Drug Experience
Drug Safety 5 (2): 94-108, 1990 0114-5916/90/0003-0094/$07.50/0 © ADiS Press Limited All rights reserved. MEDT03154
Amphotericin B Nephrotoxicity Ramzi Sabra and Robert A. Branch Division of Clinical Pharmacology, Departments of Pharmacology and Medicine, Vanderbilt University School of Medicine, Nashville, Tennessee, USA
Contents
Summary ....................................................................................................................................... 94 1. Clinical Manifestations of Nephrotoxicity ............................................................................. 95 1.1 Azotaemia ........................................................................................................................... 95 1.2 Urinary Concentrating Ability .......................................................................................... 96 1.3 Hypokalaemia and Hypomagnesaemia ............................................................................ 97 1.4 Renal Tubular Acidosis ..................................................................................................... 97 1.5 Pathological Findings ......................................................................................................... 97 2. Mechanisms of Nephrotoxicity ............................................................................................... 97 2.1 Effect on Membrane Sterols ............................................................................................. 98 2.2 Physiological Effects .......................................................................................................... 99 2.2.1 Short Term Effects .................................................................................................... 99 2.2.2 Long Term Effects .................................................................................................. 102 3. Measures to Reduce Nephrotoxicity..................................................................................... 103 3.1 Mannitol Administration ................................................................................................ 103 3.2 Other Formulations of Amphotericin B ........................................................................ 104 3.3 Salt Supplementation ....................................................................................................... 104 3.3.1 Case Reports ............................................................................................................ 104 3.3.2 Retrospective Study ................................................................................................ lOS 3.3.3 Prospective Study.................................................................................................... lOS 4. Recommendations for Minimising Nephrotoxicity............................................................. 105
Summary
The frequency of fungal infections is increasing. Amphotericin B remains the antifungal drug of choice for most systemic infections, but a limiting factor for its use is the development of nephrotoxicity. Amphotericin B-induced nephrotoxicity is manifested as azotaemia, renal tubular acidosis, impaired renal concentrating ability and electrolyte abnormalities like hypokalaemia and sodium and magnesium wasting. All these abnormalities occur to varying degrees in almost all patients receiving the drug. Upon withdrawal oftherapy renal function gradually returns to baseline, although in some instances permanent damage is sustained, especially when the cumulative dose exceeds 5g. Salt depletion enhances the development of nephrotoxicity. The mechanism of nephrotoxicity involves direct cell membrane actions to increase permeability, as well as indirect effects secondary to activation of intrarenal mechanisms (tubuloglomerular feedback) and/or release of mediators (thromboxane A2). The latter effects are presumably responsible for the observed acute decreases in renal blood flow and filtration rate, responses that are inhibited by several physiological and pharmacological interventions. Changes in intracellular calcium levels may also contribute to the observed effects.
Amphotericin B Nephrotoxicity
95
In the clinical situation, and in long term models of nephrotoxicity in the rat, salt loading protects against deterioration in renal function; recommendations are made for the optimisation of amphotericin B therapy by salt loading. New preparations of the drug, such as Iiposomal amphotericin B, may also prove useful in minimising nephrotoxicity while maintaining antifungal activity, but further research is needed with both salt loading and liposomal amphotericin B to confirm or deny their protective effect on kidney function.
The past few years have seen an increasing incidence of mycotic diseases in the USA. Fungal organisms have been isolated from I to 12% of hospitalised patients, and now account for approximately 5% of all cases of primary septicaemia (McGowan 1985). The main reason for this rising incidence of mycotic diseases is the higher number of hospitalised patients with risk factors for opportunistic and fungal infections, i.e. compromised immune defence mechanisms. Foremost among the diseases responsible is the acquired immunodeficiency syndrome (AIDS); other factors include such therapeutic manoeuvres as cancer chemotherapy, radiotherapy and iatrogenic immunosuppression after organ transplantation. After more than 30 years of clinical use, and despite the introduction of several newer antifungal agents, amphotericin B remains the antimicrobial agent of choice for the treatment of most systemic fungal infections. It continues to have potent fungicidal and fungistatic activities against many strains of fungi, few of which have developed resistance to it. The formidable effectiveness of the drug is tempered by the variety and frequency of adverse effects it causes, which include chills, fever, headache, nausea, vomiting, phlebitis, anaemia, thrombocytopenia, generalised pains, convulsions and renal toxicity. Nephrotoxicity is manifested as azotaem.ia, renal tubular acidosis and decreased urine concentrating ability, as well as electrolyte disturbances such as urinary potassium wasting and hypokalaemia, and magnesium wasting and hypomagnesaemia. The most limiting of these side effects, and a reportedly common one, is renal toxicity. In general, the other disturbances are easily mana~eable and do not pose as important a threat
to the continued administration of the drug as does nephrotoxicity. Deterioration of renal function may be so severe as to necessitate premature or temporary discontinuation of drug therapy, or a lowering of dose, leading to progression of the disease I • and a protracted hOSPItal stay. This report aims to review the nature of amphotericin B nephrotoxicity, with special emphasis on the recent advances in understanding the mechanisms involved, the role of renal autacoids and homeostatic mechanisms and their implications for clinical practice. Recommendations are offered for minimising the risk of amphotericin B-induced nephrotoxicity.
