Chapter 75. Acute Renal Failure

• Prerenal azotemia and acute tubular necrosis account for the overwhelming majority of hospital-acquired acute renal failure cases, whereas acute glomerulonephritis and vasculitides are more common causes of acute renal failure developing outside the hospital.
• Acute renal failure occurs in at least 10% to 30% of patients admitted to an ICU and is associated with a mortality rate of about 50% despite advances in supportive care and technology.
• The most important diagnostic classification to be made in the evaluation of patients with acute renal failure is based on the site of the renal lesion (pre-, intra-, or postrenal).
• Since there are few specific therapies available in patients with established acute tubular necrosis, the major clinical focus is on prevention of ARF by identification of subjects at highest risk.
• All aspects of treatment of acute tubular necrosis, including renal replacement therapy, are basically supportive. The nondialytic measures of greatest importance are maintenance of nutritional, volume, and electrolyte homeostasis.
• Emergent renal replacement therapy is indicated in the management of acute renal failure when pulmonary edema, hyperkalemia, refractory acidosis, or symptomatic uremia develops.
• Prophylactic renal replacement therapy should be considered in patients with sustained anuria, persistent oliguria with progressive azotemia and glomerular filtration rate <10 mL/ min, and to prevent uncontrolled positive fluid balance.

Acute renal failure (ARF) is defined as rapid decline (over hours to days) in glomerular filtration rate (GFR). This manifests as a rapid increase in blood urea nitrogen (BUN; azotemia) and serum creatinine, and may or may not be accompanied by a decline in urine output; specific definitions (e.g., absolute or fractional increase of serum creatinine) vary widely between studies.1 ARF occurs in 2% to 5% of general medical-surgical admissions, but up to 10% to 30% of ICU admissions. Acute renal failure can be classified into three broad etiologic/anatomic categories: prerenal ARF, intrarenal ARF, and postrenal ARF. Prerenal ARF is caused by renal hypoperfusion with intact renal parenchyma (55% to 60% of instances of ARF). Intrarenal ARF is caused by parenchymal renal diseases (35% to 40% of instances of ARF). Postrenal ARF is caused by urinary tract obstruction (∼5% of instances of ARF).2

Most ARF in the ICU is caused by prerenal azotemia, and the rest predominantly by renal parenchymal injury with acute tubular necrosis (ATN); the former may convert to the latter. Despite advances in critical care and dialysis technologies, the mortality of ARF requiring dialysis in critically ill patients remains 50% to 80%.3–5 Although there is evidence that isolated ARF itself increases mortality, it is clear that mortality of ICU ARF increases with every additional nonrenal organ failure present at the time of initiation of renal replacement therapy (RRT).3,5 Emerging data suggest that renal ischemia-reperfusion injury and the uremic milieu actually contribute to the development of distant organ injury (increased pulmonary vascular permeability and cardiac and splanchnic organ apoptosis).6–8 Outcome is particularly poor in septic ARF; in one study, mortality of septic ARF was 74%, compared to 45% in nonseptic cases.9 ARF in the presence of shock (septic or cardiogenic) is an increasingly common occurrence in modern ICUs, and has driven the development of continuous renal replacement therapy (CRRT) to permit control of azotemia and fluid balance in hemodynamically unstable patients.5,10 Although a broad differential diagnosis and therapeutic plan should be considered for every ICU patient with ARF, in most cases the approach is to reverse any element of prerenal azotemia, provide supportive care in the presence of ATN, and intervene with effective RRT when indicated, in anticipation of probable renal recovery in the event of patient survival.

Prerenal Acute Renal Failure

Prerenal azotemia is the most common cause of ARF in hospitalized patients. The main feature of prerenal ARF is the presence of intact renal parenchymal tissue and the rapid correction of GFR with restoration of renal perfusion. Uncorrected prerenal azotemia predisposes to the development of ATN. Prerenal ARF is caused by any condition leading to renal hypoperfusion, including systemic hypoperfusion with hypovolemia, cardiac failure, or vasodilatory shock, and/or regional hypoperfusion caused by renal vasoconstriction (Table 75-1; see Chap. 21).

Renal blood flow and GFR are relatively maintained during mild hypoperfusion, due to compensatory mechanisms.11 Renal perfusion is largely preserved within a range of mean arterial pressure (MAP) between 80 to 180 mm Hg, if cardiac output is adequate. As MAP falls below 80 mm Hg there is a precipitous fall in renal blood flow and GFR.12 There are two major mechanisms of renal blood flow autoregulation: a myogenic reflex and tubuloglomerular (TG) feedback. The myogenic reflex is mediated by stretch receptors in the afferent arterioles, which detect a decrease in perfusion pressure, leading to autoregulatory relaxation of afferent arterioles and vasodilation. TG feedback defends renal perfusion as follows: chloride concentration is continuously sensed in the tubular lumen by the macula densa, just distal to the thick ascending loop of Henle. When luminal chloride decreases (presumably reflecting decreased renal blood flow, GFR, intravascular volume, or a combination of these), a vasodilatory signal is transduced to the corresponding afferent arteriole (and vice versa if flow increases). Together these mechanisms autoregulate renal blood flow in the face of hypotension or hypertension. A third mechanism additionally helps autoregulate GFR. Increased renin secretion stimulated by hypotension/hypovolemia sensed in the afferent arteriole helps maintain GFR (but not renal blood flow) during hypotension, through the efferent arteriolar action of angiotensin II (discussed further below).

Of course, in addition to local autoregulation, systemic neurohumoral influences also play a prominent role in determining the renal response to shock. Systemic hypoperfusion activates the sympathetic nervous system and renin-angiotensin-aldosterone axis. Norepinephrine and angiotensin II are systemic vasoconstrictors, and tend to increase renal blood flow by preserving renal perfusion pressure. On the other hand, both hormones are renal vasoconstrictors, though they differ in their glomerular hemodynamic effects. Angiotensin II preferentially constricts efferent arterioles, and helps preserve glomerular filtration, increasing filtration fraction (the ratio of GFR to renal plasma flow) by creating “back-pressure” to augment net filtration pressure in the glomerular capillary. Norepinephrine causes balanced afferent and efferent arteriolar constriction, similarly increasing filtration fraction in the face of decreased renal blood flow, but to a lesser extent than angiotensin II. Both angiotensin II and norepinephrine stimulate intrarenal vasodilator prostaglandin production, thus attenuating their simultaneous effect of afferent arteriolar vasoconstriction and helping preserve renal perfusion.

Drugs may unfavorably alter the glomerular hemodynamic response to renal hypoperfusion. Nonsteroidal anti-inflammatory drug (NSAIDs) administration to patients with decreased effective arterial blood volume (due to hypovolemia or congestive heart failure) or renal vasoconstriction (due to cirrhosis) leads to a decline in renal blood flow and GFR. These patients are dependent on vasodilator prostaglandins to maintain renal perfusion, so NSAIDs leave renal vasoconstrictor influences unopposed. Angiotensin-converting enzyme inhibitors (ACEIs) and angiotensin receptor blockers (ARBs) can lead to prerenal azotemia in patients who are dependent on angiotensin II for maintenance of GFR. This phenomenon is most commonly seen in patients receiving ACEIs or ARBs in the presence of hypovolemia, bilateral renal artery stenosis, or unilateral renal artery stenosis with a solitary kidney.

Prerenal azotemia also leads to avid renal tubular sodium and water reabsorption throughout the nephron. Catecholamines and angiotensin II directly increase sodium transport and reabsorption in the proximal and distal nephron. Efferent arteriolar constriction by angiotensin II and increased filtration fraction simultaneously lead to decreased peritubular capillary hydraulic pressure and increased peritubular capillary oncotic pressure. The combination of high oncotic pressure and low hydraulic pressure in peritubular capillaries increases sodium and water absorption in the proximal tubule, a process termed glomerulotubular balance. Angiotensin II also leads to downstream production of aldosterone, another salt-retaining influence. Severe hypovolemia/ hypotension (>10% to 15% decrease in MAP or blood volume) leads to nonosmotic vasopressin secretion, and avid water reabsorption in the collecting duct, along with systemic vasoconstriction. Finally, in hypovolemic patients decreased atrial stretch downregulates production of atrial natriuretic peptide, also favoring sodium retention (the opposite is true if renal hypoperfusion is caused by congestive heart failure). Thus the combination of glomerulotubular balance and the tubular effects of catecholamines, angiotensin II, aldosterone, and vasopressin mediate the salt and water retention which is the hallmark of prerenal azotemia. Accordingly, patients with prerenal azotemia tend to have oliguria, low urine sodium, and concentrated urine with a urine osmolality exceeding 500 mOsm/kg. Low urine sodium (and fractional excretion of sodium; see below) and increased urine osmolality (with a high urine:plasma creatinine ratio) are not seen in patients who have prerenal ARF due to renal losses (ongoing diuretic therapy, salt-wasting nephropathies, osmotic diuresis, adrenal insufficiency, and central or nephrogenic diabetes insipidus). Other common laboratory features of prerenal ARF are increased serum BUN:creatinine ratio (caused by low tubular flow and increased urea reabsorption), decreased fractional excretion of urea (see below), polycythemia/high serum albumin (hemoconcentration), mild hypercalcemia, hyperuricemia, and acid-base abnormalities (metabolic acidosis from diarrhea or shock or lactic acidosis; metabolic alkalosis from diuretics or vomiting). Hyponatremia may also be present, depending on abnormalities in water balance (see Chap. 76). The renal response to volume challenge or vasoactive drug initiation may also be used to determine the presence or absence of a reversible, “prerenal” etiology of ARF.

Postrenal Acute Renal Failure

Postrenal ARF (often called obstructive uropathy) accounts for 5% of all cases of ARF, and is more common in the elderly. Unilateral obstruction is not sufficient alone to cause ARF. Renal insufficiency due to obstruction occurs only when the obstruction involves a site that affects both kidneys, or a single functioning kidney. The most common cause of postrenal obstruction is bladder neck obstruction (prostatic hypertrophy, prostate cancer, and neurogenic bladder). Postrenal obstruction is also caused by bilateral ureteral obstruction or unilateral obstruction in patients with solitary kidney (stones, clots, sloughed renal papillae, retroperitoneal fibrosis, or retroperitoneal masses). Intratubular obstruction can be caused by crystals like uric acid, calcium oxalate, calcium phosphate, acyclovir, sulfadiazine, indinavir, methotrexate, or by paraprotein (myeloma cast nephropathy). Volume expansion, sometimes with urinary alkalinization (uric acid, methotrexate, or myeloma), is the primary treatment for these causes of intratubular obstruction.

Obstruction to urine flow by increased tubular hydrostatic pressure is only partly responsible for the reduced GFR of obstructive uropathy. ARF is also caused and sustained by renal vasoconstriction that occurs in response to ureteral obstruction, mediated by thromboxanes. Obstruction should be suspected in patients with recurrent urinary tract infections, nephrolithiasis, prostate disease, or pelvic tumor. Causes of obstructive uropathy are listed in Table 75-1. These patients usually have a preceding history of obstructive symptoms followed by sudden onset of anuria or oliguria. Polyuria and nocturia due to renal concentrating defect may be seen in patients with partial or intermittent obstruction. Other features of ARF secondary to obstruction are increased BUN:creatinine ratio, hyperkalemia, and defective urinary acidification with metabolic acidosis.

