Impact of Continuous Renal Replacement Therapy with Polyacrylonitrile-Derived Filter on Caspofungin Concentration: A Retrospective Study

Abstract

Introduction

Caspofungin pharmacokinetics may be altered in critically ill patients. In vitro studies suggest significant drug adsorption with polyacrylonitrile (PAN) membranes during continuous renal replacement therapy (CRRT).

Methods

This study retrospectively analyzed 66 plasma caspofungin concentrations (pCAS_conc) from 35 ICU patients between 2021 and 2024, comparing those on PAN-CRRT (n = 19) versus those without (n = 47).

Results

Caspofungin is mainly prescribed for candidemia (40%) and invasive Candida infections (25.7%). Median pCAS_conc at 12 h was similar between groups, but at 18 and 24 h, it was significantly higher in the PAN-CRRT group [5.3 (4.1–6.9) vs. 3.3 (1.9–3.7) mg/L, p = 0.04 and 5.3 (3.9–5.8) vs. 3.1 (2.5–4.5) mg/L, p = 0.01, respectively]. ECMO use was more frequent in PAN-CRRT patients (75% vs. 21.7%, p = 0.01). Persistent candidemia occurred in three patients (one in PAN-CRRT, two in non-PAN-CRRT).

Conclusion

This study found no evidence of caspofungin depletion with PAN-CRRT, warranting further research on clinical outcomes.

Key Summary Points

Why carry out this study?

Caspofungin pharmacokinetic may be altered in intensive care unit patients.

Previous studies show no effect of continuous renal replacement therapy on caspofungin pharmacokinetic but few explored the effect of polyacrylonitrile (PAN) membrane dialysis.

Recent in vitro study shows a high adsorption level of caspofungin with PAN membrane and therefore we aim to evaluate if the high adsorption properties of PAN membrane alter the caspofungin concentration in ICU patients.

What was learn from the study?

We did not find total plasma caspofungin concentrations decrease in patients treated with PAN-CRRT compared to those without.

Because our study was retrospective and low powered, further research is needed to confirm these results and to explore the potential clinical impact of PAN-CRRT on caspofungin treatment.

Introduction

Caspofungin is a first-line treatment for candidemia [1], and may be used for invasive aspergillosis when polyenes and triazoles are contraindicated. Due to its high plasma protein binding (> 97%) [2] and low renal clearance, dose adjustments are not recommended in acute kidney injury. However, intensive care unit (ICU) patients show reduced caspofungin exposure compared to healthy subjects [3], influenced by factors such as body weight and albuminemia [4]. Optimization strategies include weight-based dosing [3], higher loading or daily doses [5], and therapeutic drug monitoring [6].

Continuous renal replacement therapy (CRRT) typically does not alter plasma caspofungin concentrations (pCAS_conc), regardless of modality (hemofiltration, hemodiafiltration) [7–10]. However, caspofungin adsorption by dialysis membranes remains controversial. Clinical studies have only assessed caspofungin exposure during CRRT with polysulfone membranes, which exhibit minimal adsorptive capacity, while polyacrylonitrile (PAN) membranes, known for their high adsorptive properties, lack clinical evaluation.

Baud et al. recently conducted two experimental studies demonstrating a significant reduction in caspofungin concentrations in a crystalloid solution after 6 h of PAN-CRRT, due to adsorption rather than effluent elimination, even at higher doses [11, 12]. Based on these in vitro findings and despite the absence of clinical data, the French National Agency for the Safety of Drugs and Health Products (ANSM) advised against the use of caspofungin in PAN-CRRT patients, due to potential subtherapeutic drug concentrations and treatment failure [13].

We hypothesized that pCAS_conc would be lower in PAN-CRRT patients. This study aimed to compare caspofungin concentrations in PAN-CRRT patients to those not undergoing PAN-CRRT.

Methods

This retrospective observational study was conducted from January 1, 2021 to October 2, 2024, in the medical and surgical ICUs of a French University hospital. Adult patients with at least one plasma caspofungin concentration (pCAS_conc) measurement within 24 h of the last administration were included. Patients on intermittent hemodialysis (IHD) were classified in the non-PAN-CRRT group, as IHD in our units does not use PAN membranes. The French Intensive Care Society Ethics Committee approved the study (CE SRLF no. 24–093, IRB no. 00014135), and written informed consent was waived per French regulations. The study was performed in accordance with the Helsinki Declarations and all its amendments.