1. Clinical Manifestations of Nephrotoxicity Amphotericin B nephrotoxicity is manifested as disturbances in both glomerular and tubular function. Azotaemia of varying degrees occurs in almost all patients treated with the drug, and is frequently accompanied by mild renal tubular acidosis and impairment in urinary concentrating ability. Renal sodium and potassium wasting with hypokalaemia are fairly common occurrences, sometimes associated with hypomagnesaemia. 1.1 Azotaemia
An increase in blood urea nitrogen (BUN) and creatinine levels has been reported to occur in over 80% of patients receiving amphotericin B. In a survey of 56 patients conducted between 1956 and 1963, at the time of the introduction of this agent, 93% of patients taking amphotericin B developed values of BUN exceeding 200 mg/L, and 83% had
96
serum creatinine levels greater than 15 mg/L (Butler et at. 1964a). This high incidence of azotaemia has also been reported in other studies (Burgess & Birchall 1972; Holeman & Einstein 1963). A more recent report indicates that in almost every patient treated the glomerular filtration rate (GFR) falls approximately 40% within the first 2 to 3 weeks of therapy, stabilises at 20 to 60% of normal, and remains at this level for the remainder of the treatment (Medoff & Kobayashi 1980). Renal blood flow is also reported to be reduced (Bell et at. 1962; Burgess & Birchall 1972; Sanford et al. 1962). Whereas GFR tends to return to baseline I to 2 months after stopping therapy, renal blood flow requires a much longer time. A severe insult may occur in some patients, raising the serum creatinine concentration to values greater than 35 mg/L. In such cases a brief cessation of therapy (2 to 5 days) is warranted, and usually leads to an improvement in renal function, although in some cases dysfunction may be permanent. The previously administered dose can then be reinstated after due consideration of the clinical status of the patient, the level of infection remaining and alternative therapeutic options. The relationship between the extent of permanent deterioration in renal function and the total cumulative dose employed is not completely settled. Initial reports indicated a significant correlation between the net rise in serum creatinine levels and the total dose of amphotericin B administered (Butler et at. I 964a). This was later challenged by Miller and Bates (1969) who failed to observe such a relationship, and suggested instead that renal toxicity is an individualised and unpredictable response. Burgess and Birchall (1972) were also unable to delineate a dose-response relationship, although none of their patients received an amphotericin B dose greater than 2g. Experience at Vanderbilt University Medical Center also argues against a dose-dependent nephrotoxic response (Branch et al. 1987; Heidemann et al. 1983b). Patients have been observed to develop azotaemia at doses ranging from 100mg to 1.5g, with no significant increase in frequency as the cumulative dose increased. As a general statement, it has been
Drug Safety 5 (2) 1990
proposed that with total cumulative doses less than 5g, the risk of permanent kidney damage is lower than with doses exceeding that amount (Butler et at. 1964; Winn 1963). In these series, 58 and 88%, respectively, of patients receiving more than 5g of amphotericin B had persistent renal impairment. The frequency of dosing as a risk factor for nephrotoxicity has also been addressed, with several studies showing that administration of the drug on alternate days lowers the incidence compared with the same total dose given on a daily basis (Littman et al. 1958; Rubin et at. 1989). Age has been suggested as a risk factor for development of azotaemia (Miller & Bates 1969). A recent study seems to support this contention, although it is complicated by the fact that all patients reported were receiving aminoglycoside antibiotics in addition to amphotericin B (Stein et at. 1988). An important risk factor for the development of azotaemia in amphotericin B-treated patients is their state of hydration. Several reports have appeared in the last few years addressing this question (Branch 1988; Branch et at. 1987; Heidemann et at. 1983b; Stein et at. 1988). This subject is covered in more detail in section 3, but for the present suffice it to say that salt depletion appears to potentiate amphotericin B nephrotoxicity while salt loading protects against it. 1.2 Urinary Concentrating Ability
In an early study, it was shown that amphotericin B could induce a loss of concentrating ability of the kidney which was apparent as early as 2 weeks after the start of therapy, and lasted for the remainder of the treatment (Holeman & Einstein 1963). This abnormality occurred in all patients examined in that study. However, 3 months after therapy was discontinued, the kidneys regained their ability to concentrate urine. Similar results were reported by Butler and co-workers (1964a) and by Bullock et al. (1976). Barbour et al. (1979) reported 3 cases of patients who developed an inability to concentrate the urine due to a defect in free water reabsorption even under maximal stimuli, indicating a failure in the vasopressin response
97
Amphotericin B Nephrotoxicity
of the medullary collecting tubule. This disturbance in concentrating ability frequently precedes the fall in GFR (Beard et al. 1960), and therefore is unlikely to be explained by it.
1.3 Hypokalaemia and Hypomagnesaemia Hypokalaemia is a common and characteristic complication associated with amphotericin B treatment, occurring in approximately 75% of patients (Burgess & Birchall 1972). Supplements of potassium chloride are frequently necessary to maintain normal levels of potassium. A dose-dependent response has been proposed by some investigators, but the mechanisms by which urinary potassium wasting takes place are not clear. It is most probably due to direct toxic effects on distal tubules which increase their permeability for potassium (Butler 1966); alternatively, this could be a result of activation of sodium/potassium exchange in the distal tubule. Magnesium wasting may also occur with amphotericin therapy (Barton et al. 1984; Douglas & Healy 1969). In the study by Barton et al. the lowest serum levels and the largest fractional excretion of magnesium were observed by the fourth week of therapy, after a cumulative dose of approximately 500mg. These changes may occur, however, as early as the first 2 weeks of therapy. Despite the continuation of the drug, no further magnesium wasting occurred, suggesting a 'plateau-ing' of the effect. This abnormality, however, was fully reversible, evidenced by the normal serum and urinary magnesium levels measured I year after discontinuation of therapy. In the event of hypomagnesaemia occurring in association with hypokalaemia, correction of the former abnormality is necessary before potassium replacement can restore normal potassium levels. 1.4 Renal Tubular Acidosis This is another common complication of the use of amphotericin B (Douglas & Healy 1969; McCurdy et al. 1968). McCurdy et al. studied urinary acid secretion in 7 patients with amphotericin B
nephropathy. Five had defects in acid secretion similar to distal renal tubular acidosis. The impaired titratable acid excretion seen was greater than can be accounted for by depression of glomerular filtration, and suggests a tubular defect (Gouge & Andriole 1971; McCurdy et al. 1964). It also occurs in the absence of hypokalaemia, which itself may limit the minimal urinary pH achieved. Therefore, this defect also appears to be a specific tubular effect of amphotericin B. 1.5 Pathological Findings
As suggested by the above clinical findings, there are 2 aspects to amphotericin B nephrotoxicity: glomerular and tub~lar. Pathological examination of kidney biopsy specimens and postmortem examination of kidneys of patients who are or were receiving amphotericin B treatment do not reveal a consistent pattern of change. In general, changes can be described as diffuse and nonspecific. Tubular changes include fragmentation and thickening of basement membranes, necrosis of distal tubular epithelium, nephrocalcinosis, and vacuolisation of distal tubular cells (Butler et al. 1964a; Holeman & Einstein 1963; McCurdy et al. 1968; Takacs et al. 1963). Glomerular changes include hyalinisation and fibrosis of the basement membrane, hypercellularity, and vacuolisation of the medial smooth muscle of glomerular arterioles (Bullock et al. 1976; McCurdy et al. 1968). Analysis of the urinary sediment may (rarely) reveal haematuria, pyuria, tubular cell, casts and (very rarely) proteinuria.
