Population Pharmacokinetic-Pharmacodynamic Analysis of Anidulafungin in Adult Patients with Fungal Infections

Abstract

To evaluate the exposure-response relationships for efficacy and safety of intravenous anidulafungin in adult patients with fungal infections, a population pharmacokinetic-pharmacodynamic (PK-PD) analysis was performed with data from 262 patients in four phase 2/3 studies. The plasma concentration data were fitted with a previously developed population PK model. Anidulafungin exposures in patients with weight extremities (e.g., 40 kg and 150 kg) were simulated based on the final PK model. Since the patient population, disease status, and efficacy endpoints varied in these studies, the exposure-efficacy relationship was investigated separately for each study using logistic regression as appropriate. Safety data from three studies (n = 235) were pooled for analysis, and one study was excluded due to concomitant use of amphotericin B as a study treatment and different disease populations. The analysis showed that the same dosing regimen of anidulafungin can be administered to all patients regardless of body weight. Nonetheless, caution should be taken for patients with extremely high weight (e.g., >150 kg). There was a trend of positive association between anidulafungin exposure and efficacy in patients with esophageal candidiasis or invasive candidiasis, including candidemia (ICC); however, adequate characterization of the effect of anidulafungin exposure on response could not be established due to the relatively small sample size. No threshold value for exposure could be established, since patients with low exposure also achieved successful outcomes (e.g., area under the curve < 40 mg · h/liter in ICC patients). There was no association between anidulafungin exposure and the treatment-related adverse events or all-causality hepatic laboratory abnormalities.

INTRODUCTION

Anidulafungin is an echinocandin antifungal agent approved as an intravenous (i.v.) treatment for invasive candidiasis, including candidemia (ICC) and esophageal candidiasis (EC), in adult patients (1). The spectrum of activity of anidulafungin in vitro includes Candida spp., Aspergillus spp., and Pneumocystis jirovecii and isolates of Candida spp. that are resistant to azoles or amphotericin B but exhibit no cross-resistance to anidulafungin (2–5). The pharmacokinetics (PK) of i.v. anidulafungin have been evaluated for healthy adult subjects, subjects with renal or hepatic impairment, patients with fungal infections, and immunocompromised pediatric subjects (6–10). Anidulafungin has predictable linear PK with low intersubject variability, and PK parameters in patients with fungal infections were similar to those in healthy subjects (1). Previously, a population PK analysis was performed on the pooled sparse concentration data from 225 adult patients in four phase 2/3 studies (including two ongoing studies), which demonstrated that PK parameters of anidulafungin were not affected by age, race, or the presence of concomitant medications that are known metabolic inhibitors or inducers of P450 enzymes (8). Although body weight, gender, and study effect (participation in an ICC study versus other studies) were identified as sources of variability in clearance (CL), together they accounted for less than 20% of intersubject variability and were deemed to be of a little clinical relevance (8).

The following additional analyses, which will be presented here, were performed on these four phase 2/3 studies after the two ongoing studies were completed. To support product labeling, simulations were performed to further understand the PK profiles of anidulafungin in patients having different demographic characteristics (with a special focus on weight extremities) and evaluate if the same dosing regimen could be recommended for all patients; simulations were also performed with the same patient population receiving the recommended dosing regimens to obtain the corresponding exposure parameters. Furthermore, a population PK-pharmacodynamic (PD) analysis was performed to evaluate the association between anidulafungin exposure and efficacy or safety endpoints. Based on the preclinical data, both PK/PD indices, the area under the curve over a 24-h dosing interval/MIC ratio (AUC_0-24/MIC) and the peak concentration (C_max)/MIC ratio, appeared to be associated with treatment efficacy (11). To confirm the preclinical findings, the PK/PD index was evaluated in our analysis.

