Evaluation of Total Body Weight versus Adjusted Body Weight Liposomal Amphotericin B Dosing in Obese Patients

ABSTRACT

Liposomal amphotericin B (LAmB) is used for various fungal infections, but it is unclear which dosing weight to use in obese patients. The purpose of this study was to compare clinical outcomes of adjusted body weight (adjBW) versus total body weight (TBW) dosing of LAmB. This single-center, retrospective cohort study included patients who received LAmB for definitive therapy and whose TBW exceeded 120% of their ideal body weight (IBW). Analyses were conducted for 3 mg/kg for adjBW versus TBW and 5 mg/kg for adjBW versus TBW. A total of 238 patients were included. For the 68 patients who received LAmB at 3 mg/kg, there were no differences in safety or efficacy outcomes. For the 170 patients who received LAmB at 5 mg/kg, significantly more patients in the TBW group experienced the primary outcome of nephrotoxicity (57% versus 35% [P value of 0.016]) and had significantly higher rates of early discontinuation of LAmB due to toxicity (33% versus 17% [P = 0.030]). There was a trend toward increased 90-day mortality in the adjBW group (60% versus 45% [P = 0.079]); however, adjBW dosing was not associated with increased mortality in an adjusted model. Given the lower rates of nephrotoxicity but a possible trend toward increased mortality in patients whose TBW exceeds 120% of their IBW, dosing LAmB by adjBW may be reasonable in patients who are not critically ill and who have lower-risk infections. In critically ill patients or those with fungal pathogens or sites of infection associated with higher mortality risk, dosing by TBW can be considered.

INTRODUCTION

Liposomal amphotericin B (LAmB) is an antifungal agent with activity against a wide variety of fungi, including endemic fungi, yeasts, and molds. The typical dose is 3 to 6 mg/kg of body weight once daily depending on the infection, but whether dosing should be based on total body weight (TBW) or adjusted body weight (adjBW) in obese patients is uncertain. The volume of distribution of LAmB is low, at 0.1 to 0.16 liters/kg (1), indicating a lack of extensive distribution to the fat. This finding suggests that adjBW dosing may be sufficient in obese patients.

Animal studies (2–4) have shown significant increases in amphotericin B serum concentrations in obese rats and rabbits, with increased rates of nephrotoxicity. These data suggest that ideal body weight (IBW) or adjBW dosing may be appropriate. In one human pharmacokinetic study of 16 obese patients given LAmB, the authors recommended, based on Monte Carlo simulations, a dose cap of 100 kg (e.g., 300 mg for 3-mg/kg dosing and 500 mg for 5-mg/kg dosing) (5). Taken together, these limited previous studies suggest that dosing by adjBW in obese patients may help minimize toxicity while having similar efficacy outcomes. However, to our knowledge, there have been no studies to date comparing clinical outcomes in patients receiving adjBW versus TBW dosing of LAmB for invasive fungal infections. Therefore, the purpose of this study was to compare safety and efficacy outcomes in obese patients receiving LAmB dosed by adjBW versus TBW.

RESULTS

A total of 808 patients with orders for LAmB between 1 June 2008 and 31 November 2019 for definitive therapy were screened. From this list, 238 patients were ultimately included, with the most common exclusions occurring as a result of the TBW not exceeding 120% of the IBW or receipt of LAmB doses other than 3 or 5 mg/kg (Fig. 1).

FIG 1.

Study population. q24h, every 24 h.

Baseline characteristics are shown in Table 1. Patients had a median age of 57 years and were primarily Caucasian. The median weight was 93 kg, and the median body mass index (BMI) was 31 kg/m². The majority of patients were immunosuppressed, most commonly because of cancer on chemotherapy or receipt of immunosuppressants. Most patients received an infectious disease consult, and 42% were admitted to the intensive care unit (ICU). The most common infections were invasive candidiasis, cryptococcosis, histoplasmosis, and aspergillosis, with central nervous system (CNS) infections comprising 14% of all infections. In the 3-mg/kg comparator groups, more patients dosed by TBW were immunosuppressed, while patients in the adjBW group had higher Charlson comorbidity indexes; the 3-mg/kg groups were otherwise similar. In the 5-mg/kg comparator groups, patients dosed by adjBW had higher median weights and BMIs as well as higher Charlson comorbidity indexes but fewer Candida infections than patients dosed by TBW. The 5-mg/kg groups were otherwise similar. The median inpatient duration of LAmB was 6 days, with 22% of patients discharged from the hospital on LAmB (Table 2).

