Acute Renal Failure: Finally, Some Progress
- Validation of a consensus definition of ARF
- Identification of biomarkers that promise earlier diagnosis
- Pharmacologic and dialytic interventions that have shown im-proved survival in randomized controlled trials
Getting the definition right
The classic definition of ARF is an “acute and sustained reduction in renal function.” Unfortunately, translating this into a specific research protocol requires defining renal function, acute, and sustained. This ambiguity has blossomed into more than 35 specific definitions of ARF. These definitions span from mild deteriorations in renal function to severe episodes requiring renal replacement therapy (RRT). An example of the former is contrast nephropathy, where an event is defined by a 0.5 mg/dL or 25% increase in serum creatinine. The use of RRT to define ARF often is used in retrospective studies because it is easily gleaned from procedure codes and is unambiguous as well as unanimously perceived as an important end point.
A recent contribution to the definition of ARF is the realization that modest increases in serum creatinine greatly increase mortality. Lassnigg et al found that, in cardiac surgery patients, an increase in creatinine of 0.1-0.5 mg/dL nearly tripled 30-day mortality compared to patients with no increase or a modest decrease in creatinine (J Am Soc Nephrol. 2004;15:1597-1605).
Similar findings were found in a more generalizable cohort of 9,205 patients admitted to Brigham and Women's Hospital in
The New RIFLE criteria
In August 2000, a group of inter-ested intensivists and nephrolo-gists convened the first meeting
of the Acute Dialysis Quality Initiative (ADQI). This expert panel was brought together to develop evidence-based, clinical practice guidelines regarding ARF. In 2004 ADQI published a consensus definition of ARF called the RIFLE criteria:
- Risk: Increase in creatinine of 1.5-2.0 times baseline (loss of more than 25% of glomerular filtration rate [GFR]) or urine output less than 0.5 mL/kg/hr for more than six hours.
- Injury: Increase in creatinine of 2-3 times baseline (loss of 50% of GFR) or urine output loss of more than 0.5 mL/kg/hr for more than 12 hours.
- Failure: Increase in creatinine more than three times baseline (loss of 75% of GFR), an increase in serum creatinine to greater than 4 mg/dL, urine output less than 0.3 mL/kg/hr for more than 24 hours, or anuria for more than 12 hours.
- Loss of function: Persistent renal failure (i.e., need for dialysis) for more than four weeks.
- End-stage renal disease: Persistent renal failure (i.e., need for dialysis) for more than three months.
Recently, a few experiments validating the RIFLE criteria have been published. Uchino et al retrospectively examined 20,000 adult patients admitted to a single center over two years. They found a linear increase in hospital mortality with increasing degrees of renal failure as assessed by the RIFLE criteria: 4.4% for no renal failure, 15.1% for risk of failure, 29.9% for renal injury, and 53.9% for renal failure. These relationships persisted with multivariate logistic regression (Crit Care Med. 2006;34:1913-1917). Hoste et al similarly reported on one year's worth of admissions to seven ICUs at a single center and also found a graded increase in mortality with increasing severity of ARF as assessed by the RIFLE criteria: 5.5% for no ARF, 8.8% for risk of ARF, 11.4% for renal injury, and 26.3% for renal failure (Crit Care. 2006;10:R73). Interestingly, when they looked at markers of severity of illness excluding the renal system there was no difference between the groups, suggesting that the difference in mortality could be accounted solely by the increasing severity of renal failure.
Though the RIFLE criteria depend on changes in serum creatinine as a proxy for changes in GFR, there are a number of limitations with using creatinine to define ARF. Serum creatinine levels are dependent on multiple variables besides renal function. Age, gender, race, catabolic state, and muscle mass all influence serum creatinine, so it may be unclear if an abnormal creatinine is due to altered renal function or an extra-renal factor. Likewise, a normal creatinine could mask significantly altered renal function if there is concurrent alteration of these extra-renal influences of creatinine. Equations such as the MDRD equation and the Cockroft-Gault formula, which try to normalize creatinine for age, gender, race and body size, have never been validated in ARF.
Serum creatinine and other models of GFR are limited in assessing ARF because they are functional markers of injury. One goal of current research is to find a biomarker of renal injury itself. In hepatology, physicians use a mixture of biomarkers (transaminases) and functional markers (bilirubin, albumin, and coagulation time) to assess liver damage. The biomarkers rise rapidly in response to injury while changes in the functional markers lag and are altered only following profound damage. In nephrology, no biomarkers are currently available, so we are limited to functional markers, i.e., various assessments of GFR. Fortunately, the search for a “renal troponin” has resulted in multiple biomarker candidates. The two most promising are KIM-1 and NGAL, both of which now have limited human data.
The cystatin C marker
Like creatinine, cystatin C is a functional marker; however, it avoids much of the inter-patient variability that compromises creatinine. Cystatin C is released by all nucleat-ed cells regardless of muscle mass, gender, race, or age. It is likewise unaffected by inflammation or cirrhosis. It is freely filtered at the glomerulus; however, unlike traditional markers of GFR that are ideally neither reabsorbed nor secreted, cystatin C is completely reabsorbed and metabolized in the proximal tubule. Traditional clearance calculations of urinary excretion per time, divided by the plasma concentration, do not work with cystatin C. Decreased renal function simply increases plasma cystatin C. Cystatin C is better at predicting cardiovascular mortality than either serum creatinine or the four-variable MDRD. Among ICU patients, cystatin C is better at modeling GFR than serum creatinine. Cystatin C was able to diagnose ARF 1.5 days before serum creatinine was increased. Important-ly, the test was highly specific; only 1 in 20 elevations in cystatin C was a false positive (Kidney Int. 2004;60:1115-1122).
Therapy or poison?