2. Mechanisms of Nephrotoxicity The exact mechanisms involved in amphotericin B-induced nephrotoxicity have not been clearly defined. The polyene antibiotics, including amphotericin B, have been known for over 2 decades to interact with sterols found in cell membranes. This induces the formation of pores within the membrane, resulting in leakage of ions and cellular constituents (Andriole 1973). In addition, administration of the drug is associated with severe vaso-
98
Drug Safety 5 (2) 1990
constriction and a decrease in renal blood flow, which can result in ischaemic injury to the kidney and reduction in GFR. Evidence suggests that this haemodynamic response occurs secondary to the activation of renal homeostatic mechanisms, or release of renal vasoconstrictor autacoids (see section 2.2). Such mechanisms may at least in part explain the nephrotoxicity of the drug. The relative contribution of the direct cellular effects of amphotericin B and its activation of intrarenal mechanisms has not been determined. Most of the emphasis in this section is centred on the recent findings regarding the role of indirect mechanisms in amphotericin B-induced reductions in the GFR. Readers who wish to have a more detailed description of the 'sterol hypothesis' of amphotericin B action are referred to several excellent reviews (Andriole 1973; Kerridge 1979; Medoff & Koyabashi 1980; Medoff et al. 1983; Norman 1976). 2.1 Effect on Membrane Sterols Amphotericin B is a polyene antibiotic containing a hydrophilic region, made up of an hydroxylated hydrocarbon chain, and a sequence of conjugated double bonds which is lipophilic (fig. I). This unique structure allows for the incorporation of the drug into cellular membranes. A large body
of literature suggests that this interaction with the cell membrane, specifically with sterol moieties in it, is the basis of the antifungal, and possibly also the toxic, effects of the drug. In the early I 960s, several studies demonstrated that polyene antibiotics induced changes in cellular permeability that resulted in the leakage of important cellular constituents, followed by lysis and death (Caltrider & Gottlieb 1961; Kinsky 196Ia,b; Marini et al. 1961; Sutton et al. 1961). It was also discovered that the toxic effect of the drug on cells was dependent on the presence of sterols in the cell membranes, and that addition of sterols to the growth media of certain fungi prevented the polyene-induced inhibition of growth and permeability changes (Caltrider & Gottlieb 1961; Gottlieb et al. 1958; Kotler-Brajtburg et al. 1974). The increased permeability has been documented in both artificial and natural membranes (Andriole 1973). In the case of amphotericin B, the membranes containing ergosterol were found to be more sensitive to the antibiotic than those containing cholesterol, which accounted for its relatively lower toxicity to mammalian cells compared with other polyenes (Kotler-Brajtburg et al. 1974; Teerlink et al. 1980). Nevertheless, the interaction with cholesterol is of such a degree as to lead to the formation of aqueous pores in the membrane. These consist of an annulus of polyene and sterol OH
Fig. 1. Chemical structure of amphotericin B.