The two most important diagnostic tests when obstruction is suspected are bladder catheterization and renal ultrasonography. If urinary tract obstruction is strongly suspected, but ultrasound results are equivocal, then a “stone protocol” noninfused computed tomography (CT) scan should be performed. In some cases where false-negative ultrasound or CT scan results are suspected, cystoscopy and retrograde pyelograms may be required to definitively exclude the diagnosis of obstructive uropathy. For example, we would request retrograde pyelograms despite normal ultrasound images in a patient with anuric, hyperkalemic ARF and extensive pelvic tumor, potentially encasing the ureters and preventing dilation and hydronephrosis. Retroperitoneal fibrosis can similarly cause obstructive ARF without hydronephrosis. Early diagnosis is essential, as the extent of parenchymal damage is dependent on the duration of obstruction; complete recovery is possible up until 10 to 14 days of obstruction.

Intrinsic Acute Renal Failure

Intrinsic acute renal failure can be categorized anatomically, according to the site of the lesion: vascular, glomerular, or tubulointerstitial. We will discuss tubular and interstitial causes first, because they are far more common in hospitalized patients.

Acute Tubular Necrosis

The most common cause of intrinsic ARF in hospitalized patients is acute tubular necrosis (ATN). ATN is caused by ischemia, nephrotoxins, or a combination of both, and accounts for approximately 85% to 90% of intrinsic ARF cases.2,13 Ischemic ATN is commonly seen in patients with sepsis or severe cardiac failure, or postoperative patients, particularly after cardiac and aortic surgeries. Massive trauma or cardiac arrest are other causes of ATN. Prerenal failure can result in ischemic ATN if renal hypoperfusion is severe and not reversed by timely therapy. Although improving renal perfusion may reverse prerenal ARF (by definition), and diminish ischemic contributions to the pathogenesis of ATN, it is quite conceivable that in many cases ATN develops despite appropriate resuscitation and adequate renal perfusion. Zager has shown in an endotoxemic rat model of septic ARF that paired combinations of insults (renal cross clamp, systemic endotoxin, aminoglycoside, and temperature elevation) cause azotemia and renal pathologic findings of ATN, but these insults individually cause no renal dysfunction or injury.13 We suspect that this synergistic injury model accurately reflects the pathogenesis of much ARF in the ICU. Positive pressure mechanical ventilation alters renal perfusion and function through a variety of mechanisms, both hemodynamic and inflammatory.14 Other experimental data have shown that endotoxin, tumor necrosis factor, and numerous other inflammatory mediators are directly cytotoxic to renal endothelial and tubular cells.15

The diagnosis of ATN is usually made by clinical exclusion of alternative diagnoses such as obstruction or prerenal azotemia, and a lack of features suggestive of other intrinsic renal lesions. Characteristic urinalysis and urine chemistry findings frequently support the clinical diagnosis (see below). Ischemic ATN is usually reversible by tubular regeneration in surviving patients with previously normal renal function. Failure to recover prompts consideration of a differential diagnostic list including bilateral cortical necrosis, renal atheroembolism, renal artery stenosis/thrombosis/dissection, and severe forms of other intrinsic lesions such as rapidly progressive glomerulonephritis (GN). Bilateral cortical necrosis causes irreversible renal failure and is associated with profound shock with disseminated intravascular coagulation, obstetric complications, hemolytic uremic syndrome, or snake bites.16

Cellular mediators that play a role in the pathogenesis of ATN include calcium, reactive oxygen species, phospholipases, proteases, adhesion molecules, and nitric oxide (NO).17 ATN has several phases: prerenal, initiation, extension, maintenance, and repair.18 It is not intuitive that renal tubular injury should cause decreased glomerular filtration and ARF. A decrease in glomerular ultrafiltration coefficient has been shown in several animal models of ARF, but this is a minor contributor to the observed decrement in GFR.19,20 The pathophysiologic mechanisms that explain the reduction of GFR in ATN are hemodynamic abnormalities, tubular obstruction, and tubular backleakage of glomerular filtrate (Fig. 75-1).21 Renal vasoconstriction is seen in ARF,21,22 caused by activation of tubuloglomerular feedback; increased distal chloride delivery past injured tubular segments is sensed by the macula densa, causing vasoconstriction of the corresponding afferent arteriole. This reversible, functional mechanism seems to be the major cause of decreased GFR in ATN, and is in part protective. Severe hypovolemia would rapidly result if injured tubules failed to reabsorb the bulk of filtered sodium and water; thus the term “acute renal success” has been used to describe the development of decreased GFR (“acute renal failure”) in the presence of tubular necrosis.22 Furthermore, reabsorption of filtered sodium accounts for the bulk of renal oxygen consumption; continued glomerular filtration of sodium in ATN may aggravate hypoxic damage to sublethally injured tubules. The phenomenon of medullary hypoxia plays an important role in the pathogenesis of ATN. Low medullary blood flow is required for urinary concentration.23 Reabsorption of sodium chloride by the medullary thick ascending limb of the loop of Henle (mTAL) is the major determinant of medullary oxygen consumption, resulting in a hypoxic environment under normal circumstances.23,24 mTAL is vulnerable to ischemic injury if increased oxygen requirement is associated with decreased oxygen delivery. In addition, the inflammatory mechanisms that dominate as ATN progresses from initiation to extension and maintenance phases of ATN result in medullary congestion and hypoperfusion.18

The tubular factors that are also involved in the reduction of GFR in ARF are tubular obstruction and tubular backleakage. Necrotic cell debris incorporated into casts causes obstruction of proximal and distal tubules and has been shown to play a significant role in experimental ARF.25,26 Backleakage of tubular fluid across denuded basement membranes and injured proximal tubule cells has been demonstrated in several experimental models of ARF.27 Subsequently it has been shown that tubular backleakage and intratubular obstruction are important factors contributing to the reduction of GFR in human ischemic ARF.28,29

Nephrotoxic injury is the second major cause of ATN. Nephrotoxic ATN is caused by drugs (aminoglycosides, cisplatin, amphotericin, and chemotherapy), radiocontrast agents, heme pigments (myoglobin and hemoglobin), and myeloma light chain proteins. ARF due to aminoglycosides and radiocontrast agents accounts for most cases of nephrotoxic ATN. Nephrotoxicity of cancer chemotherapy is discussed in Chap. 74; malignancy and ARF are further discussed below. Three nephrotoxic ARF syndromes will be further discussed here: aminoglycoside nephrotoxicity, radiocontrast nephropathy, and ARF caused by nonsteroidal anti-inflammatory drugs (NSAIDs).

Aminoglycoside Nephrotoxicity

Aminoglycosides like tobramicin, gentamicin, amikacin, and netilmicin are widely used for the treatment of gram-negative infections. Although they are effective antibiotics, therapy with aminoglycosides is complicated by nephrotoxicity in 10% to 20% of patients.30 Aminoglycosides are excreted by glomerular filtration and are reabsorbed by proximal tubular cells. The mechanism of aminoglycoside-induced renal injury is not well understood. Accumulation of aminoglycosides in the proximal tubules in high concentration results in disruption of a variety of intracellular processes. Aminoglycosides are tubular toxins, and the earliest morphologic changes consist of vacuolization of proximal tubules, loss of brush border, and the presence of myeloid bodies within proximal tubule cells. Clinical evidence also attests to the tubular toxicity of aminoglycosides; maximum urine osmolality falls, and renal wasting of Mg2 + and K+ ensues. The relationship between this tubule damage and reduced GFR remains unclear, although in experimental models using high doses of aminoglycosides, tubule obstruction and backleakage can be demonstrated. In experimental models of aminoglycoside nephrotoxicity, relatively small doses decrease glomerular permeability, while larger doses cause renal vasoconstriction. The relevance of these hemodynamic changes to human aminoglycoside nephrotoxicity is not known.

ARF generally develops 7 to 10 days after aminoglycoside therapy is started. Aminoglycoside nephrotoxicity is associated with nonoliguric ARF, due to a concentrating defect caused by tubular injury. Aminoglycoside nephrotoxicity is also accompanied by potassium and magnesium wasting. Risk factors for aminoglycoside nephrotoxicity are summarized in Table 75-2. In patients with preexisting renal insufficiency the dosing interval should be adjusted and levels monitored to minimize the risk of ototoxicity and nephrotoxicity. Once-daily dosing of aminoglycosides decreases nephrotoxicity with no apparent loss of effectiveness.31–33 In patients with acutely deteriorating renal function who require aminoglycoside therapy, monitoring may be very difficult and require frequent reassessment. Aminoglycosides should be discontinued whenever renal dysfunction develops; increasing trough levels and decreased calculated aminoglycoside clearance may signal decreased GFR before serum creatinine increases. Recovery of renal function is expected in most cases when nephrotoxicity is recognized early and aminoglycosides held, but may be delayed for days to weeks.

Radiocontrast Nephropathy

ARF due to radiocontrast agents occurs within 24 to 48 hours of intravenous radiocontrast administration, and is termed radiocontrast nephropathy. Vasoconstriction and direct tubular toxicity due to generation of oxygen free radicals are thought to be the pathogenetic mechanisms of radiocontrast nephrotoxicity. In most cases the ARF is mild and recovery typically begins with stabilization of serum creatinine at 3 to 5 days postcontrast.34 Renal failure is usually nonoliguric. However, in patients with preexisting severe chronic renal failure radiocontrast nephropathy may be severe, irreversible, and require chronic dialysis.

The risk of nephrotoxicity due to contrast agents varies with the type and dose of agent. Low- or iso-osmolal agents are less nephrotoxic than ionic high-osmolal agents.35,36 The incidence of ARF due to radiocontrast agents is less than 2% in patients with normal renal function, but is inversely related to GFR in patients with chronic renal insufficiency; in high-risk patients the incidence is as high as 60%.37 There are several important risk factors for radiocontrast nephropathy, including chronic renal insufficiency, diabetic nephropathy, severe congestive heart failure, intravascular volume depletion, high contrast dose, and multiple myeloma (Table 75-3). Of note, the risk of radiocontrast nephropathy at any GFR level of chronic renal impairment is approximately double in diabetics compared to nondiabetics. The group at risk for radiocontrast nephropathy can be reliably identified, and a variety of prophylactic strategies have been used, several successfully.