Therapeutic drug monitoring of caspofungin is routine in both ICUs. Samples were mainly taken within the hour before the next infusion (C_min), a validated surrogate for AUC_0-24. Since collection times varied, measurements were also analyzed in predefined intervals (H12 ± 1 h, H18 ± 1 h, H24).

Total pCAS_conc was determined using liquid chromatography-tandem mass spectrometry (Quantis; ThermoFisher) with the Chromsystem Antimycotic Drugs kit. All quality control parameters met European Medicines Agency (EMA) bioanalytical validation standards.

CRRT was performed using the Prismax/Prismaflex^® (Vantive) system with the PAN ST150 membrane. Hemodiafiltration or hemofiltration was at the clinician’s discretion. Clinical, biological, and CRRT characteristics, along with caspofungin dosing and clinical outcomes, were extracted from electronic records.

Persistent candidemia was defined as a positive blood culture with the same Candida species ≥ 5 days after the initial culture [14].

Quantitative data are presented as medians (IQRs) and compared using Student’s t test or Mann–Whitney U test. Qualitative variables were analyzed using the chi-squared or Fisher’s exact test. Statistical significance was set at p < 0.05. Analyses were performed using SPSS (v.18.0; SPSS) and GraphPad Prism^® (v.9.0; GraphPad Software).

Results

Patients’ Characteristics

Between January 2021 and October 2024, 40 patients with 76 samples were screened. The main reason for exclusion was incorrect sampling time (n = 5 patients and 10 samples). A total of 35 patients with 66 available pCAS_conc were included. The cohort comprised 20 female patients (57.1%) and 15 male patients (42.9%), with a median age of 44 (41–69) years. The primary ICU admission diagnoses were acute respiratory distress syndrome (ARDS) (10/35, 28.6%) and septic shock (10/35, 28.6%). The median Simplified Acute Physiology Score II (SAPSII) at ICU admission was 54 (33–70) (Table 1).

Table 1.

Patient’s characteristics

	Missing data	All (n = 35)
Biological sex, n (%)	0
Male		15 (42.9%)
Female		20 (57.1%)
Age (years), median [IQR]	0	44 [41–69]
SAPS2, median [IQR]	0	54 [33–70]
Weight (kg), median [IQR]	0	80 [65–94]
Diagnosis at ICU admission	0
ARDS		10 (28.6%)
Septic shock		10 (28.6%)
Toxic epidermal necrolysis		3 (8.6%)
Cardiogenic shock		3 (8.6%)
Respiratory failure		2 (5.7%)
Cardiac arrest		2 (5.7%)
Sickle disease complications		2 (5.7%)
Hemorragic shock		1 (2.9%)
Post-operative cardiac surgery		1 (2.9%)
Pancreatitis		1 (2.9%)
Types of fungal infection	0
Candidemia		14 (40%)
Probabilistic		7 (20%)
Invasive aspergillosis		5 (14.3%)
Mediastinitis		2 (5.7%)
Necrotizing soft tissue infection		2 (5.7%)
Peritonitis		2 (5.7%)
Invasive intrasvascular device infection		2 (5.7%)
Pleural effusion		1 (2.9%)
Identified pathogen	0
Candida albicans		9 (25.7%)
Candida parapsilosis		5 (14.3%)
Aspergillus fumigatus		5 (14.3%)
Nakaseomyces glabratus		4 (11.4%)
Pichia kudriavzevii		3 (8.6%)
Other (Clavispora lusitaniae, Candida tropicalis)		2 (5.7%)
Saprochaete capitata		1 (2.9%)
No documentation		6 (17.1%)
Multiple fungi identified		3 (8.6%)
Persisting candidemia	0	3/14 (21.4%)
Septic shock at caspofungin introduction	0	22 (62.9%)
Death during ICU	0	12 (34.3%)

n number of cases, SAPS2 Simplified Acute Physiology Score II, ARDS acute respiratory distress syndrome

Plasma Caspofungin Concentrations Data

Caspofungin was prescribed for proven candidemia (14/35, 40%), empirical treatment (7/35, 20%), combination therapy for aspergillosis (5/35, 14.3%), and invasive candidiasis (n = 9/35, 25.7%). The main species identified were Candida albicans and C. parapsilosis. At caspofungin initiation, 22 patients (62.9%) had septic shock, and ICU mortality was 34.3%.

