Clinicians who prescribe amphotericin B should keep in mind the potential for nephrotoxicity. Lipid-based formulations can be less toxic than the conventional form but may also be more expensive. Patient outcomes and cost-effectiveness of the various formulations are discussed
Keywords: amphotericin B, invasive fungal infections, nephrotoxicity, adverse events, hospital length of stay
Abstract
Purpose:
Patients with invasive fungal infections are often treated initially with conventional amphotericin B deoxycholate (CAB), followed by a switch to lipid-based formulations of amphotericin B (LF-AMB). Our study examined adverse events and hospital length of stay (LOS) among adults who received LF-AMB exclusively or CAB followed by LF-AMB (CAB/LF-AMB).
Methods:
Data were extracted from the Cerner Health Facts database. The study included adults with evidence of infection by Aspergillus, Candida, or Cryptococcus in addition to either renal insufficiency, a clinical condition suggesting intolerance to CAB, or CAB exposure within 90 days of admission. Nephrotoxicity was defined as a serum creatinine (SCr) level exceeding a 100% increase from baseline and an absolute level above 1.2 mg/dL. We used a hierarchical mixed-effect logistic regression model with nephrotoxicity as the outcome for the multivariate analysis.
Results:
The study included 327 LF-AMB and 81 CAB/LF-AMB patients with similar demographics and baseline SCr values. The mean pre-to-post percentage increase in SCr levels was greater for CAB/LF-AMB (122.9%) compared with LF-AMB (62.2%) (P < 0.001). The multivariate-adjusted odds ratio of nephrotoxicity was 5.93, for a 95% confidence interval of 2.92 to 12.05 (P < 0.001) for CAB/LF-AMB compared with LF-AMB. Hypokalemia, hypomagnesemia, and infusion-related reactions were more frequent with CAB/LF-AMB.
Compared with the LF-AMB group, the CAB/LF-AMB patients had a longer post-amphotericin B LOS (24.1 days vs. 15.7 days, respectively; P < 0.001), with a marginal effect of 4.5 days longer for those receiving CAB/LF-AMB (P = 0.016).
Conclusion:
In this retrospective study, we noted a significantly longer post-amphotericin B LOS and a greater frequency of adverse events, including nephrotoxicity, for patients whose initial treatment was CAB and who were switched to LF-AMB, compared with patients who received LF-AMB only.
INTRODUCTION
Conventional amphotericin B deoxycholate (CAB), a broad-spectrum polyene, has long been a cornerstone of therapy for patients with invasive fungal infections. Current guidelines from the Infectious Disease Society of America (IDSA) include CAB (Fungizone, Bristol-Myers Squibb) as an acceptable therapy for invasive candidiasis if intolerance occurs or if the availability of other antifungal agents (grade A1) is limited;1 these guidelines also recommend CAB for cryptococcal meningitis (grade A2).2 In addition, the guidelines take into account the potential toxicity of CAB, which has been associated with adverse effects in 50% to 90% of cases, including infusion-related reactions requiring treatment, hypokalemia, hypomagnesemia, and acute renal failure.3–6
Lipid-based formulations of amphotericin B (LF-AMB), including liposomal amphotericin B (L-AmB) (e.g., AmBisome, Astellas) and amphotericin B lipid complex (ABLC) (Abelcet, Enzon/Sigma Tau), have shown equivalent efficacy and reduced toxicity, compared with CAB, in various clinical settings.3,7–13 In a meta-analysis of 13 studies, the absolute risk reduction in nephrotoxicity associated with LF-AMB versus CAB was 18%.14 The clinical use of LF-AMB has grown in recent years, and IDSA guidelines state that LF-AMB may be used as an alternative therapy for patients with invasive candidiasis1 and for those with cryptococcal meningitis and coexisting renal impairment.2
The key clinical and economic differences between amphotericin B products lie in their adverse-event profiles and drug-acquisition costs. Clinicians may select LF-AMB when a patient’s renal impairment or unacceptable toxicity precludes the use of CAB (Table 1).12,15–23 Although published data documenting the extent of this practice are limited, clinicians may be encouraged to prescribe CAB as a first-line agent as a result of:
current IDSA guidelines and formulary recommendations.
the more restrictive FDA-approved labeling for LF-AMB (e.g., use among patients intolerant of CAB rather than among all patients with invasive fungal infections).
the relatively lower drug-acquisition costs of CAB.
Table 1.
Clinical Conditions That Preclude Treatment With Conventional Amphotericin B Deoxycholate
|
Patients whose initial therapy is CAB may be switched to LF-AMB if adverse events arise during treatment. Regardless of the underlying rationale, this treatment strategy might have an overall adverse effect on patients and health care systems.
In the study, we observed nephrotoxicity, other adverse events, and hospital length of stay (LOS) in patients whose clinical presentation might warrant the use of LF-AMB for the treatment of Candida, Aspergillus, or Cryptococcus infections. We examined patients receiving LF-AMB exclusively or CAB, followed by LF-AMB (CAB/LF-AMB).
METHODS
Study Design and Data Source
Data for our retrospective cohort study were extracted from the Health Facts electronic medical record (EMR) database (Cerner Corp., Kansas City, Missouri), which had been created from comprehensive clinical records in the health care system. The information in Health Facts, derived from hospitals across the U.S., included time-stamped pharmacy, laboratory, and admissions data as well as billing and registration forms from all affiliated patient-care locations. Health Facts operates under established Health Insurance Portability and Accountability Act (HIPAA) policies and procedures, and statistical methods were used to establish de-identification.
Extensive security controls were in place to minimize the chance of unauthorized access to or modifications to protected health information.
Study Population
The study population included patients 18 years of age or older who were hospitalized between January 2001 and June 2010. Each patient had at least one order for intravenous (IV) L-AmB and/or amphotericin B lipid complex (ABLC). For patients with multiple eligible encounters, only the first encounter was considered.
