Comparative analysis of the nephrotoxicity of liposomal amphotericin B and amphotericin B lipid complex in hematological cancer patients: A multicenter retrospective study

Abstract

Background

The incidence rates of acute kidney injury (AKI) associated with amphotericin B lipid complex (ABLC) and liposomal amphotericin B (L-AMB) are inconsistent across studies.

Aims

This study aimed to assess the AKI incidence rates in the largest cohort of patients undergoing ABLC and L-AMB treatment for invasive fungal diseases (IFD) in individuals with hematological cancers.

Methods

This was a multicenter, international, retrospective cohort study involving patients treated with either L-AMB or ABLC for IFD. All patients had a diagnosed malignant hematological disease. Various clinical and epidemiological variables were examined, including the concurrent use of nephrotoxic drugs. The primary outcome was the incidence of AKI.

Results

A total of 637 patients were included in the study, with 294 patients in the ABLC group and 343 patients in the L-AMB group. The most common diagnosis was acute leukemia (56%), followed by lymphoma (22%). The predominant classification of IFD was probable (43%). ABLC was associated with an increased likelihood of inducing grade 1 and 2 nephrotoxicity compared to L-AMB (p < 0.001). Multivariate analysis identified age, vancomycin, and polymyxin use as independent risk factors for AKI. However, serum creatinine levels returned to baseline in 95.3% of patients.

Conclusion

Nephrotoxicity associated with L-AMB was lower than that associated with ABLC in patients with hematological cancer. Most AKI cases were mild to moderate and did not have significant short-term impact.

Resumen

Antecedentes

La tasa de incidencia de la insuficiencia lesión renal aguda (IRA) debida tanto al complejo lipídico de anfotericina B (ABLC) como a la anfotericina B liposomal (L-AMB) son divergentes en varios estudios.

Objetivos

Este estudio ha tenido como objetivo evaluar las tasas de incidencia de IRA en pacientes con cáncer hematológico e infección fúngica invasiva (IFI) tratados con ABLC y L-AMB.

Métodos

Se ha realizado un estudio de cohortes retrospectivo, multicéntrico e internacional con pacientes con IFI bajo tratamiento con L-AMB o ABLC. Todos los pacientes presentaban diagnóstico de enfermedades hematológicas malignas. Se examinaron diversas variables clínicas y epidemiológicas, incluido el uso concomitante de fármacos nefrotóxicos. La variable principal fue la aparición de IRA.

Resultados

Se incluyeron en el estudio 637 pacientes, 294 en el grupo ABLC y 343 en el grupo L-AMB. La enfermedad más frecuente fue la leucemia aguda (56%), seguida del linfoma (22%), y la IFI se clasificó como “probable” en el 43% de los casos. La ABLC se asoció con una mayor probabilidad de sufrir nefrotoxicidad en los grados 1 y 2 (p < 0,001). Un análisis multivariante reveló que la edad, y el uso de vancomicina y polimixina, eran factores de riesgo independientes para desarrollar IRA. Sin embargo, las concentraciones séricas de creatinina volvieron a los valores basales en el 95,3% de los casos.

Conclusiones

La nefrotoxicidad de la L-AMB fue menor que la de la ABLC en pacientes con cáncer hematológico. La mayoría de los casos de IRA fueron de leves a moderados, sin repercusiones significativas a corto plazo.

In patients with invasive fungal diseases (IFDs), treatment with amphotericin B can lead to nephrotoxicity, which increases hospital stay, healthcare costs, and the risk of mortality.¹⁶ Some studies have reported nephrotoxicity rates exceeding 50% in patients receiving the conventional deoxycholate amphotericin B formulation (dAMB).^21,25

Lipid-based formulations of amphotericin B are associated with lower nephrotoxicity compared to dAMB.¹⁹ The most commonly used lipid formulations are liposomal amphotericin B (L-AMB) and amphotericin B lipid complex (ABLC).¹⁰ Previous studies comparing the incidence of acute kidney injury (AKI) among patients treated with lipid formulations and those receiving dAMB have shown that L-AMB and ABLC reduce the risk of AKI by approximately 50% compared to dAMB.⁴ However, comparisons between ABLC- and L-AMB-associated AKI have yielded inconsistent results. Some studies suggest that ABLC and L-AMB are associated with varying degrees of AKI.¹⁸

