The lipids have been developed in four major forms [4]. The first is amphotericin B in true liposomes (AMB1); liposomes are microspheres composed of lipid membranes surrounding an aqueous central core. Amphotericin B is located in the membrane, at a relatively low 10% concentration. The second form is amphotericin B lipid complex (ABLC), which includes an approximately 33% concentration of amphotericin B in microparticulate polymorphic sheets and ribbons. The third form is amphotericin B colloidal dispersion (ABCD), in which amphotericin B is intercalated in a 1:1 ratio with cholesteryl sulfate in disc-like particles. All three preparations have been evaluated in open clinical studies, usually at dosages of 1, 3, or 5 mg/kg of body weight per day. Even at higher dosages, these preparations are consistently less nephrotoxic than AMBd at 1 mg/kg per day [1, 2, 4, 6, 7].

Most of our current knowledge derives from open clinical trials; many patients have been enrolled in these trials because of clinical failure of AMBd or renal toxicity from AMBd [6-10]. Treatment with any of these preparations has been associated with less impairment of renal function than is seen with AMBd. For example, AMB1 was remarkably benign in 197 courses (average dose, 1.5 mg/kg) given to 187 transplant recipients who were treated with cyclosporine for immune suppression [7]. One third of the patients had hypokalemia or increased serum creatinine levels, but these outcomes were directly attributed to AMB1 in only 7% of affected patients. Therapy with AMB1 was discontinued because of toxicity in only 3% of patients. Another large study [6] showed a similar reduction in nephrotoxicity. Although all three preparations cause infusion-related fever and rigors, AMB1 appears to produce fewer reactions than the other two preparations. Accordingly, because of clear evidence of reduced toxicity and some evidence of clinical efficacy [8, 10-18], all three preparations are now licensed in several European countries. However, use of these drugs has been significantly limited by their costs, which may reach $500 to $1000 per day for doses of 5 mg/kg.

Partially driven by the high costs of these preparations and the difficulties in obtaining them, some clinicians in Europe and the United States have prepared their own fourth variant by vigorously mixing AMBd with Intralipid (Baxter Healthcare, Glendale, California) [2]. Intralipid-amphotericin B is not standardized, and, in most studies, doses of 1.5 mg/kg or less have been used. Even at these doses, nephrotoxicity is frequent, and the value of this homemade preparation is uncertain.

Although the three "defined" preparations clearly reduce nephrotoxicity, they are not licensed in the United States. The primary reason is that their efficacy in relation to AMBd is only partially defined. Scattered reports have been published of patients treated successfully for acute disseminated candidiasis, chronic hepatosplenic candidiasis, aspergillosis, coccidioidomycosis, cryptococcosis, leishmaniasis, and other infections [10-17]. In these open studies, diagnostic criteria for mycotic infection are varied and sometimes inadequate, and criteria for treatment are often not clearly defined. In one report of liposomal AMB1 treatment, only 2 of 56 patients considered to have candidiasis had candidemia, and 30 patients received a diagnosis of candidiasis on the arguable basis of throat and fecal cultures [9]. In general, the clinical efficacy noted in these reports is in the range of that reported elsewhere for AMBd. For disseminated candidiasis, for example, clinical efficacy ranges from 70% to 80% [8-10].

There is particular interest in the use of lipid-associated amphotericin B to treat acute invasive aspergillosis. In one series of 21 neutropenic patients treated for probable or definitive invasive aspergillosis, 61% of the patients responded to AMB1 [9]; this response rate is similar to that in another series of 28 evaluable patients [10]. In another, larger, series [8], 60% of 72 patients responded to treatment with ABLC at dosages as high as 5 mg/kg per day. However, it is difficult to compare the outcomes in such mixed groups of patients with outcomes in patients with the worst-case scenario—allogenic bone marrow transplant recipients who have graft-versus-host disease and invasive aspergillosis, in whom mortality may exceed 90%. Prevention of aspergillosis and other systemic mycoses through the early use of a relatively nontoxic drug is perhaps more attractive than late-salvage efforts among patients with established disease. In one randomized study of bone marrow recipients [19], AMB1 was no more effective than placebo in preventing fungal infection, but the study consisted of only 76 patients and thus was not conclusive. Other patients that fare poorly with invasive mycoses are liver transplant recipients. In a double-blind study comparing AMB1 with placebo for prophylaxis of mycotic infections [20], none of 40 AMB1 recipients compared with 6 of 37 controls receiving placebo developed well-documented invasive mycoses (P < 0.05).

A few other comparative studies have published data. In two phase II randomized dose-finding studies [13, 14], ABLC was compared with AMBd for treating coccidioidomycosis and cryptococcosis. Nineteen patients with cryptococcosis were given more than twelve 5-mg/kg doses of ABLC; in 42% of these patients, cerebrospinal fluid cultures converted to negative. Fourteen patients with cryptococcosis were treated with AMBd at 0.7 to 1.2 mg/kg per dose; cerebrospinal fluid cultures of 7 of these patients converted to negative. Although the efficacy of ABLC appeared similar to that of amphotericin B, these studies were too small to be definitive. Other open studies of cryptococcal meningitis [11, 17] have found that cryptococcosis responds to ABLC. In one study [11], AMB1 caused cerebrospinal fluid culture conversions in 4 of 7 patients with the acquired immunodeficiency syndrome (in 3 of the 4, conversion occurred within 1 week of initiating treatment). A larger study of 23 patients had similar response rates [17].

The most definitive study comparing AMBd with a lipid-associated preparation has not yet been published in full-length form [18]. In this study, 231 patients with invasive candidiasis (84% of whom had candidemia) were randomly assigned in a 2:1 ratio to receive ABLC (5 mg/kg per day) or AMBd (0.6 to 1.0 mg/kg per day). The two groups had similar predisposing factors, including hematologic cancer, solid tumors, recent surgery, corticosteroid use, immunologic disorders, and neutropenia (< 500 polymorphonuclear leukocytes/mm³). Of the 194 patients evaluated, 65% of ABLC recipients and 61% of AMBd recipients responded to the respective drugs. It is curious that despite this large and apparently well-conducted trial showing that the drugs have similar efficacies, ABLC has been licensed in the United States only as second-line therapy for aspergillosis.

The current patchwork collection of these and other reports is encouraging. More data are required to allow firm comparisons of the efficacies of lipid-associated amphotericin B and AMBd. Several large trials comparing lipid-associated amphotericin B preparations with AMBd in neutropenic patients who have suspected or documented mycoses are now nearing completion. Comparative multicenter studies of histoplasmosis and cryptococcosis are beginning. However, even if the lipid-amphotericin B preparations show efficacy similar to that of AMBd, the formidable challenge of the marketplace remains. Some studies currently in progress are exploring doses of less than 5 mg/kg, an important issue given the high prices of these agents. With affordable costs and further data showing similar efficacy, lipid amphotericin B derivatives could rapidly supplant AMBd. However, without effective, lower doses or lower production costs, only a few of us may be able to purchase new clothes for the emperor, however attractive they appear.