Faculty – Nephrology Dept. of Nephrology (Renal)
KEM Hospital, Pune
A 65-year-old male with acute decompensated heart failure secondary to a urinary tract infection is admitted to the ICU. Over 24 hours, he develops AKI with rising creatinine and oliguria, alongside signs of fluid overload (APE). The team faces a pivotal decision: initiate or delay RRT, aiming to balance renal recovery with patient stability.
(Click / Tap on Questions to Reveal Content)
Multiple studies have shown an association between AKI, fluid overload and mortality. Earlier studies have utilised body weight and cumulative fluid balance from intake and output charts as measures of fluid overload, with an increase in body weight of 5% being defined as fluid overload. The limitation that occurs is the absence of baseline weight and the state of fluid balance at ICU admission in a large number of patients. However a systematic review (2020) of 31 observational studies in 31,076 patients utilised these definitions and found an adjusted relative risk for mortality with fluid overload was 8.83 (95% CI, 4.03-19.33), and for cumulative fluid balance 2.15 (95% CI, 1.51-3.07), 3 days after ICU admission. At any time point, fluid overload increased the risk for mortality to 2.79 (95% CI, 1.55-5.00) and cumulative fluid balance to 1.39 (95% CI, 1.15-1.69). Additionally greater the degree of fluid overload greater the mortality as shown in multiple studies. The contribution of Fluid overload to mortality in AKI persists after starting and stopping RRT with HR of 2.0 (1.27 – 3.37) and 2.52 (1.55 – 4.08) respectively. Hence maintenance of normal fluid balance and avoiding fluid overload is paramount and optimisation and stabilization of fluid status should be targeted with diuretics and/or RRT used, to reduce fluid overload in patients with AKI. Diuretic responsiveness has been shown in the frusemide stress test to predict the need for RRT with patients who develop a diuresis after a 1mg/kg dose improving, and those who did not produce a urine output of 200 ml at 2 hours more often needing RRT. The frusemide stress test (diuretic responsiveness) therefore becomes a “biomarker” for predicting the development of AKI-3 and need for RRT.
Large studies [Lee(2022) in 1815 patients] proved that worsening acidosis was associated both with worsening AKI and also with higher mortality with the most severe metabolic acidosis also associated with the highest mortality. A multicentre study in 2021 in 414 patients showed that blood pH was a strong predictor of mortality in AKI patients receiving RRT, while multiple small studies have shown a deleterious effect of metabolic acidosis on cardiac output and MAP, and an improvement in MAP and vasopressor need (norepinephrine dose ) with acidosis correction by RRT. Finally the BICAR-ICU study showed that in patients with Acute Kidney Injury (AKIN) score of 2-3 (n=182), there was a significant difference in both day 28 mortality and one or more organ failure at day 7, in patients supplemented with bicarbonate to correct pH.
Thus optimizing fluid balance to avoid overload, assessing diuretic responsiveness in oliguric patients adequately resuscitated and correcting acidosis should be priorities in patients with AKI.
The clinical setting or background in which olguria develops may provide a clue to it being a manifestation of compensation for prerenal factors rather than of irreversible tubular damage. Sepsis, dehydration, drug combinations like ACEI/ARB and NSAIDs, haemorrhage low cardiac ouput states, nephrotic syndrome or liver cirrhosis suggest a depleted effective arterial blood volume especially if a chart review reveals a temporal relationship between oliguria or raised creatinine and a potential reversible insult. A normal urinalysis and a low fractional excretion of sodium have been time honoured tests of distinguishing pre renal oliguria from intrinsic parenchymal damage, but multiple confounders like vasoactive medications and diuretics can diminish their value. However significant proteinuria (urinary protein/creatinine ratio > 0.5 mg/mg), hematuria especially glomerular, casts or leukocytes on microscopic examination, are almost always indicators of intrinsic renal parenchymal injury and should prompt more extensive investigation. ATN is characterized by muddy brown casts. Urine microscopy has been shown to have a high positive predictive value and low negative predictive value in patients with a high pretest probability of ATN and a high negative predictive value in patients with a low pretest probability.