In 2010, Safdar et al. conducted a meta-analysis comparing AKI incidence between L-AMB and ABLC therapies. While they noted heterogeneity in patient populations and outcome measures, they did not reach a definitive conclusion.¹⁷ One clinical trial found a higher incidence of AKI with ABLC compared to L-AMB; however, this finding was not consistent with several retrospective studies. Notably, all previous studies were conducted in centers located in the United States or Europe,³¹ and most were single-center studies. AKI incidence rates are likely to vary across institutions depending on patient populations, concomitant medication use, and supportive care protocols, including hydration and electrolyte supplementation.⁸

Although several systematic reviews, with and without meta-analyses, have been published, none have included a sufficiently large population to robustly evaluate the difference in AKI incidence rates between ABLC and L-AMB therapies. This multicenter study aimed to assess the incidence of AKI in patients with possible, probable, or confirmed IFDs undergoing ABLC or L-AMB therapy following hematopoietic stem cell transplantation or chemotherapy for acute leukemia.

1

Methods

1.1

Study design

This was a multicenter, international retrospective cohort study including patients recruited between January 2003 and January 2021 from 18 centers across eight countries: Argentina, Brazil, Chile, Greece, Iran, the Kingdom of Bahrain, Turkey, and Peru. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist for methodological reporting.

1.2

Ethical considerations

The primary ethical concern in this retrospective, non-interventional study was the protection of patients’ personal data. All personal information was de-identified in the dataset. The study was approved by the local Ethics Review Boards at participating centers, which waived the requirement for written informed consent due to the retrospective and observational nature of the study. The study was conducted in accordance with the Basic & Clinical Pharmacology & Toxicology policy for experimental and clinical studies.²⁹

1.3

Inclusion and exclusion criteria

The patients had to meet all the following inclusion criteria: aged > 16 years; have had a treatment with L-AMB or ABLC between 2003 and 2020; have had a diagnosis of acute myeloid leukemia, or myelodysplastic syndrome, having received hematopoietic stem cell transplantation; and have been treated with L-AMB or ABLC for IFD (possible, probable or confirmed) for at least 5 days. Drug selection was determined by the treating physician at each center.

Exclusion criteria included prior use of dAMB, missing data necessary for calculating AKI incidence, and the use of ABLC or L-AMB as prophylaxis.

1.4

Variables

The following clinical and epidemiological variables were collected: sex, concomitant use of nephrotoxic agents (furosemide, aminoglycosides, vancomycin, and polymyxins), comorbidities, concomitant therapies, body weight (measured or estimated based on height), specification of fungal disease if not empirical, daily serum creatinine levels, and duration and dosage of ABLC or L-AMB therapy. Fungal disease was classified according to the European Organization for Research and Treatment of Cancer/Invasive Fungal Infections Cooperative Group/National Institute of Allergy and Infectious Diseases Mycoses Study Group (EORTC/MSG) criteria.⁷

The primary outcome was AKI, defined according to the Kidney Disease Improving Global Outcomes (KDIGO) criteria.¹⁴ Patients were monitored for 15 days following the initiation of amphotericin B therapy. This time frame was selected to evaluate AKI associated with initial therapy, excluding the cumulative dose. AKI was defined as an absolute increase in serum creatinine of ≥0.3 mg/dL or a ≥50% increase from baseline. Patients diagnosed with AKI were further stratified into three categories:

• •

  Group 1 (mild AKI): serum creatinine increase ≥0.3 mg/dL or ≥150–200% from baseline;

• •

  Group 2 (moderate AKI): serum creatinine increase > 200–300%;

• •

  Group 3 (severe AKI): serum creatinine increase > 300% or ≥4.0 mg/dL, based on KDIGO criteria.

1.5

Data sources

All data were extracted from patient medical records at the participating centers.

1.6

Statistical analysis

A sample size calculation indicated that a non-inferiority design comparing AKI incidence required 300 patients per group (ABLC and L-AMB), assuming prior AKI rates as reported in the literature.^2,6 The study was powered at 80% with a 95% confidence interval and a non-inferiority margin of 10%.