A total of 19 samples in 9 patients were collected during PAN-CRRT and 47 samples in 31 patients without PAN-CRRT (22 without RRT and 9 on IHD). Samples were distributed as follows: 17 pCAS_conc at 12 h, 18 at 18 h, and 31 at 24 h. Ten non-PAN-CRRT samples were collected during an IHD period with a median of 12 (7–18) h after the end of the last IHD session.

Among PAN-CRRT samples (n = 19), 12 (63.2%) were collected during continuous veno-venous hemofiltration (CVVH) and 7 (36.8%) during continuous veno-venous hemodiafiltration (CVVHDF), with a median dialysis dose of 25 (25–28) mL/kg/h. The median exposure to CRRT between caspofungin infusion and sampling time was 18 (17–22) h. The median duration of filter use prior to sampling, available for 10/19 (52.6%) PAN-CRRT samples, was 31.5 (21.5–63.5) h (35.5 (23.8–47.3) h at H12, n = 2; 48.5 (36.5–55.5) h at H18 n = 4; and 36.8 (21.8–31.3) h at H24, n = 4). Patients without PAN-CRRT and without IHD had a median plasma creatinine of 43 (36–74) µmol/L and a glomerular filtration rate of 125 (93–144) mL/min.

Table 2 summarizes pCAS_conc and patients characteristics by group. Caspofungin treatment duration and doses did not significantly differ between groups. In the 24-h measurement group, more PAN-CRRT patients were treated with extra-corporeal membrane oxygenation (ECMO) and aspartate aminotransferase was significantly higher. In the 18-h measurement groups, transaminases and bilirubin levels were higher in PAN-CRRT patients, though not significantly. Albuminemia was similar between groups.

Table 2.

Characteristics at different times of pCAS_conc sampling by CRRT status

	Missing data	Without PAN-CRRT (No. of samples = 47)	With PAN-CRRT (No. of samples = 19)	p value
Characteristics of pCASconc at H12		n = 12	n = 5
Days between caspofungin initiation and sample collection	0	9 [5–13]	17 [6–19]	0.37
Dose (mg/day)	0	78 [60–100]	70 [70–70]	0.55
Dose (mg/kg/j)	0	0.94 [0.8–1.1]	1 [1–1]	0.75
ECMO at measurement	0	3 (25%)	4 (80%)	0.10
SGOT (UI/L)	0	54 [36–78]	45 [37–71]	0.87
SGPT (UI/L)	0	57 [36–120]	35 [27–39]	0.29
Bilirubin (µmol/L)	0	12.5 [6.5–37]	6 [5–16]	0.49
PT (%)	2	75 (64–83)	66 (53–74)	0.412
Albumin (g/L)	9	19 [15–24]	16 [13–23]	0.77
Characteristics of pCASconc at H18		n = 12	n = 6
Days between caspofungin initiation and sample collection	0	8.5 [3–24.5]	9.5 [3–18]	0.96
Dose (mg/day)	0	70 [50–70]	70 [50–70]	0.80
Dose (mg/kg/j)	0	0.83 [0.76–1]	0.84 [0.78–1]	> 0.99
ECMO at measurement	0	4 (33.3%)	1 (16.7%)	0.62
SGOT (UI/L)	1	34 [28–50]	177 [25–248]	0.17
SGPT (UI/L)	1	37 [25–70]	212 [22–274]	0.27
Bilirubin (µmol/L)	1	8 [5–9]	254 [6–425]	0.12
PT (%)	3	77 (73–89)	71 (59–79)	0.254
Albumin (g/L)	3	27 [15–34]	25 [21–27]	0.77
Characteristics of pCASconc at H24		n = 23	n= 8
Days between caspofungin initiation and sample collection	0	8 [4–22]	9 [5–20]	0.80
Dose (mg/day)	0	70 [70–70]	70 [70–110]	0.32
Dose (mg/kg/j)	0	0.89 [0.85–1]	1 [0.91–1]	0.39
ECMO at measurement	0	5 (21.7%)	6 (75%)	0.01
SGOT (UI/L)	0	47 [30–67]	141 [69–250]	0.02
SGPT (UI/L)	0	40 [32–54]	66 [32–139]	0.33
Bilirubin (µmol/L)	0	7 [4–18]	18 [7–34]	0.33
PT (%)	2	72 (64–82)	67 (60–72)	0.354
Albumin (g/L)	9	19 [17–27]	17 [14–22]	0.23

