Daptomycin Pharmacokinetics and Pharmacodynamics in a Pooled Sample of Patients Receiving Thrice-Weekly Hemodialysis

Abstract

While the pharmacokinetic (PK) properties of daptomycin in hemodialysis (HD) patients have been evaluated previously by three groups, resultant dosing recommendations have varied. To address this clinical conundrum, this study combined concentration-time data from these PK evaluations and derived uniform dosing recommendations among patients on HD receiving daptomycin. A two-compartment model with separate HD and non-HD clearance terms was fit to the PK data from these studies by using BigNPAG. Embedded with PK parameters from the population PK analysis, 5,000-subject Monte Carlo simulations (MCS) were performed to identify HD dosing schemes that provided efficacy (cumulative and daily area under the concentration-time curve [AUC] values) and toxicity (trough concentrations of ≥24.3 mg/liter) profiles comparable to those from simulations employing the daptomycin PK model derived from the Staphylococcus aureus bacteremia–infective endocarditis (SAB-IE) study. Separate HD dosing schemes were sought for the two weekly interdialytic periods (48 and 72 h). For the 48-h interdialytic period, intra- and post-HD dosing provided the most isometric cumulative and daily AUCs. For the 72-h interdialytic period, all HD dosing schemes provided noncumulative AUC values from 48 to 72 h (AUC_48–72) that were <50% of the SAB-IE AUC_48–72 values. Increasing the parent dose by 50% intra- or post-HD provided comparable AUC_48–72 values, while maintaining acceptable trough concentration (C_min) values. When efficacy and toxicity profiles were evaluated for each individual study, higher probabilities for C_min reaching ≥24.3 mg/liter were observed in one of the three studies. Given the high probability of C_min being ≥24.3 mg/liter in one of the three studies, more intensive creatine phosphokinase (CPK) monitoring may be warranted in HD patients receiving daptomycin.

INTRODUCTION

Within the United States alone, there are over 450,000 individuals with end-stage kidney disease (1). Of these, more than 300,000 receive thrice-weekly hemodialysis (HD) as their primary mode of kidney replacement therapy. Infections among patients on HD are common and account for nearly 30% of all deaths in this population (2, 3). One of the most common pathogens associated with infections in the HD population is methicillin-resistant Staphylococcus aureus (MRSA) (4, 5). The risk of invasive MRSA infections is 100-fold higher among patients on HD relative to individuals in the general population (4). Furthermore, infections due to MRSA among patients on HD prolong hospitalizations and increase inpatient costs (6).

Daptomycin is a lipopeptide approved for the treatment of Staphylococcus aureus bacteremia and complicated skin and skin structure infections (7), both of which are common in patients receiving HD (1, 2, 6, 8, 9). While the pharmacokinetics (PK) of daptomycin among patients receiving HD have been described by several groups, there are inconsistencies in the HD dosing recommendations offered (10–13). Regardless of the interdialytic period (48 h versus 72 h) during thrice-weekly HD, one study recommended administering a 50% increase in the parent daptomycin dose (4 or 6 mg/kg of body weight, depending on the indication) intra-HD, while another study advocated post-HD administration. In contrast, the study by Patel and colleagues recommended administering the parent dose of daptomycin intra- or post-HD administration for the 48-h interdialytic period and increasing the parent dose by 50% for the 72-h interdialytic period. In light of the discordant thrice-weekly HD dosing recommendations, the objectives of this study were to (i) combine concentration-time data from these previous HD PK evaluations and (ii) derive uniform pharmacokinetic/pharmacodynamic (PK/PD)-optimized HD dosing recommendations. With respect to the HD dosing recommendations, the goal was to identify total body weight-based HD dosing schemes for each FDA-approved daptomycin dosing regimen with efficacy and toxicity profiles comparable to those obtained from Monte Carlo simulations (MCS) employing the daptomycin population PK model derived from patients enrolled in the Staphylococcus aureus bacteremia–infective endocarditis (SAB-IE) clinical study (13–15). Furthermore, optimized dosing schemes were sought for the two interdialytic periods of a typical thrice-weekly HD schedule (48 h and 72 h).

MATERIALS AND METHODS

Pharmacokinetic data acquisition.