Because clinicians might select LF-AmB for patients who are intolerant of CAB or whose infections are refractory to CAB, all patients had to have evidence of renal insufficiency, a clinical condition suggesting unacceptable toxicity or intolerance to CAB (see Table 1) or had to have received CAB within 90 days of the encounter. Patients also had to have evidence of infection with Aspergillus, Candida, and/or Cryptococcus during the index encounter or within 90 days before the index encounter through a positive blood culture and/or relevant International Classification of Diseases (ICD-9-CM) codes 112.x, 117.3, 117.5, 484.6, or 495.4 as a discharge diagnosis.
Outcomes and Other Measures
Outcome measures were nephrotoxicity, hypokalemia, hypomagnesemia, new-onset aspartate aminotransferase (AST) elevations, infusion-related reactions, and hospital LOS. Baseline laboratory values were defined as the earliest results within 48 hours of admission. Pre-amphotericin B levels were defined as the laboratory results drawn closest to the first amphotericin B order. Post-amphotericin B hospital LOS was measured from the first order for amphotericin B to hospital discharge.
Nephrotoxicity was assessed following the first order for amphotericin B. The primary nephrotoxicity endpoint was defined as a minimum of a two-fold increase in serum creatinine (SCr) level at least 24 hours following an order for amphotericin B and an absolute post-amphotericin B SCr level greater than 1.2 mg/dL, as defined by Wingard et al.22 Sensitivity analyses included other peak levels and the percentage of change in SCr levels from baseline to the last post-amphotericin B order during the encounter.
During amphotericin B therapy, hypokalemia and hypomagnesemia were defined as a value below the lower limit of normal (according to the individual facility’s laboratory) obtained more than 24 hours after the first amphotericin B order and while the patient was receiving an amphotericin B agent.
New-onset AST elevation was restricted to patients with a preamphotericin B AST level within normal limits and was defined as a value greater than the upper limit of normal (ULN) obtained more than 24 hours after the first amphotericin B order.
An infusion reaction requiring treatment was defined as therapy with IV corticosteroids, diphenhydramine (any route), or epinephrine within 6 hours after the first amphotericin B order.
We derived patient clinical characteristics and comorbidities of interest from administrative records (e.g., ICD-9-CM codes, discharge) and clinical records (e.g., pharmacy, laboratory) of encounters within the previous 12 months, including the current encounter. Patients were classified as surgical or medical according to the Diagnosis-Related Group (DRG) system.
Measures of organ dysfunction were obtained within a 48-hour window surrounding the time of admission using a Sepsis-related Organ Failure Assessment (SOFA) score of 2 or above.24 Critical care exposure was defined as two or more orders from an intensive-care unit (ICU) 12 or more hours apart, mechanical ventilation, or orders for vasopressors.
We calculated total exposure of the first amphotericin B agent as the factor of the daily dose (in milligrams) times the total duration in days. A 3:1 ratio was used to convert to adjusted dosages of LF-AMB for purposes of dose–response relationships to nephrotoxicity, as described by Ullman et al.25 Drugs considered nephrotoxic or those that place patients at an additional risk of nephrotoxicity are presented in Table 2.
Table 2.
Medications With Direct Nephrotoxic Effects*
| Antiviral agents |
|
| Aminoglycosides |
|
| Other antibiotics |
|
| Calcineurin inhibitors |
|
| Anticancer drugs |
|
| Radiocontrast agents |
|
| Miscellaneous antimicrobials |
|
Systemic routes only, including oral/nasogastric tube/feeding tube, intra-muscular, and intravenous.
Analysis
Descriptive statistics were used to summarize patients’ demographic and clinical characteristics and treatment outcomes. We conducted a multivariate analysis to compare outcomes between the two study groups. The LF-AMB patients were defined as the reference group. For continuous outcomes, we used a generalized linear model (GLM) with a gamma distribution and a logarithmic link to account for the skewed distributions of post-amphotericin B LOS, and the robust variance estimates were used to adjust for a clustering effect at the hospital level.
We also provided marginal effects (i.e., a measure of the amount of change in observed units in the outcome produced by a 1-unit change in the exposure of interest, with other variables held at their means). For binary outcomes, we used a multilevel (hierarchical) mixed-effects logistic regression model.
To avoid misspecification, predictors (i.e., potential covariates) selected for the GLM occurred before the initiation of the first antifungal agent, and care was taken to avoid overfitting the data given the sample size. To determine which predictors had the strongest association with the various outcomes, we applied a bootstrapping procedure to a stepwise regression algorithm26 after first excluding predictors with a P value greater than 0.2527 from the univariate analysis.
Although overfitting was a concern, our objective was to estimate rather than predict the effect. Subsequently, when faced with nested models of similar fit, we selected the model with more clinically relevant predictors.28 Missing data were rare, and no imputation was performed. Only patients with complete data were considered in the multivariate analyses.
RESULTS
Patient Characteristics
We identified 327 LF-AMB patients (105 receiving L-AmB, 222 receiving ABLC) and 81 CAB/LF-AMB patients (nine receiving L-AmB, 72 receiving ABLC) (Figure 1 and Table 3). LF-AMB patients (51.7% were men; mean, 52.9 years of age) and CAB/LF-AMB patients (61.7% were men; mean, 48.6 years of age) were similar in age, sex, and racial distribution; however, the CAB/LF-AMB patients were more likely to have had an urgent hospital admission (76.5% vs. 64.5%, respectively; P = 0.040) and to be surgical patients (54.3% vs. 35.5%, respectively; P = 0.007).
Figure 1.
Study population with invasive fungal infections. ABLC = amphotericin B lipid complex; CAB = conventional amphotericin B deoxycholate; LF-AMB = lipid-based formulations of amphotericin B; L-AmB= liposomal amphotericin B.
Table 3.