Categorical variables were expressed as percentages, while continuous variables were presented as arithmetic means with standard deviations (SD), according to data distribution. Associations with AKI were assessed using Student's t-test, Chi-square test, or Fisher's exact test as appropriate. A p-value < 0.05 was considered statistically significant.

Multivariate analysis was performed using binary logistic regression, including variables that were significant in univariate analysis. Linear regression was used to analyze temporal changes in serum creatinine, with R² values calculated. Multiple models were constructed, each including a single covariate, to identify potential independent risk factors for AKI. Statistical analyses were conducted using SPSS version 23.

2

Results

A total of 637 patients were included in the study. Among them, 294 received ABLC therapy, while 343 received L-AMB therapy. The mean doses were 3.92 ± 1.18 mg/kg for ABLC and 4.13 ± 1.22 mg/kg for L-AMB. The baseline clinical characteristics of the two groups were generally comparable and are summarized in Table 1. Most patients were male, with a mean age of 44.5 ± 17.8 years in the ABLC group and 42 ± 18.6 years in the L-AMB group. Acute leukemia was the most frequent underlying disease (148 in the ABLC group and 207 in the L-AMB group), followed by lymphoma (98 and 50 patients, respectively). The most common classification of IFD was empirical, followed by confirmed (83 in the ABLC group and 90 in the L-AMB group). The most frequent IFDs were candidiasis and aspergillosis.

Table 1.

Clinical characteristics of patients treated with amphotericin B lipid complex (ABLC) or liposomal amphotericin B (L-AMB). Key variables include demographic data, underlying hematologic malignancies, type of invasive fungal disease (IFD), and concomitant use of nephrotoxic agents.

Data	ABLC	%	L-AMB	%	p-value
Male gender	170	58%	200	58%	0.483
Chemotherapy
Bussulfan	11	4%	20	6%	0.204
Ciclofosfamide	97	33%	89	26%	0.003
Timoglobulin	16	5%	16	5%	0.283
Fludarabin	39	13%	63	18%	0.137
Melfalan	2	1%	35	10%	<0.001
Ciclosporin	21	7%	53	15%	0.005
MMF	12	4%	26	8%	0.103
Rituximab	25	9%	15	4%	0.008
Cisplatin	1	0%	6	2%	0.122
Citarabin	108	37%	133	39%	0.224
Daunorubicin	71	24%	60	17%	<0.001
Antibiotics
Amikacin	58	20%	163	48%	<0.001
Vancomycin	199	68%	241	70%	0.251
Polymyxin B	148	50%	106	31%	<0.001
Diuretics	237	81%	147	43%	<0.001
Tranplant
Autologous	7	2%	56	16%	<0.001
Alogenic	33	11%	35	10%	0.516
Stem cells	7	2%	38	11%	0.106
Underlying disease
Acute leukemia	148	50%	207	60%	0.006
Chronic leukemia	22	7%	8	2%	0.002
Lymphoma	91	31%	50	15%	<0.001
Myeloma	18	6%	19	6%	0.444
Aplastic anemia	6	2%	12	3%	0.192
Others	17	5%	40	11%
Lenght of stay	45.6 ± 97.2		55.6 ± 74.1		0.143
Age	44.5 ± 17.8		42.0 ± 18.6		0.096
Weight	64.8 ± 16.9		63.7 ± 17.1		0.981
Invasive fungal disease classification
Confirmed	83	28%	90	26%
Probably	20	7%	65	19%
Possible	35	12%	71	21%
Empirical	156	53%	117	34%
Etiology
Candidiasis	29	10%	14	4%
Aspergillosis	13	4%	25	7%
Cryptococal disease	11	4%	3	1%
Fusariosis	10	3%	10	3%
Mucormycosis	2	1%	26	8%
Others	18	6%	12	3%

Although the overall clinical characteristics of the ABLC and L-AMB groups were similar, some differences were observed regarding the concomitant use of potentially nephrotoxic agents during amphotericin administration. The L-AMB group had a higher frequency of amikacin use, whereas the ABLC group had a greater use of polymyxin B and loop diuretics. Additionally, the L-AMB group included more patients with lymphoma and chronic leukemia, which led to differences in chemotherapy regimens between the groups. While there were differences in IFD etiology, the frequencies were not statistically significant.