This PK/PD systems analysis was performed utilizing concentration-time data from three previous evaluations in patients receiving thrice-weekly HD (reported by Salama et al., Benziger et al., and Patel et al. [10–13]). Details of these studies have been published previously (10–13) and are provided in Table 1. Briefly, two of the studies were single-dose (6 mg/kg on total body weight post-HD, infused over 30 min) PK studies (10, 11, 13). In contrast, the study by Benziger et al. evaluated the PK profile of daptomycin (6 mg/kg of total body weight post-HD, infused over 30 min) administered post-HD after 3 consecutive HD sessions (12). In the study by Benziger et al., patients were excluded if their body mass index was ≤18.5 or ≥40 kg/m² (12). The studies by Salama et al. had similar weight restrictions and only included patients that were within 40% of ideal body weight and >40 kg (10, 11). In the study by Patel et al., patients were eligible for inclusion if their pre-HD weight was within 10% of their prescribed dry weight within 1 week of the study (13). The number of PK blood samples collected per patient differed between studies (Salama et al., 11 samples/patient; Benziger et al., 28 samples/patient; Patel et al., 23 samples/patient) (10–13).

Table 1.

Demographics of hemodialysis patients and dialysis filter characteristics from each study

Parameter	Salama et al. (n = 7)	Benziger et al. (n = 7)	Patel et al. (n = 12)	Overall (n = 26)
Mean (SD) age (yrs)	56 (15)	55.1 (11.7)	61.8 (14.1)	57.7 (13.8)
No. (%) of males	4 (57.1)	5 (71.4)	7 (58.3)	16 (66.7)
No. (%) of Caucasians	2 (28.6)	2 (28.6)	10 (83.3)	14 (58.3)
No. (%) of patients of African racial origin	4 (57.1)	5 (71.4)	2 (16.7)	11 (45.8)
No. (%) of nonanuric patientsa	NRd	NR	4 (33.3)	NA
Mean (SD) wt (kg)	63.1 (13.4)	86.8 (20.3)	81.2 (11.9)	77.8 (17.1)
Dosage regimen	6 mg/kg post-HD, 1 time	6 mg/kg thrice weekly post-HD, 3 times	6 mg/kg post-HD, 1 time
No. of PK samples/patient	11	28	23
Blood/dialysis flow rates (ml/min)	400/700	NR	300/600
Dialysis duration (h)	4	3.5–4	3.5
Characteristics of dialysis filter used
Dialysis filter	Optiflux F-200NR	Various	Optiflux F-180NR
B12 clearance (ml/min)b	206	Varied (mean, 166)	168
KoA for ureac	1,317	Varied (mean, 1,185)	1,239

^aProduction of ≥100 ml of urine over 24-h collection period.

^bDialysis clearance of vitamin B₁₂ for blood and dialysis flow rates of 300 ml/min and 500 ml/min, with the exception of the study by Salama et al., for which values reported are for blood and dialysis flow rates of 400 ml/min and 800 ml/min.

^cMaximum theoretical urea clearance of dialyzer at infinite blood and dialysis flow rates.

^dNR, not reported; NA, not available.

Dialysis regimens varied across studies but were consistent with current standards of care. All dialysis filters used in these studies were considered high-flux dialysis membranes. In the study by Salama et al., patients received a standardized 4-hour HD regimen in which average blood and dialysate flow rates were 400 and 700 ml/min, respectively. An F-200NR dialysis filter (polysulfone membrane; surface area, 2.0 m²; ultrafiltration coefficient [Kuf] of 62 ml/h/mm Hg; Fresenius Medical Care North America) was used (10, 11). Dialysis regimens ranged from 3.5 to 4 h in the study by Benziger et al. The specific type of dialysis filter and blood and dialysate flow rates in the study by Benziger et al. varied among patients as well (12). In the study by Patel et al., patients were standardized to a 3.5-hour HD regimen in which blood and dialysis flow rates were fixed at 300 and 600 ml/min, respectively, and an F-180NR dialysis filter (polysulfone membrane; surface area, 1.8 m², Kuf, 59 ml/h/mm Hg; Fresenius Medical Care North America) was used (13). Additional characteristics of the dialysis filters are provided in Table 1.

Pharmacokinetic data analysis.

All HD PK data were analyzed in a population PK model using BigNPAG (16). PK data from the three studies were pooled in a combined analysis and were also modeled individually. The structural PK model was parameterized as a two-compartment model with zero-order infusion and first-order intercompartmental transfer and elimination. Both non-HD clearance (CL) and HD clearance (CL_HD) were modeled simultaneously, where CL_HD was treated as a piece-wise input function, turned on during HD and off during non-HD. The model also included a differential equation to account for the amount of daptomycin recovered from dialysate generated during HD on day 3 (13). The ordinary differential equations for these models were the following: dX₁/dt = R₁ − X₁ × {K₁₂ + (CL/V_c) + [(CL_HD/V_c) × R₂]} + K₂₁ × X₂ dX₂/dt = (K₁₂ × X₁) − (K₂₁ × X₂), and dX₃/dt = R₂ × (CL_HD/V_c) × X₁, where X₁ is the amount (in mg) of drug in the central compartment; X₂ is the amount of drug in the peripheral compartment; X₃ is the amount of daptomycin recovered in the dialysate during hemodialysis; CL is the nondialytic clearance from the central compartment (in liters per hour); CL_HD is the dialytic clearance from the central compartment; K₁₂ and K₂₁ are first-order intercompartmental transfer rate constants (in inverse hours); and V_c is a scalar and represents the volume of the central compartment (liters). R₁ is the time-delimited zero-order drug input rate (piece-wise input function) into the central compartment (in mg/h). R₂ is the rate constant, constrained to 0 (HD turned off) or 1 (HD turned on).