TABLE 1.

Baseline patient and disease characteristics by treatment group^a

Characteristic	Value for group
	All	3 mg/kg			5 mg/kg
		TBW (57 patients)	adjBW (11 patients)	P value	TBW (116 patients)	adjBW (54 patients)	P value
Median age (yrs) (IQR)	57 (47–64)	56 (49–66)	55 (47–60)	0.708	59 (48–64)	56 (44–64)	0.366
No. (%) of male patients	123 (51.7)	28 (49.1)	7 (63.6)	0.514	62 (53.4)	26 (48.1)	0.520
No. (%) of patients of race				0.458			0.734
White	184 (77.3)	44 (77.2)	7 (63.6)		93 (80.2)	40 (74.1)
Black	35 (14.7)	9 (15.8)	2 (18.2)		14 (12.1)	10 (18.5)
Asian	3 (1.3)	0	0		2 (1.7)	2 (1.9)
Unknown	16 (6.7)	4 (7)	2 (18.2)		7 (6)	3 (5.6)
Median wt (kg) (IQR)	93 (81–104)	95 (80–103)	93 (88–108)	0.532	90 (80–102)	98 (81–108)	0.028
Median BMI (kg/m2) (IQR)	31 (29–35)	31 (29–35)	32 (28–36)	0.980	31 (29–34)	32 (30–37)	0.080
No. (%) of immunosuppressed patients	139 (58.4)	38 (64.9)	3 (27.3)	0.020	64 (55.2)	34 (63)	0.339
Cancer on chemotherapy	85 (35.7)	19 (33.3)	1 (9.1)	0.155	42 (36.2)	23 (42.6)	0.425
Immunosuppressant	64 (26.9)	24 (42.1)	3 (27.3)	0.506	25 (21.6)	12 (22.2)	0.921
HIV	12 (5)	1 (1.8)	0	1.000	7 (6)	4 (7.4)	0.745
High-dose steroid	10 (4.2)	1 (1.8)	0	1.000	6 (5.2)	3 (5.6)	1.000
SOT	20 (8.4)	9 (15.8)	1 (9.1)	1.000	6 (5.2)	4 (7.4)	0.727
No. (%) of patients with other comorbidities
Malignancy	102 (42.9)	20 (35.1)	4 (36.5)	1.000	51 (44)	27 (50)	0.462
CKD	46 (19.3)	7 (12.3)	3 (27.3)	0.347	22 (19)	14 (25.9)	0.301
BMT	35 (14.7)	6 (10.5)	3 (27.3)	0.154	16 (13.8)	10 (18.5)	0.494
Cirrhosis	20 (8.4)	2 (3.5)	2 (18.2)	0.120	12 (10.3)	4 (7.4)	0.779
Median Charlson comorbidity index (IQR)	4 (2–5)	3 (1–5)	5 (3–6)	0.031	4 (2–5)	5 (3–7)	0.042
No. of patients with ID consult/total no. of patients (%)b	193/214 (90.2)	50/54 (92.6)	7/9 (77.8)	0.201	94/103 (91.3)	42/48 (87.5)	0.472
No. (%) of patients with pathogen isolatedc
Candida spp.	67 (28.2)	19 (33.3)	3 (27.3)	1.000	25 (21.6)	20 (37)	0.041
Cryptococcus	47 (19.7)	13 (22.8)	1 (9.1)	0.437	18 (15.5)	15 (27.8)	0.060
Histoplasma	41 (17.2)	17 (29.8)	2 (18.2)	0.715	19 (16.4)	3 (5.6)	0.053
Aspergillus spp.	32 (13.4)	4 (7)	0	1.000	20 (17.2)	8 (14.8)	0.825
Zygomycetes	20 (8.4)	1 (1.8)	1 (9.1)	0.299	14 (12.1)	4 (7.4)	0.432
Coccidioides	6 (2.5)	2 (3.5)	1 (9.1)	0.416	3 (2.6)	0	0.552
Fusarium	15 (6.3)	2 (3.5)	1 (9.1)	0.416	9 (7.8)	3 (5.6)	0.754
Blastomyces	15 (6.3)	2 (3.5)	1 (9.1)	0.416	10 (8.6)	2 (3.7)	0.343
Other	7 (2.9)	1 (1.8)	1 (9.1)	0.299	4 (3.4)	1 (1.9)	1.000
Unknown organism	8 (3.4)	0	0		6 (5.2)	2 (3.7)	1.000
No. (%) of patients with site of infectionc
CNS infection	33 (13.9)	5 (8.8)	1 (9.1)	1.00	16 (13.5)	11 (20.4)	0.367
Pulmonary	96 (40.3)	21 (40.1)	2 (18.2)	0.194	47 (40.5)	24 (44.4)	0.629
Other site	142 (59.7)	35 (61.4)	8 (72.7)	0.734	70 (60.3)	29 (53.7)	0.414
No. (%) of patients with ICU admission	99 (41.6)	19 (33.3)	4 (36.4)	1.00	49 (42.2)	27 (50)	0.344
Median initial SCr (mg/dl) (IQR)	0.83 (0.65–1.07)	0.87 (0.73–1.15)	1.0 (0.61–1.21)	0.952	0.80 (0.57–1.00)	0.90 (0.66–1.19)	0.100