A number of recent trials have been conducted to reduce the incidence or severity of ARF. No agent has shown consistent efficacy, though some agents show promise. These include loop diuretics, such as furosemide. Furosemide in ARF has been variously theorized to be both therapeutic and poisonous. Proponents point to its ability to decrease metabolic activity in the ascending loop in order to better align oxygen supply and demand in ischemic ARF (Intensive Care Med. 2005;31:79-85). Opponents point to repeated retrospective analysis showing worse outcomes with loop diuretics in ARF (JAMA. 2002;288:2547-2553).
A randomized controlled double-blind trial sought to find the truth, at least in regard to speeding recovery of renal failure requiring dialysis (Am J Kidney Dis. 2004;44:402-409). Researchers administered high-dose furosemide (25 mg/kg/day in a single daily dose until death or renal recovery) in patients on RRT for ARF. The primary end point was survival. No difference was seen in survival or time on dialysis, though the diuretic group achieved urine output of 2 L/day sooner than the placebo group (5.7 vs 7.8 days).
Using a hypertension agent
Another drug thought to have benefit in ARF patients is fenoldopam, a selective dopamine-1 receptor agonist that causes systemic and renal vasodilatation. It is approved for the IV treatment of hypertension. Because of its ability to cause renal vasodilatation and its lack of either alpha or beta sympathetic activity, it was thought to have promise in the treatment of ischemic ARF.
Despite a failure in the prevention of contrast nephropathy, use of fenoldopam in ICU-associated acute tubular necrosis has showed some promise. In a well-conceived and executed trial, Tumlin et al found a non-significant 11% reduction in death or dialysis at 21 days in patients with ARF (Am J Kidney Dis. 2005;46:26–34). The study had a number of strengths, including the fact that it was a multicenter, double-blind, randomized controlled trial with the criteria for dialysis pre-specified in the protocol. Two predetermined subgroups did have significant benefit: There was a reduction in the primary end point for both diabetic patients and patients who were post-cardiac surgery. The cardiac surgery subgroup data are backed up by some prospectively gathered cohort data, but not by randomized controlled trials.
Morelli et al looked at prophylactic administration of fenoldopam in septic ICU patients prior to developing ARF. ARF is common in sepsis (50% prevalence in septic shock) and carries a mortality rate as high as 70%, so a prophylactic strategy is rational. This study successfully reduced ARF (19% vs. 34%) and ICU days (7 vs. 9 days), and had a nearly-significant 47% reduction in need for dialysis (Crit Care Med.2005;33:2451-2456).
Researchers also have been testing human atrial natriuretic peptide (hANP), which reduces renal injury in lab animals. Single-center trials in humans initially showed benefit, but a pair of large, multicenter trials failed to replicate those results. These trials were criticized for a variety of reasons: late administration of study drug, using a dose of hANP that induced hypotension, using the drug for only a single
24-hour infusion, and not having a protocol in place for starting dialysis, the primary end point.
In 2004, Sward revisited hANP in a randomized, placebo-controlled trial of patients with ARF following coronary artery bypass graft surgery (Crit Care Med. 2004;32:1310-1315). The research- ers treated patients early in renal failure. The study drug was administered until the patient got better, died, or started dialysis. The researchers used a lower dose for a longer period of time than in the earlier trials. As a result, hypotension was not an issue. This protocol showed a significant reduction in the need for dialysis (21% hANP vs. 47% in the placebo group) and a reduction in the composite end point of death or dialysis at 21 days (28% vs. 57%). An interesting note is that all patients received furosemide. This differentiates this protocol from the three large (and negative) hANP trials.
Dopamine still doesn't work
Despite an adequately powered, randomized controlled trial (328 patients) that showed no benefit, dopamine continues to be examined for the treatment and prevention of ARF (Lancet. 2000;356:2139-2143). A recent meta-analysis reconfirms the lack of benefit from dopamine. The investigators found no increase in survival, no decreased need for dialysis, and no increase in renal function (Ann Intern Med. 2005;142:510-524).
Another approach that has been explored is higher-dose dialysis. Multiple studies have suggested that higher delivered doses of dialysis resulted in improved outcomes in ARF. This was initially found in retrospective studies, and two subsequent prospective, randomized controlled trials have supported this dose effect of dialysis in ARF. Ronco's study on dose of continuous venous-venous hemodialysis (CVVH) (Lancet. 2000;356: 26-30) showed that 35 mL/kg/hr of effluent resulted in improved survival compared with 25 mL/kg/hr. Schiffle published a study (
J Med. 2002;346:305-310) showing that daily intermittent dialysis reduced mortality and accelerated renal recovery among survivors. While identified as a trial of frequency, the small molecule clearance was much greater in the daily dialysis group, resulting in ambiguity regarding whether frequency or improved clearance was responsible for the benefit.
A recent randomized controlled trial adds to this data. A single center randomized 206 patients with ARF requiring dialysis to either CVVH with an effluent rate of 25 mL/kg/hr (the standard dose at their institution) or the same CVVH prescription plus an additional 1.0-1.5 L/hr of dialysate (CVVHDF). The CVVH-DF group had better urea and creatinine reduction in the first 24 hours. CVVHDF improved survival at 28 days: 59% in the CVVHDF group vs. 39% in the CVVH group (Kidney Int. 2006;70. Advance online publication July 19).
Patient outcomes in ARF have not improved for the last few decades but recent developments hold great promise. Looking forward, patients will benefit from consensus guidelines to disseminate research and new biomarkers to allow earlier and more accurate diagnosis. In terms of therapeutics, a mixture of new and old agents continue to be evaluated; though promising, no agent has yet consistently proven to be effective. In dialysis the theory of a dose response to increasing clearance continues to gain weight with additional data.
Dr. Topf is a clinical nephrologist at