Apart from Cardiorenal and Hepatorenal syndromes most other pre renal conditions are characterized by a diuresis in response to volume loading. A BUL/creatinine ratio > 20 :1 is common in prerenal azotemia and diminishes over 24 hours in response to volume loading while in ATN the value is < 15:1. However disproportionate rise in BUL may be seen in patients with gastrointestinal bleeding, steroid or diuretic use and dyspropionately low value in patients with liver disease, where urea synthesis is decreased.
Although multiple studies have shown that biomarkers of renal injury increase before the functional markers of creatinine and urine output, their clinical utility in deciding about renal replacement therapy has not been established. Only the ELAIN study used a NGAL value > 150 as a criteria to initiate RRT and interestingly this was the only study that has ever shown a benefit of early initiation.
All high quality studies (AKIKI, IDEAL-ICU and STARRT AKI) which evaluated the optimum timing of initiating AKI, established that conventional indications for initiating RRT (often considered as consequences or complications of AKI) were appropriate in ICU patients, and that initiating RRT in patients without the above indications, failed to improve mortality at 60 or 90 days, and was associated with an increased incidence of access related infections, adverse events, delayed renal recovery and longer dialysis dependence. Given the strong association of fluid overload with mortality a reasonable approach might be to perform a frusemide stress test in patients who have fluid overload > 5% of starting body weight accompanied by oliguria and consider initiating RRT in patients unresponsive to frusemide to decrease fluid overload. A post hoc analysis of the PICARD study showed that mortality progressively increased with persisting fluid overload and the degree of fluid overload that persisted after initiating RRT. Patients with crush injuries and rhabdomyolyis are similarly at high risk of developing life threatening hyperkalemia and would probably benefit from RRT early in oliguria.
Conversely in patients considered at risk for hypotension during RRT without life threatening indications, RRT may be withheld with careful monitoring as 2 studies (Silversides 2014) and (Teheranian 2019) have shown that hypotension significantly increased the risk of dying. The AKIKI 2 study also established that duration of oliguria before RRT was not associated with a poorer outcome and that a BUL > 300 mg/dl was associated with a higher 60 day mortality.
Significant organ edema, venous congestion and decreased tissue perfusion characterize cardiorenal syndrome cases and mortality increases in such situations. Additionally myocardial function may be worsened by metabolic acidosis, hyperkalemia, and hypocalcemia all of which may be improved by renal replacement therapy. Hence an early start of RRT as a renal and more importantly cardiac support may benefit these patients. Once again the ELAIN trial which showed a significant mortality benefit had a large proportion of cardiac surgery patients. However in patients with severe myocardial ischemia, hypotension due to RRT may cause additional myocardial stunning or injury, as could arrhythmias triggered by rapid electrolyte shifts and caution must be exercised while dialyzing such patients. IDEAL-ICU which included a significant number of septic patients and the subset of septic patients in STAART AKI did not indicate any benefit from an early start to RRT and septic patients may need to “earn” their RTT
In the BICAR – ICU study the infusion of isotonic bicarbonate to patients with AKI 2 & 3, and severe acidosis (pH < 7.20) reduced the composite end point of death or new organ failure delayed and significantly averted the need for RRT, though in the cohort as a whole , there was no significant difference. Most of the patients studied were on pressor agents and mechanical ventilation and so at risk for adverse events and hemodynamic instability during RRT. Hence manouveres like bicarbonate infusion may delay the need for RRT and some patients may avoid it altogether. Similarly in patients responding to the frusemide stress test with a brisk diuresis RRT initiation may be delayed and in a significant number of cases avoided altogether.
Overenthusiastic fluid resuscitation in a patient, no longer fluid responsive increases venous congestion. Not only does this venous overfilling fail to improve organ perfusion, it may actually worsen it as perfusion pressure is the difference between mean arterial pressure and right atrial pressure, a marker of venous filling. However right atrial pressure measurement in patients on mechanical ventilation or with co-morbidities like COPD may not be measured accurately and may need to be derived from point of care ultrasonographic measurements like IVC diameter and VEXUS scores. Recently increasing evidence has been obtained to show that excessive fluid administration is linked to increased incidence of acute kidney injury, increased morbidity and mortality.