^aAbbreviations: adjBW, adjusted body weight; BMI, body mass index; BMT, bone marrow transplant; CKD, chronic kidney disease; CNS, central nervous system; HIV, human immunodeficiency virus; ICU, intensive care unit; ID, infectious diseases; IQR, interquartile range; SOT, solid-organ transplant; TBW, total body weight.

^bMissing data for 21 (9.8%) patients.

^cSome patients had >1 pathogen and/or >1 site of infection.

TABLE 2.

Characteristics of liposomal amphotericin B and concurrent medications^a

Characteristic	Value for group
	All	3 mg/kg			5 mg/kg
		TBW (57 patients)	adjBW (11 patients)	P value	TBW (116 patients)	adjBW (54 patients)	P value
Median inpatient duration (days) (IQR)	6 (3–14)	4 (2–7)	5 (4–14)	0.182	7 (3–15)	8 (3–14)	0.718
No. (%) of patients discharged on amphotericin	52 (21.8)	15 (26.3)	3 (27.3)	1.000	28 (24.1)	6 (11.1)	0.048
Median saline bolus (ml/day) (IQR)	625 (82–1,000)	733 (0–1,000)	225 (0–1,000)	0.401	636 (216–1,000)	500 (19–1,000)	0.420
No. (%) of patients with concurrent medications
Other antifungal	99 (41.6)	18 (31.6)	3 (27.3)	1.000	53 (45.7)	25 (43.3)	0.941
Other nephrotoxin	115 (48.3)	29 (50.9)	5 (45.5)	1.000	51 (44)	30 (55.6)	0.159
Other HypoK inducers	80 (33.6)	16 (28.1)	3 (27.3)	1.000	39 (33.6)	22 (40.7)	0.368
Other HypoMg inducers	80 (33.6)	25 (43.9)	5 (45.5)	1.000	46 (39.7)	24 (44.4)	0.555

^aAbbreviations: adjBW, adjusted body weight; HypoK, hypokalemia; HypoMg, hypomagnesemia; TBW, total body weight.

Outcomes for the 3-mg/kg and 5-mg/kg analyses are shown in Table 3. For the 3-mg/kg analysis, 28% of the TBW group experienced nephrotoxicity, compared to 33% of the adjBW group. The difference was not statistically significant, with a P value of 0.71. There were also no significant between-group differences for any of the secondary outcomes (Table 3 and Fig. 2).