Amphotericin B Nephrotoxicity
in which the hydrophilic region of the drug molecule faces the interior of the pore (Andreoli 1973; DeKruijiff & Demel 1974; Holz 1974). Amphotericin B increases the permeability of the toad urinary bladder to urea, potassium and chloride (Gatzy et aI. 1979; Lichtenstein & Leaf 1965, 1966), of erythrocyte and liposomes to potassium (Butler et al. 1965; Teerlink et al. 1980), and of the erythrocyte to sodium and chloride (Kinsky 1963; Kinsky et al. 1962). It also induces changes in permeability to hydrogen of the turtle bladder (Finn et al. 1977; Steinmetz & Lawson 1970) and of purified renal brush border membrane vesicles (Capasso et al. 1986; Schell et al. 1989). Considering the hypokalaemia, renal sodium wasting and renal tubular acidosis that are observed in clinical practice with this drug, it is reasonable to suggest that part or all of these effects may be explained by a direct effect on the transport processes in the renal tubules. In fact, administration of amphotericin B in vivo results in binding to sterols in most tissues, with the highest levels documented in the kidney (Hoeprich 1978). Moreover, binding of amphotericin B to the luminal membrane of the renal tubular cells appears to be necessary for the nephrotoxic effect on these cells (Bolard et al. 1980). In agreement with the proposed tubular effects of the drug, Cheng et al. (1982) demonstrated an increase in tubular permeability to inulin in vivo in rats receiving an acute infusion of amphotericin B, resulting in a back-diffusion of inulin. 2.2 Physiological Effects
2.2.1 Short Term Effects Infusions of amphotericin B, either intravenously or directly into the renal artery, induce short term reductions in renal blood flow and GFR, and an increase in renal vascular resistance, in rats and dogs (Butler et al. 1964b; Gerkens et al. 1983; Heidemann et al. 1983a). The exact mechanisms mediating these changes have not been fully elucidated. In theory, I of 2 mechanisms is possible. The drug can either act directly on the vascular smooth muscle to induce contraction, or it may act via release of other vasoconstrictor agents. Evi-
99
dence in the literature suggests that the latter may be an important mechanism. Most of the support for this hypothesis comes from studies which demonstrated inhibition of the short term renal effects of amphotericin B by a variety of physiological and pharmacological interventions. These studies also led to the suggestion that the renal homeostatic mechanism known as tubuloglomerular feedback (TGF) may playa role in amphotericin B nephrotoxicity. TGF was discovered in the early 1970s through single nephron microperfusion experiments (Schnermann 1975). It is a regulatory mechanism that contributes to the integration of proximal and distal tubular function. In brief, an increased delivery of sodium chloride to the distal nephron, specifically to the region of the macula densa cells, results in the activation of TGF, leading to release of agents that induce afferent arteriolar constriction and a decrease in single nephron GFR. It has been suggested that the transport of sodium chloride across the macula densa cells is what triggers the activation of TGF (Wright & Schnermann 1974). The effector mechanism is thought to be the release of adenosine which, in the kidney, is a vasoconstrictor (Osswald et al. 1980, 1982). The resultant decrease in GFR will reduce the rate of delivery, and consequently the concentration of sodium chloride at the macula densa, thus achieving a new steady-state. In normal circumstances, therefore, TGF is a protective mechanism for conserving electrolytes, which may be lifesaving in the shock state (Thurau & Boylan 1976). Its contribution to renal function in pathophysiological states is still controversial, but worth discussing in the context of amphotericin B nephrotoxicity. It has already been established that amphotericin B can cause changes in tubular cell permeability both in vivo and in vitro. Such changes might activate TGF in I of 2 ways. Impairment of sodium reabsorption in the proximal renal tubule can increase the delivery of sodium and chloride ions to the distal tubule, which has been shown to activate TGF. Alternatively, a direct action of amphotericin B on the distal tubular cells may lead to an increased permeability to sodium chloride in
100
that area, thus facilitating entry of these ions into the cells. In both cases a functional reduction in GFR is achieved. Several physIological and pharmacological interventions can influence the activity of TGF. Salt depletion potentiates the fall in GFR induced by hyperperfusion of the distal nephron, while salt loading attenuates it (Thurau 1975). High ceiling diuretics like furosemide (frusemide) and uncoupiers of oxidative phosphorylation like cyanide also inhibit the response, presumably by interfering with the signal, i.e. chloride reabsorption in the ascending limb of the loop of Henle (Wright & Schnermann 1974). Finally, adenosine receptor antagonists such as xanthines inhibit TGF by interfering with the efferent limb of the reflex, while inhibitors of adenosine uptake or metabolism potentiate it (Osswald et al. 1982; Schnermann et al. 1977). If TGF is indeed involved in the acute renal effects of amphotericin B, then inhibition of the former should alleviate the impairment in renal function due to the latter. Support for a role for TGF in amphotericin B-induced nephrotoxicity is derived from studies which specifically addressed this question. Administration of the drug to saltdepleted dogs resulted in reductions in renal blood flow and GFR, as expected. These changes were inhibited in salt-loaded dogs, or in salt-depleted dogs given an acute salt load (Gerkens & Branch 1980). That same study showed that continuous administration of furosemide to dogs with adequate replacement of urinary losses also inhibited the effects of amphotericin B. Finally, aminophylline was shown to prevent the fall in renal blood flow and GFR due to amphotericin B, an effect which could not be reproduced with nitroprusside, suggesting that it was not due to nonspecific vasodilation (Gerkens et al. 1983; Heidemann et al. 1983a). In a recent study conducted in this laboratory, the authors examined the effect of short term infusions of amphotericin B on the renal microcirculation in rats. The results revealed that the drug decreased both the total renal blood flow and GFR. Micropuncture measurements showed that the single nephron GFR was decreased by 2 mechan-
Drug Safety 5 (2) 1990
isms. There was a decrease in single nephron plasma flow, brought about mainly by vasoconstriction of the afferent arteriole, but also of the efferent arteriole (table I). This was associated with a significant reduction in the glomerular capillary ultrafiltration coefficient (kr), an effect which is currently thought to be mediated by mesangial cell contraction (Schlondortf 1987). When these results are compared with the pattern seen with TGF activation (Ichikawa 1982) the similarities are apparent, supporting the TGF hypothesis of amphotericin B nephrotoxicity. Recent unpublished studies from this laboratory, however, provide evidence contrary to the above. Since most of the studies on TGF were performed in rats, while most of the amphotericin B results mentioned above were reported in dogs, the present authors decided to re-examine the influence of these various interventions on amphotericin B nephrotoxicity in rats. The demonstration of similar results would in that case lend greater support to the hypothesis. The results, however, revealed that in the model used (the in situ renal perfusion model) neither salt loading nor salt depletion influenced the response (table II). Furthermore, the inhibitory effect of aminophylline was limited to the renal blood flow response, while the decrease in GFR was unaffected. This effect of
