The best timing for renal replacement therapy (RRT) in intensive care unit (ICU) patients with acute kidney injury (AKI) is unknown. The investigators will conduct a multicenter prospective randomized open-label trial to compare two strategies in ICU patients (mechanically ventilated and/or receiving catecholamine infusion) with severe AKI defined as RIFLE F classification. These patients will be randomly allocated to one of the following strategies:

1. an "early" strategy where RRT is started immediately when a RIFLE F status is documented
2. a "delayed" strategy where RRT (in patients who also present RIFLE F renal failure) is started only in case of occurrence of one or more of the following events ("Alert Criteria"): oliguria or anuria lasting for more than 72 hours after randomization, serum urea concentration > 40 mmol /L, serum potassium concentration > 6 mmol /L, serum potassium concentration > 5.5 mmol /L that persists despite well-conducted medical treatment with at least sodium bicarbonate and / or glucose-insulin infusion, arterial pH < 7.15 in the context of pure metabolic acidosis (PaCO2 <35 mmHg) or in the context of mixed acidosis with PaCO2> 50 mmHg without possibility of lowering this PaCO2 value, acute overload pulmonary edema generating severe hypoxemia requiring oxygen flow> 5L/min in spontaneously breathing patients or FiO2> 50% in mechanically (invasive or noninvasive) ventilated to maintain SpO2> 95%, despite diuretic therapy.

The primary endpoint is overall survival, measured from the date of randomization to the date of death, regardless of the cause. The minimum duration of each patient's follow-up will be 60 days.

Background:

Acute kidney injury (AKI) is a common complication in intensive care unit (ICU) patients. Renal replacement therapy (RRT) is the major supportive treatment of AKI. Despite progress in RRT management, mortality remains high and the timing of its initiation remains open to debate when no metabolic disorder (severe hyperkalemia or metabolic acidosis) or major fluid overload threaten short-term prognosis. Such abnormalities mandate RRT and are non-inclusion criteria of our study. Whereas many studies have focused on RRT modalities, no prospective randomized study has evaluated the criteria for initiating RRT in ICU in the absence of the above-mentioned life-threatening disorders. In other words, whether duration of oliguria/anuria and/or value of serum urea/creatinine are an adequate indication for RRT is unknown. Given the lack of high quality data, it is not surprising that survey of practices showed wide variation in the timing of RRT initiation and that no precise guidelines could be drawn by expert recommendation as to the optimal start of RRT, making a randomised controlled study of timing of RRT both desirable and ethical.

Objective:

The main objective of this study is to compare two strategies of RRT initiation in terms of overall survival in ICU patients (mechanically ventilated and/or receiving catecholamine infusion) with severe AKI defined as RIFLE F classification. These patients will be randomly allocated to one of the following strategies:

1. an "early" strategy where RRT is started immediately when a RIFLE F status is documented
2. a "delayed" strategy where RRT (in patients who also present RIFLE F renal failure) is started only in case of occurrence of one or more of the following events ("Alert Criteria"): oliguria or anuria lasting for more than 72 hours after randomization, serum urea concentration > 40 mmol /L, serum potassium concentration > 6 mmol /L, serum potassium concentration > 5.5 mmol /L that persists despite well-conducted medical treatment with at least sodium bicarbonate and / or glucose-insulin infusion, arterial pH < 7.15 in the context of pure metabolic acidosis (PaCO2 <35 mmHg) or in the context of mixed acidosis with PaCO2> 50 mmHg without possibility of lowering this PaCO2 value, acute overload pulmonary edema generating severe hypoxemia requiring oxygen flow> 5L/min in spontaneously breathing patients or FiO2> 50% in mechanically (invasive or noninvasive) ventilated to maintain SpO2> 95%, despite diuretic therapy.

Design:

Prospective, multicenter, randomized, open-label trial comparing two RRT initiation strategies in terms of overall survival.

Primary endpoint:

Overall survival, measured from the date of randomization to the date of death, regardless of the cause. The minimum duration of each patient's follow-up will be 60 days.

Secondary endpoints:

Survival rate at day 28, percentage of patients requiring who did not require RRT in the "delayed" strategy, time until cessation of RRT therapy, rate of adverse events potentially related to the AKI or to RRT (e.g; RRT catheter-related complications, hemorrhage due to anticoagulation required for RRT etc…), rate of nosocomial infections, number of ventilator-free days of RRT-free days and of vasopressors free days, length of stay in ICU and hospital, rate of limitations of treatment for futility, total cost of consumables (including RRT catheters and lines among others) related to RRT between day 1 and day 28.

Number of subjects required:

We hypothesized that the "delayed" strategy would prove beneficial to the patients and would translate into increased survival. The study is designed to prove superiority (and not noninferiority) of this strategy over the "early" one.

The 60 days survival rate with the "early" strategy is estimated to be 45%. It is necessary to include 620 patients (310 per arm) to obtain a power of 90% to detect a survival improvement of 14% at day 60 with the "delayed" strategy (log-rank two tailed test, global significance level of 5%), with two blind interim analyses by independent observers at 90 and 180 deaths (group sequential approach of O'Brien-Fleming), and a estimated dropout rate of 10%.

Duration of study:

Inclusion: 18 months Minimum participation of each patient: 60 days Analysis and report: 10 months

Inclusion criteria The following five criteria are required for inclusion

1. Hospitalized in intensive care unit

2. Age ≥ 18 years

3. Acute kidney injury compatible with the diagnosis of acute tubular necrosis defined by a clinical ischemic or toxic insult context

4. Have an AKI classified as RIFLE F, that is to say, with at least one of the following three criteria:

   • creatinine> 354 mmol / l or > 3 times the baseline creatinine
   • anuria for more than 12 hours
   • oliguria defined as urine output < 0.3 ml / kg / h or < 500ml/d for more than 24 hours

5. Mechanical ventilation and/or catecholamines infusion (noradrenaline or/and adrenaline)

Non-inclusion criteria

One or more of the following criteria:

• Chronic renal failure (defined as creatinine clearance < 30 ml / min)

• Patients already enrolled in the study

• Inclusion criteria number 4 present for more than 5 hours

• Acute renal failure due to:

  • urinary tract obstruction
  • renal vessels obstruction
  • tumor lysis syndrome
  • thrombotic microangiopathy
  • acute glomerulonephritis

• Intoxication with a dialyzable product

• Child-Pugh class C liver cirrhosis

• Renal transplant

• Cardiac arrest without awakening at time of potential inclusion

• Moribund state

• Decision to limit treatment

• RRT already started for the current episode of AKI

• Presenting (at the time of potential inclusion) a strong indication for immediate RRT

  • oligoanuria for more than 3 days
  • serum urea concentration > 40 mmol / l serum potassium concentration > 6 mmol /L, serum potassium concentration > 5.5 mmol /L that persists despite well-conducted medical treatment with at least sodium bicarbonate and / or glucose-insulin infusion, arterial pH < 7.15 in the context of pure metabolic acidosis (PaCO2 <35 mmHg) or in the context of mixed acidosis with PaCO2> 50 mmHg without possibility of lowering this PaCO2 value, acute overload pulmonary edema generating severe hypoxemia requiring oxygen flow> 5L/min in spontaneously breathing patients or FiO2> 50% in mechanically (invasive or noninvasive) ventilated to maintain SpO2> 95%, despite diuretic therapy.

• Under cardiopulmonary bypass

• Included in another clinical trial on RRT modalities.