TABLE 3.

Outcomes by treatment group^a

Outcome	% missing	Value for group
		3 mg/kg			5 mg/kg
		TBW (57 patients)	adjBW (11 patients)	P value	TBW (116 patients)	adjBW (54 patients)	P value
No. (%) of patients with nephrotoxicity		15/53 (28.3)	3/9 (33.3)	0.71	58/102 (56.9)	15/43 (34.9)	0.016
Mild		5/15 (33.3)	1/3 (33.3)	1.00	19/58 (32.8)	8/15 (53.3)	0.229
Moderate		5/15 (33.3)	1/3 (33.3)	1.00	25/58 (43.1)	5/15 (33.3)	0.567
Severe		5/15 (33.3)	1/3 (33.3)	1.00	14/58 (24.1)	2/15 (13.3)	0.497
No. (%) of patients with hypokalemia	0.8	16 (28.1)	4 (3.2)	0.719	70/114 (61.4)	36/54 (66.7)	0.51
Median potassium nadir (mmol/liter) (IQR)		2.9 (2.6–3.1)	2.8 (2.6–3.0)	0.536	2.9 (2.7–3.0)	3 (2.7–3.2)	1.0
No. (%) of patients with hypomagnesemia	1.7	1/55 (1.8)	1/11 (9.1)	0.308	18/114 (15.8)	4/54 (7.4)	0.15
Median magnesium nadir (mg/dl) (IQR)b					1.2 (1–1.3)	1.2 (0.9–1.3)	0.859
No. (%) of patients with composite safety data	0.8	25 (43.9)	7 (63.6)	0.325	89/114 (77.4)	39/54 (72.2)	0.465
No. (%) of patients with dose/frequency reduction for toxicity	4.2	3/54 (5.6)	1/10 (10)	0.50	8/110 (7.3)	3/54 (5.6)	1.00
No. (%) of patients with early discontinuation for toxicity	8.9	13/52 (25)	1/10 (10)	0.43	36/110 (32.7)	9/54 (16.7)	0.030
60-day mortality rate [no. (%) of patients]	4.6	23/53 (43.4)	4/11 (36.4)	0.748	49/112 (43.8)	27/51 (52.9)	0.275
90-day mortality rate [no. (%) of patients]	5.5	24/53 (45.3)	5/11 (45.5)	1.00	50/111 (45)	30/50 (60)	0.079
In-hospital all-cause mortality rate [no. (%) of patients]		22 (38.6)	4 (36.4)	1.00	42 (36.2)	26 (48.1)	0.14
In-hospital fungal-infection-associated mortality rate [no. (%) of patients]	0.4	10 (17.5)	3 (27.3)	0.428	19/115 (16.5)	10/54 (18.5)	0.748
No. (%) of patients with readmission within 60 days		11/35 (31.4)	4/7 (57.1)	0.20	29/74 (39.2)	14/28 (50)	0.37
No. (%) of patients with readmission within 90 days		11/35 (31.4)	4/7 (57.1)	0.20	32/74 (43.2)	15/28 (53.6)	0.38
Median length of stay (days) (IQR)		12 (7–31)	18 (11–54)	0.109	25 (12–43)	31 (16–49)	0.077

^aAbbreviations: adjBW, adjusted body weight; TBW, total body weight.

^bMagnesium nadir data are not presented for the 3-mg/kg analyses as there was only one patient in each group with hypomagnesemia, and there were therefore no median (IQR) data for magnesium nadirs.

FIG 2.

Kaplan-Meier survival curve for 3 mg/kg LAmB.