Table I. Effects of amphotericin B (0.05 mg/kg/min) on various parameters (after Sabra et al. 1989) Parameter
Before
After
p Value
SNGFR (nl/min)
35.3 ± 2.2
22.8 ± 2.8
< 0.0005
QA (nl/min)
142 ± 12
89 ± 14
Drug Safety 5 (2): 94-108, 1990 0114-5916/90/0003-0094/$07.50/0 © ADiS Press Limited All rights reserved. MEDT03154
Amphotericin B Nephrotoxicity Ramzi Sabra and Robert A. Branch Division of Clinical Pharmacology, Departments of Pharmacology and Medicine, Vanderbilt University School of Medicine, Nashville, Tennessee, USA
Contents
Summary ....................................................................................................................................... 94 1. Clinical Manifestations of Nephrotoxicity ............................................................................. 95 1.1 Azotaemia ........................................................................................................................... 95 1.2 Urinary Concentrating Ability .......................................................................................... 96 1.3 Hypokalaemia and Hypomagnesaemia ............................................................................ 97 1.4 Renal Tubular Acidosis ..................................................................................................... 97 1.5 Pathological Findings ......................................................................................................... 97 2. Mechanisms of Nephrotoxicity ............................................................................................... 97 2.1 Effect on Membrane Sterols ............................................................................................. 98 2.2 Physiological Effects .......................................................................................................... 99 2.2.1 Short Term Effects .................................................................................................... 99 2.2.2 Long Term Effects .................................................................................................. 102 3. Measures to Reduce Nephrotoxicity..................................................................................... 103 3.1 Mannitol Administration ................................................................................................ 103 3.2 Other Formulations of Amphotericin B ........................................................................ 104 3.3 Salt Supplementation ....................................................................................................... 104 3.3.1 Case Reports ............................................................................................................ 104 3.3.2 Retrospective Study ................................................................................................ lOS 3.3.3 Prospective Study.................................................................................................... lOS 4. Recommendations for Minimising Nephrotoxicity............................................................. 105
Summary
The frequency of fungal infections is increasing. Amphotericin B remains the antifungal drug of choice for most systemic infections, but a limiting factor for its use is the development of nephrotoxicity. Amphotericin B-induced nephrotoxicity is manifested as azotaemia, renal tubular acidosis, impaired renal concentrating ability and electrolyte abnormalities like hypokalaemia and sodium and magnesium wasting. All these abnormalities occur to varying degrees in almost all patients receiving the drug. Upon withdrawal oftherapy renal function gradually returns to baseline, although in some instances permanent damage is sustained, especially when the cumulative dose exceeds 5g. Salt depletion enhances the development of nephrotoxicity. The mechanism of nephrotoxicity involves direct cell membrane actions to increase permeability, as well as indirect effects secondary to activation of intrarenal mechanisms (tubuloglomerular feedback) and/or release of mediators (thromboxane A2). The latter effects are presumably responsible for the observed acute decreases in renal blood flow and filtration rate, responses that are inhibited by several physiological and pharmacological interventions. Changes in intracellular calcium levels may also contribute to the observed effects.
Amphotericin B Nephrotoxicity
95
In the clinical situation, and in long term models of nephrotoxicity in the rat, salt loading protects against deterioration in renal function; recommendations are made for the optimisation of amphotericin B therapy by salt loading. New preparations of the drug, such as Iiposomal amphotericin B, may also prove useful in minimising nephrotoxicity while maintaining antifungal activity, but further research is needed with both salt loading and liposomal amphotericin B to confirm or deny their protective effect on kidney function.
The past few years have seen an increasing incidence of mycotic diseases in the USA. Fungal organisms have been isolated from I to 12% of hospitalised patients, and now account for approximately 5% of all cases of primary septicaemia (McGowan 1985). The main reason for this rising incidence of mycotic diseases is the higher number of hospitalised patients with risk factors for opportunistic and fungal infections, i.e. compromised immune defence mechanisms. Foremost among the diseases responsible is the acquired immunodeficiency syndrome (AIDS); other factors include such therapeutic manoeuvres as cancer chemotherapy, radiotherapy and iatrogenic immunosuppression after organ transplantation. After more than 30 years of clinical use, and despite the introduction of several newer antifungal agents, amphotericin B remains the antimicrobial agent of choice for the treatment of most systemic fungal infections. It continues to have potent fungicidal and fungistatic activities against many strains of fungi, few of which have developed resistance to it. The formidable effectiveness of the drug is tempered by the variety and frequency of adverse effects it causes, which include chills, fever, headache, nausea, vomiting, phlebitis, anaemia, thrombocytopenia, generalised pains, convulsions and renal toxicity. Nephrotoxicity is manifested as azotaem.ia, renal tubular acidosis and decreased urine concentrating ability, as well as electrolyte disturbances such as urinary potassium wasting and hypokalaemia, and magnesium wasting and hypomagnesaemia. The most limiting of these side effects, and a reportedly common one, is renal toxicity. In general, the other disturbances are easily mana~eable and do not pose as important a threat
to the continued administration of the drug as does nephrotoxicity. Deterioration of renal function may be so severe as to necessitate premature or temporary discontinuation of drug therapy, or a lowering of dose, leading to progression of the disease I • and a protracted hOSPItal stay. This report aims to review the nature of amphotericin B nephrotoxicity, with special emphasis on the recent advances in understanding the mechanisms involved, the role of renal autacoids and homeostatic mechanisms and their implications for clinical practice. Recommendations are offered for minimising the risk of amphotericin B-induced nephrotoxicity.