Baseline Patient Characteristics and Exposures During an Encounter
| LF-AMB (n = 327) | CAB/LF-AMB (n = 81) | P Valuea | |
|---|---|---|---|
| Demographics | |||
| Age, years (mean [SD]) | 52.9 [19.0] | 48.6 [17.9] | 0.067 |
| Male (%) | 51.7 | 61.7 | 0.105 |
| Race (%) | |||
| Caucasian | 64.8 | 63.0 | 0.630 |
| African-American | 26.6 | 30.9 | |
| Other/unknown | 8.6 | 6.2 | |
| DRG category (%) | |||
| Medical | 61.2 | 44.4 | 0.007 |
| Surgical | 35.5 | 54.3 | |
| Unknown | 3.4 | 1.2 | |
| Urgent/emergent admission | 64.5 | 76.5 | 0.040 |
| Comorbidities, current or 12 months prior (%)b | |||
| Impaired immune function | 76.2 | 65.4 | 0.049 |
| Hypertension | 30.9 | 14.8 | 0.004 |
| Hematologic malignancy | 28.8 | 28.4 | 0.950 |
| Blood dyscrasias | 28.8 | 13.6 | 0.005 |
| Chronic respiratory conditions | 24.2 | 27.2 | 0.575 |
| Diabetes | 19.6 | 19.8 | 0.971 |
| Chronic kidney disease | 19.0 | 7.4 | 0.013 |
| HIV/AIDS | 16.5 | 19.8 | 0.489 |
| Solid tumor | 13.8 | 3.7 | 0.011 |
| Major solid organ transplantation | 13.8 | 7.4 | 0.122 |
| Dyslipidemia | 13.2 | 7.4 | 0.155 |
| Heart failure | 12.5 | 16.1 | 0.404 |
| Cardiac dysrhythmias (other than atrial fibrillation) | 11.0 | 12.4 | 0.733 |
| Atrial | 9.8 | 8.6 | 0.754 |
| Coronary artery disease | 8.9 | 2.5 | 0.060 |
| Autoimmune disorders | 8.3 | 3.7 | 0.223 |
| Stroke/transient ischemic attack | 8.3 | 3.7 | 0.223 |
| Dialysis dependence | 5.8 | 1.2 | 0.145 |
| Cirrhosis/chronic liver disease | 5.5 | 1.2 | 0.141 |
| Stem-cell transplant | 4.9 | 7.4 | 0.370 |
| Nephrotoxic drug exposure (one or more agents, %) | |||
| Prior to encounter (90 days) | 26.9 | 25.0 | 0.729 |
| During encounter | 79.5 | 76.5 | 0.558 |
| Antifungal exposure (%) | |||
| Antifungal therapy prior to encounter (90 days) | 19.0 | 18.5 | 0.928 |
| CAB exposure (90 days) | 0.6 | 2.5 | 0.178 |
| Encounter event (%) | |||
| Critical care use during encounter | 56.7 | 58.0 | 0.852 |
| Chemotherapy | 16.5 | 14.8 | 0.710 |
| Organ system dysfunction (within 48 hours of admission) | 53.5 | 55.6 | 0.742 |
| Respiratory dysfunction | 12.8 | 14.8 | 0.639 |
| Hematological dysfunction | 25.1 | 25.9 | 0.875 |
| Hepatic dysfunction | 13.5 | 4.9 | 0.033 |
| Cardiovascular dysfunction | 12.8 | 16.1 | 0.450 |
| Renal dysfunction | 20.5 | 14.8 | 0.247 |
| Antifungal treatment during encounter | |||
| Time from presentation to initial antifungal, days (mean [SD]) | 7.0 [10.5] | 6.0 [7.4] | 0.417 |
| Time from presentation to amphotericin B, days (mean [SD]) | 13.3 [17.3] | 10.6 [12.0] | 0.191 |
| Initial antifungal class (%) | |||
| Echinocandin | 9.8 | 1.2 | 0.010 |
| Amphotericin B | 37.3 | 49.4 | 0.047 |
| Azole, first generation | 41.9 | 45.7 | 0.538 |
| Azole, second generation | 8.6 | 0.0 | 0.002 |
| Flucytosine | 2.4 | 3.7 | 0.462 |
| Total exposure of first amphotericin B agent (mg)c | 2,460.7 [2,702.0] | 479.7 [703.9] | <0.001 |
| Total adjusted exposure LF-AMB (3:1 ratio, mg) mean [SD] (Ullman et al.25)c | 820.2 [900.7] | ||
| Total duration of first amphotericin B agent (days)c | 8.3 [8.8] | 9.3 [11.7] | 0.412 |
| Laboratory measures, first, within 48 hours of admissiond | |||
| Serum creatinine, mg/dL (mean [SD]) | 1.5 [1.3] | 1.3 [1.2] | 0.391 |
| eGFR, mL/minute/1.73 m2 (mean [SD]) | 71.5 [54.4] | 78.9 [44.3] | 0.258 |
| Laboratory measures, pre-amphotericin Bd | |||
| Serum creatinine, mg/dL (mean [SD]) | 1.5 [1.4] | 1.2 [0.9] | 0.046 |
CAB = conventional amphotericin B deoxycholate; DRG = Diagnosis-Related Group; eGFR = estimated glomerular filtration rate; HIV/AIDS = immunodeficiency human virus/acquired syndrome; LF-AMB = lipid-based formulations of amphotericin B; SD = standard deviation.
Based on blood culture results and/or relevant International Classification of Disease (ICD-9-CM) codes.
Comorbidities are listed in descending order based on the percentage of LF-AMB patients with the condition.
N = 325 for LF-AMB and N = 78 for CAB/LF-AMB
Baseline laboratory measures were evaluated at two different time points: (1) at presentation, defined as the earliest result within 48 hours of presentation, and (2) pre-amphotericin B, defined as the result obtained closest to and before the first amphotericin B order.