For the 5-mg/kg analysis (Table 3), TBW dosing was associated with significantly more nephrotoxicity than adjBW dosing: 57% versus 35% (P value of 0.016). In the adjBW group, the majority of nephrotoxicity cases were mild, whereas for TBW dosing, there was a numerically higher proportion of moderate and severe nephrotoxicity cases than in the adjBW group, although these differences did not achieve statistical significance. The Kaplan-Meier curve for the probability of nephrotoxicity showed an increased risk for the TBW group at every time point, with a greater increase in risk as the LAmB duration increased; however, this was not statistically significant (Fig. 3). There were no differences in incidences of hypokalemia, hypomagnesemia, or time to hypokalemia or hypomagnesemia (Kaplan-Meier curves not shown). There was, however, a significantly higher early-discontinuation rate due to LAmB toxicity in the TBW group: 33% versus 17% (P = 0.03). For the efficacy outcomes, there were no significant differences between adjBW dosing and TBW dosing for any of the mortality, time-to-mortality (Fig. 4), readmission, or length-of-stay outcomes. Although not a statistically significant difference, there was a trend toward increased mortality on the Kaplan-Meier curve and increased 90-day mortality in the adjBW group.

FIG 3.

Nephrotoxicity Kaplan-Meier curve for 5 mg/kg LAmB.

FIG 4.

Kaplan-Meier survival curve for 5 mg/kg LAmB.

In the competing-risk analysis for nephrotoxicity in the 5-mg/kg groups (Table 4), after adjusting for confounders, dosing according to TBW was found to significantly increase the risk for nephrotoxicity, with a hazard ratio of 1.65 (95% confidence interval [CI], 1.10 to 2.48) (P value of 0.015). Other significant associations included a lower risk of nephrotoxicity in patients with Candida infections and in non-ICU patients and a higher risk of nephrotoxicity in those receiving a concomitant aminoglycoside.

TABLE 4.

Nephrotoxicity competing-risk analysis for the 5-mg/kg dose^a

Variable	Hazard ratio	95% CI	P value
Dosing wt: TBW vs adjBW	1.65	1.10–2.48	0.015
Candida infection	0.55	0.32–0.95	0.033
Admitting unit: non-ICU	0.63	0.42–0.95	0.027
Aminoglycoside	2.13	1.16–3.94	0.015

^aAbbreviations: adjBW, adjusted body weight; ICU, intensive care unit; TBW, total body weight.

A Cox regression analysis was conducted for mortality in the 5-mg/kg comparisons to confirm whether trends toward increased mortality persisted in this analysis. The results are shown in Table 5. Dosing by adjBW was not significantly associated with increased mortality (hazards ratio, 1.09 [95% CI, 0.61 to 1.94]) (P value of 0.779). Variables that were associated with increased mortality risk include increased BMI, cirrhosis, bone marrow transplant status, pulmonary infection, and ICU admission.

TABLE 5.

Multivariate Cox analysis of the 5-mg/kg dose^a

Characteristic	Hazard ratio	95% CI	P value
Dosing wt: adjBW vs TBW	1.09	0.61–1.94	0.779
Age (yrs)
18–40	1.00
41–60	1.58	0.96–6.38	0.299
>60	2.47	0.62–4.03	0.061
BMI	1.08	1.02–1.14	0.007
Cancer on chemotherapy	1.55	0.70–3.42	0.281
Immunocompromised	1.35	0.59–3.12	0.479
Cirrhosis	3.04	1.31–7.07	0.010
BMT	2.06	1.07–3.96	0.030
CCI	1.09	0.98–1.20	0.122
ID consult	0.85	0.39–1.86	0.691
Cryptococcus or endemic fungi	0.73	0.36–1.49	0.388
CNS infection	1.09	0.43–2.78	0.852
Pulmonary infection	2.07	1.18–3.64	0.012
ICU admission	2.34	1.26–4.31	0.007
Concomitant other antifungal	0.72	0.39–1.33	0.298
Length of stay	0.99	0.99–1.00	0.306

^aAbbreviations: adjBW, adjusted body weight; BMI, body mass index; BMT, bone marrow transplant; CCI, Charlson comorbidity index; CNS, central nervous system; ICU, intensive care unit; ID, infectious diseases; TBW, total body weight [χ²(16) = 55 (P < 0.001)].