A number of measures are used for prevention of radiocontrast nephropathy. Alternative imaging approaches (ultrasound, noninfused CT, or gadolinium-enhanced magnetic resonance angiography) are preferred to radiocontrast studies in patients with a high risk of radiocontrast nephropathy. Nonionic low-osmolal agents35 and nonionic iso-osmolol agents36 are less nephrotoxic, and these agents are used to decrease the incidence of ARF due to radiocontrast agents. Standard recommendations suggest that IV hydration with 0.45% saline is beneficial in reducing the incidence and severity of radiocontrast nephropathy,38 but recently normal saline (1 mL/kg per hour for 12 hours pre- and postcontrast) was demonstrated superior for this purpose.39 Most recently, Merten and colleagues showed that administration of an equimolar sodium bicarbonate solution was superior to normal saline for radiocontrast nephropathy prophylaxis, infused as 3 mL/kg per hour for 1 hour precontrast, then 1 mL/kg per hour for 6 hours.40 In some studies the administration of acetylcysteine, a thiol-containing antioxidant, in combination with saline hydration has been shown to be beneficial in reducing the incidence of contrast nephropathy when administered in various oral regimens (1200 mg once or 600 mg every 12 hours before and after radiocontrast).41–43 One successful study used intravenous acetylcysteine for radiocontrast nephropathy prophylaxis.44 Although other studies did not show any benefit,45,46 a recent meta-analysis of eight randomized controlled trials involving 855 patients reported that the use of acetylcysteine reduced the risk of radiocontrast by 59%.46 However, further doubt has been cast over these inconsistent findings by emerging experimental data that suggest that the apparent efficacy of acetylcysteine in these studies may have been artifactual. Specifically, acetylcysteine causes a decrement in serum creatinine (but not cystatin C) by a GFR-independent mechanism,47 perhaps by inhibiting creatinine phosphokinase function.48 Nevertheless, given the available data, oral administration of acetylcysteine to patients at high risk of developing contrast nephropathy appears warranted, and should do no harm in combination with adequate volume expansion. Other approaches such as vasodilation with dopamine49 or fenoldopam50 failed to prevent radiocontrast nephropathy, and mannitol and furosemide appeared to increase the rate of ARF postcontrast in one study.38 Prophylactic hemodialyis has not been shown to be beneficial in the prevention of contrast nephropathy.51 More recently, Marenzi and colleagues compared the effectiveness of hemofiltration with intravenous isotonic saline administration in patients with chronic renal insufficiency undergoing coronary interventions.52 BUN and serum creatinine were measured at baseline, immediately before the procedure, and then daily for the following 3 days and at hospital discharge. In this study there was less likelihood of an increase in serum creatinine (the primary end point), less likelihood of requiring emergent hemodialysis, and less in-hospital and 1-year mortality in the hemofiltration group.52 Adulteration of the primary end point by hemofiltration of creatinine in one arm is a fatal flaw of this study. Other flaws include an unbalanced level of care between the groups (floor vs. hemofiltration in the ICU), and more regulation of fluid balance in the hemofiltration group (and more pulmonary edema in the fluids-alone arm). Since hemofiltration is also expensive, associated with a variety of risks, and a limited resource, we do not recommend this approach to radiocontrast nephropathy prevention unless confirmed by additional prospective controlled studies.

Nonsteroidal Anti-Inflammatory Drugs and Acute Renal Failure

NSAIDs can cause hemodynamically mediated ARF. Vasodilatory prostaglandins (prostacyclin and prostaglandin E2 [PGE2]) are essential for the maintenance of renal blood flow and GFR in states of effective volume depletion, such as congestive heart failure, cirrhosis of the liver, nephrotic syndrome, and in states of true volume depletion.53 Prostaglandins counterbalance the effects of vasoconstrictors such as angiotensin II and catecholamines. NSAIDs inhibit prostaglandins and thus would lead to unopposed effect of angiotensin II and catecholamines, leading to reduced GFR. Hyperkalemia may be prominent because NSAIDs impair renin secretion. Patients with chronic renal insufficiency are similarly at risk for NSAID-induced ARF, because vasodilator prostaglandins maintain hyperfiltration in remnant nephrons. Hemodynamically-mediated ARF is seen within days of taking NSAIDs in high-risk patients. Cyclooxygenase-2 (COX-2) selective drugs have the same renal effects as nonselective NSAIDs, because COX-2 is constitutively expressed and physiologically active in the kidney.54 The second type of acute renal injury is allergic interstitial nephritis, as discussed below. Chronic NSAID use is also associated with papillary necrosis and is thought to occur in patients who are taking multiple analgesics.

Acute Tubulointerstitial Nephritis

ARF due to acute interstitial nephritis (AIN) is most often caused by allergic reaction to various drugs (allergic AIN), but there are also a variety of infectious (Legionnaire's disease, cytomegalovirus, and Hantavirus), autoimmune (lupus), alloimmune (renal transplant rejection), and infiltrative (sarcoidosis, leukemia, and lymphoma) disorders that can cause AIN (see Table 75-1). The most common drugs that cause allergic AIN are penicillins, cephalosporins, ciprofloxacin, rifampin, sulfonamides (furosemide, thiazide diuretics, and trimethoprim-sulfamethoxazole), cimetidine, allopurinol and NSAIDs.

Patients with AIN classically present with ARF temporally related to drug therapy or infection, associated with a triad of fever, rash, and eosinophilia.55 Urinalysis findings include leukocyturia with eosinophiluria, leukocyte casts, and low-grade proteinuria. Although all these signs are present in the majority of patients with AIN, absence of these does not exclude the diagnosis of AIN. In particular, these findings are usually absent in NSAID-induced AIN. Proteinuria is mild to moderate except in NSAID-induced AIN, in which nephrotic syndrome caused by a minimal change lesion has been described. Tubular dysfunction (e.g., Fanconi's syndrome, distal renal tubular acidosis, or hyperkalemia) occurs in the majority of patients with AIN.

AIN is usually suspected from the history and laboratory findings. In the absence of urinary tract infection, detection of large numbers of urinary eosinophils (>5%) strongly suggests the diagnosis of drug-induced tubulointerstitial nephropathy (TIN). Hansel's stain of the urine is more sensitive than Wright's stain for the detection of urinary eosinophils.56 The Hansel method correctly identified 10 of 11 patients with TIN, as opposed to only 2 of 11 correctly classified using Wright's stain. False-positive results with the Hansel technique are most commonly caused by rapidly progressive GN or acute prostatitis. Diagnosis can be confirmed by renal biopsy if ARF is progressive and treatment of an alternative diagnosis is considered (e.g., rapidly progressive GN), or if there is no recovery of renal function after discontinuation of the medication suspected to have caused AIN.

No therapy is recommended in patients with mild renal insufficiency, and in patients who respond after discontinuation of the offending medication. AIN is usually reversible after withdrawal of the offending agent and treatment of underlying disease; corticosteroid therapy is unproven, and generally recommended only in patients with biopsy-confirmed acute allergic interstitial nephritis who do not respond to conservative management, and have no contraindication to immunosuppression.

Rapidly Progressive Glomerulonephritis

Glomerulonephritis such as postinfectious glomerulonephritis, lupus nephritis, and Goodpasture's syndrome can cause acute or subacute renal failure (see Table 75-1). Rapidly progressive glomerulonephritis (RPGN) causing ARF is an emergency. The combination of ARF with an “active” urine sediment (heavy proteinuria, hematuria, leukocyturia, and erythrocyte/leukocyte/mixed cellular casts) should prompt urgent evaluation with renal biopsy and serologies. The underlying pathologic lesion is classically necrotizing crescentic glomerulonephritis, treated with high-dose pulse corticosteroid therapy to stabilize renal function, followed by intensive immunosuppression (with or without plasmapheresis). Causes include inflammatory disorders with immune complex deposition (lupus, postinfectious, or cryoglobulinemia), anti–glomerular basement membrane antibodies (Goodpasture's syndrome, when associated with pulmonary hemorrhage), and pauci-immune glomerulonephritis (usually with small-vessel vasculitis; see below). Associated clinical features such as pulmonary hemorrhage, sinus involvement, leukocytoclastic vasculitis, neuropathy, or other stigmata of autoimmune disease may increase the index of suspicion for this diagnosis, but it must be emphasized that a skilled urinalysis demonstrating an “active” sediment in a patient with ARF may make this diagnosis alone. In fact the presence of erythrocyte casts is pathognomonic of GN; associated with ARF, this is essentially diagnostic of RPGN.

Vascular Causes of Acute Renal Failure

Vascular causes of ARF are categorized into small-vessel diseases and large-vessel diseases (see Table 75-1). Diseases involving small vessels include microscopic polyarteritis (vasculitis in polyarteritis nodosa and Kawasaki's disease involves medium-sized vessels), Wegener's granulomatosis, mixed cryoglobulinemia, and conditions that are categorized as thrombotic microangiopathies, including thrombotic thrombocytopenic purpura, hemolytic uremic syndrome, scleroderma renal crisis, malignant hypertension, and antiphospholipid antibody syndrome. Large-vessel renal vascular diseases include thromboembolic diseases and renal vein thrombosis. Atheroembolic disease should be considered in patients who develop ARF after instrumentation of the aorta, particularly in patients with known atherosclerotic disease.57 Renal vein thrombosis is usually a complication of nephrotic syndrome, and if bilateral can cause ARF. Abdominal compartment syndrome (see Chap. 42) is present when the intra-abdominal pressure (IAP) reaches 20 to 25 cm H2O, and unless decompressed, irreversible organ failure may result.58 The pathogenesis of oliguria and ARF in abdominal compartment syndrome involves venous compression (decreased venous return and renal vein compression), ureteral compression with obstructive uropathy, and possibly changes in renovascular resistance and intrarenal blood flow distribution. Regardless of the underlying cause, a reduction in urine output with or without azotemia in the presence of a measured intra-abdominal pressure over 15 cm H2O is cause for concern and should prompt intervention.

Monitoring Renal Function in the ICU

Attempts to develop strategies to prevent or treat ARF in the ICU have been hampered by the lack of sensitive real-time methods to monitor renal perfusion and function in the clinical setting.59 There are no direct measures of renal perfusion in clinical use. Of the available research techniques for this purpose, para-aminohippuric acid (PAH) clearance is not a valid method to assess renal plasma flow because renal PAH extraction is impaired in critical illness, after cardiac surgery, post–renal transplantation, and in ARF.60 Clinical indices such as plasma concentrations of urea nitrogen and creatinine, urine output, urine chemistries, and urinalysis may be assessed in combination to monitor renal perfusion and function, but alterations in these markers are insensitive, indirect, and often delayed manifestations of renal hypoperfusion and injury.59 Although relatively insensitive, it should be recognized that progressive elevation of serum creatinine is a specific sign of decreased GFR and diagnostic of ARF. As depicted in Fig. 75-2, daily serum creatinine increments of 0.5 to 1 mg/dL (depending on muscle mass, and in the absence of rhabdomyolysis) signal severe depression of GFR.61 In the absence of obstruction or renal hypoperfusion, this is usually diagnostic of ATN.