Most patients had evidence of impaired immune function. Comorbidity profiles were generally similar, but LF-AMB patients were significantly more likely to have hypertension, blood dyscrasias, chronic kidney disease, and solid tumors compared with CAB/LF-AMB patients.
All patients had similar rates of exposure to nephrotoxic drugs before (26%) and during (79%) the encounter. Mean baseline SCr levels were similar between the groups (1.5 mg/dL for LF-AMB and 1.3 mg/dL for CAB/LF-AMB), and there was a small difference in pre-amphotericin B SCr levels (P = 0.046).
The mean time from hospital admission to the receipt of any antifungal treatment was 7.0 days for LF-AMB patients and 6.0 days for CAB/LF-AMB patients (P = 0.417). The mean time from admission to the first amphotericin B treatment (CAB or LF-AMB) was 13.3 days for LF-AMB and 10.6 days for CAB/LF-AMB (P = 0.191).
The mean total duration of the first amphotericin B agent was similar between patient groups (8.3 days for LF-AMB and 9.3 days for CAB in the CAB/LF-AMB group; P = 0.412). The total mean exposure for the first amphotericin B agent was 2,460.7 mg for LF-AMB (820.2 mg using a 3:1 dose-adjustment ratio) and 479.7 mg for CAB. Orders for antibiotics were pervasive (more than 90%), with an average of four antibiotic classes ordered per patient.
Adverse Drug Events
Descriptive (unadjusted) and multivariate (adjusted) results for adverse events following treatment with LF-AMB or CAB/LF-AMB are presented in Tables 4 and 5, respectively.
Table 4.
Nephrotoxicity and Comparison of Outcomes: Conventional and Lipid-Based Amphotericin Versus Lipid-Based Amphotericin (Unadjusted)
| LF-AMB (n = 327) | CAB/LF-AMB (n = 81) | P Valuea | |
|---|---|---|---|
| Nephrotoxicity, SCr, and Related Outcomes | |||
| Nephrotoxicity, primary outcomeb | |||
| Post-amphotericin B SCr > 100% increase from the pre-amphotericin B level and an absolute level of > 1.2 mg/dLc (n [%]) | 47 [18.6] | 28 [49.1] | < 0.001 |
| SCr, post-amphotericin B peak (absolute value)b | |||
| Peak > 1.5 mg/dL (n [%]) | 120 [47.4] | 36 [63.2] | 0.032 |
| Peak > 2.0 mg/dL (n [%]) | 90 [35.6] | 23 [40.4] | 0.498 |
| Peak > 2.5 mg/dL (n [%]) | 63 [24.9] | 18 [31.6] | 0.300 |
| Peak > 3.0 mg/dL (n [%]) | 46 [18.2] | 13 [22.8] | 0.422 |
| SCr, post-amphotericin B peak (relative change)b | |||
| Peak 1.5 × pre-amphotericin B (n [%]) | 91 [36.0] | 38 [66.7] | <0.001 |
| Peak 1.5 × pre-amphotericin B and 1.5 ULN (n [%]) | 51 [20.2] | 21 [36.8] | 0.007 |
| Peak 2 × pre-amphotericin B (n [%]) | 53 [21.0] | 26 [45.6] | < 0.001 |
| Peak 3 × pre-amphotericin B (n [%]) | 21 [8.3] | 11 [19.3] | 0.014 |
| SCr, post-amphotericin B peakb | |||
| Peak (absolute value) (mean [SD]) | 2.1 [1.7] | 2.3 [1.5] | 0.400 |
| % change (relative change) (mean [SD]) | 62.2% [107.5] | 122.9 [131.7] | <0.001 |
| Other Adverse Events | |||
| Hypokalemia during amphotericin B therapyd (n [%]) | 144 [59.3] | 55 [71.4] | 0.055 |
| Hypomagnesemia during amphotericin B therapye (n [%]) | 82 [39.8] | 42 [58.3] | 0.007 |
| New-onset AST elevation following amphotericin B therapyf (n [%]) | 25 [16.2] | 12 [24.5] | 0.192 |
| Infusion reaction requiring treatmentg (n [%]) | 63 [19.3] | 24 [29.6] | 0.041 |
| Mortalityh | 104 [32.1] | 26 [32.5] | 0.945 |
| Hospital LOS | |||
| Overall hospital LOS (days) (mean [SD]) | 29.0 [27.3] | 34.8 [26.9] | 0.088 |
| Post-amphotericin B hospital LOS (mean [SD]) | 15.7 [18.9] | 24.1 [21.9] | <0.001 |
AST = aspartate transaminase; CAB = conventional amphotericin B deoxycholate; LF-AMB = lipid-based amphotericin B; LOS = length of stay; SCr = serum creatinine; SD = standard deviation; ULN = upper limit of normal.
In the setting of multiple outcomes, a Bonferroni or similar adjustment should be made when making inferences (e.g., if alpha = 0.05, adjust alpha downward by dividing by the number of outcomes).
Note: The number of patients in each analysis varied according to the availability of the specific outcome measure (e.g., mortality data, availability of pre-amphotericin B and post-amphotericin B laboratory values). Denominators for each analysis are indicated by a superscripted letter.
LF-AMB = 253, CAB/LF-AMB = 57.
Definition based on Wingard et al., 2000.22
LF-AMB = 243, CAB/LF-AMB = 77.
LF-AMB = 206, CAB/LF-AMB = 72.
LF-AMB = 154, CAB/LF-AMB = 49.
LF-AMB = 327, CAB/LF-AMB = 81.
LF-AMB = 323, CAB/LF-AMB = 80
Table 5.