DISCUSSION

In this retrospective cohort analysis, for LAmB dosed at 5 mg/kg, dosing by adjBW was found to be safer, with less nephrotoxicity and lower rates of medication discontinuation for toxicity, than dosing by TBW. The lower rates of nephrotoxicity with adjBW dosing persisted in the competing-risk analysis. Dosing by adjBW was also not significantly different in efficacy compared to TBW dosing in the 5-mg/kg analyses. There was, however, a trend toward increased 90-day mortality and increased mortality in the Kaplan-Meier analysis in the adjBW group. A Cox regression analysis was therefore conducted on the 5-mg/kg groups and found no increased association with mortality based on dosing weight, but the possibility remains that the trend toward increased mortality with adjBW dosing reflects a real risk that we were underpowered to detect. For LAmB dosed at 3 mg/kg, there were no differences in any safety or efficacy outcomes between adjBW and TBW dosing. This finding may be due to smaller sample sizes and the resultant lack of power. Overall, the study results suggest that dosing LAmB by adjBW may be reasonable in patients who are not critically ill and who have lower-risk infections. Given the trend toward increased mortality with adjBW dosing, in critically ill patients or those with fungal pathogens or sites of infection that are associated with higher mortality risk, dosing by TBW can be considered.

To our knowledge, this is the first study comparing clinical outcomes in patients dosed by adjBW versus TBW for LAmB. Previous studies were either animal studies or human pharmacokinetic studies. In a study by Vadiei et al. (2), obese rats given amphotericin B deoxycholate were found to have high areas under the curve (AUCs) and resultant increases in serum creatinine (SCr) from the baseline compared to normal-weight rats. Similarly, in a rabbit study by Koldin et al. (3), hypercholesteremic rabbits given amphotericin B lipid complex were also found to have higher AUCs and increased SCr from the baseline. In another rabbit model, however, hypercholesteremic rabbits given amphotericin B deoxycholate had no differences in SCr elevations or mortality compared to rabbits with normal cholesterol (4). Collectively, these studies suggest that obese animals given the same weight-based doses as nonobese animals have significantly increased amphotericin B exposures with associated increased toxicities and that dosing by adjBW may therefore be preferred. However, overall results were mixed; in several of the studies, animals were hypercholesteremic rather than obese, and there are limitations to the ability to extrapolate animal data to humans.

There has been one pharmacokinetic study of LAmB in obese patients, in which Monte Carlo simulations were done on 16 morbidly obese patients given LAmB dosed by TBW (5). Based on their models, those authors recommended a dose cap at 100 kg of body weight (e.g., 300 mg for 3-mg/kg dosing and 500 mg for 5-mg/kg dosing). These results similarly suggest that for obese patients, TBW dosing may not be the optimal strategy. This study was small, however, and did not assess clinical outcomes.

Consistent with the results of these previous studies, our study found that dosing by TBW in obese patients was associated with significantly higher rates of nephrotoxicity, with this association also being seen after adjusting for confounders in the competing-risk multivariate analysis. Rates of nephrotoxicity were relatively high in our study, occurring in 44% of patients overall. In previous studies of LAmB dosed from 3 to 6 mg/kg daily (6–8), rates of nephrotoxicity from LAmB ranged from 18.7% to 47%, depending on the indication and dosage. The incidence in our study is consistent with the ones in these previous studies but falls within the higher end of the range, possibly due to a sicker patient population or the inclusion of obese patients only and therefore higher absolute doses.