1. Clinical Manifestations of Nephrotoxicity Amphotericin B nephrotoxicity is manifested as disturbances in both glomerular and tubular function. Azotaemia of varying degrees occurs in almost all patients treated with the drug, and is frequently accompanied by mild renal tubular acidosis and impairment in urinary concentrating ability. Renal sodium and potassium wasting with hypokalaemia are fairly common occurrences, sometimes associated with hypomagnesaemia. 1.1 Azotaemia
An increase in blood urea nitrogen (BUN) and creatinine levels has been reported to occur in over 80% of patients receiving amphotericin B. In a survey of 56 patients conducted between 1956 and 1963, at the time of the introduction of this agent, 93% of patients taking amphotericin B developed values of BUN exceeding 200 mg/L, and 83% had
96
serum creatinine levels greater than 15 mg/L (Butler et at. 1964a). This high incidence of azotaemia has also been reported in other studies (Burgess & Birchall 1972; Holeman & Einstein 1963). A more recent report indicates that in almost every patient treated the glomerular filtration rate (GFR) falls approximately 40% within the first 2 to 3 weeks of therapy, stabilises at 20 to 60% of normal, and remains at this level for the remainder of the treatment (Medoff & Kobayashi 1980). Renal blood flow is also reported to be reduced (Bell et at. 1962; Burgess & Birchall 1972; Sanford et al. 1962). Whereas GFR tends to return to baseline I to 2 months after stopping therapy, renal blood flow requires a much longer time. A severe insult may occur in some patients, raising the serum creatinine concentration to values greater than 35 mg/L. In such cases a brief cessation of therapy (2 to 5 days) is warranted, and usually leads to an improvement in renal function, although in some cases dysfunction may be permanent. The previously administered dose can then be reinstated after due consideration of the clinical status of the patient, the level of infection remaining and alternative therapeutic options. The relationship between the extent of permanent deterioration in renal function and the total cumulative dose employed is not completely settled. Initial reports indicated a significant correlation between the net rise in serum creatinine levels and the total dose of amphotericin B administered (Butler et at. I 964a). This was later challenged by Miller and Bates (1969) who failed to observe such a relationship, and suggested instead that renal toxicity is an individualised and unpredictable response. Burgess and Birchall (1972) were also unable to delineate a dose-response relationship, although none of their patients received an amphotericin B dose greater than 2g. Experience at Vanderbilt University Medical Center also argues against a dose-dependent nephrotoxic response (Branch et al. 1987; Heidemann et al. 1983b). Patients have been observed to develop azotaemia at doses ranging from 100mg to 1.5g, with no significant increase in frequency as the cumulative dose increased. As a general statement, it has been
Drug Safety 5 (2) 1990
proposed that with total cumulative doses less than 5g, the risk of permanent kidney damage is lower than with doses exceeding that amount (Butler et at. 1964; Winn 1963). In these series, 58 and 88%, respectively, of patients receiving more than 5g of amphotericin B had persistent renal impairment. The frequency of dosing as a risk factor for nephrotoxicity has also been addressed, with several studies showing that administration of the drug on alternate days lowers the incidence compared with the same total dose given on a daily basis (Littman et al. 1958; Rubin et at. 1989). Age has been suggested as a risk factor for development of azotaemia (Miller & Bates 1969). A recent study seems to support this contention, although it is complicated by the fact that all patients reported were receiving aminoglycoside antibiotics in addition to amphotericin B (Stein et at. 1988). An important risk factor for the development of azotaemia in amphotericin B-treated patients is their state of hydration. Several reports have appeared in the last few years addressing this question (Branch 1988; Branch et at. 1987; Heidemann et at. 1983b; Stein et at. 1988). This subject is covered in more detail in section 3, but for the present suffice it to say that salt depletion appears to potentiate amphotericin B nephrotoxicity while salt loading protects against it. 1.2 Urinary Concentrating Ability
In an early study, it was shown that amphotericin B could induce a loss of concentrating ability of the kidney which was apparent as early as 2 weeks after the start of therapy, and lasted for the remainder of the treatment (Holeman & Einstein 1963). This abnormality occurred in all patients examined in that study. However, 3 months after therapy was discontinued, the kidneys regained their ability to concentrate urine. Similar results were reported by Butler and co-workers (1964a) and by Bullock et al. (1976). Barbour et al. (1979) reported 3 cases of patients who developed an inability to concentrate the urine due to a defect in free water reabsorption even under maximal stimuli, indicating a failure in the vasopressin response
97
Amphotericin B Nephrotoxicity
of the medullary collecting tubule. This disturbance in concentrating ability frequently precedes the fall in GFR (Beard et al. 1960), and therefore is unlikely to be explained by it.
1.3 Hypokalaemia and Hypomagnesaemia Hypokalaemia is a common and characteristic complication associated with amphotericin B treatment, occurring in approximately 75% of patients (Burgess & Birchall 1972). Supplements of potassium chloride are frequently necessary to maintain normal levels of potassium. A dose-dependent response has been proposed by some investigators, but the mechanisms by which urinary potassium wasting takes place are not clear. It is most probably due to direct toxic effects on distal tubules which increase their permeability for potassium (Butler 1966); alternatively, this could be a result of activation of sodium/potassium exchange in the distal tubule. Magnesium wasting may also occur with amphotericin therapy (Barton et al. 1984; Douglas & Healy 1969). In the study by Barton et al. the lowest serum levels and the largest fractional excretion of magnesium were observed by the fourth week of therapy, after a cumulative dose of approximately 500mg. These changes may occur, however, as early as the first 2 weeks of therapy. Despite the continuation of the drug, no further magnesium wasting occurred, suggesting a 'plateau-ing' of the effect. This abnormality, however, was fully reversible, evidenced by the normal serum and urinary magnesium levels measured I year after discontinuation of therapy. In the event of hypomagnesaemia occurring in association with hypokalaemia, correction of the former abnormality is necessary before potassium replacement can restore normal potassium levels. 1.4 Renal Tubular Acidosis This is another common complication of the use of amphotericin B (Douglas & Healy 1969; McCurdy et al. 1968). McCurdy et al. studied urinary acid secretion in 7 patients with amphotericin B
nephropathy. Five had defects in acid secretion similar to distal renal tubular acidosis. The impaired titratable acid excretion seen was greater than can be accounted for by depression of glomerular filtration, and suggests a tubular defect (Gouge & Andriole 1971; McCurdy et al. 1964). It also occurs in the absence of hypokalaemia, which itself may limit the minimal urinary pH achieved. Therefore, this defect also appears to be a specific tubular effect of amphotericin B. 1.5 Pathological Findings
As suggested by the above clinical findings, there are 2 aspects to amphotericin B nephrotoxicity: glomerular and tub~lar. Pathological examination of kidney biopsy specimens and postmortem examination of kidneys of patients who are or were receiving amphotericin B treatment do not reveal a consistent pattern of change. In general, changes can be described as diffuse and nonspecific. Tubular changes include fragmentation and thickening of basement membranes, necrosis of distal tubular epithelium, nephrocalcinosis, and vacuolisation of distal tubular cells (Butler et al. 1964a; Holeman & Einstein 1963; McCurdy et al. 1968; Takacs et al. 1963). Glomerular changes include hyalinisation and fibrosis of the basement membrane, hypercellularity, and vacuolisation of the medial smooth muscle of glomerular arterioles (Bullock et al. 1976; McCurdy et al. 1968). Analysis of the urinary sediment may (rarely) reveal haematuria, pyuria, tubular cell, casts and (very rarely) proteinuria.