For all models, the inverse of the estimated assay variance was used as the first estimate for weighting. Weighting was accomplished by making the assumption that total observation variance was proportional to assay variance, which was determined on a between-day basis. The analysis was performed with adaptive γ, a scalar that multiplies the polynomial described above and is optimized with each cycle to produce the best approximation to the homoscedastic assumption.

Upon attaining convergence, Bayesian estimates for each patient were obtained using the “population of one” utility in BigNPAG (16). For each model, the mean, median, and modal values were employed as measures of central tendency for the population parameter estimates and were evaluated in a maximum a posteriori (MAP) Bayesian analysis. Scatter plots were examined for individual patients and for the population as a whole. Goodness of fit was assessed by regression with an observed-versus-predicted plot, coefficients of determination, and log likelihood values. Predictive performance evaluations were based on weighted mean bias and bias-adjusted weighted mean precision.

Mean PK parameter estimates from the individual PK models were compared by one-way analysis of variance (ANOVA). Linear regression was used to determine the association between estimated CL_HD and dialysis filter characteristics outlined in the manufacturer package inserts (B₁₂ clearance, the KoA [dialyzer mass transfer area coefficient] for urea, surface area, and Kuf). All statistical analyses were computed using STATA (version 11.0; College Station, TX).

Monte Carlo simulations.

Embedded with the final population PK HD model, a series of 5,000-subject MCSs using the ADAPT 5 package of programs of D'Argenio and Schumitzky (17) were performed. The goal was to identify total body weight-based (in mg/kg) HD dosing schemes that provide efficacy and toxicity profiles comparable to those obtained from MCSs that employ the SAB-IE daptomycin population PK model (15). Separate dosing schemes were sought for both FDA-approved daptomycin dosing regimens (4 mg/kg intravenously [i.v.] every 24 h [Q24H] and 6 mg/kg i.v. Q24H) and the two interdialytic periods observed in a thrice-weekly HD schedule (48 h and 72 h).

The primary exposure target for efficacy in this study was the area under the concentration-time curve (AUC). Daptomycin has been shown to be a concentration-dependent antibiotic, and the AUC/MIC ratio is the PD parameter that is most closely linked with antibacterial activity (18–21). However, the AUC/MIC ratio associated with maximal effect has varied in the literature (18–21). Since the literature is inconclusive on the definitive AUC/MIC threshold, HD dosing schemes should ideally result in AUC distributions (PD target most closely linked to effect) comparable to that observed among non-HD patients. To accomplish this, our study used the distribution of AUC values generated from an MCS embedded with the SAB-IE population PK model as our referent or non-HD AUC exposure distribution (14, 15). For each FDA-approved daptomycin dosing regimen, this study focused on identifying an HD dosing scheme in mg/kg (based on total body weight) that generated an AUC exposure profile (cumulative and noncumulative daily AUCs) comparable to the non-HD MCSs (14, 15).

Regarding toxicity, the focus was on identifying HD regimens that minimized the probability of having trough concentrations in excess of 24.3 mg/liter. Early animal data suggested that daptomycin trough concentrations were associated with creatine phosphokinase (CPK) elevations (22, 23). This relationship has been corroborated in a more recent analysis, which demonstrated that elevations in daptomycin troughs, especially above 24.3 mg/liter, are associated with an increased probability of a CPK elevation (14). For each FDA-approved daptomycin dosing regimen and interdialytic period (48 h and 72 h), the candidate HD administration scheme that best achieved the exposure targets associated with efficacy and minimal toxicity was selected as the optimal HD regimen.

For the HD MCSs, the FDA-approved daptomycin doses (4 and 6 mg/kg) as one-time doses were initially evaluated. Alternative HD doses were only considered if the simulated exposure profiles generated with the parent doses in the HD model were not comparable to those obtained in the non-HD MCSs. The actual amount (in milligrams) of daptomycin used in the MCS was determined by using the average distribution of persons' weights observed in the United States (27). The simulated administration times for the candidate HD regimens included the following: 1 h prior to HD (pre-HD), 2.5 h into HD (intra-HD), and immediately following HD (post-HD). These dosing administration strategies were selected since they have been used for a variety of different antibiotics (24, 25). The duration of HD was set at 3.5 h for all simulations.