Direct measurement of GFR is probably the most relevant marker of renal function to monitor in the ICU, and can be used as an index of adequate systemic perfusion, a marker of organ dysfunction, a guide to medication dosing, and to determine timing of initiation of renal replacement therapy. GFR measurement methods more sensitive than monitoring serum creatinine and urea nitrogen have been validated in this population, but are not yet in widespread clinical use.59,62 Abbreviated creatinine clearance measurements (2 to 4 hours) provide a more accurate estimate of GFR than serum creatinine alone, and are more timely than 24-hour urine collections.63 Emerging data suggest that serial monitoring of serum cystatin C levels may provide a more sensitive index of early ARF than serum creatinine or BUN. Cystatin C is a nonglycosylated 13-kDa basic protein that is a member of the cystatin superfamily of cysteine protease inhibitors. It is produced by all nucleated cells, and its production rate is unaltered by inflammatory conditions or muscle mass; only thyroid dysfunction has been shown to alter serum levels independent of GFR in studies to date.64–66 It undergoes purely renal excretion. Serum cystatin C levels increase before serum creatinine levels in patients with progressive chronic kidney disease.64,66 Emerging data suggest that plasma cystatin C levels similarly increase 1 to 2 days before serum creatinine in patients developing ARF in a variety of settings,66 including radiocontrast nephropathy,67 renal transplantation,68,69 cirrhotic ARF,70 malarial ARF,71 and ARF in the ICU.72,73 In the latter study of 85 critically-ill patients, Herget-Rosenthal and colleagues showed that ARF defined by elevation of cystatin C occurred 1 to 2 days before serum creatinine elevation; the area under the curve in receiver operating curve analysis was 0.82 and 0.97 on the 2 days prior to creatinine-defined ARF.73

In patients with azotemia and/or oliguria, traditional urine chemistry markers may provide useful information, interpreted in combination with other clinical and laboratory parameters (see below), including microscopic urinalysis.59,74 Hopefully, more sensitive markers of early renal hypoperfusion and injury will prove clinically useful in the future.75,76 Meanwhile, all clinically available renal function indices should be used in combination for assessment of renal perfusion and function in the ICU.

Diagnostic Approach to Acute Renal Failure

The diagnostic approach to ARF involves assignment of the cause to prerenal, renal, or postrenal categories, with further refinement of the diagnosis based on additional laboratory testing.

History and Physical Examination in Acute Renal Failure

Any decrease in effective perfusion of the kidneys can result in the syndrome of prerenal ARF. This may be the result of an absolute decrease in the extracellular fluid (ECF) volume, redistribution of ECF from vascular to interstitial locations (third-spacing), or impaired delivery of blood to the kidneys such as can occur in patients with renal arterial stenosis, vasculitis, or depressed cardiac function. Third-space losses should be suspected in the presence of severe burns, pancreatitis, peritonitis, or recent abdominal surgery. Absolute decreases in the ECF volume are most common in the setting of gastrointestinal fluid losses or in patients receiving excessive doses of diuretics.

Decreases in weight, if known, can provide some information about the degree of ECF loss. However, weight changes are subject to misinterpretation if the nature of fluid loss is not taken into account (Fig. 75-3). The ability of pure water loss (which is spread out across the total-body water) to cause volume depletion is only one-third as great as an equivalent loss of isotonic fluid (all of which must come from the smaller ECF compartment). Thus the importance of changes in weight should be assessed relative to changes in serum sodium concentration.

The cardinal signs of ECF volume depletion are changes in hemodynamic parameters, jugular venous pressure, and in the skin. An orthostatic increase in pulse of 15 beats per minute or a decrease in diastolic blood pressure of 10 mm Hg can detect losses of 5% of the ECF volume. A postural increase in pulse (supine to standing) of at least 30 beats per minute is 96% specific for clinically-significant volume depletion, whereas systolic pressure may fall 20 mm Hg upon standing in 10% of normal individuals and in up to 30% of patients > age 65.77 The inability of a patient to stand because of severe lightheadedness is a relatively specific sign of hypovolemia. In uncomplicated hypovolemia, estimated central venous pressure by examination of jugular venous pressure should be low. Skin changes that accompany volume depletion include cool, mottled extremities, dry mucous membranes and axillae, and skin tenting (particularly over the forehead and sternum, where age-related changes in skin elasticity are not as pronounced as elsewhere).

Obstruction of the urinary tract must be considered in every patient with an acute deterioration of renal function. The symptoms of acute urinary tract obstruction (severe flank pain, hematuria, and changes in urine flow) are often mistaken for urinary tract infection. Of more importance from a historical standpoint is the identification of preexisting conditions that predispose to urinary tract obstruction. Some of these are listed in Table 75-1. Physical findings suggestive of obstruction include palpably enlarged kidneys, pelvic or abdominal masses, bladder enlargement, prostatic hypertrophy, aneurysmal dilation of the aorta, and signs of inflammatory bowel disease. If oliguria or anuria develops in a critically ill patient with a Foley catheter in place, possible catheter occlusion should be assessed by sterile flushing and if necessary a catheter change.

Intrinsic ARF can be the final result of many diverse renal insults. While space limitations do not permit a thorough review of all aspects of the history and physical examination in intrinsic ARF, some points deserve comment. ARF due to therapeutic or recreational drugs (e.g., cocaine-induced rhabdomyolysis) is so common that a detailed drug history is mandatory.78 The presence of a skin rash should suggest the possibility of a systemic vasculitis with renal involvement or acute tubulointerstitial nephritis. Palpable purpura due to leukocytoclastic vasculitis is characteristic of Henoch-Schönlein purpura. One of the pulmonary-renal syndromes should be considered if prominent thoracic complaints accompany ARF. These include, among others, Goodpasture's syndrome, Wegener's granulomatosis, microscopic polyarteritis, systemic lupus erythematosus, and ChurgStrauss syndrome.

Diagnostic Tests in Acute Renal Failure

The majority of cases of ARF can be diagnosed by history and physical examination. However, in a significant minority the cause remains obscure after initial assessment, and further evaluation is necessary.

Daily urine volume must be measured in all patients with ARF. Bladder catheterization is both diagnostic and therapeutic in patients with obstruction at the level of the bladder neck or urethra. Urine volume is determined by the requirement to excrete the daily obligate solute load (electrolytes and nitrogenous wastes) in appropriately concentrated urine. Assuming maximal urine concentrating ability (1400 mOsm/kg), the minimum daily urine output required to excrete the average daily solute load is 400 mL, below which positive solute balance and azotemia develop, thus the standard definition of oliguria (<400 mL/24 hours). In terms of monitoring urine output, if urine is maximally concentrated (1400 mOsm/kg), and excretion of 10 mOsm/kg per day (700 mOsm/d in a 70-kg person) is required to avoid solute retention, this mandates urine output of 500 mL daily (21 mL/h, or 0.3 mL/kg per hour). Of course, if solute appearance increases (patient size, hypercatabolism, or hyperalimentation) or maximal urinary concentrating ability is diminished (renal dysfunction or advanced age), higher urine volumes are required to maintain adequate solute excretion. Since such conditions are more the rule than the exception, it seems more appropriate to expect solute retention at urine outputs below the more typical ICU monitoring target of 0.5 to 1 mL/kg per hour (840 to 1680 mL/d). Of course, urine output targets must be sufficient to control fluid balance as well as solute excretion, so higher urine output values may be required for patients with large obligate fluid intakes.

ARF is classified as anuric (urine output <100 mL/d), oliguric (urine output <400 mL/d), or nonoliguric (urine output >400 mL/d). Causes of ARF associated with various urine flow patterns are listed in Table 75-4. Prerenal ARF with polyuria may be seen very rarely if excessive urine losses are the cause of the prerenal state. This occurs in adrenal or mineralocorticoid deficiency states and excessive diuresis. Although occasional polyuric patients with urinary indices suggestive of prerenal ARF have been described,79 it is believed that the majority of them in fact have polyuric ATN rather than prerenal ARF. The continued use of an indwelling bladder catheter after the cause of ARF has been determined is frequently unnecessary and merely increases the risk of nosocomial urinary tract infection. This is particularly true in the oligo-anuric patient. Intermittent bladder catheterization once or twice daily can provide useful information with a lower risk of urosepsis. An external condom-type catheter does not provide sufficient information to replace the Foley catheter in persons with ARF. Because it is also associated with an increased risk of urinary infection, it cannot be recommended in this setting.

Urinalysis is also useful in patients with ARF. The urinary specific gravity tends to be >1.020 in patients with prerenal failure. On the other hand, patients with intrinsic or postrenal ARF are generally isosthenuric, with a urine specific gravity of approximately 1.010. Substantial proteinuria (3 g/d or more) strongly suggests the possibility of a glomerular disease, with nephrotic-range proteinuria (>3.5 g per 24 hours) pathognomonic of glomerular rather than tubular disease; this may be confirmed with a “spot” urine protein:creatinine ratio (>3 suggests nephrotic-range proteinuria will be confirmed by 24-hour urine collection). Glycosuria in the absence of hyperglycemia strongly suggests proximal tubular injury with Fanconi's syndrome. A positive reaction for blood in the urine is consistent with acute glomerular or tubular injury, urinary tract infection, or nephrolithiasis. If blood is present on dipstick but not microscopically, or if the findings are disproportionate (e.g., 4+ blood on dipstick with rare erythrocytes on microscopy), a pigment nephropathy (hemoglobinuria or myoglobinuria) should be considered. The urine sediment is usually unremarkable in prerenal and postrenal azotemia, except for occasional hyaline casts. In postrenal ARF due to stones, blood and crystals can be seen. Intrinsic ARF is often associated with a characteristic (or even diagnostic) urine sediment. A careful microscopic examination frequently can distinguish between GN, AIN, ATN, and TIN. Erythrocyte casts, often accompanied by proteinuria and numerous erythrocytes and leukocytes, are pathognomonic of GN. Detection of large numbers of leukocytes, leukocyte casts, and eosinophils in uninfected urine strongly suggests the diagnosis of drug-induced AIN. ATN is suggested by findings including muddy brown granular casts, free renal tubular cells, and tubular cell casts.

Several measurements of urine composition have been suggested as ways to differentiate between prerenal azotemia and intrinsic ARF in the oliguric patient.80 Urine electrolytes are most useful in this regard, especially the fractional excretion of sodium (FENa), calculated as

where UNa and PNa are urine and plasma sodium concentrations, respectively, and UCr and PCr are urine and plasma creatinine concentrations, respectively. Values of FENa <0.01 (1%) in oliguric patients suggest avid tubular sodium reclamation and prerenal azotemia with functioning renal tubules, whereas values >.03 (3%) suggest tubular injury. The FENa is less useful in patients who are not oliguric.81 Contrary to common belief, however, it may be useful in diuretic-treated patients. Although an elevated value may be a result of ATN or the effects of the diuretic, a low level in the face of diuretic therapy strongly suggests volume depletion and prerenal ARF. Some causes of ARF presenting with a low FENa are listed in Table 75-5. A low UNa (<10 mEq/L) as an isolated measurement is often used as evidence of a prerenal state. However, this measurement depends exquisitely on the state of water balance in addition to sodium balance. It cannot be said that it is any easier to use than FENa, since an independent evaluation of water balance must be made to interpret it. Therefore it is not recommended as an isolated measurement in the routine evaluation of ARF.