Mean creatinine levels during the first 14 days of treatment in both groups are shown in Fig. 1. Patients were stratified by the degree of nephrotoxicity. Initial creatinine levels were significantly higher in the ABLC group (1.14 ± 1.04 mg/dL) compared to the L-AMB group (0.97 ± 0.93 mg/dL; p = 0.020). Given the complexity of patients and the peak in serum creatinine observed between days 4 and 8—followed by stabilization—linear regression analysis was conducted. In patients who developed AKI, serum creatinine increased in both groups: ABLC (R² = 0.07) and L-AMB (R² = 0.003). In patients who did not develop AKI, creatinine levels decreased significantly in both groups: ABLC (R² = 0.05) and L-AMB (R² = 0.001).

Fig. 1.

Mean serum creatinine levels over 14 days in patients receiving ABLC or L-AMB, stratified by the presence or absence of AKI. Dashed lines indicate linear regression trends. Patients with AKI showed a significant increase in creatinine over time, while those without AKI demonstrated a significant decrease (p < 0.05).

The incidence of AKI was 35% in the ABLC group and 20.1% in the L-AMB group. According to nephrotoxicity grading, ABLC was significantly more likely to cause grade 1 AKI (≥50% increase from baseline) compared to L-AMB (35% vs 20.1%; p < 0.001). The incidence of grade 2 AKI (≥200% of baseline creatinine) was also higher in the ABLC group (21.1%) than in the L-AMB group (6.7%; p < 0.001). A binary logistic regression model was used for this unadjusted analysis, accounting for the fact that more patients in the ABLC group received nephrotoxic agents such as diuretics and polymyxin B, while more patients in the L-AMB group received amikacin. The mean doses of ABLC and L-AMB were not associated with increased AKI risk. Grade 3 AKI (≥300% baseline creatinine without dialysis) occurred in 5.5% and 4.7% of patients in the ABLC and L-AMB groups, respectively. Overall, 6.4% of patients required dialysis (6.8% in the ABLC group and 6.1% in the L-AMB group).

Additional risk factors for AKI beyond amphotericin use are detailed in Table 2. Univariate analysis identified significant associations with age, certain chemotherapeutic agents (e.g., fludarabine, cyclophosphamide), and nephrotoxic drugs (e.g., loop diuretics, polymyxin B, vancomycin). In multivariate binary logistic regression, age, vancomycin, and polymyxin B were independently associated with increased AKI risk. When nephrotoxic co-medications were analyzed by AKI grade (AKI 1 and AKI 2) for each amphotericin formulation, L-AMB was consistently associated with a lower risk of AKI, even in the presence of amikacin, polymyxin B, or diuretics (Table 3).

Table 2.

Risk factors associated with acute kidney injury (AKI) in patients treated with ABLC or L-AMB. The results of the univariate and multivariate analyses evaluating the demographic and treatment-related variables are included.

	Without AKI (n = 465)		With AKI (n = 172)
Data	n	%	n	%	p-value
Male gender	268	58%	102	59%	0.387
Chemotherapy
Bussulfan	25	5%	6	3%	0.27
Ciclofosfamide	127	27%	59	34%	0.019
Timoglobulin	22	5%	10	6%	0.289
Fludarabin	85	18%	17	10%	0.012
Melfalan	29	6%	8	5%	0.365
Ciclosporin	58	12%	16	9%	0.242
MMF	28	6%	10	6%	0.527
Rituximab	31	7%	9	5%	0.402
Cisplatin	5	1%	2	1%	0.572
Citarabin	182	39%	59	34%	0.325
Daunorubicin	97	21%	34	20%	0.365
Antibiotics
Amikacin	156	34%	65	38%	0.183
Vancomycin	309	66%	131	76%	0.008
Polymyxin B	164	35%	90	52%	<0.001
Diuretics	257	55%	127	74%	<0.001
Tranplant
Autologous	50	11%	13	8%	0.14
Alogenic	50	11%	18	10%	0.521
Stem cells	36	8%	9	5%	0.289
Underlying disease
Acute leukemia	263	57%	92	53%	0.246
Chronic leukemia	17	4%	13	8%	0.037
Lymphoma	96	21%	45	26%	0.091
Myeloma	26	6%	11	6%	0.42
Aplastic anemia	14	3%	4	2%	0.434
Others	41	8%	15	8%	0.268
Age	41.7 ± 18.2		46.9 ± 18.1		<0.001

Table 3.