For each simulated HD dosing scheme, the AUC, the serum drug concentration immediately prior to the next dose (C_min), and the serum drug concentration 1 h post-completion of infusion (C_max) were determined. For AUCs, both cumulative (AUC_0–48 and AUC_0–72) and noncumulative daily (AUC_0–24, AUC_24–48, and AUC_48–72) values were simulated. The C_min values simulated were at the end of each interdialytic period, 48 h (C_{min 48}) and 72 h (C_{min 72}). The simulated AUC, C_min, and C_max values for each interdialytic period (48 h and 72 h) were compared to those obtained from the MCS derived from the SAB-IE PK model for three consecutive doses of 4 and 6 mg/kg i.v. Q24H given as a 30-min i.v. infusion. The C_min values from the SAB-IE PK model were simulated 24 h after the second dose (C_{min 48}) and 24 h after the third dose (C_{min 72}).

For all simulations, the population simulation without process noise option was utilized. Both normal and log-normal distributions were evaluated, and the distribution selected was based on the ability to best recreate the original mean parameter vector and associated distribution.

RESULTS

Twenty-six patients were included in the pooled analysis (Table 1). The average age was 57.7 years, and 16 (61.5%) of the 26 subjects were male. Patients were predominantly Caucasian in the study by Patel et al. (83.3%), whereas patients were primarily African-American in the studies by Salama et al. (57.1%) and Benziger et al. (71.4%). In comparison to an average weight of ∼80 kg in the studies by Benziger et al. and Patel et al., patients in the study by Salama et al. had an average weight of 63.1 kg.

Population pharmacokinetic analysis.

The mean (with standard deviation [SD]) population parameter estimates identified by BigNPAG in the combined and individual analyses are provided in Table 2. As a point of reference, the SAB-IE population PK model is also provided in Table 2 (14, 15). Regression line equations from the observed-versus-predicted plots and measures of bias and precision for each HD population PK model are summarized in Table 3. Using the population mean parameter values as the measure of central tendency, the overall fits of the models to their respective data after the Bayesian step were good, and the plots of predicted versus observed concentrations showed slopes and intercepts that were highly acceptable.

Table 2.

PK parameters from population PK analyses

Parametera	Mean (SD) value for model
	HD Salama	HD Benziger	HD Patel	HD overall	SAB-IE
Vc (liters)b	4.99 (1.30)	7.39 (1.59)	4.77 (1.08)	5.52 (1.90)	6.56 (3.10)
CL (liters/h)	0.20 (0.07)	0.26 (0.02)	0.25 (0.06)	0.24 (0.06)	0.96 (0.47)
CLHD (liters/h)b	2.05 (0.67)	1.27 (0.34)	0.87 (0.23)	1.30 (0.63)
K12 (h−1)	1.55 (1.10)	0.44 (0.28)	3.37 (4.20)	1.61 (1.91)	1.67 (3.94)
K21 (h−1)	0.95 (0.51)	0.51 (0.26)	3.37 (4.45)	1.17 (1.13)	1.34 (3.40)

^aCL, nondialytic clearance from the central compartment; K₁₂, transfer rate constant from central compartment to peripheral compartment; K₂₁, transfer rate constant from peripheral compartment to central compartment; V_c, volume in the central compartment.

^bSignificant differences (P < 0.05) between individual population PK models.

Table 3.

Goodness of fit and predictive performance of daptomycin HD population PK models

Parameter	HD Salama	HD Benziger	HD Patel	HD Overall
Regression linea	(1.00 × predicted) + 0.15	(1.00 × predicted) + 0.20	(1.01 × predicted) + 0.26	(1.01 × predicted) + 0.21
R2b	0.99	0.98	0.98	0.98
Mean weighted error (mg/liter)	−0.06	−0.07	−0.11	−0.09
Bias-adjusted mean weighted square error (mg/liter)2	1.00	1.00	1.01	1.00

^aBest-fit regression line for the observed-predicted plot after the Bayesian step.

^bCoefficient of determination for the best-fit linear regression line for the observed-predicted plot after the Bayesian step.

Mean non-HD CL in the combined HD model was approximately one-quarter the estimated mean CL in the SAB-IE model. In contrast, CL_HD in the HD model was about 25% greater than the mean CL in the SAB-IE model. Mean PK parameter estimates from individual evaluations were generally similar, with the exception of V_c, CL_HD, and the K₁₂/K₂₁ ratio (P < 0.05). Although not statistically significant (P = 0.14), the mean CL observed in the study by Salama et al. was approximately 25% less than the estimated mean CL in the studies by Benziger et al. and Patel et al.