The objectives of these analyses were as follows: (i) to evaluate the PK of anidulafungin in patients with different demographics or at different dosing regimens using the developed population PK model, (ii) to explore the anidulafungin exposure-response relationship for efficacy and safety endpoints using a nonlinear mixed-effects approach as appropriate, including the evaluation of the PK-PD index, and (iii) to explore potential covariates that may be predictive of PD responses.

MATERIALS AND METHODS

Study design.

The study design, study population, dosing regimen, timing of blood samples, and efficacy endpoints varied among the four phase 2/3 clinical studies and are summarized in Table 1. The safety and efficacy results from studies 1 to 3 and a detailed description of the study design have been reported previously (12–14). Anidulafungin was administered at an infusion rate of no more than 1.1 mg/min. The anidulafungin maintenance dose (MD) ranged from 50 mg to 100 mg, given once daily. The regimen started with a loading dose (LD) on day 1 that was twice the daily MD given on subsequent days.

Table 1.

Anidulafungin study design and data included for analysis

Study	Design (reference)	Dose (LD/MD); duration	PK sampling plan	PK data	Primary and secondary efficacy endpoint/PK-PD pair (na)b
1 (EC study)	Phase 3, randomized, double-blind, double-dummy, noninferiority study of patients with EC in comparison with fluconazole (12)	100/50 mg; 14–21 days	Days 7 and 21, predose; day 3, postdose (0–4 h following end of infusion); day 14, delayed postdose (8–12 h following end of infusion)	129 patients (350 concns)	ENDE and ENDF; all treated, efficacy-AUC (129) and efficacy-AUC/MIC (100); clinically evaluable, ENDF-AUC (102) and ENDF-AUC/MIC (78)
2 (ICC study)	Phase 2, open-label, randomized, dose-ranging study in patients with ICC (NCT00037219) (13)	100/50 mg, 150/75 mg, 200/100 mg; 15–42 days	Days 6 and 20, predose; day 3, postdose (0–4 h following end of infusion); day 13, delayed postdose (8–12 h following end of infusion)	87 patients (211 concns)	GLBE and GLBF; all treated, efficacy-AUC (87) and efficacy-AUC/MIC (71); clinically evaluable, GLBF-AUC (50) and GLBF-AUC/MIC (43)
3 (EC and OCc study)	Phase 2, open-label, noncomparative study of patients with azole-refractory mucosal candidiasis, including EC and OC (NCT00041704) (14)	100/50 mg; 14–21 days	Days 1, 7, and 14, predose; day 7, intensive sampling (9 samples over 24-h dosing interval)	19 patients (180 concns)	CLNE and CLNF; ENDE was also evaluated in patients diagnosed with EC (13), but ENDF was not due to insufficient data; efficacy-AUC (19), CLNF-AUC/MIC (12), and ENDE-AUC (13)
4 (IA study)	Phase 2, open-label, noncomparative study of patients with IA in combination with amphotericin B (AmBisome) (NCT00037206)	200/100 mg; up to 90 days	Similar to study 1	27 patients (78 concns)

^an, number of data pairs.

^bEfficacy represents both the primary and secondary endpoints unless specified.

^cOC, oropharyngeal candidiasis.

The Candida isolates were collected from most of the patients in studies 1 to 3, while few isolates were collected in study 4 due to the technical challenge of obtaining Aspergillus spp. The anidulafungin MIC (95% to 100% growth inhibition after a 24-h incubation) values at the baseline for Candida species were used for PK-PD index evaluation.

Since the patient population, disease status, and efficacy endpoints varied among the four studies, the exposure-efficacy and PK-PD index-efficacy relationships were investigated separately for each study.

The safety data from studies 1 to 3 were pooled for analysis of treatment-related adverse events (AEs), treatment-related gastrointestinal (GI) AEs (the single most frequently reported incidence), and all-causality hepatic laboratory abnormalities (i.e., alanine transaminase [ALT], aspartate transaminase [AST], gamma-glutamyl transferase [GGT], alkaline phosphatase, and total bilirubin). The criteria for the definition of hepatic laboratory abnormalities were based on the protocol definition. The patient population studied had a high incidence of hepatic abnormalities at the baseline. Study 4 was excluded from the AE analysis for two reasons: concomitant use of amphotericin B as study treatment and a different disease population (invasive aspergillosis [IA]) with serious underlying condition, which would confound the interpretation of the data.