Most other urinary diagnostic indices do not show any clear-cut superiority over FENa in distinguishing prerenal azotemia from ATN; however, they are independently useful in the assessment of tubular function. Recent data suggest urinary fractional excretion of urea (FEUN) is superior to FENa to distinguish prerenal azotemia from ATN, particularly in diuretic-treated patients. FEUN is calculated as

where UUN and BUN are urine and serum urea nitrogen concentrations, respectively, and UCr and PCr are urine and plasma creatinine concentrations, respectively. The normal FEUN is 50% to 65%, reflecting reabsorption of approximately 50% of filtered urea in the proximal tubule; urea reabsorption is trivial in the thick ascending limb and distal convoluted tubule. Hypovolemia results in increased urea absorption, decreased urea clearance, and thus a lower FEUN.82 Loop and thiazide diuretics, which act at the thick ascending limb and distal convoluted tubule, do not interfere directly with urea reaborption and should not alter FEUN. However proximal tubule diuretics and osmotic diuresis decrease proximal reabsorption of urea and may produce an inappropriately high FEUN. Carvounis and colleagues prospectively evaluated 102 hospitalized patients with ARF.74 Patients were divided into three groups: 50 were deemed prerenal; 27 were deemed prerenal with diuretics given up to the day of consultation (details were not provided as to whether diuretics were given 1 or 23 hours prior to the urine sample); and 25 were diagnosed with ATN. Patients with AIN, GN, and obstructive nephropathy were excluded. FeNa was <1%, as expected, in 92% of group 1 patients, but in only 48% of the prerenal patients treated with diuretics. In contrast, 90% of the group 1 patients and 89% of those given diuretics had a FEUN <35%. The ATN patients evidenced a mean FEUN of 59%. A FEUN <35% had 85% sensitivity, 92% specificity, 99% positive predictive value and 75% negative predictive value for a prerenal state. In this study, the urine:plasma creatinine ratio also performed better than FENa to distinguish prerenal azotemia from ATN. One other urine chemistry test may be useful in hyperuricemic patients with ARF and possible urate nephropathy (tumor lysis syndrome and hypovolemia with hyperuricemia and acid urine): urine uric acid:urine creatinine ratios >1.1 are consistent with acute urate nephropathy.83

Several radiographic studies are useful in the evaluation of patients with ARF. Plain films of the abdomen can assess kidney size, detect >90% of renal stones, and detect skeletal abnormalities of secondary hyperparathyroidism, which imply chronic renal insufficiency rather than ARF. In our view, the potential hazards of intravenous pyelography make this test of little benefit in the work-up of ARF. Renal ultrasound is a sensitive and specific method for detecting hydronephrosis. It is probably indicated in nearly every patient with ARF unless obstruction can be proven more quickly in another manner (e.g., by bladder catheterization in a patient with symptoms of bladder neck obstruction) or if some diagnosis other than obstruction is made with certainty early in the evaluation. If clinical suspicion of obstruction persists despite an apparently negative ultrasound, retrograde pyelography is the definitive diagnostic maneuver.84 Computed tomography of the kidneys does not have any advantages over ultrasound and retrograde pyelography in the detection of obstruction, but may help clarify its cause. Radionuclide scans of the kidneys may be useful in the evaluation of patients with suspected vascular accidents of the kidneys.

The great majority of the time, volume status can be ascertained by history and physical examination alone. Occasionally one is presented with a patient in whom the ECF volume status cannot be determined with certainty on clinical grounds, particularly subjects with significant third-space losses. A typical example would be a patient with sepsis, hypotension, noncardiogenic pulmonary edema, anasarca, and oliguria. In this case, the clinician is often faced with the contradictory choice of using either diuretics or intravenous fluids. It is obviously inappropriate to treat a patient as if volume overload and hypovolemia were present simultaneously; in this setting, invasive hemodynamic monitoring with a central venous pressure (and central venous oxygen saturation) or a pulmonary artery catheter (and mixed venous oxygen saturation) can be an invaluable diagnostic adjunct. Complementary information may be obtained from simultaneous echocardiography.

There are a number of diseases that commonly lead to ARF. In many cases of ARF in the ICU, prerenal azotemia combines with exposure to nephrotoxins (endogenous or exogenous) to cause ATN. However, the suspicion of other causes such as obstruction or specific parenchymal lesions is increased in certain patient populations.

Malignancy and Acute Renal Failure

ARF is seen in patients with malignancy and could be caused by malignancy itself or its treatment with chemotherapeutic agents (Table 75-6). The full differential diagnosis of ARF must be considered in cancer patients, including common causes such as prerenal azotemia, drug nephrotoxicity, and obstructive uropathy, as well as rarer causes such as tumor infiltration of the kidneys and radiation nephritis. Urinary tract obstruction leading to ARF is commonly seen in patients with malignancy. Ureteral obstruction is seen in patients with abdominal and pelvic metastasis. ARF is also caused by intratubular obstruction. This is usually seen in patients with acute uric acid nephropathy as part of tumor lysis syndrome, intratubular obstruction by light chains in patients with multiple myeloma, and in patients who received high-dose methotrexate.

A variety of chemotherapeutic agents are potentially nephrotoxic, including cisplatin, nitrosoureas, and methotrexate, as discussed in Chap. 74. In addition, mitomycin is known to cause hemolytic uremic syndrome thus leading to renal failure.

If large numbers of malignant cells suddenly die, as in spontaneous necrosis or successful chemotherapy of large tumors, ARF can occur as a result of tumor lysis syndrome. This is noted most commonly in patients with germ cell tumors or hematologic malignancies because of their rapid turnover.85 The tumor lysis syndrome is due to the toxic effects of intracellular constituents, which cause ARF, hyperkalemia, hyperphosphatemia, hypocalcemia, and hyperuricemia. The most common pathogenetic mechanism involved is acute urate nephropathy. When serum uric acid levels exceed about 20 mg/dL, the risk of ARF is high. The urinary ratio of uric acid to creatinine is >1.1.83 Uric acid crystals may be seen in the urinary sediment. Synergistic factors in the development of the acute nephropathy include volume depletion and acidosis. The usual pathophysiology of urate nephropathy is intratubular deposition of urate crystals, causing intrarenal obstruction.86 Less commonly, ureteral obstruction may be seen.

Once oliguria occurs, diuretics are seldom useful in restoring urine flow. Allopurinol and alkaline diuresis are effective measures when instituted prophylactically, but do not reverse established urate nephropathy.87Rasburicase, a recombinant form of urate oxidase, has been shown to be effective in reducing plasma uric acid levels within 4 hours after the first dose;88 urate oxidase converts uric acid to allantoin, which is more soluble than uric acid. Severe hyperphosphatemia (levels >20 mg/dL) can be associated with tumor lysis and subsequent ARF.89 The pathogenesis of the renal dysfunction can be either intratubular crystallization of phosphate or metastatic calcification in the renal parenchyma. It is important to make the distinction between this entity and urate nephropathy, because urinary alkalinization promotes calcium phosphate crystallization. If hyperuricemia is prevented or controlled by volume expansion, allopurinol, or uricase, it may be preferable to avoid or stop urinary alkalinization if severe hyperphosphatemia develops, to avoid hyperphosphatemia ARF and the risk of tetany (hypocalcemia combined with metabolic alkalosis). Renal replacement therapy in tumor lysis syndrome is indicated in patients in whom the resolution of ARF is unlikely, and is also indicated in patients with life-threatening electrolyte abnormalities. Normalization of uric acid and phosphorus levels is required for recovery of renal function. Hemodialysis is far more effective in the clearance of uric acid compared to peritoneal dialysis.85 Continuous renal replacement therapies (CRRT) have been used in the management of tumor lysis syndrome. The solute clearances vary depending on the specific prescription. In these patients the uric acid and phosphorus clearances were, respectively, 45 mL/min and 47 mL/min in continuous venovenous hemodiafiltration and 39 mL/min and 40 mL/min in continuous arteriovenous hemodialysis.90,91 These clearances were achieved with high dialysate and high replacement fluid rates. CRRT cannot correct electrolyte abnormalities as rapidly as intermittent hemodialysis, but it is a much more effective treatment for severe hyperphosphatemia, because rebound elevation of serum phosphorus postdialysis is prevented. Sequential hemodialysis to control severe evolving electrolyte abnormalities, followed by CRRT to prevent rebound may be the best approach. If RRT does not result in diuresis within a week, ureteral catheterization should be performed to exclude ureteral obstruction.

Thrombotic Microangiopathy and Acute Renal Failure

Hemolytic uremic syndrome (HUS) and the closely related entity thrombotic thrombocytopenic purpura (TTP) are characterized by renal involvement early in their course. Renal disease tends to be more fulminant in HUS than in TTP. Nearly 100% of patients with HUS have ARF at some point in their course (which is severe in at least 60%) as compared with a total incidence of renal failure of even mild severity in TTP that is <50%. Clinical features of these disorders are more thoroughly discussed in Chap. 70.

In the syndrome of disseminated intravascular coagulation (DIC), the pathologic picture is cortical necrosis. This can be prevented in some experimental models by the infusion of the angiotensin antagonist saralasin. It is unknown if activated protein C will decrease the contribution of microvascular injury and DIC to the pathogenesis of septic ARF. Numerous other causes of microangiopathic hemolytic anemia and thrombocytopenia can also cause ARF, including malignant hypertension, scleroderma renal crisis, and antiphospholipid syndrome.

Pregnancy and Acute Renal Failure

The incidence of ARF complicating pregnancy has been decreasing for the past 25 years. This is attributable to improvements in prenatal care and obstetric science and fewer complications related to septic abortions. The timing of ARF in pregnancy has a bimodal distribution.92 The early peak, which occurs during the first 20 weeks of gestation, is due to septic abortions, while the later peak (at 36 to 40 weeks) is secondary to preeclampsia and bleeding complications.

Preeclampsia and eclampsia are accompanied by decreases in GFR of 20% to 25% that are seldom clinically significant (see Chap. 105). However, in a small proportion of pregnant women, the renal dysfunction progresses to severe ARF. Women with preeclampsia are at increased risk for the development of cortical necrosis. Cortical necrosis also can occur as a complication of placental abruption (often with concealed blood loss) or prolonged intrauterine death. Although severe, irreversible renal failure may be seen, partial recovery of renal function due to patchy cortical necrosis is more common.93 The diagnosis of complete cortical necrosis should be established by renal biopsy or arteriography to exclude patchy cortical necrosis and ATN, from either of which a degree of functional recovery is to be expected.

Coexisting ARF and fulminant hepatic failure have been described in gravid subjects. In women with acute fatty liver of pregnancy, ARF is present in over half. Although the mortality rate for both mother and fetus in this condition has been reported to be at least 70%, the prognosis may be improving because of increased recognition of less severe cases. The syndrome known by the acronym HELLP (hemolysis, elevated liver enzymes, and low platelets) is a variant of unusually severe preeclampsia. The great majority of these patients have evidence of hepatic dysfunction and sinusoidal congestion that may culminate in liver rupture. The mean creatinine clearance in these women is about 55 mL/min (less than half normal), while 10% have severe ARF with a creatinine clearance of <20 mL/min.