When individualized Bayesian estimates of CL_HD were regressed to various dialysis filter characteristics, a significant relationship was observed between CL_HD and manufacturers' estimates of B₁₂ clearance (R² = 0.51; P < 0.001), KoA for urea (R² = 0.19; P = 0.025), and surface area (R² = 0.32; P = 0.003). There was no association found between CL_HD and manufacturers' estimates of Kuf (R² = 0.02; P = 0.457). In addition, no other baseline covariates were found to be associated with CL_HD (age, sex, race, or weight).

Monte Carlo simulations.

The AUC distributions for the initial candidate HD regimens are displayed in Fig. 1 and 2 (see also Table S1 in the supplemental material). For the 48-h interdialytic period, all candidate dosing schemes resulted in cumulative AUC exposures similar to the SAB-IE AUC_0–48 values (Fig. 1a and 2a). However, the partitioned AUC values were not as comparable. While mean AUC_0–24 values were well matched, administration of the same parent dose pre-HD or intra-HD resulted in lower mean AUC_24–48 values relative to the referent SAB-IE simulations. In contrast, administering the dose post-HD resulted in more isometric AUC_24–48 values.

Fig 1.

Mean daptomycin HD AUC values for 4 mg/kg dose during the 48-h interdialytic period (a), 4 mg/kg during the 72-h interdialytic period (b), and 6 mg/kg (50% dose augmentation) during the 72-h interdialytic period (c). Simulations for HD candidate dosing regimens were based on one dose. The simulated administration times for the candidate HD regimens included 1 h prior to HD (pre-HD), 2.5 h into HD (intra-HD), and immediately following HD (post-HD). The duration of HD was set at 3.5 h for all simulations. Simulations for SAB-IE (non-HD) regimens were 4 mg/kg every 24 h for 3 days, irrespective of the HD dose (in mg/kg). All pairwise comparisons resulted in P values of <0.05. Statistical significance does not indicate clinical differences.

Fig 2.

Mean daptomycin HD AUC values for 6 mg/kg dose during the 48-h interdialytic period (a), 6 mg/kg during the 72-h interdialytic period (b), and 9 mg/kg (50% dose augmentation) during the 72-h interdialytic period (c). Simulations for HD candidate dosing regimens were based on one dose. The simulated administration times for the candidate HD regimens included 1 h prior to HD (pre-HD), 2.5 h into HD (intra-HD), and immediately following HD (post-HD). The duration of HD was set at 3.5 h for all simulations. Simulations for SAB-IE (non-HD) regimens were 6 mg/kg every 24 h for 3 days, irrespective of HD dose (in mg/kg). All pairwise comparisons were associated with P values of <0.05. Statistical significance does not indicate clinical differences.

For the 72-h interdialytic period, administration of the same parent dose, irrespective of HD administration time, resulted in suboptimal partitioned AUC exposures (Fig. 1b and 2b). Most notably, the mean AUC_48–72 values associated with pre-HD and intra-HD dosing were more than 2-fold lower than the referent exposure groups. Even post-HD administration resulted in mean AUC_48–72 values that were only half the SAB-IE AUC_48–72 values (Fig. 1b and 2b). To achieve more comparable partitioned AUC exposures during the 72-h interdialytic period, we evaluated the impact of increasing the parent dose by 50% (Fig. 1c and 2c). While this strategy resulted in higher AUC_0–24 values, the distributions of AUC_0–72 and AUC_24–48 values for each HD dosing regimen were similar to the referent group. For the partitioned period from 48 to 72 h, administering a 50% increase in the dose intra- or post-HD resulted in mean AUC exposures that were 67% and 76% of the SAB-IE values, respectively.

The results of the C_max and C_min MCS analyses are shown in Table 4. Mean C_max values for each candidate HD dosing regimen were similar to the SAB-IE simulation C_max values for both 4 and 6 mg/kg. Administration of daptomycin pre-HD led to mean C_{min 48} values similar to the SAB-IE simulations. Although intra- and post-HD dosing resulted in higher mean C_{min 48} values relative to the referent SAB-IE simulation C_{min 48} values, the probability of C_{min 48} values exceeding 24.3 mg/liter was low for all candidate HD dosing regimens (Tables 4 and 5). All candidate dosing regimens also yielded lower mean C_{min 72} values relative to the referent SAB-IE simulations. Likewise, the probability of exceeding a C_{min 72} value of 24.3 mg/liter was lower for all HD candidate dosing regimens than with the SAB-IE model. Analyses of C_max and C_{min 72} values for a 50% supplemental dose during the 72-h interdialytic period are also shown in Tables 4 and 5. While mean C_max values were greater, C_{min 72} values were comparable to the referent SAB-IE simulations (Table 4). The probability of C_{min 72} values exceeding 24.3 mg/liter was lower than the SAB-IE model for all HD candidate regimens (Table 5).

Table 4.