5 patients are both in the without PAN-CRRT and the with PAN-CRRT samples groups. Bold p-values are considered significant

CRRT continuous renal replacement therapy, ECMO extra-corporeal membrane oxygenation, n number of samples, PT prothrombin time data are expressed as number (percentage), median [IQR], n (%)

Figure 1 shows pCAS_conc across sampling times groups. In the 12-h measurement group, median pCAS_conc did not significantly differ between patients with and without PAN-CRRT [6.0 (5.7–7.7) mg/L vs. 5.7 (3.4–7.8) mg/L, p = 0.45]. However, median pCAS_conc was higher in PAN-CRRT patients in the 18-h group [5.3 (4.1–6.9) mg/L vs. 3.3 (1.9–3.7) mg/L, p = 0.04] and in the 24-h group [5.3 (3.9–5.8) mg/L vs. 3.1 (2.5–4.5) mg/L, p = 0.01].

Fig. 1.

pCAS_conc at H12, H18, and H24 according to CRRT status. CRRT continuous renal replacement therapy, pCAS_concplasma caspofungin concentration, H12, H18, and H24 number of hours between last infusion and measurement. *p < 0.05; ns = p > 0.05

To explore the potential impact of treatment duration on plasma caspofungin concentrations, we assessed the correlation between the number of days of caspofungin treatment and pCAS_conc at H12, H18, and H24 in patients with and without PAN-CRRT (Fig. 2). No significant correlation was observed in any subgroup. Spearman correlation coefficients were low and not statistically significant in any panel, suggesting that treatment duration did not influence pCAS_conc at these time points.

Fig. 2.

Correlation between days of caspofungin treatment and plasma caspofungin concentrations (pCAS_conc) at H12, H18, and H24 in patients with and without PAN-CRRT. Scatter plots show the relationship between the duration of caspofungin treatment (in days) and plasma caspofungin concentrations (pCAS_conc, in mg/L) at different sampling time points. a, c, and e correspond to patients without PAN-CRRT at H12, H18, and H24, respectively; b, d, and f correspond to patients with PAN-CRRT at the same time points. Each figure part includes a linear trend line and reports the Spearman correlation coefficient (r) along with the two-tailed p value

Among the 14 patients with candidemia, 11 were in the group without PAN-CRRT and 3 in the group with PAN-CRRT. Persistent candidemia occurred in 3 patients: 2/11 (18.2%) in the group without PAN-CRRT and 1/3 (33%) in the group with PAN-CRRT. The latter had an intravascular device infection (ECMO cannula) that could not be removed.

Discussion

This is the first clinical study assessing plasma caspofungin concentrations (pCAS_conc) in patients undergoing CRRT with PAN membranes. Contrary to in vitro data, we found no decrease in pCAS_conc and no apparent increase in persistent candidemia rates.

Unexpectedly, pCAS_conc at H18 and H24 were significantly higher in PAN-CRRT patients. Several hypotheses may explain this finding. First, liver impairment is known to affect caspofungin pharmacokinetics [15], and liver function tests differ between groups, although non-significantly, but may have contributed. Second, a PAN filter drug-release mechanism, previously demonstrated with gentamicin [16], could also apply to caspofungin and, since the filters had been in use for extended periods prior to sampling (i.e., more than 30 h), membrane saturation may also have limited drug adsorption. However, these data were limited and prevent coming to a definitive conclusion about adsorption kinetic over time. Lastly, selection bias due to the small sample size cannot be excluded.