The rare but well-defined clinical entity of idiopathic postpartum ARF is characterized by the onset of ARF in the peripartum period following a previously normal pregnancy and childbirth.94 The serum creatinine level rises rapidly within a few days to several weeks following parturition. There is often an associated consumptive coagulopathy. Other common clinical features include malignant hypertension, lethargy, seizures, and a dilated cardiomyopathy. The pathophysiology of this disorder is unknown. The outcome is poor; most patients have severe chronic renal failure, require dialysis, or die.

Liver Disease and Acute Renal Failure

ARF frequently occurs in the setting of severe liver disease. Prerenal azotemia can be a consequence of ascitic redistribution of the ECF (or its excessive treatment with diuretics), GI fluid losses (e.g., vomiting, diarrhea, or hemorrhage), or cardiac dysfunction (e.g., alcoholic cardiomyopathy). If a coagulopathy is present, blood clots in the collecting system can cause obstructive uropathy. Cirrhosis may predispose to the occurrence of papillary necrosis. Intrinsic ARF in subjects with liver disease is classified into disorders that simultaneously injure liver and kidney and those in which the renal disease is a consequence of the hepatic process (Table 75-7). The kidney is generally much more sensitive to the effects of potential nephrotoxic agents if jaundice is present. A relatively specific glomerular lesion called cirrhotic glomerulosclerosis has been identified and is usually asymptomatic; however, it probably predisposes to ARF of other causes. More commonly, patients with hepatitis B or C with proteinuria have membranous nephropathy or membranoproliferative (cryoglobulinemic) glomerulonephritis. Hepatitis B is also associated with polyarteritis nodosa. The most serious of coexisting hepatic and renal diseases is the hepatorenal syndrome (HRS).

HRS is difficult to distinguish from prerenal ARF in the setting of advanced liver failure; indeed, many authorities consider the two conditions to be part of a single spectrum. Perhaps the best operational definition of HRS is severe prerenal azotemia that does not respond to volume repletion. The principal functional abnormalities that have been described are renal vasospasm in the setting of peripheral vascular resistance that is abnormally low;95 there is no “typical” structural lesion of HRS. Oliguria and a low FENa are frequent, if not universal, findings; the diagnosis of HRS should be considered untenable in their absence.

The therapy of HRS has been very disappointing.95,96 A trial of empirical volume expansion with colloid solutions, with an end point of clear-cut increases in ascites and peripheral edema, can be used diagnostically to differentiate prerenal ARF from HRS. Attempts to modify low systemic vascular resistance have generally been unsuccessful in reversing this condition, as have attempts at selectively increasing renal perfusion. Emerging data suggest potential utility of combination treatment with midodrine and octreotide in hepatorenal syndrome;97 vasopressin and analogues may also be useful for this indication.96 There is little information on the use of transjugular intrahepatic portosystemic shunt in patients with HRS. Dialysis is only recommended in patients who are awaiting liver transplantation or in whom recovery of the underlying liver lesion is possible.

Paracentesis, ascitic reinfusion, and use of peritoneovenous shunts have not produced substantial improvements in patient survival in the few available controlled studies. Attempts to redistribute ascitic fluid into the vascular tree and then excrete it with various combinations of albumin, furosemide, and dopamine have not improved long-term results. Hemodialysis and related modalities have not been shown to improve survival and generally cannot be recommended except in two general settings. If there is any doubt about whether the patient may have potentially reversible ATN (such as by demonstration of a FENa >1%) or if the patient is a candidate for hepatic transplantation, all supportive measures including dialysis should be considered.

Some diseases simultaneously affect both liver and kidney. These diseases are summarized in Table 75-7.

Acute Renal Failure in Renal Transplantation

The approach to the transplant patient with ARF is no different from that in any other patient, with the exception that several unique entities must be considered. It is simplest to consider these in relation to the time since transplantation occurred. Within the first few hours or days after surgery, technical problems are the first consideration. In addition to hypovolemia and ATN, these include vascular thrombosis, ureteral stenosis, urinary leaks, and obstructive fluid collections such as hematomas or lymphoceles. Hyperacute rejection, though often apparent at the time of surgery, may not be recognized for several hours or days. Thorough diagnostic evaluation is mandatory, including renal ultrasound, confirmatory tissue typing (particularly the direct cross-match), and occasionally angiography or transplant biopsy. Surgical intervention is often required in addition to the usual supportive measures.

During the period beginning approximately a week after surgery and continuing for the next several months, drug effects, acute rejection, and infectious processes are of particular concern. It is during this time that antirejection drug dosages are at their highest levels, and as a result complications related to these drugs are most frequent. The immunosuppressive drugs cyclosporine and tacrolimus are frequent causes of dose-dependent acute nephrotoxicity, and levels should be monitored closely; thrombotic microangoiopathy is a rarer adverse effect of these calcineurin inhibitors. The clinical diagnosis of acute rejection is often difficult to make, frequently requiring histologic confirmation or empirical antirejection therapy. Although an acute rejection episode occurs in the majority of renal transplant recipients within the first year following engraftment, it is increasingly uncommon thereafter. Thus a diagnosis of acute rejection several years after transplantation is less likely as long as the patient complies with therapy.

One of the most frequent and severe infections compromising renal function in transplant recipients is cytomegalovirus (CMV), which can cause dysfunction in many organ systems, including the central nervous system, lungs, liver, and kidneys. CMV is often suspected clinically on the basis of fever, multisystem organ involvement (including ARF), and progressive leukopenia. It is more common in patients who have received intensive immunosuppression for severe or recurrent rejection episodes.

Late causes of ARF in transplant recipients include recurrence of the patient's original renal disease, de novo transplant glomerulopathy, infections, transplant artery stenosis, and urologic problems such as stricture or rejection of the ureter. In addition to renal ultrasound, transplant biopsy is often useful in defining the cause of ARF late in the course of a kidney transplant.

Frequently, ARF develops in hospitalized patients in whom it is predictable and can be prevented or ameliorated. To intervene effectively, it is important to identify the patients at risk (Table 75-8). There is an additive interaction among the risk factors. Unfortunately, most attempts to modify the course of ARF are probably too late if tubular damage has already begun4.

Although ARF may be prevented or ameliorated by judicious use and monitoring of nephrotoxic drugs, and perhaps evolving cytoprotective and anti-inflammatory therapies, the major focus in ARF prophylaxis and therapy remains optimization of renal perfusion. The primary causes of renal hypoperfusion differ between the major types of shock, and therapies vary accordingly. The role of hemodynamic monitoring and support with fluids and vasoactive drugs in the prevention of ARF in the ICU is important, but there is little high-grade evidence to guide these therapeutic choices. As discussed above, renal perfusion is optimized by using fluids and vasoactive drugs to seek an adequate perfusion pressure and cardiac output. There is generally no adequate study to guide the choice of specific fluid or vasoactive drug to improve renal function. It appears that colloids are not superior to crystalloids for prevention of ARF in critically ill patients. A large randomized, controlled prospective trial of albumin versus saline in almost 7000 critically ill patients found no demonstrable effect of one over the other on mortality, renal function, or the frequency of renal replacement therapy.98 Of note, patients with cirrhosis were excluded from this trial, and limited data suggest that albumin is useful to prevent ARF in cirrhotic patients with spontaneous bacterial peritonitis,99 or those undergoing large-volume paracentesis. As discussed above, for prevention of radiocontrast nephropathy normal saline is superior to half-normal saline,39 and an equimolar solution of sodium bicarbonate is superior to normal saline.40 Otherwise, despite the importance of fluid therapy in the prevention of ARF, the nature of the optimum fluid resuscitation regimen remains a disputed topic. Similarly, there are no proven advantages of any specific inotrope or vasopressor with respect to renal function. In patients with septic shock, the use of early goal-directed therapy, steroids in nonresponders to adrenocorticotropic hormone stimulation, and activated protein C may lead to a decreased incidence or severity of ARF, but there is no evidence in the literature that improved survival in the pivotal studies supporting these therapies is associated with a decreased incidence of ARF. Limited evidence suggests that tight glycemic control and lung-protective ventilation strategies may decrease the incidence of ARF in critically ill patients.4,14

Prevention of nephrotoxic injury is the other mainstay of ARF prevention in critically ill patients. Avoidance of potential nephrotoxins such as intravenous radiocontrast, aminoglycosides, amphotericin, and NSAIDs is prudent, when possible. Single daily dosing of aminoglycoside antibiotics is associated with a lower risk of nephrotoxicity and equivalent antimicrobial efficacy compared to multiple dosing strategies.31–33 Drug modifications such as nonionic radiocontrast36,37 and lipid-emulsified amphotericin B may also reduce the incidence of ATN. Aggressive diuresis must be avoided when possible, particularly in conjunction with the use of ACEIs or ARBs, and must be accompanied by careful monitoring of fluid balance and renal function. Monitoring the serum levels of potentially nephrotoxic drugs such as aminoglycosides, cyclosporine A, tacrolimus, and vancomycin is recommended, although studies proving that therapeutic drug monitoring decreases the incidence of ATN are lacking.

Numerous studies have sought to demonstrate efficacy of pharmacologic interventions for primary prevention of ARF (prophylaxis), or secondary prevention in ARF (ATN therapy to decrease dialytic requirement and improve outcome). Broadly speaking, these studies have been trials of diuretics or renal vasodilators. There have been other negative studies of agents including insulin-like growth factor and thyroxine, but these will not be further discussed here.

Diuretics

Furosemide is a loop diuretic and vasodilator that may decrease oxygen consumption in the loop of Henle by inhibiting sodium transport, thus potentially lessening ischemic injury. By increasing urinary flow, it may also reduce intratubular obstruction and backleakage of filtrate. Based on these properties, furosemide might be expected to prevent ARF. However, there are few data to support its use. Furosemide was found to be ineffective or harmful when used to prevent ARF after cardiac surgery, and to increase the risk of ARF when given to prevent contrast nephropathy.38,100

Similarly, there is little evidence of benefit from diuretic therapy in established ARF. A single recent study in 100 patients with oliguric ARF after cardiac surgery suggested that a cocktail infusion of furosemide, mannitol, and dopamine improved renal function post–cardiac surgery compared to intermittent loop diuretics alone.101 However, numerous studies in patients with established ATN have found no benefit of loop diuretics.102–107 Their use did not accelerate renal recovery, decrease the need for dialysis, or reduce mortality. It was shown that the mortality rate of oliguric patients who responded to furosemide with a diuresis was lower than those who did not.106–107 However, the clinical characteristics, severity of renal failure, and mortality rates were similar in patients with either spontaneous non-oliguric ARF or patients who became non-oliguric after furosemide. This implies that those patients able to respond to furosemide have less severe renal damage than nonresponders, rather than deriving any true therapeutic benefit from furosemide administration. Although administration of furosemide might facilitate improved fluid management if it induces a diuresis, a retrospective review of a recent trial in critically ill patients with ATN raised concerns of possible harm from loop diuretics in ARF. The authors found that diuretic use was associated with an increased risk of death and nonrecovery of renal function.108 Most of the increased risk, however, was seen in those patients unresponsive to high doses of diuretics, implying they had more severe disease. Therefore, diuretics should be used with caution in critically ill patients, and iatrogenic hypovolemia and superimposed prerenal azotemia must be avoided. Diuretics should be withdrawn if there is no response, to avoid ototoxicity. Our current maximal, synergistic diuretic challenge combines furosemide 200 mg IV with chlorothiazide 500 mg IV. In patients who experience an increase in urine output, hypotension must be avoided, since kidneys with ATN are susceptible to further damage from decreases in perfusion pressure. To maintain the diuresis, a continuous infusion of drug is probably preferable to intermittent bolus administration.109 Although there are no large randomized controlled trials, the overall evidence suggests that continuous infusion of diuretics as opposed to bolus administration is more effective and associated with less toxicity and delayed development of diuretic resistance.