C_max and C_min values from Monte Carlo simulation analyses for the 4-, 6-, and 9-mg/kg doses for each HD candidate dosing regimen

Regimen and parametera	Mean (SD) value (μg/ml) for dose ofb:
	4 mg/kg	6 mg/kg	9 mg/kg
Cmax
Pre-HD	28.97 (16.76)	43.46 (25.14)	65.19 (37.71)
Intra-HD	27.51 (15.58)	41.26 (23.37)	61.89 (35.06)
Post-HD	32.79 (19.99)	49.18 (29.99)	73.77 (44.98)
SAB-IE (non-HD)	27.66 (18.97)	41.48 (28.46)	NA
Cmin 48
Pre-HD	6.18 (2.77)	9.27 (4.15)	NA
Intra-HD	7.93 (3.37)	11.89 (5.05)	NA
Post-HD	9.08 (3.86)	13.62 (5.79)	NA
SAB-IE (non-HD)	5.44 (4.64)	8.16 (6.95)	NA
Cmin 72
Pre-HD	4.07 (2.22)	6.11 (3.33)	9.16 (4.99)
Intra-HD	5.05 (2.54)	7.58 (3.82)	11.37 (5.73)
Post-HD	5.74 (2.85)	8.61 (4.28)	12.92 (6.41)
SAB-IE (non-HD)	6.27 (5.45)	9.40 (8.17)	NA

^aSimulations for HD candidate dosing regimens were based on one dose. Simulations for SAB-IE (non-HD) regimens were based on dosing every 24 h for 3 days.

^bNA, not applicable, as the 9-mg/kg dose was not evaluated. All pairwise comparisons were associated with a P value of <0.05. Statistical significance does not indicate clinical differences.

Table 5.

Probabilities of C_min values exceeding 24.3 mg/liter from Monte Carlo simulation analyses for the 4-, 6-, and 9-mg/kg doses for each HD candidate dosing regimen

Regimen and parametera	Probability (%) for dose of:
	4 mg/kg	6 mg/kg	9 mg/kg
Cmin 48
Pre-HD	0.00	0.16	NAb
Intra-HD	0.02	1.76	NA
Post-HD	0.12	4.16	NA
SAB-IE (non-HD)	0.52	2.96	NA
Cmin 72
Pre-HD	0.00	0	0.38
Intra-HD	0.00	0.04	1.86
Post-HD	0.00	0.16	4.74
SAB-IE (non-HD)	1.10	5.46	NA

^aSimulations for HD candidate dosing regimens were based on one dose. Simulations for SAB-IE (non-HD) regimens were based on dosing every 24 h for 3 days.

^bNA, not applicable, as the 9-mg/kg dose was not evaluated.

Efficacy and toxicity profiles for each study that was modeled individually are shown in Fig. 3 and 4. For all dosing schemes evaluated, distributions of AUC values from the MCS analyses of the individual studies were generally similar to the combined analysis. While AUC distributions were well matched across studies and were comparable to the combined analysis, the probability of exceeding a C_min of 24.3 mg/liter varied (Fig. 4). The most notable differences were observed with the Salama et al. study, particularly for post-HD daptomycin administration (Fig. 4). As expected, since daptomycin was found to have a linear PK profile, comparisons of exposure profiles across each study were the same for a parent dose of 4 mg/kg (data not shown).

Fig 3.

Mean daptomycin AUC values from 5,000-subject Monte Carlo simulations for each HD study and the SAB-IE model for intra-HD dosing (2.5 h into HD) (a) and post-HD (immediately following HD) dosing (b). Simulations for HD candidate dosing regimens were based on one dose. For the HD simulations, 6 mg/kg was administered during the 48-h interdialytic period, and 9 mg/kg (50% dose augmentation) was administered during the 72-h interdialytic period. Simulations for SAB-IE (non-HD) regimens were 6 mg/kg every 24 h for 3 days, irrespective of HD dose (in mg/kg). All pairwise comparisons were associated with P values of <0.05. Statistical significance does not indicate clinical differences.

Fig 4.

Probabilities of C_min values exceeding 24.3 mg/liter for each HD study and the SAB-IE model. Simulations for HD candidate dosing regimens were based on one dose. For the HD simulations, 6 mg/kg was administered during the 48-h interdialytic period, and 9 mg/kg (50% dose augmentation) was administered during the 72-h interdialytic period. Simulations for SAB-IE (non-HD) regimens were 6 mg/kg every 24 h for 2 days for the C_{min 48} analysis and for 3 days for C_{min 72} analysis. The dotted line represents the probability of C_min values exceeding 24.3 mg/liter for the SAB-IE model at each C_min threshold (48 and 72 h).