Comparison of Conventional Amphotericin B Versus Lipid-Based Amphotericin B (Adjusted)
| Outcome | Model (Statistic) | Effect Estimate | 95% CI | P Value | |
|---|---|---|---|---|---|
| CAB/LF-AMB versus LF-AMB (referent) | Nephrotoxicity* | Logistic regression (odds ratio) | 5.93 | 2.92–12.05 | <0.001 |
| Post-amphotericin B hospital LOS | Generalized linear model (coefficient) | 1.31 | 1.08–1.60 | 0.006 | |
| Marginal effect (mean, days) | 4.5 | 0.8–8.3 | 0.016 |
CAB = conventional amphotericin B deoxycholate; CI = confidence interval; LF-AMB = lipid-based amphotericin B; LOS = length of stay.
Post-amphotericin B serum creatinine level of more than 100% increase from the pre-amphotericin B level and an absolute level of more than 1.2 mg/dL (definition based on Wingard et al., 200022).
Serum Creatinine and Nephrotoxicity
We used the subset of patients with both pre-amphotericin B and post-amphotericin B SCr values (253 patients receiving LF-AMB and 57 patients receiving CAB/LF-AMB) to evaluate SCr-based outcomes. Nephrotoxicity was more than twice as common among CAB/LF-AMB patients (49.1%) than among LF-AMB patients (18.6%) (P < 0.001).
These findings remained statistically significant following the multivariate analysis (Tables 5 and 6). The analysis demonstrated that the odds of nephrotoxicity occurring with CAB/LF-AMB were 5.93 times higher than with LF-AMB (95% CI, 2.92–12.05; P < 0.001).
Table 6.
Nephrotoxicity* (Primary Outcome) Multivariate Analysis, Full Model Results (n = 305)
| Covariate | Odds Ratio | 95% CI | P Value |
|---|---|---|---|
| CAB/LF-AMB treatment during index encounter (referent = LF-AMB treatment) | 5.93 | 2.92–12.05 | <0.001 |
| Critical-care use before antifungal treatment during index encounter | 1.50 | 0.79–2.86 | 0.217 |
| Chemotherapy during index encounter | 2.14 | 1.03–4.46 | 0.041 |
| Platelet count ≤100 × 103/mm3 within 48 hours of admission | 2.50 | 1.24–5.03 | 0.010 |
| Hypertension (current or 12 months prior) | 0.49 | 0.21–1.14 | 0.099 |
| Atrial (current or 12 months prior) | 0.48 | 0.16–1.41 | 0.182 |
| Alveolitis, pneumonitis, pneumoconiosis, pulmonary fibrosis, or idiopathic pulmonary hemosiderosis (current or 12 months prior) | 3.81 | 0.81–17.92 | 0.091 |
| Prior IV antibiotic use (90 days) | 0.29 | 0.14–0.62 | 0.001 |
| Total exposure of first antifungal (per 100 mg) | 1.01 | 1.00–1.02 | 0.170 |
CAB = conventional amphotericin B deoxycholate; CI = confidence interval; IV = intravenous; LF-AMB = lipid-based formulations of amphotericin B.
Post-amphotericin B serum creatinine level above a 100% increase from the pre-amphotericin B level and an absolute level of more than 1.2 mg/dL.
More CAB/LF-AMB patients had a peak SCr level above 1.5 mg/dL (63.2% vs. 47.4%, respectively; P = 0.032) compared with the LF-AMB group. The mean percentage increase in SCr levels (before and after amphotericin B therapy) was greater with CAB/LF-AMB (122.9%) than with LF-AMB (62.2%) (P < 0.001).
A greater number of CAB/LF-AMB patients also experienced significant relative elevations in SCr concentrations.
Other Adverse Events and Mortality
During amphotericin B therapy, both hypokalemia (71.4% for CAB/LF-AMB vs. 59.3% for LF-AMB; P = 0.055) and hypomagnesemia (58.3% for CAB/LF-AMB vs. 39.8% for LF-AMB; P = 0.007) were more frequent among patients receiving CAB/LF-AMB than among those receiving LF-AMB. Infusion reactions requiring treatment also were more frequent with CAB/LF-AMB (29.6%) than with LF-AMB (19.3%) (P = 0.041). No significant differences were observed between the groups for new-onset AST elevations or mortality.
Length of Stay
Compared with the LF-AMB patients, the CAB/LF-AMB group had a somewhat longer overall hospital LOS (P = 0.088) but a significantly longer post-amphotericin B LOS (P < 0.001). For the CAB/LF-AMB patients, the mean overall LOS was 34.8 days, with approximately 70% (24.1 days) occurring after the first order for amphotericin B. For the LF-AMB group of patients, the mean overall LOS was 29.0 days, 54% (15.7 days) occurring after the first order for amphotericin B. This finding remained statistically significant following the multivariate analysis (Tables 5 and 7) and showed a marginal effect of 4.5 additional days in post-amphotericin B LOS for patients receiving CAB/LF-AMB (P = 0.016).
Table 7.
Post-Amphotericin B Hospital Length of Stay, Multivariate Analysis, Full Model Results (n = 408)
| Covariate | Coefficient | 95% CI | P Value |
|---|---|---|---|
| CAB/LF-AMB treatment during index encounter (referent = LF-AMB treatment) | 1.315 | 1.081–1.599 | 0.006 |
| Male sex (referent = female sex) | 1.200 | 1.024–1.405 | 0.024 |
| Medical DRG (referent = surgical DRG) | 0.654 | 0.557–0.769 | <0.001 |
| Coronary artery disease (current or 12 months prior) | 0.827 | 0.656–1.044 | 0.110 |
| No. of antibiotic classes during index encounter (per unit) | 1.121 | 1.071–1.173 | <0.001 |
| Critical-care use before antifungal treatment during index encounter | 1.045 | 0.895–1.220 | 0.578 |
| Voriconazole use prior to amphotericin B during index encounter | 0.694 | 0.553–0.870 | 0.002 |
| Polymicrobial blood culture | 1.274 | 1.034–1.568 | 0.023 |
| Baseline (constant) | 10.320 | 7.738–13.763 | <0.001 |
CAB = conventional amphotericin B deoxycholate; CI = confidence interval; DRG = Diagnosis-Related Group; LF-AMB = lipid-based formulations of amphotericin B.