Incidence of AKI (Grade 1 and Grade 2, based on KDIGO criteria) among ABLC and L-AMB patients receiving concomitant nephrotoxic drugs. A differential nephrotoxic impact is observed in the presence of polymyxin, loop diuretics, and amikacin.

Variable	OR (CI95%)
	AKI 1	AKI 2
L-AMB	0.47 (032–0.67)	0.35 (0.23–0.53)
Subgroup
Without amikacin	0.32 (0.20–0.53)	0.29 (0.16–0.52)
With amikacin	0.59 (0.31–1.1)	0.32 (0.16–0.63)
Without polymyxin	0.50 (0.31–0.82)	0.33 (0.18–0.6)
With polymyxin	0.54 (0.32–0.92)	0.48 (0.27–0.87)
Without diuretics	0.50 (0.25–1.02)	0.49 (0.21–1.13)
With diuretics	0.61 (0.39–0.96)	0.44 (0.34–0.59)

Among the 637 patients, 85 developed grade 2 AKI, representing an overall incidence of 13.3% (22.1% in the ABLC group and 6.7% in the L-AMB group). Serum creatinine levels returned to baseline in 81 of these 85 patients (95.3%), indicating that AKI was generally transient and resolved without specific intervention. Among these 85 patients, 31 (36.5%) discontinued treatment: 19 (30.1%) in the ABLC group and 12 (52.2%) in the L-AMB group.

3

Discussion

Lipid formulations of amphotericin B have long been used as alternatives to the conventional, nephrotoxic deoxycholate amphotericin B (dAMB).^11,13 Several clinical guidelines recommend the use of ABLC or L-AMB over dAMB in patients with hematologic malignancies.⁵ Although various meta-analyses and narrative reviews have evaluated the nephrotoxicity of these two lipid formulations,^3,17 most lacked sufficient statistical power due to limited sample sizes. The only randomized controlled trial directly comparing ABLC and L-AMB found higher nephrotoxicity in the ABLC group.³¹ However, that study was criticized for not reflecting real-world conditions. In contrast, the present multicenter retrospective cohort study evaluated over 600 patients from diverse geographic and clinical backgrounds to assess the nephrotoxic profiles of ABLC and L-AMB in patients with hematologic cancers.

Our findings demonstrate that ABLC was associated with a significantly higher incidence of nephrotoxicity, including mild and moderate AKI, even after adjusting for confounding variables such as concomitant nephrotoxic drug use. These results align with those of Wingard et al., who prospectively evaluated patients with IFD and reported greater nephrotoxicity with ABLC compared to L-AMB.³¹ The enhanced nephrotoxic potential of ABLC is likely due to its increased accumulation in renal parenchyma during the initial days of treatment.^9,15,30

Patients with hematologic malignancies are predisposed to renal injury from multiple sources, including cancer-related mechanisms such as thrombotic microangiopathy, and the nephrotoxic effects of radiation or chemotherapy.¹² These patients often receive multiple nephrotoxic agents.^2,29 For example, conditioning regimens for hematopoietic stem cell transplantation (HSCT) can influence AKI risk, with up to 33% of patients undergoing myeloablative HSCT requiring renal replacement therapy (RRT), compared to only 4% of those receiving reduced-intensity conditioning.¹²

In our cohort, the widespread use of nephrotoxic antibiotics—particularly vancomycin, polymyxin B, and amikacin—likely reflects the limited availability of less nephrotoxic alternatives in the participating countries. Agents such as linezolid, ceftazidime/avibactam, and ceftolozane/tazobactam, which have more favorable renal safety profiles, remain prohibitively expensive or unavailable in many settings.^22,26–28

The higher nephrotoxicity observed in the ABLC group may also be explained in part by higher baseline serum creatinine levels, indicating possible pre-existing renal impairment. Despite these baseline differences, the higher incidence of AKI in the ABLC group remained statistically significant after adjusting for nephrotoxic drug use. Nonetheless, formulas used to estimate glomerular filtration rate (e.g., Cockcroft–Gault, MDRD, Wright) are known to have limited accuracy in certain patient populations, which is a limitation of this and other retrospective studies.^1,20