2. Mechanisms of Nephrotoxicity The exact mechanisms involved in amphotericin B-induced nephrotoxicity have not been clearly defined. The polyene antibiotics, including amphotericin B, have been known for over 2 decades to interact with sterols found in cell membranes. This induces the formation of pores within the membrane, resulting in leakage of ions and cellular constituents (Andriole 1973). In addition, administration of the drug is associated with severe vaso-
98
Drug Safety 5 (2) 1990
constriction and a decrease in renal blood flow, which can result in ischaemic injury to the kidney and reduction in GFR. Evidence suggests that this haemodynamic response occurs secondary to the activation of renal homeostatic mechanisms, or release of renal vasoconstrictor autacoids (see section 2.2). Such mechanisms may at least in part explain the nephrotoxicity of the drug. The relative contribution of the direct cellular effects of amphotericin B and its activation of intrarenal mechanisms has not been determined. Most of the emphasis in this section is centred on the recent findings regarding the role of indirect mechanisms in amphotericin B-induced reductions in the GFR. Readers who wish to have a more detailed description of the 'sterol hypothesis' of amphotericin B action are referred to several excellent reviews (Andriole 1973; Kerridge 1979; Medoff & Koyabashi 1980; Medoff et al. 1983; Norman 1976). 2.1 Effect on Membrane Sterols Amphotericin B is a polyene antibiotic containing a hydrophilic region, made up of an hydroxylated hydrocarbon chain, and a sequence of conjugated double bonds which is lipophilic (fig. I). This unique structure allows for the incorporation of the drug into cellular membranes. A large body
of literature suggests that this interaction with the cell membrane, specifically with sterol moieties in it, is the basis of the antifungal, and possibly also the toxic, effects of the drug. In the early I 960s, several studies demonstrated that polyene antibiotics induced changes in cellular permeability that resulted in the leakage of important cellular constituents, followed by lysis and death (Caltrider & Gottlieb 1961; Kinsky 196Ia,b; Marini et al. 1961; Sutton et al. 1961). It was also discovered that the toxic effect of the drug on cells was dependent on the presence of sterols in the cell membranes, and that addition of sterols to the growth media of certain fungi prevented the polyene-induced inhibition of growth and permeability changes (Caltrider & Gottlieb 1961; Gottlieb et al. 1958; Kotler-Brajtburg et al. 1974). The increased permeability has been documented in both artificial and natural membranes (Andriole 1973). In the case of amphotericin B, the membranes containing ergosterol were found to be more sensitive to the antibiotic than those containing cholesterol, which accounted for its relatively lower toxicity to mammalian cells compared with other polyenes (Kotler-Brajtburg et al. 1974; Teerlink et al. 1980). Nevertheless, the interaction with cholesterol is of such a degree as to lead to the formation of aqueous pores in the membrane. These consist of an annulus of polyene and sterol OH
Fig. 1. Chemical structure of amphotericin B.