DISCUSSION
In this retrospective study of patients for whom lipid-based formulations of amphotericin B (LF-AMB) therapy is warranted, the exclusive use of LF-AMB resulted in lower rates of nephrotoxicity and other adverse events. In addition, a shorter LOS was observed with LF-AMB therapy alone when compared with sequential conventional amphotericin B deoxycholate (CAB)/LF-AMB therapy.
Although baseline renal function was similar in both patient groups, the multivariate analyses suggested that patients who received CAB before LF-AMB had 5.93 times higher odds (95% CI, 2.92–12.05) of developing nephrotoxicity compared with patients who received only LF-AMB. Other SCr-based measures of renal function also indicated an advantage of LF-AMB therapy alone in univariate analyses.
Additional adverse events that occurred more frequently with CAB/LF-AMB included electrolyte imbalances (hypokalemia and hypomagnesemia) and infusion reactions requiring treatment. Although a detailed examination of the epidemiology of amphotericin B–associated infusion reactions was beyond the scope of this study, our data suggest that they are common (occurring in as many as 28% of the CAB patients) despite pretreatment with medications such as corticosteroids and diphenhydramine (e.g., Benadryl, McNeil).
Institutional policies regarding pre-treatment may vary, and such data were not analyzed. As in many previous studies of patients with invasive fungal infections, we found that more than 30% of patients died during hospitalization.
Patients receiving sequential CAB/LF-AMB therapy had a longer post-amphotericin B LOS than LF-AMB patients, resulting in a marginal effect of 4.5 additional days. Although we did not evaluate a direct association between nephrotoxicity and hospital LOS in this study, previous studies have demonstrated a longer LOS and higher hospital costs in amphotericin B– treated patients who experienced nephrotoxicity compared with patients who did not.29–31 Nephrotoxicity observed in the CAB/LF-AMB cohort might have contributed to the differences in LOS observed between our two study groups.
Studies have demonstrated that switching from CAB to LF-AMB immediately following the development of nephrotoxicity can ameliorate deteriorating renal function and reduce hospital LOS.23,32 In a multicenter study of compassionate-use LF-AMB therapy, 17 of 50 patients with initially high SCr levels recovered normal renal function after switching from CAB to LF-AMB despite a relatively high total dose of LF-AMB.23
A randomized trial that compared early and late switching from CAB to LF-AMB also supports these findings.32 Following initial treatment with CAB for a mean of 4.8 days, patients were randomly assigned to undergo an immediate switch to LF-AMB (an early switch) or to a late switch if nephrotoxicity developed. The late switch was made only if SCr levels doubled or exceeded 170 μmol/L. The mean changes in SCr levels, from randomization until the end of treatment, were −2.3% for the early-switch patients and 16.2% for the late-switch patients (P = 0.04). The late-switch group also had a 68% longer LOS, although this was not statistically significant (25.1 vs. 14.9 days; P = 0.13).
Despite their equivalent efficacy and superior safety profile, LF-AMB is still considered a second-line choice after CAB in many patients because of FDA labeling requirements and drug-acquisition costs.33 Although P&T committees sometimes make choices based on the relative cost of these therapies, most decision-makers realize the need to consider the secondary costs of impaired renal function and other adverse effects in estimating the true cost-effectiveness of these agents.23–25 Hospitals and health care systems have developed policies and practices to help balance the direct costs of lipid-based agents with the indirect costs of CAB-induced toxicity.
Strategies that tend to reduce CAB-related nephrotoxicity while preserving CAB’s clinical benefits include 24-hour continuous infusion, diligent patient monitoring with aggressive hydration, and swift correction of electrolyte imbalances.34–36 However, these methods might not account for the impact of moderate, apparently reversible decreases in renal function that affect hospital-based outcomes or long-term outcomes following patient discharge. Because continuous infusions and aggressive electrolyte management entail additional resources and costs, more research is needed to assess the full economic impact of these approaches.
STUDY LIMITATIONS AND STRENGTHS
When interpreting results, clinicians should consider the strengths and limitations of our study. To our knowledge, this was the first study to use real-world data to compare adverse events and length of stay (LOS) in patients with invasive fungal infections with clinical conditions warranting use of LF-AMB who began CAB therapy but later were switched to LF-AMB, compared with patients whose only amphotericin B therapy was LF-AMB.
Although a formal cost evaluation was not within the scope of this work, our findings regarding adverse events and LOS might be relevant to future pharmacoeconomic analyses comparing these treatment groups.
As with any observational study, the potential exists for residual confounding following multivariate analysis. Selection bias might have affected the selection of the initial therapy beyond the specific variables that we measured; such unobserved confounders might play a role. However, we minimized confounding by including a wide variety of demographic characteristics, comorbidities, and encounter events as possible independent variables in the models.
Our EMR–based data source included treatment characteristics that are not available in administrative claims-based analyses, such as drug-based timing and dosing variables (e.g., post-amphotericin B LOS, total dose of amphotericin B). The EMR data also supported the evaluation of relationships between drug therapy and clinical laboratory results.
A limitation of this study was the absence of explicit documentation as to why patients who initiated therapy with CAB were subsequently switched to LF-AMB; the HIPAA-compliant database does not contain progress notes. We could infer that the switch was made because of nephrotoxicity or other adverse events, but it is possible that the switched patients might not have been representative of all patients who began with CAB. In addition, the policies for use of LF-AMB are likely to differ in various hospitals contributing to the database and this could have affected treatment patterns of patients in facilities where there were barriers to accessing LF-AMB agents.
Finally, as for infusion reactions, institutions may have different policies regarding pretreatment, which could have influenced our findings. We did not have information about these practice patterns.