Baud et al.’s in vitro study [11] assessed only the free fraction of caspofungin in a protein-free solution, limiting its clinical relevance. While our study did not directly evaluate free fraction adsorption, evidence suggests that echinocandin membrane adsorption is much lower in blood than in saline [17]. Further studies assessing the free fraction in PAN-CRRT patients are needed.

Our study has several strengths. All the patients received a higher-than-standard caspofungin dose, aligning with ICU pharmacokinetics recommendations. In both groups, samples were collected predominantly from day 8 onwards after caspofungin initiation, which likely ensured steady-state concentrations and reduced the risk of underestimation [18]. However, this timing may not fully capture the early phase of treatment, during which significant pharmacokinetic variations can occur even if we found no correlation between days of treatment and variation of pCAS_conc. Key pharmacokinetics factors such as body weight and albuminemia [4] were analyzed, showing no significant differences. Additionally, persistent candidemia rates were comparable between groups, with no apparent negative impact of PAN-CRRT, despite the low number of cases.

However, this study has limitations. As a retrospective study, it is subject to inherent methodological biases. Multiple confounding factors, such as fluid loading or ECMO-related caspofungin adsorption [19], may have influenced pCAS_conc, but the small sample size prevented further subgroup analysis. Furthermore, none of the pCAS_conc samples were collected within the 12 h following PAN filter replacement, preventing assessment of caspofungin pharmacokinetic during this early phase. Further studies should evaluate caspofungin concentrations at different time points after filter change to better characterize the kinetic of adsorption. Finally, we measured only total plasma caspofungin concentrations, as recommended [6], without assessing unbound drug levels. While caspofungin is highly protein-bound (97%), some argue that only the free fraction is responsible for antifungal activity. Previous studies have indicated that reduced plasma AUC₀₋₂₄ or C_min are not always linked to lower efficacy in ICU patients, who often have hypoalbuminemia and may exhibit a higher free fraction [20].

Conclusion

This study is the first to assess plasma caspofungin concentrations in ICU patients on PAN-CRRT. The findings suggest no pCAS_conc reduction, but the lack of free fraction data limits evaluation of PAN membranes’ impact. Larger prospective studies are needed to clarify caspofungin pharmacokinetics, clinical outcomes, and treatment failure risks. Until further research, the French National Agency for the Safety of Drugs and Health Products (ANSM) recommendations should be followed.

Author Contributions

Romain Arrestier and Keyvan Razazi contributed to the conceptualization and writing of the article, data collection and statistical analysis. Claire Pressiat, Anne Hulin and Laura Bouabdallah contributed to data collection and reviewing of the article. Armand Mekontso Dessap, Paul Masi and Nicolas Mongardon contributed to reviewing of the article. All the authors had access to the data. Romain Arrestier and Keyvan Razazi accessed the original data and vouch for its authenticity. All authors read and approved the final manuscript.

Funding

No funding or sponsorship was received for this study. The journal’s Rapid Service fee was funded by the Assistance Publique – Hôpitaux de Paris (APHP).

Data Availability

All data generated or analyzed during this study are included in this published article.

Declarations

Conflict of Interest

Romain Arrestier reports personal fees from Pharma Dom, MSD and Baxter outside the submitted work. Armand Mekontso Dessap reports grants from Fisher&Paykel, Baxter, Philips, Ferring and GSK, personal fees from Air Liquide, Baxter, Amomed, Getinge and Addmedica, outside the submitted work. Keyvan Razazi received lecture fees from MSD, Shionogi, and a travel grant from Pfizer outside the submitted work. Nicolas Mongardon received personal fees from AOP Health and Baxter, and research grants from Air Liquide outside the submitted work.

Ethical Approval

The French Intensive Care Society Ethics Committee approved the study (CE SRLF n° 24–093, IRB n° 00014135), and written informed consent was waived per French regulations. The study was performed in accordance with the Helsinki Declarations and all its amendments.