The consideration of a lower incidence of nephrotoxicity development with adjBW dosing should be balanced with the trend toward increased mortality risk seen in our study. The numbers were overall small, and it is possible that we were simply underpowered to detect a statistically significant difference in mortality. Our study population was also highly heterogeneous in terms of illness severity and fungal pathogen; the possibility of increased mortality risk is of particular concern in critically ill patients and those with higher-mortality-risk infections such as mucormycoses. In a previous trial of LAmB in patients with cryptococcal meningitis (6), LAmB dosed at 3 mg/kg daily was associated with rates of 10-week mortality similar to those of LAmB at 6 mg/kg daily (13.9% versus 9.5%); although statistical analyses between these two groups were not conducted, the authors of that study concluded that 3 mg/kg daily may be sufficient for cryptococcal meningitis. The mortality rates in this previous study, however, were much lower than the 48% overall 90-day mortality risk seen in our study. Dosing differences likely matter more when mortality risk is higher. In another study of 44 immunocompromised patients with Aspergillus, Fusarium, and Zygomycetes infections, patients received either 7.5 mg/kg, 10 mg/kg, 12.5 mg/kg, or 15 mg/kg of LAmB daily (9). The groups were too small to conduct statistical analyses on safety outcomes between groups. Overall, 50% of patients experienced nephrotoxicity, but no patients required renal replacement therapy, and there was no dose-limiting nephrotoxicity in any of the dosing groups. While nephrotoxicity risk is an important consideration when choosing LAmB doses, when the mortality risk is very high, the balance of risk versus benefit may favor more aggressive dosing strategies. Therefore, dosing LAmB by adjBW may be reasonable in patients who are not critically ill and who have lower-risk infections. In critically ill patients or those with fungal pathogens or sites of infection that are associated with higher mortality risk, dosing by TBW can be considered.

Adding to the currently available literature, our study was the first to examine clinical outcomes of different dosing weights in a relatively large cohort of patients. Our study population also included a substantial portion of critically ill and immunocompromised patients, increasing the applicability of the results to these patient populations.

Our study did, however, have several limitations. First, this was a single-center retrospective study with the inherent associated limitations and possible confounders. There were several between-group differences in baseline characteristics that may have affected outcomes. Although our heterogeneous patient population increased study applicability, the inclusion of many different fungal pathogens and patient populations may also have increased the risk of confounding. To account for these limitations, we conducted a competing-risk multivariate analysis for nephrotoxicity and a multivariate analysis for mortality, which confirmed the results seen in the primary analyses. In addition, we found differences in safety outcomes for the 5-mg/kg analyses but not for the 3-mg/kg analyses, likely due to the very small size and, therefore, the lack of power in the 3-mg/kg groups. Finally, laboratory values to analyze rates of nephrotoxicity, hypomagnesemia, and hypokalemia were assessed only for the inpatient duration of LAmB due to limited access to retrospective outpatient data. It is possible that the development of nephrotoxicity, hypomagnesemia, or hypokalemia may have been missed in patients who developed these outcomes as outpatients. However, a minority of patients, 22%, were discharged on LAmB. In addition, we most likely missed only minor cases of outpatient LAmB adverse events. If any of these toxicities were severe enough to warrant readmission, unless this readmission occurred at a facility outside the health system, this toxicity would have then been captured and counted. A final consideration is the TBW of patients in our study: the median TBW was 93 kg, with a maximum weight of 160 kg. Dosing beyond this weight was not able to be assessed in our study; additional studies may be warranted to determine appropriate dosing strategies in patients beyond this body weight.

In conclusion, dosing liposomal amphotericin B by adjBW was associated with less nephrotoxicity but also a trend toward increased mortality compared to dosing by TBW in obese patients. Based on our findings, for patients whose TBW exceeds 120% of their IBW, dosing LAmB by adjBW may be reasonable in patients who are not critically ill and who have lower-risk infections. In critically ill patients or those with fungal pathogens or sites of infection that are associated with higher mortality risk, dosing by TBW can be considered.

MATERIALS AND METHODS

This was a single-center, retrospective, observational cohort study conducted at Barnes-Jewish Hospital in St. Louis, MO. The study was approved by the Washington University Institutional Review Board.

All adult patients who had orders for LAmB between 1 June 2008 and 30 November 2019 for definitive therapy were screened. Definitive therapy was defined as treatment for a fungal infection confirmed by positive microbiology, including the following: positive fungal cultures except for Candida species in respiratory or urinary specimens; positive Histoplasma or Cryptococcus antigen; or positive Coccidioides, Histoplasma, or Blastomyces antibodies. Patients with ICD-9 or ICD-10 codes for invasive fungal infection were also included to capture patients who had positive fungal microbiology in the outpatient setting or at outside hospitals prior to transfer.