CONCLUSION
When an amphotericin B agent is required to treat invasive fungal infections, clinicians should consider the consequences of well-established, treatment-specific adverse events, especially nephrotoxicity, as well as the impact on hospital LOS. Although lipid-based amphotericin B (LF-AMB) agents have shown equivalent efficacy and reduced toxicity compared with conventional amphotericin B deoxycholate (CAB), they are often reserved as secondary agents, presumably because of the differences in drug-acquisition costs.
The question remains as to whether a strategy of initiating with CAB and then switching to LF-AMB, once renal or other toxicity occurs, provides a cost-effective approach to the use of these agents. In this study, there were clinical and economic consequences for patients with clinical circumstances warranting the use of LF-AMB but who initially received CAB and were later switched to LF-AMB. Further evaluation of the clinical and overall economic impact of specific therapeutic approaches is needed to help inform treatment decisions.
Footnotes
Disclosure. This work was supported by Astellas Pharma US. Paresh Chaudhari is a full-time employee of Astellas Pharma US. Rolin L. Wade and Robert J. Taylor are employees and stockholders of Cerner Research, which owns the Health Facts database and provides consulting services to pharmaceutical and biotechnology companies. Cerner Research received payment for consulting services delivered to Astellas in connection with the conduct of this study and the development of this manuscript. Jaime L. Natoli is a senior consultant at Kaiser Permanente in Los Angeles and was an employee and stockholder of Cerner Research at the time this study was conducted. Brian H. Nathanson’s company, OptiStatim LLC, has a paid consulting agreement with Cerner Research. David Horn served as a paid consultant to Astellas during the study and manuscript preparation.
REFERENCES
- 1.Pappas PG, Kauffman CA, Andes D, et al. Clinical practice guidelines for the management of candidiasis: 2009 update by the Infectious Diseases Society of America. Clin Infect Dis. 2009;48(5):503–535. doi: 10.1086/596757. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Perfect JR, Dismukes WE, Dromer F, et al. Clinical practice guidelines for the management of cryptococcal disease: 2010 update by the Infectious Diseases Society of America. Clin Infect Dis. 2010;50(3):291–322. doi: 10.1086/649858. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Enoch DA, Ludlam HA, Brown NM. Invasive fungal infections: A review of epidemiology and management options. J Med Microbiol. 2006;55(Part 7):809–818. doi: 10.1099/jmm.0.46548-0. [DOI] [PubMed] [Google Scholar]
- 4.Gallis HA, Drew RH, Pickard WW. Amphotericin B: 30 years of clinical experience. Rev Infect Dis. 1990;12(2):308–329. doi: 10.1093/clinids/12.2.308. [DOI] [PubMed] [Google Scholar]
- 5.Grasela TH, Jr, Goodwin SD, Walawander MK, et al. Prospective surveillance of intravenous amphotericin B use patterns. Pharmacotherapy. 1990;10(5):341–348. [PubMed] [Google Scholar]
- 6.Clements JS, Jr, Peacock JE., Jr Amphotericin B revisited: Reassessment of toxicity. Am J Med. 1990;88(5N):22N–27N. [PubMed] [Google Scholar]
- 7.Saliba F, Dupont B. Renal impairment and amphotericin B formulations in patients with invasive fungal infections. Med Mycol. 2008;46(2):97–112. doi: 10.1080/13693780701730469. [DOI] [PubMed] [Google Scholar]
- 8.Walsh TJ, Hiemenz JW, Seibel NL, et al. Amphotericin B lipid complex for invasive fungal infections: Analysis of safety and efficacy in 556 cases. Clin Infect Dis. 1998;26(6):1383–1396. doi: 10.1086/516353. [DOI] [PubMed] [Google Scholar]
- 9.Mehta J, Kelsey S, Chu P, et al. Amphotericin B lipid complex (ABLC) for the treatment of confirmed or presumed fungal infections in immunocompromised patients with hematologic malignancies. Bone Marrow Transplant. 1997;20(1):39–43. doi: 10.1038/sj.bmt.1700842. [DOI] [PubMed] [Google Scholar]
- 10.Sharkey PK, Graybill JR, Johnson ES, et al. Amphotericin B lipid complex compared with amphotericin B in the treatment of cryptococcal meningitis in patients with AIDS. Clin Infect Dis. 1996;22(2):315–321. [PubMed] [Google Scholar]
- 11.Hiemenz JW, Walsh TJ. Lipid formulations of amphotericin B: Recent progress and future directions. Clin Infect Dis. 1996;22(Suppl 2):S133–S144. doi: 10.1093/clinids/22.supplement_2.s133. [DOI] [PubMed] [Google Scholar]
- 12.Clark AD, McKendrick S, Tansey PJ, et al. A comparative analysis of lipid-complexed and liposomal amphotericin B preparations in haematological oncology. Br J Haematol. 1998;103(1):198–204. doi: 10.1046/j.1365-2141.1998.00934.x. erratum, 1998;103(1):1215. [DOI] [PubMed] [Google Scholar]
- 13.Wong-Beringer A, Jacobs RA, Guglielmo BJ. Lipid formulations of amphotericin B: Clinical efficacy and toxicities. Clin Infect Dis. 1998;27(3):603–618. doi: 10.1086/514704. [DOI] [PubMed] [Google Scholar]