Importantly, the overall clinical impact of nephrotoxicity was modest, as the need for dialysis was similar in both groups. Therefore, amphotericin B formulation alone cannot be considered the sole contributor to renal injury. The slightly lower cost of ABLC compared to L-AMB²³ warrants consideration in cost-benefit analyses, particularly in resource-limited settings.²⁴

This study has several limitations. As a retrospective cohort, it was subject to heterogeneity in disease types, chemotherapy protocols, and nephrotoxic medication use. However, the large sample size adds strength to the analysis. Even after adjusting for nephrotoxic drug use (amikacin, polymyxin, and vancomycin), L-AMB was associated with a significantly lower risk of grade 1 and 2 AKI. One limitation was the omission of hypokalemia, a key and early indicator of distal tubular injury that can be refractory and treatment-limiting despite aggressive supplementation. Severe hypokalemia has been reported as a major renal adverse event in earlier clinical trials.⁶ Finally, although amphotericin-induced AKI may be dose-dependent, this study focused on the short-term effects of therapy within the initial 15 days.

4

Conclusion

In summary, L-AMB demonstrated a more favorable renal safety profile than ABLC in patients with hematologic malignancies. Most AKI cases were mild to moderate and had no significant short-term consequences. Concomitant use of nephrotoxic agents such as vancomycin and polymyxin significantly contributed to renal injury risk and should be carefully considered when selecting antifungal therapy.

ICMJE authorship criteria

All authors contributed to the writing of the final manuscript. Specific contributions include:

• •

  Susanne Edinger – Draft and final version of the manuscript.

• •

  Edson Abdala – Data acquisition.

• •

  Umran Şumeyse Ertürk – Data acquisition.

• •

  Serhat Çelik – Data acquisition.

• •

  Leylagül Kaynar – Data acquisition.

• •

  Ricardo Rabagliati – Data acquisition.

• •

  Hamid Badali – Data acquisition.

• •

  Ali Amanati – Data acquisition.

• •

  Alexis Manuel Holguin Ruiz – Data acquisition.

• •

  Joseph Meletiadis – Data acquisition.

• •

  Maria Stamouli – Data acquisition.

• •

  Jessica Fernandes Ramos – Data acquisition.

• •

  Alexandre Vargas Schwarzbold – Data acquisition.

• •

  Juan Pablo Caeiro – Data acquisition.

• •

  Guillermo N. Giordano – Data acquisition.

• •

  Maria N. Gamaletsou – Data acquisition.

• •

  Hugo Morales – Data acquisition.

• •

  Leonardo Filipetto Ferrari – Data acquisition.

• •

  Joao Paulo Telles – Critical review of draft and statistical analysis.

• •

  Felipe Francisco Tuon – Idealization and statistical analysis, draft and final version of the manuscript.

• •

  Luiz Cesar Guarita-Souza – Review final version of the manuscript.

Ethical approval

This study was approved by the Ethical Committee from UFPR (CAAE: 81162817.2.0000.0020).

Consent to participate

Waived by the Ethical Committee.

Consent to publish

All authors approved the final version of this manuscript.

Funding

This study was supported by TEVA.

Conflict of interests

• •

  Felipe Francisco Tuon (TEVA, MSD, Pfizer).

• •

  Juan Pablo Caeiro (Tevas and Gador).

• •

  Jessica Fernandes Ramos (Biotoscana-Knight, Merck Sharp and Dohme, Pfizer).

• •

  Alexandre Vargas Schwarzbold (Fundação Oswaldo Cruz, Pfizer, Glaxo SmithKline, AstraZeneca, Merck Sharp & Dohme, Enanta, F2G, Moderna, Clover).

• •

  Joseph Meletiadis (Gilead, Pfizer).

• •

  Maria Stamouli (Msd, Pfizer, Astellas).

• •

  Edson Abdala (TEVA, Knighttx).

• •

  Ricardo Rabagliati (Pfizer, Biotoscana-Knight, Gador, Gilead).

Other authors have no conflicts of interest to declare.

Availability of data and materials

All data are available from the corresponding author upon reasonable request.

Acknowledgments

• •

  Hugo Morales – Hospital Erasto Gaertner, Curitiba, PR, Brazil.