Mannitol is an osmotic diuretic that can decrease cell swelling, scavenge free radicals, and cause renal vasodilatation by inducing intrarenal prostaglandin production.110 It may be beneficial when added to organ preservation solutions during renal transplantation and may protect against ARF caused by rhabdomyolysis if given extremely early.110–112 Otherwise, mannitol has not been shown to be useful in the prevention of ARF. In fact, mannitol may aggravate ARF due to radiocontrast agents.38 Furthermore, mannitol may precipitate pulmonary edema if given to volume-overloaded patients who remain oliguric, can exacerbate the hyperosmolar state of azotemia, and may even cause acute renal failure (“osmotic nephrosis”).113

Renal Vasodilators

Atrial Natriuretic Peptide

Atrial natriuretic peptide (ANP) has been investigated in clinical trials for the treatment or prevention of ARF. ANP can lead to an increase in GFR due to arteriolar dilatation and also has a natriuretic effect. ANP improved renal function in animal studies and a phase II human ATN study.114–115 However, larger randomized trials failed to show any benefit of ANP compared to placebo.116–117 Hypotension was significantly more common in ANP-treated patients in these studies, and may have negated any potential benefit of renal vasodilation in these patients. Hence there is no convincing evidence to support the use of ANP in the prevention or treatment of ARF. A new, promising, but underpowered (61 patients) positive study of ANP to treat ARF immediately following cardiac surgery showed a decreased rate of postoperative renal replacement therapy compared to placebo-treated patients;118 a larger prospective trial in this setting appears warranted.

Dopamine

Low-dose dopamine administration (1 to 3 μg/kg per minute) to normal individuals causes renal vasodilation and increased GFR, and acts as a proximal tubular diuretic. Due to these effects, numerous studies have used low-dose dopamine to either prevent or treat ATN in a variety of clinical settings. It has been given as prophylaxis for ARF associated with radiocontrast administration, repair of aortic aneurysms, orthotopic liver transplantation, unilateral nephrectomy, renal transplantation, and chemotherapy with interferon.119–122 Yet despite more than 20 years of clinical experience, prevention trials with low-dose dopamine all have been small, inadequately randomized, of limited statistical power, and with end points of questionable clinical significance. Denton and associates reviewed the data on the use of renal doses of dopamine and concluded that there is no convincing data that renal-dose dopamine either prevents ARF or improves outcome in patients with established ARF.120 Meta-analyses done by Kellum and Decker121 and by Marik122 similarly did not show any benefit of dopamine in the prevention or treatment of ARF. The Australian and New Zealand trial (ANZICS) studied the effects of low-dose dopamine in ICU patients with systemic inflammatory response syndrome and ARF.123 This is a double-blind randomized study. The patients were randomly assigned to receive either dopamine at a dose of 2 μg kg−1 min−1 or placebo. The investigators found no difference in the primary outcome measure, which was the peak serum creatinine concentration during trial infusion (Fig. 75-4). The proportion of patients reaching serum creatinine values above 3.4 mg/dL and proportion of patients requiring renal replacement therapy was similar in both groups (Fig. 75-5). There was no difference in the length of ICU and hospital stay, or mortality and time to renal recovery. This study convincingly demonstrated that low-dose dopamine does not improve outcome in critically ill patients with early ARF. Furthermore, there is concern for the potential harmful effects of dopamine, even at low doses. It can trigger tachyarrhythmias and myocardial ischemia, decrease intestinal blood flow, cause digital necrosis and hypothyroidism, and suppress T-cell function.119 It has also been shown to increase the risk of radiocontrast nephropathy when given prophylactically to patients with diabetic nephropathy.49

When a diagnosis of intrinsic ARF has been firmly established and potentially reversible causes have been excluded or treated, the patient is then monitored for early detection of complications. Conservative measures consist of hospital observation with attention to blood pressure, volume status, neurologic function, and evidence of hemorrhagic or infectious complications. Electrolytes (Na+, K+, Cl− , HCO32−, Ca2+, and PO43−), BUN, and creatinine should be monitored daily, although the hyperkalemic patient will require more frequent monitoring.

Fluid intake should be adjusted to replace urine and insensible losses, while Na+ and K+ are allowed in amounts to replace urine and GI losses. Obviously, fluid and electrolyte restriction will be more severe in the oliguric patient. A small daily reduction in weight is expected in the patient with ARF, but weight loss >1 kg daily indicates severe catabolism or volume loss. On the other hand, it should be emphasized that maintenance of weight or weight gain indicates volume expansion.

Clinical trials to determine the benefit of nutritional therapy with amino acids to improve renal failure show conflicting results.124,125 The guiding principle of nutritional management in the nondialyzed case of ARF has traditionally been slowing the accumulation of nitrogen waste products and avoiding severe electrolyte perturbations. We recommend limiting potassium and phosphorus intake to avoid severe electrolyte disturbances, but we do not limit protein intake specifically to limit production of nitrogenous waste and avoid dialysis. Whether a patient with ARF is in the period before or after initiation of renal replacement therapy, protein intake should be prescribed to provide 1.0 to 1.5 g/kg per day (higher in more catabolic patients, and in those receiving CRRT, who have additional amino acid removal). Calories can be provided in the form of simple carbohydrates such as jelly and sugar. Caloric requirements increase in proportion to the severity of the underlying illness, and intake of 3000 kcal/d or more may be necessary to blunt catabolism. Should the patient be incapable of eating due to anorexia or other reasons, a small, soft feeding tube may be inserted into the stomach and tube feeding administered by constant infusion or intermittent bolus. In general, tube feedings with lower K+ and Na+ concentrations are chosen. For the patient who cannot tolerate enteral feeding, total parenteral nutrition (TPN) becomes necessary. Because of the volume of fluid required for TPN, earlier institution of dialysis and ultrafiltration is often necessary. A reasonable approach is to give sufficient calories to prevent negative nitrogen balance but not to overfeed patients to the point that hepatic and platelet complications develop or that dialysis imposes additional risks.

The management of various electrolyte abnormalities often associated with ARF is described in greater detail in Chap. 76 and will only be touched on here. Administration of excessive amounts of free water is the most common cause of hyponatremia in patients with ARF. Less important causes of hyponatremia in this setting include water production from carbohydrate metabolism and water release from injured tissue. Judicious fluid management will prevent any untoward complications from hyponatremia. Hyperkalemia is the most grave of the electrolyte perturbations that may complicate ARF. The clinical consequences of hyperkalemia are largely confined to the cardiovascular and neuromuscular systems. The earliest hyperkalemic effects are manifest in the electrocardiogram (ECG). ECG changes are uniformly present above a potassium level of 8 mEq/L; below 7 mEq/L, changes may not be evident. The treatment of hyperkalemia is summarized in Table 75-9.

Disturbances of divalent ion metabolism are common in ARF. Hyperphosphatemia is an almost universal accompaniment of oliguric ARF. Control of hyperphosphatemia in the acute setting is achieved by oral administration of aluminum hydroxide gels. Short-term administration of aluminum-containing compounds to patients with renal failure does not result in aluminum intoxication. However, concurrent administration of oral citrate compounds must be avoided because citrate enhances gut absorption of aluminum and accelerates the development of aluminum intoxication.

Hypocalcemia is an expected complication of ARF but is generally of no clinical significance and does not require intervention. However, severe depression of serum Ca2 + may complicate rhabdomyolysis-induced ARF. Nevertheless, calcium salts are contraindicated except as part of the treatment of life-threatening hyperkalemia or if hypocalcemic tetany develops. Hypercalcemia may be observed when rhabdomyolysis-induced ARF enters the diuretic phase, and Ca2 + deposited in damaged muscle is released. It is usually self-limited.

Uric acid levels may rise by 1 to 2 mg/dL per day in cases of ARF not due to rhabdomyolysis, but rarely does the level exceed 15 mg/dL. This plateau is thought to result from enhanced metabolism of uric acid by gut bacteria. Extreme elevations in uric acid levels have been observed in cases of exertional rhabdomyolysis and following treatment of some lymphomas.

Metabolic acidosis occurs commonly in ARF, and in cases of trauma or sepsis the fall in plasma bicarbonate may exceed 15 mEq per 24 hours. In the catabolic patient with ARF, acid is released from metabolism of sulfur- and phosphorus-containing compounds, which contributes to a rise in the anion gap. Acidosis may be treated with sodium bicarbonate; however, the additional volume is hazardous in the patient with compromised renal function, the buffering requires increased ventilation, and adverse intracellular effects may result. Uncontrollable acidosis may mandate dialysis therapy.

Metabolic alkalosis may complicate ARF in the patient with nasogastric suction, but is a less frequent acid-base disturbance in these patients. Although nasogastric suction–induced metabolic alkalosis normally responds to saline supplementation, this maneuver is ineffective in ARF and carries the hazard of potential volume overload. Severe metabolic alkalosis in ARF may require dialysis using a reduced concentration of bicarbonate or acetate in the bath. Alternatively, 0.1 N hydrochloric acid may be administered by slow intravenous infusion. Development of metabolic alkalosis in the patient on nasogastric suction may be prevented or attenuated by use of histamine blockers such as cimetidine.

Infection continues to be a leading cause of death in the patient with ARF, and pneumonia, urinary tract infection, and wound infection are most common. Infection has been the cause of death in up to 70% of cases of ARF and has contributed to 50% of deaths occurring during the diuretic phase.126 In several large series of ARF, the incidence of septicemia exceeds 50%, and intra-abdominal sepsis is known to be a particularly important determinant of survival. In the septic patient without an obvious source, intra-abdominal sepsis must be vigorously excluded. Imaging procedures of value for detecting an abdominal source include CT scanning and ultrasound, with CT the better of the two. The severity of infection in ARF is undiminished despite the advent of new antibiotics and other advances in general medical care. Prophylactic antibiotics are not recommended for the patient with ARF and may be harmful.