Amphotericin B Nephrotoxicity
in which the hydrophilic region of the drug molecule faces the interior of the pore (Andreoli 1973; DeKruijiff & Demel 1974; Holz 1974). Amphotericin B increases the permeability of the toad urinary bladder to urea, potassium and chloride (Gatzy et aI. 1979; Lichtenstein & Leaf 1965, 1966), of erythrocyte and liposomes to potassium (Butler et al. 1965; Teerlink et al. 1980), and of the erythrocyte to sodium and chloride (Kinsky 1963; Kinsky et al. 1962). It also induces changes in permeability to hydrogen of the turtle bladder (Finn et al. 1977; Steinmetz & Lawson 1970) and of purified renal brush border membrane vesicles (Capasso et al. 1986; Schell et al. 1989). Considering the hypokalaemia, renal sodium wasting and renal tubular acidosis that are observed in clinical practice with this drug, it is reasonable to suggest that part or all of these effects may be explained by a direct effect on the transport processes in the renal tubules. In fact, administration of amphotericin B in vivo results in binding to sterols in most tissues, with the highest levels documented in the kidney (Hoeprich 1978). Moreover, binding of amphotericin B to the luminal membrane of the renal tubular cells appears to be necessary for the nephrotoxic effect on these cells (Bolard et al. 1980). In agreement with the proposed tubular effects of the drug, Cheng et al. (1982) demonstrated an increase in tubular permeability to inulin in vivo in rats receiving an acute infusion of amphotericin B, resulting in a back-diffusion of inulin. 2.2 Physiological Effects
2.2.1 Short Term Effects Infusions of amphotericin B, either intravenously or directly into the renal artery, induce short term reductions in renal blood flow and GFR, and an increase in renal vascular resistance, in rats and dogs (Butler et al. 1964b; Gerkens et al. 1983; Heidemann et al. 1983a). The exact mechanisms mediating these changes have not been fully elucidated. In theory, I of 2 mechanisms is possible. The drug can either act directly on the vascular smooth muscle to induce contraction, or it may act via release of other vasoconstrictor agents. Evi-
99
dence in the literature suggests that the latter may be an important mechanism. Most of the support for this hypothesis comes from studies which demonstrated inhibition of the short term renal effects of amphotericin B by a variety of physiological and pharmacological interventions. These studies also led to the suggestion that the renal homeostatic mechanism known as tubuloglomerular feedback (TGF) may playa role in amphotericin B nephrotoxicity. TGF was discovered in the early 1970s through single nephron microperfusion experiments (Schnermann 1975). It is a regulatory mechanism that contributes to the integration of proximal and distal tubular function. In brief, an increased delivery of sodium chloride to the distal nephron, specifically to the region of the macula densa cells, results in the activation of TGF, leading to release of agents that induce afferent arteriolar constriction and a decrease in single nephron GFR. It has been suggested that the transport of sodium chloride across the macula densa cells is what triggers the activation of TGF (Wright & Schnermann 1974). The effector mechanism is thought to be the release of adenosine which, in the kidney, is a vasoconstrictor (Osswald et al. 1980, 1982). The resultant decrease in GFR will reduce the rate of delivery, and consequently the concentration of sodium chloride at the macula densa, thus achieving a new steady-state. In normal circumstances, therefore, TGF is a protective mechanism for conserving electrolytes, which may be lifesaving in the shock state (Thurau & Boylan 1976). Its contribution to renal function in pathophysiological states is still controversial, but worth discussing in the context of amphotericin B nephrotoxicity. It has already been established that amphotericin B can cause changes in tubular cell permeability both in vivo and in vitro. Such changes might activate TGF in I of 2 ways. Impairment of sodium reabsorption in the proximal renal tubule can increase the delivery of sodium and chloride ions to the distal tubule, which has been shown to activate TGF. Alternatively, a direct action of amphotericin B on the distal tubular cells may lead to an increased permeability to sodium chloride in
100
that area, thus facilitating entry of these ions into the cells. In both cases a functional reduction in GFR is achieved. Several physIological and pharmacological interventions can influence the activity of TGF. Salt depletion potentiates the fall in GFR induced by hyperperfusion of the distal nephron, while salt loading attenuates it (Thurau 1975). High ceiling diuretics like furosemide (frusemide) and uncoupiers of oxidative phosphorylation like cyanide also inhibit the response, presumably by interfering with the signal, i.e. chloride reabsorption in the ascending limb of the loop of Henle (Wright & Schnermann 1974). Finally, adenosine receptor antagonists such as xanthines inhibit TGF by interfering with the efferent limb of the reflex, while inhibitors of adenosine uptake or metabolism potentiate it (Osswald et al. 1982; Schnermann et al. 1977). If TGF is indeed involved in the acute renal effects of amphotericin B, then inhibition of the former should alleviate the impairment in renal function due to the latter. Support for a role for TGF in amphotericin B-induced nephrotoxicity is derived from studies which specifically addressed this question. Administration of the drug to saltdepleted dogs resulted in reductions in renal blood flow and GFR, as expected. These changes were inhibited in salt-loaded dogs, or in salt-depleted dogs given an acute salt load (Gerkens & Branch 1980). That same study showed that continuous administration of furosemide to dogs with adequate replacement of urinary losses also inhibited the effects of amphotericin B. Finally, aminophylline was shown to prevent the fall in renal blood flow and GFR due to amphotericin B, an effect which could not be reproduced with nitroprusside, suggesting that it was not due to nonspecific vasodilation (Gerkens et al. 1983; Heidemann et al. 1983a). In a recent study conducted in this laboratory, the authors examined the effect of short term infusions of amphotericin B on the renal microcirculation in rats. The results revealed that the drug decreased both the total renal blood flow and GFR. Micropuncture measurements showed that the single nephron GFR was decreased by 2 mechan-
Drug Safety 5 (2) 1990
isms. There was a decrease in single nephron plasma flow, brought about mainly by vasoconstriction of the afferent arteriole, but also of the efferent arteriole (table I). This was associated with a significant reduction in the glomerular capillary ultrafiltration coefficient (kr), an effect which is currently thought to be mediated by mesangial cell contraction (Schlondortf 1987). When these results are compared with the pattern seen with TGF activation (Ichikawa 1982) the similarities are apparent, supporting the TGF hypothesis of amphotericin B nephrotoxicity. Recent unpublished studies from this laboratory, however, provide evidence contrary to the above. Since most of the studies on TGF were performed in rats, while most of the amphotericin B results mentioned above were reported in dogs, the present authors decided to re-examine the influence of these various interventions on amphotericin B nephrotoxicity in rats. The demonstration of similar results would in that case lend greater support to the hypothesis. The results, however, revealed that in the model used (the in situ renal perfusion model) neither salt loading nor salt depletion influenced the response (table II). Furthermore, the inhibitory effect of aminophylline was limited to the renal blood flow response, while the decrease in GFR was unaffected. This effect of
Table I. Effects of amphotericin B (0.05 mg/kg/min) on various parameters (after Sabra et al. 1989) Parameter
Before
After
p Value
SNGFR (nl/min)
35.3 ± 2.2
22.8 ± 2.8
< 0.0005
QA (nl/min)
142 ± 12
89 ± 14