DISCUSSION

Given the morbidity and mortality associated with infections in HD patients, knowledge of the most optimal dosing scheme for a given antimicrobial is critical for ensuring the highest probability of success (3). For a number of antibiotics, multiple dosing recommendations in HD patients have been published (25). This creates a major dilemma for clinicians when trying to treat potentially fatal infections in this vulnerable patient population. Such a situation is exemplified by daptomycin, as three separate groups have described its PK in HD patients and subsequently published different dosing recommendations (10–13). Interestingly, these groups observed similar concentration-time profiles in their original analyses. However, because different PD exposures for efficacy and toxicity were targeted, various HD dosing schemes were proposed (10–13), and the inconsistent recommendations have led to confusion as to which HD dosing scheme to use in clinical practice.

Given the uncertainty in the literature, this study combined the concentration-time data from these three previous PK evaluations to derive uniform HD dosing schemes for both FDA-approved daptomycin dosing regimens and the two interdialytic periods observed in a thrice-weekly HD schedule. Overall, the collective results of the combined PK/PD analyses support the original dosing recommendations offered by Patel et al. (13). For the 48-h interdialytic period, the analyses indicated that administration of the same parent dose (4 mg/kg or 6 mg/kg) intra- or post-HD is preferred. These administration strategies led to the most comparable AUCs (cumulative and partitioned), C_max, and C_{min 48} values relative to those obtained from the SAB-IE simulations. Furthermore, the probability of C_{min 48} values exceeding 24.3 mg/liter was extremely low with both intra- and post-HD dosing.

Dose selection, however, was not as straightforward for the 72-h interdialytic interval. While cumulative AUC exposures were acceptable when the parent dose was given intra- or post-HD, the mean AUC_48–72 values were at least 50% lower than the AUC_48–72 values derived from the SAB-IE model. To compensate for the suboptimal HD partitioned exposures, we evaluated the impact of increasing the parent dose (4 or 6 mg/kg) by 50%. This alternative dosing scheme was assessed based on a previous evaluation (13) as well as the expected simplicity and convenience for both patients and providers in the clinical setting. Increasing the parent dose by 50% for the 72-h interdialytic period intra- or post-HD resulted in mean AUC_48–72 values that were approximately 70% of the parent dose for non-HD AUC_48–72 values. Although this was not completely isometric with non-HD exposures, 70% is an improvement over the partitioned AUC exposures observed with parent dose HD regimens. Achieving more comparable exposures from 48 to 72 h with this dosing scheme was at the expense of higher AUC_0–24 values. Since AUC_0–24 values have not been associated with adverse events, we do not believe this to be a major clinical concern. More importantly, from a CPK toxicity perspective, the probability of C_{min 72} values exceeding 24.3 mg/liter was acceptably low (Table 5). However, caution is still advised with this dosing scheme, and increased CPK monitoring may be warranted. Evaluation of doses of higher than a 50% increase (12 mg/kg) resulted in more comparable mean AUC_48–72 values (457.0 mg · h/liter for intra-HD and 521.4 mg · h/liter for post-HD, compared to 514.7 mg · h/liter for the SAB-IE model [see Table S1 in the supplemental material]), but these regimens were associated with unacceptably high probabilities of exceeding a C_{min 72} of 24.3 mg/liter (11.9% for intra-HD and 19.1% for post-HD). Of note, no HD patients in the 3 analyses experienced CPK elevations (10–13).

When PK data were analyzed for each study separately, heterogeneity was observed. While PK estimates generally lined up in the studies by Benziger et al. and Patel et al., some differences were observed. Most notably, CL_HD, on average, was greater in the Salama et al. study than in the other two studies. In contrast, the estimated mean non-HD CL was ∼ 20% lower in the Salama et al. study. The exact reason(s) for the differences in PK estimates observed across studies is not entirely clear; it may be a function of heterogeneous patient populations. In the study by Salama et al., patients weighed ∼20 kg less and most were African-American (10, 11). Methodological differences between studies might further explain the differences in PK parameters as well. In the study by Salama et al., more HD was delivered; higher blood and dialysate flow rates were employed (10, 11). In addition, based on manufacturer package insert data, the dialysis filters used in each study demonstrated potentially important clearance differences. These data suggest that clearance of larger molecules (i.e., vitamin B₁₂) in an in vitro model was greater for the filter used by Salama et al. and may have contributed to the differences observed for CL_HD. While the aforementioned reasons are intuitive, further data are warranted to quantify the contributions of these differences in methods and of patient factors on PK estimates.

