or catecholamine infusion
3. Availability of both equipment IHD and CRRT (in the investigational center at the time of inclusion)
4. One of the 2 following situations 4.a: Either at least one of the 3 following complications of AKI (whatever the KDIGO stage): persistent severe hyperkalaemia despite medical treatment, persistent severe metabolic acidosis despite medical treatment or severe pulmonary edema due to fluid overload despite diuretic therapy 4.b: Or an AKI stage 3 of KDIGO with one of the 2 following criteria: serum urea concentration>40mmol/L or persistence of oligo-anuria>3 days
5. Affiliation of social security system
6. Written consent obtained from the patients (from a support person, family member or a close relative if the patient is not able to expressing and sign consent) or inclusion without initial consent in case of emergency, if the patient is not able to express his/her consent and in the absence of support person, family member or a close relative

Non-inclusion criteria

• Moribund state (patient likely to die within 24h)
• Previous inclusion in the study
• Subject deprived of freedom, or under a legal protective measure (example: patients under guardianship or curatorship)
• Subject receiving state medical aid
• Pregnancy or breastfeeding woman
• Patient included in another research trial on AKI
• Advanced chronic kidney disease (CKD) defined by an estimated GFR<20 mL/min/1.73 m2
• Presence of a drug overdose or of a dialyzable toxin that necessitates RRT (because IHD may be preferable in these conditions).• Presence or clinical suspicion of renal obstruction, rapidly progressive glomerulonephritis, vasculitis, thrombotic microangiopathy or acute interstitial nephritis
• Brain injured patients or other causes of increased intracranial pressure
• Fulminant hepatic failure

The investigators plan to include 1000 participants. Adjusted difference between the proportions of the two arms, of patients who will meet one or more criteria for a major adverse kidney event at day 90 (MAKE90), will be estimated with 95% two-sided confidence interval (95% CI) by the Cochran-Mantel-Haenszel method, stratified by fluid overload and vasopressor dose. If the upper limit of the 95% CI is < 7.5%,the investigators will conclude the IHD is not inferior to CRRT. If the 95% confidence interval does not only lie entirely below 7.5% but also adjusted difference is less than zero, a superiority test will be carried out using the same method. A subgroup analysis will be done according to fluid overload (≥ 6 liters from admission ICU to randomization) and to vasopressor dose (> 0.5 mcg/kg/min).

Medico-economic analysis: The cost estimation will be expressed with mean and 95% confidence intervals obtained by the bootstrapping method. The incremental cost-effectiveness ratio is defined as the ratio of the difference between mean costs and the difference on the primary endpoint. It will also be estimated by a mean and the 95% confidence interval using bootstrap method.

Inclusion Criteria:

1. Adults (> or= 18 years old) in ICU

2. Receiving (or who have received) invasive mechanical ventilation and/or catecholamine infusion

3. Availability of both equipment IHD and CRRT (in the investigational center at the time of inclusion)

4. One of the 2 following situations 4.a: Either at least one of the 3 following complications of AKI* (whatever the KDIGO stage): persistent severe hyperkalaemia despite medical treatment, persistent severe metabolic acidosis despite medical treatment or severe pulmonary edema due to fluid overload despite diuretic therapy 4.b: Or an AKI stage 3 of KDIGO with one of the 2 following criteria: serum urea concentration>40mmol/L or persistence of oligo-anuria>3 days

   *Definitions of these complication are provided in the main text (Section 7.1)

5. Affiliation of social security system

6. Written consent obtained from the patients (from a support person, family member or a close relative if the patient is not able to expressing and sign consent) or inclusion without initial consent in case of emergency, if the patient is not able to express his/her consent and in the absence of support person, family member or a close relative

Exclusion Criteria:

• Moribund state (patient likely to die within 24h)
• Previous inclusion in the study
• Subject deprived of freedom, or under a legal protective measure (example: patients under guardianship or curatorship)
• Subject receiving state medical aid
• Pregnancy or breastfeeding woman
• Patient included in another research trial on AKI
• Advanced chronic kidney disease (CKD) defined by an estimated GFR<20 mL/min/1.73 m2
• Presence of a drug overdose or of a dialyzable toxin that necessitates RRT (because IHD may be preferable in these conditions).
• Presence or clinical suspicion of renal obstruction, rapidly progressive glomerulonephritis, vasculitis, thrombotic microangiopathy or acute interstitial nephritis
• Brain injured patients or other causes of increased intracranial pressure
• Fulminant hepatic failure