The Venous Excess Ultrasound (VExUS) score, allows quantification of systemic venous congestion, particularly in patients with fluid overload who are at a greater risk of organ damage. VExUS grade 3, has been shown to be associated with an increased risk of acute kidney injury (AKI) in cardiac surgery. Both VExUS grade 2 (moderate venous congestion) and grade 3 (severe venous congestion) are also independently associated with mortality (grade 2: adjusted hazard ratio: 4.03, 95%confidence interval [CI]: 1.81–8.99; grade 3: adjusted hazard ratio: 2.70, 95% CI: 1.10–6.65; p = 0.03). VExUS grade 3 had a significantly better positive predictive value than a RAP value ≥12 mm Hg for congestive kidney injury.
As a semi quantitative tool VEXUS also assesses severity of congestion and real-time monitoring of the response to decongestive therapy, influencing recovery.
VExUS-guided decongestion, compared to usual care, significantly increased the likelihood of achieving decongestion more than 2-fold faster by 2 days (OR: 2.5, 1.3–3.4, p = 0.01). Severe congestion (VExUS grade 3) prevalence was 48% and 40% in the intervention and control groups at admission.
Thus use of bedside USG and Vexus allows quantitation of venous fluid overload predicts the response of organ injury like AKI to fluids and allows monitoring of fluid removal and recovery.
The two largest trials on dose/intensity of RRT, ATN and RENAL failed to show an effect on renal recovery. Similarly one systematic review which included both these studies reached the same conclusion. Hypotension has been shown to delay renal recovery and at least one analysis of ultrafiltration rates showed that ultrafiltration rates > 1.75 ml/kg/hour were associated with lower renal recovery than rates of 1.0 to 1.75 ml/kg/hour in patients receiving CRRT. It is conceivable that higher net UF rates may produce more hypotension and delay renal recovery. Similarly a systematic review from 2021 found that PD, the least intense modality, was associated with better renal recovery than CRRT, PIRRT or IHD and that therefore a “gentler less intense” form of RRT may promote renal recovery, an approach that would need to be adopted with careful monitoring to avoid complications of underdialysis. The STAART AKI study also demonstrated that patients who received early initiation of RRT were more likely to be dialysis dependant at 90 days. Taken in conjunction with the findings that hypotension prolongs renal recovery it appears that starting RRT in patients with more unstable hemodynamic conditions and more aggressive ultrafiltration may be counterproductive and allowing more time of dialysis may allow spontaneous return of function, however many of the conclusions are based on sub analysis rather than being the primary outcome of any study.
The K-DIGO recommendation to “discontinue RRT when it is no longer required, either because intrinsic kidney function has recovered to the point that it is adequate to meet patient needs, or because RRT is no longer consistent with the goals of care.” is imprecise and unhelpful. Importantly it is not graded as at the time of recommending there was insufficient evidence to base a guideline on and wide variation in criteria for weaning existed among even RCTs. As a result the sensitivity and specificity of various weaning criteria remained unassessed. Given that RRT is associated with a significant risk of complications including an increased ICU and hospital stay, higher costs and lower chance of recoverability it is desirable to minimize the duration of RRT.
Various studies including AKIKI, ELAIN, AKI Network and BEST Kidney used criteria like spontaneous or diuretic augmented urine output and timed urine collection, measured creatinine clearance to discontinue RRT in AKI patients and found that an unassisted urine output of 400 to 1000 ml per 24 hours or a diuretic assisted output of > 2000 ml per 24 hours predicted successful weanng from RRT. Thus although KDIGO suggested not using diuretics to enhance kidney function recovery, or to reduce the duration or frequency
of RRT (2B), in a very recent study (2025), a frusemide stress test which produced a urine output to 188 ml at 2 hours was found to have a high sensitivity(0.80), specificity (0.92) and AUC of 0.913 which outperformed a 24 urinary collection (0.739). Spontaneous urine output between 400 and 500 ml per 24 hours also was found to outperform a serum creatinine in predicting renal recovery, while a measured GFR of 15 ml/minute from a timed urine collection had a positive predictive value of 84% and a negative predictive value of 88% for discontinuing RRT. Studies using timed urine collections and estimation of GFR studied GFRs between 12 and 20 ml/minute as indication of renal recovery. The figure of 20 ml/minute was found to be too high and a value of 12 to 15 ml/min may be a reasonable criteria for attempting to withhold RRT. Although studies utilised durations of 2 to 24 hours for timed collection an 8 hour collection seemed to be a reasonable balance between accuracy and patient convenience.