Although dissimilar PK estimates were noted across data sets, this did not translate into major clinical differences in resultant exposure profiles. From an efficacy standpoint, AUC exposures, both cumulative and daily partitioned, were largely comparable across studies. Given the high likelihood of morbidity and mortality in this patient population, the comparable AUC exposure profiles are reassuring and suggest no dose alterations are needed to maximize “efficacy” based on the filter type or flow rates employed. However, the probabilities of C_min values in excess of 24.3 mg/liter for both the 48- and 72-h interdialytic periods varied considerably across studies, particularly with post-HD dosing. With post-HD dosing, probabilities of exceeding a C_min of 24.3 mg/liter were similar to or lower than the “non-HD” or referent group in the studies by Benziger et al. and Patel et al., while the likelihood of exceeding this threshold was greater than the referent group in the study by Salama et al. Subsequent analyses of a 5-min infusion post-HD resulted in similar probabilities of exceeding a C_min of 24.3 mg/liter (data not shown). In a patient population that has an increased incidence of myalgias as a result of their underlying disease (26), CPK monitoring more frequently than weekly should be considered in HD patients receiving daptomycin. Given the existence of a commercially available CPK assay, this should be a relatively straightforward consideration for daptomycin use in patients on HD.

Several additional points should be noted when interpreting the results of this study. First, the PK studies included in this analysis examined noninfected patients within a reasonable, narrow weight range (Table 1). Further studies are needed in patients receiving HD that are volume overloaded and in patients on HD with active infections. While these studies are still needed, we believe our findings can be applied to these populations, since the major driver of daptomycin exposure, drug clearance (non-HD CL and CL_HD), is likely to be similar across populations. Second, the PK structural model used in this analysis did not include urinary clearance of drug. In the previous analysis by Patel et al., very little daptomycin (0.7%) was recovered from the urine of patients who were producing at least 100 ml of urine in 24 h. Because of this, we do not anticipate HD patients making urine would change our results substantially. However, daptomycin exposures in HD patients producing larger amounts of urine merit further investigation.

Third, when devising an optimal dosing scheme for daptomycin in HD patients, it is important to consider both cumulative and noncumulative exposures within each interdialytic period. As mentioned, the AUC/MIC ratio has been shown to be the PD parameter linked to antibacterial activity. However, these conclusions were based solely on a 24-h dosing interval (18–21). While this is relevant for patients who require dosing every 24 h, the literature is inconclusive regarding the acceptance of this target for patients on HD receiving their dosages every 48 to 72 h. Cognizant of this, we targeted both cumulative and partitioned AUC values for each 24-h interval within each interdialytic period as our PD exposures for efficacy. As such, the clinical significance of suboptimal partitioned AUC values during a weekly HD session is unknown and merits further investigation.

Fourth, the C_min threshold of 24.3 mg/liter was established in patients receiving daptomycin every 24 h with a creatinine clearance (CL_CR) of >30 ml/min (14). Patients with diminished renal function (CL_CR of <30 ml/min) require a dosage adjustment to every 48 to 72 h (7). Since these are two distinct dosing intervals, the C_min threshold associated with toxicity in HD patients may not be the same as it is in patients who are dosed every 24 h. Additional studies are sorely needed to quantify the exposure-response relationship in HD patients receiving daptomycin. Once these data become available, the clinical utility of our proposed HD dosing schemes should be further evaluated.

Lastly, all pairwise exposure comparisons from the MCS analyses were associated with a P value <0.05. Because these values were generated from large sample size simulations (5,000 subjects), it is not surprising that all pairwise exposure comparisons were found to be significantly different. We believe this is a situation where assessment and interpretation of the data are needed to determine which differences are clinically meaningful and relevant.

In conclusion, given the disparate dosing recommendations in the literature, this study sought to identify uniform HD dosing schemes for each interdialytic period used in a standard thrice-weekly HD schedule (Table 6). Overall, we found that administration of the same parent dose intra- or post-HD provided the most isometric exposures relative to those for the simulations derived from the SAB-IE PK model for the 48-h interdialytic period. For the 72-h interdialytic period, a 50% increase in the parent dose was needed to achieve more similar noncumulative AUC values. While this dose increase optimized exposures from an efficacy standpoint, an increased probability of exceeding a C_{min 72} of 24.3 mg/liter was seen in one of the three studies. Future studies are still needed to quantify the contribution of filter type and flow rates on PK estimates for HD patients receiving daptomycin. Until these data become available, more intensive CPK monitoring may be warranted.

Table 6.

Daptomycin dosing recommendations for patients on thrice-weekly HD

Dose group (mg/kg)	Monday/Tuesday (HD day 1, 48-h interdialytic period)	Wednesday/Thursday (HD day 2, 48-h interdialytic period)	Friday/Saturday (HD day 3, 72-h interdialytic period)
4	4 mg/kg intra- or post-HD	4 mg/kg intra- or post-HD	6 mg/kg intra- or post-HD
6	6 mg/kg intra- or post-HD	6 mg/kg intra- or post-HD	9 mg/kg intra- or post-HD