Patients were included if their TBW exceeded 120% of their IBW and if they were treated with either 3 mg/kg or 5 mg/kg of LAmB every 24 h dosed based on either adjBW or TBW. Patients were excluded if there was no recorded height available in the patient chart, if the LAmB frequency was greater than every 24 h, if LAmB was ordered but never given, or if there was no positive fungal microbiology. Patients who were on renal replacement therapy at the start of LAmB were excluded from the nephrotoxicity analysis.

For the time period included in the analysis, LAmB dosing weight was left up to the provider’s discretion, with no standardized internal guidance on which dosing weight to choose. Separate analyses were performed comparing 3-mg/kg adjBW versus TBW and 5-mg/kg adjBW versus TBW groups. Doses of between 2.5 and 3.5 mg/kg were counted as 3 mg/kg, while doses of between 4.5 and 5.5 mg/kg were counted as 5 mg/kg. Calculations were made for both adjBW and TBW, and all doses that did not fall within these parameters were excluded. The patient weight used for all calculations was the weight nearest LAmB initiation. The daily dose administered for the majority of the LAmB duration was used in the analyses.

Study definitions.

Patients were considered immunosuppressed if they had any of the following: HIV with CD4 counts of less than 200 cells/mm³, cancer on chemotherapy with resultant neutropenia, history of solid-organ transplant, receipt of immunosuppressants within the last 30 days, or receipt of methylprednisolone at 1 mg/kg daily or the equivalent dose of another steroid.

ICU admission was defined as a patient being located in the ICU at the time of the first LAmB order. Time-to-toxicity and time-to-mortality outcomes were assessed with the day of the initial LAmB order as day 1. When examining 60- and 90-day readmission data, readmissions to any hospital within the health system were captured.

Nephrotoxicity was defined as an increase in serum creatinine (SCr) to ≥1.5× the baseline and was further categorized as mild nephrotoxicity if SCr was elevated to ≥1.5× the baseline but <2× the baseline, moderate nephrotoxicity if SCr was elevated to ≥2× the baseline but <3× the baseline, or severe nephrotoxicity if SCr was elevated to ≥3× the baseline or if renal replacement therapy was initiated. Hypokalemia was defined as a potassium level of <3.3 mmol/liter at any point during inpatient LAmB treatment, while hypomagnesemia was defined as a magnesium level of <1.4 mg/dl at any time during inpatient LAmB treatment. Potassium or magnesium nadirs were analyzed only for patients who developed hypokalemia or hypomagnesemia, respectively. Laboratory values, including SCr, potassium, and magnesium, were assessed up until the last inpatient day of LAmB therapy for each patient.

Outcomes.

The primary outcome was the incidence of nephrotoxicity. Secondary safety outcomes included time to nephrotoxicity; time to and incidence of hypokalemia; time to and incidence of hypomagnesemia; magnesium and potassium nadirs; the composite safety outcome of nephrotoxicity, hypokalemia, or hypomagnesemia; dose or frequency reduction for toxicity; or early discontinuation for toxicity. Efficacy outcomes included all-cause 60- and 90-day mortality, all-cause and fungal-infection-related in-hospital mortality, time to mortality, readmission within 60 and 90 days, and hospital length of stay.

Statistical analyses.

Fisher’s exact or chi-square tests were used for categorical data, Student’s t tests or Mann-Whitney U tests were used for continuous data, and Kaplan-Meier analyses were used for time-to-event data, as appropriate. A 2-sided P value of <0.05 was considered statistically significant for all tests.

A Cox competing-risk model was used to estimate the risk of nephrotoxicity. In-hospital mortality was considered a competing risk. A multivariate Cox proportional-hazards model was also conducted for the secondary outcome of mortality. Univariate analyses were conducted on variables in Tables 1 and 2. With the exception of age, continuous covariates such as LAmB dose and initial creatinine clearance (CrCl) were included in the models as linear variables. Covariates with a P value of <0.15 on univariate analysis were retained in the final multivariate model; all covariates were assessed for collinearity. Goodness of fit was assessed using the Omnibus test of model coefficients for the Cox regression analysis.