- 14.Mistro S, Maciel IM, de Menezes RG, et al. Does lipid emulsion reduce amphotericin B nephrotoxicity? A systematic review and meta-analysis. Clin Infect Dis. 2012;54(12):1774–1777. doi: 10.1093/cid/cis290. [DOI] [PubMed] [Google Scholar]
- 15.AmBisome (amphotericin B) liposome for injection, package insert. Deerfield, Ill.: Astellas Pharma US; Revised May 2012 Available at: www.astellas.us/docs/ambisome.pdf. Accessed September 5, 2012. [Google Scholar]
- 16.Abelcet (amphotericin B lipid complex injection), package insert. Bridgewater, N.J.: Enzon Pharmaceuticals; Updated October 2010. Available at: http://dailymed.nlm.nih.gov/dailymed/archives/fdaDrugInfo.cfm?archiveid=41233. Accessed September 5, 2012. [Google Scholar]
- 17.Deray G. Amphotericin B nephrotoxicity. J Antimicrob Chemother. 2002;49(Suppl 1):37–41. doi: 10.1093/jac/49.suppl_1.37. [DOI] [PubMed] [Google Scholar]
- 18.Laniado-Laborin R, Cabrales-Vargas MN. Amphotericin B: Side effects and toxicity. Rev Iberoam Micol. 2009;26(4):223–227. doi: 10.1016/j.riam.2009.06.003. [DOI] [PubMed] [Google Scholar]
- 19.Luber AD, Maa L, Lam M, Guglielmo BJ. Risk factors for amphotericin B-induced nephrotoxicity. J Antimicrob Chemother. 1999;43(2):267–271. doi: 10.1093/jac/43.2.267. [DOI] [PubMed] [Google Scholar]
- 20.Fisher MA, Talbot GH, Maislin G, et al. Risk factors for amphotericin B–associated nephrotoxicity. Am J Med. 1989;87(5):547–552. doi: 10.1016/s0002-9343(89)80612-6. [DOI] [PubMed] [Google Scholar]
- 21.Bates DW, Su L, Yu DT, et al. Correlates of acute renal failure in patients receiving parenteral amphotericin B. Kidney Int. 2001;60(4):1452–1459. doi: 10.1046/j.1523-1755.2001.00948.x. [DOI] [PubMed] [Google Scholar]
- 22.Wingard JR, White MH, Anaissie E, et al. A randomized, double-blind comparative trial evaluating the safety of liposomal amphotericin B versus amphotericin B lipid complex in the empirical treatment of febrile neutropenia. L Amph/ABLC Collaborative Study Group. Clin Infect Dis. 2000;31(5):1155–1163. doi: 10.1086/317451. [DOI] [PubMed] [Google Scholar]
- 23.Meunier F, Prentice HG, Ringden O. Liposomal amphotericin B (AmBisome): Safety data from a phase II/III clinical trial. J Antimicrob Chemother. 1991;28(Suppl B):83–91. doi: 10.1093/jac/28.suppl_b.83. [DOI] [PubMed] [Google Scholar]
- 24.Vincent JL, de Mendonça A, Cantraine F, et al. Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: Results of a multicenter, prospective study. Working group on ‘sepsis-related problems’ of the European Society of Intensive Care Medicine. Crit Care Med. 1998;26(11):1793–1800. doi: 10.1097/00003246-199811000-00016. [DOI] [PubMed] [Google Scholar]
- 25.Ullmann AJ, Sanz MA, Tramarin A, et al. Prospective study of amphotericin B formulations in immunocompromised patients in 4 European countries. Clin Infect Dis. 2006;43(4):e29–e38. doi: 10.1086/505969. [DOI] [PubMed] [Google Scholar]
- 26.Austin PC, Tu JV. Bootstrap methods for developing predictive models. Am Stat. 2004;58(2):131–137. [Google Scholar]
- 27.Hosmer DW, Jr, Lemeshow S. Applied Logistic Regression. 2nd ed. New York: John Wiley & Sons; 2000. [Google Scholar]
- 28.Harrell FE., Jr . Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. New York: Springer-Verlag; 2001. [Google Scholar]
- 29.Bates DW, Su L, Yu DT, et al. Mortality and costs of acute renal failure associated with amphotericin B therapy. Clin Infect Dis. 2001;32(5):686–693. doi: 10.1086/319211. [DOI] [PubMed] [Google Scholar]
- 30.Greenberg RN, Cagnoni PJ, Wingard JR, et al. Factors associated with increased hospital costs in patients treated with lipid-based amphotericin B for empirical therapy. J Med Econ. 2002;5:109–118. [Google Scholar]
- 31.Cagnoni PJ, Walsh TJ, Prendergast MM, et al. Pharmacoeconomic analysis of liposomal amphotericin B versus conventional amphotericin B in the empirical treatment of persistently febrile neutropenic patients. J Clin Oncol. 2000;18(12):2476–2483. doi: 10.1200/JCO.2000.18.12.2476. [DOI] [PubMed] [Google Scholar]
- 32.Isnard F, Tilleul P, Laporte JP, et al. A benefit for renal function after early switch to liposomal amphotericin B from conventional formulation in empirical antifungal treatment (letter) Nephrol Dial Transplant. 2007;22(10):3090–3091. doi: 10.1093/ndt/gfm394. [DOI] [PubMed] [Google Scholar]
- 33.Rex JH, Walsh TJ. Estimating the true cost of amphotericin B (editorial response) Clin Infect Dis. 1999;29(6):1408–1410. doi: 10.1086/313555. [DOI] [PubMed] [Google Scholar]
- 34.Falci DR, dos Santos RP, Wirth F, Goldani LZ. Continuous infusion of amphotericin B deoxycholate: An innovative, low-cost strategy in antifungal treatment. Mycoses. 2011;54(2):91–98. doi: 10.1111/j.1439-0507.2009.01805.x. [DOI] [PubMed] [Google Scholar]
- 35.Johnson JR. Reduction of nephrotoxicity associated with amphotericin B deoxycholate. Clin Infect Dis. 2004;38(2):303–307. doi: 10.1086/380842. [DOI] [PubMed] [Google Scholar]
- 36.Schneemann M, Bachli EB. Continuous infusion of amphotericin B deoxycholate: A cost-effective gold standard for therapy of invasive fungal infections? Clin Infect Dis. 2004;38(2):303–304. doi: 10.1086/380843. [DOI] [PubMed] [Google Scholar]