Prior to the use of antacids and histamine blockers (e.g., cimetidine and ranitidine) to prevent gastric stress ulceration in ARF, GI hemorrhage was the second leading cause of death. Now many centers report a dramatic decline in GI hemorrhage.127 Recent advances in the therapy of uremic bleeding include use of 1-deamino-8-d-arginine vasopressin (DDAVP), cryoprecipitate, and conjugated estrogens.128 These drugs are most useful in patients with clinically significant bleeding episodes who lack evidence of any other (nonuremic) coagulopathy. DDAVP is typically infused over 30 minutes in a dose of 0.3 μg/kg of body weight and is effective within minutes.129 Unfortunately, because of the short duration of action of this compound (1 to 4 hours), frequent dosing may be required. The use of cryoglobulin infusions (10 units IV over 15 minutes) also may serve as a temporary expedient in patients with uremic bleeding. Conjugated estrogens (0.6 mg/kg IV daily for 5 days) have a very delayed onset of action (3 to 5 days), but the duration of the therapeutic effect can be as long as 14 days.130

Renal Replacement Therapy

In ARF, there is consensus that dialysis should be initiated if there are uremic symptoms or signs, or if volume overload, electrolyte abnormalities, or acidosis are refractory to medical management. The precise timing of RRT initiation is usually a matter of clinical judgment.3,5 Absent the above clinical indications, hemodialysis (HD) is commonly initiated when the BUN concentration reaches 100 mg/dL, and repeated to maintain a predialysis BUN below 80 mg/dL.5 This pattern of practice is based primarily on early experience suggesting that uremic bleeding diathesis and hemorrhage were reduced when hemodialysis was initiated before the BUN exceeded 100 mg/dL,5 the results of small prospective ARF studies in 1975 and 1986,131 and extrapolation from results of the National Cooperative Dialysis Study in chronic renal failure patients. This relatively prophylactic approach is balanced against potential adverse effects of RRT, including hypotension with worsening renal ischemic acute tubular necrosis, and the renal and systemic inflammatory effects of extracorporeal membrane exposure. Although it seems prudent to start RRT aggressively in critically ill catabolic patients, studies have not shown a significant benefit of prophylactic dialysis, including a recent small study using CRRT by Bouman and colleagues.132 It is universally accepted that emergent RRT is indicated in the management of ARF when pulmonary edema, hyperkalemia, refractory acidosis, or symptomatic uremia develops. In our opinion, initiation of RRT should also be considered in patients with sustained anuria, persistent oliguria with progressive azotemia and GFR <10 mL/min, and to control fluid balance, rather than waiting for complications of volume overload, electrolyte imbalance, or uremia to develop.

RRT modalities available include intermittent hemodialysis (IHD), peritoneal dialysis (PD), and continuous renal replacement therapy (CRRT). CRRT nomenclature includes elements denoting duration of therapy (C for continuous), vascular access (arteriovenous [AV], or venovenous [VV]), and the solute transport mechanism(s) used (diffusion [dialysis], convection [hemofiltration], or both [hemodiafiltration]). So the different modalities include continuous arteriovenous hemofiltration (CAVH), continuous venovenous hemofiltration (CVVH), continuous arteriovenous and venovenous hemodialysis (CAVHD and CVVHD), and continuous arteriovenous and venovenous hemodiafiltration (CAVHDF and CVVHDF). Solute clearance during RRT occurs either by diffusion or convection. Solute clearance during IHD and continuous dialysis therapies (CAVHD and CVVHD) occurs primarily by diffusion. Solute clearance during CVVH and CAVH is primarily by convection. In CAVHDF and CVVHDF both convective and diffusive solute transport occurs.

CRRT is thought to permit fluid and solute removal with greater hemodynamic stability than intermittent HD for two reasons. The first is the fact that fluid removal is done slowly and continuously, rather than in 1- to 4-L increments over 2- to 4-hour treatment sessions. Rapid diffusive solute clearance by intermittent HD results in loss of intravascular solutes and results in the reduction of plasma osmolality relative to surrounding tissues, which in turn causes transfer of plasma water into tissues. This leads to further depletion of extracellular volume which has been diminished by ultrafiltration.10 So, in hemodynamically unstable patients CRRT is preferred compared to IHD, as it allows removal of excess fluid in hypotensive patients. CRRT also allows us to better manage fluid status in patients who are receiving TPN. Apart from hypovolemia and plasma hypo-osmolality, there are a variety of other causes of intradialytic hypotension (IDH) (Tables 75-10 and 75-11).10 Although IDH is common in ICU patients, the complete differential diagnosis of this problem should be considered in every case, particularly when hypotension is profound or refractory to standard measures. Several techniques are used routinely in ICU patients receiving IHD to improve hemodynamic stability. Intermittent isolated ultrafiltration (also called IUF) sessions, which remove fluid without diffusive solute clearance, are commonly used for acute volume removal with greater hemodynamic stability than intermittent HD. The use of higher dialysate sodium concentrations, called sodium modeling (e.g., beginning HD with a 160-mEq/L sodium “bath,” decreasing to 140 mEq/L over 4 hours), similarly improves hemodynamic tolerance of intermittent HD, by ameliorating the tendency towards plasma hypo-osmolality. The use of cool dialysate (35.5°C) and increasing dialysate calcium concentration are among the other interventions proven to increase systemic vascular resistance and improve hemodynamic tolerance of HD; increased dialysate calcium also augments cardiac output. Numerous patients hemodynamically intolerant of isolated ultrafiltration, intermittent HD, their sequential combination, and the other modifications mentioned above, have been supported with CRRT. The capacity to alter fluid balance at will, even in hemodynamically unstable patients, is in large part responsible for the attractiveness of CRRT to intensivists.

Although acute therapy of severe hyperkalemia, metabolic acidosis, or intoxications is more efficiently achieved with IHD, lesser abnormalities are corrected relatively quickly and controlled effectively with CRRT . Clinical data suggest that CRRT should be strongly considered for patients with severe hyperphosphatemia or elevated intracranial pressure (ICP), and might also be a useful component of therapy for lithium intoxication.5 Continuous removal prevents the postdialytic rebound elevation of plasma concentration typically seen with IHD in patients with hyperphosphatemia (tumor lysis syndrome or rhabdomyolysis) or lithium intoxication.5 In patients with cerebral edema complicating acute liver failure, intermittent machine hemofiltration, but not continuous renal replacement therapies (CAVH or CAVHD) raised ICP and decreased cerebral perfusion pressure; similar adverse effects of intermittent HD were previously demonstrated.5 Raised ICP in this setting is due to acute solute removal and resulting plasma hypo-osmolality, causing a shift of water into the brain along with other tissues. The detrimental effect of this phenomenon on cerebral perfusion pressure may be increased by treatment-induced hypovolemia or hypotension. Despite apparent advantages over intermittent therapies in unstable patients, outcome superiority of CRRT with respect to mortality or recovery of renal function outcomes has not been demonstrated. Data from a U.S. prospective, comparative trial of intermittent HD versus CRRT in ICU ARF did not demonstrate an effect of modality on outcome.133 In this study, Mehta and colleagues excluded patients with a mean arterial pressure below 70 mm Hg. Randomization did not balance severity of illness between groups; CRRT patients had higher severity of illness and mortality, and there was no difference in mortality between HD and CRRT groups adjusted for Acute Physiology, Age, and Chronic Health Evaluation (APACHE) II scores. Ongoing studies may determine if there is an advantage of any specific modality over others in critically ill patients.

Few RRT dosing guidelines have been established in patients with ARF, but it is clear that dialysis delivery is inadequate in the ICU, even by standards set for stable end-stage renal disease outpatients. Clearance of urea, a low-molecular-weight nitrogenous waste product, is the most commonly studied marker of adequate uremic detoxification by HD. The term Kt/V is a unitless measure of HD dose (based on urea removal): K is the urea clearance of the dialysis membrane used (in milliliters per minute), t is the duration of dialysis (in minutes), and V is the volume of distribution of urea in the patient (in milliliters). Thus, Kt/V is a measure of the volume of plasma cleared of urea during a HD session (Kt) divided by the urea distribution volume (V, assumed to be total body water: 0.5 to 0.6 L/kg), and larger Kt/V values signify greater HD dose. Kt/V measurements of delivered dialysis dose are usually calculated using the ratio of postdialysis to predialysis BUN and a nomogram. In chronic hemodialysis patients the minimum acceptable dialysis dose is a Kt/V of 1.2. Extrapolation of HD data from the chronic HD population to the ARF setting is limited by numerous poorly quantifiable variables, including effects of critical illness on urea kinetic modeling. Furthermore, Kt/V measurements extrapolated from pre- and post-BUN may be very inaccurate in ARF, because rebound postdialysis increases in BUN from tissue stores hypoperfused during IHD occurs. Furthermore, numerous patient- and technique-related factors impede effective delivery of prescribed dialysis in critically ill patients. In a prospective study which examined the prescribed and delivered dialysis dose in 40 ARF patients found that 68% of ARF treatments failed to deliver the minimum dialysis dose recommended for maintenance HD of chronic renal failure patients, and that for 49% of treatments a minimally adequate dialysis dose was not even prescribed.134 Failure to adjust dialysis dose and/or achieve adequate dialysis delivery in larger patients was a major source of treatment failure.

In a study done by Paganini and colleagues, improved survival was seen in patients with higher Kt/V, and this survival benefit is particularly evident in patients with moderate-range severity of illness. In patients with least and most severity of illness the mortality was not affected by dialysis dose (Fig. 75-6).135 Another study done by Schiffl and colleagues found that patients who received daily IHD had a lower mortality compared with patients who received alternate-day IHD (28% vs. 46%), and improved renal recovery.136 In patients who received alternate-day IHD, time-averaged blood urea nitrogen level was 104 mg/dL, reflecting inadequate dialysis that is below the standard of care for most critically ill patients, so it is not surprising that their outcome was inferior to well-dialyzed patients. Nonetheless, it seems prudent to prescribe more frequent HD for ICU patients treated with intermittent dialysis, to optimize dose delivery, fluid balance, and outcome. There are also outcome data to guide CRRT dosing, provided by the landmark study of Ronco and colleagues, who conducted a large prospective study to determine the impact of hemofiltration dose on survival. In this study, patients were randomly assigned to one of three treatment doses, based on effluent volume (which is equivalent to small solute clearance by continuous hemofiltration, or by low-volume continuous hemodialysis): 20 mL h−1 kg−1 (group 1), 35 mL h−1 kg−1 (group 2), or 45 mL h−1 kg−1 (group 3). Patients assigned to group 1 had higher mortality than those in groups 2 and 3 (Fig. 75-7).137 Based on this study, we recommend a minimum effluent volume of 35 mL/kg per hour for ICU patients with ARF treated with CRRT. Current and future studies may identify alternative markers of adequate RRT dose, such as clearance of larger molecules, or biomarkers such as leukocyte response to inflammatory stimuli.59