Population PK analysis.

All modeling and simulations were performed using the software program NONMEM (version V, level 1.1; Icon Development Solutions, Ellicott City, MD). The previously developed population PK model was used to fit the concentration data from all the patients (8). This was a 2-compartment model with first-order elimination; intersubject variability in the PK parameters was modeled using multiplicative exponential random effects, and within-subject variability was modeled with a proportional error. The final PK model was used to generate the posterior Bayesian estimates of individual clearance (CL) and AUC_0–24 at steady state (AUC_ss, calculated as dose/CL). Individual AUC_ss values were subsequently used for population PK-PD analysis. A simplified term, AUC, which is equivalent to AUC_ss, is also used in the text.

In addition, simulations using the final parameter estimates of the PK model were performed to do the following: (i) estimate the anidulafungin PK profiles of a 200/100-mg (LD/MD) once-daily regimen in patients with different demographics (12 scenarios were created by combining three identified covariates with a special focus on weight extremities: weight [40, 60, and 150 kg], gender, and study effect [participation in study 2, other studies]) and (ii) estimate exposure and PK parameters in the same patient population (n = 262) receiving recommended dosing regimens (LD/MD, 100/50 mg and 200/100 mg): AUC_ss, C_max at steady state (C_max,ss), trough concentration at steady state (C_min,ss), CL, and beta-phase elimination half-life (t_1/2β).

Population PK-PD analysis.

Measures of efficacy or safety versus steady-state anidulafungin total exposure (AUC) or the PK-PD index (AUC/MIC) were modeled using a nonlinear mixed-effects approach as appropriate with NONMEM. The graphic processing of the NONMEM output was performed using the software program S-PLUS 6.2 Professional for Windows. To obtain more information from the available data, exposure-efficacy analyses of studies 1 and 2 were carried out for both clinically evaluable patients and all treated patients. Clinically evaluable patients were those who completed at least 10 days of therapy, had an end-of-therapy (EOT) assessment with a clinical outcome other than “indeterminate (for circumstances prevented an evaluation from being made),” had a result available for an EOT endoscopy, and did not have any protocol violations up to the EOT visit that impacted the assessment of efficacy. Compared with the clinically evaluable patient population, the “all treated” population was more conservative; the responses with “indeterminate” were included and treated as failures.

Since only one PD record (efficacy or safety endpoints) was obtained from each patient, a naïve pooled data analysis was considered for PK-PD analysis. The majority of the efficacy and safety endpoints were dichotomous categorical variables, and only a few efficacy endpoints were multiple categorical variables, which were translated into dichotomous variables based on the protocol definition for success and failure.

Exposure-response relationships for both efficacy and safety endpoints were first explored graphically. Logistic regression models were used to estimate the probability of response as a function of anidulafungin exposure or the PK-PD index using the Laplacian estimation method (15). A covariate modeling approach emphasizing parameter estimation rather than stepwise hypothesis testing was implemented in this analysis. Predefined covariate-parameter relationships were identified based on exploratory graphics, scientific interest, or prior knowledge, and these included age, race, and gender. The forms of the equation used in the modeling are presented below.

$$\begin{array}{cc}\log \left[\frac{p_i}{\left(1-p_i\right)}\right]={\lambda }_i& \text{Logit}\end{array}$$

$$\begin{array}{cc}\lambda {}_i={\theta }_1*{\theta }_n^{\text{Cov}}+{\theta }_2*\hspace{0.17em}\text{AUC}& \text{Categorical}\end{array}$$

$$\begin{array}{cc}\lambda {}_i={\theta }_1*{\left[\frac{\text{Cov}}{{\text{Cov}}_{\text{erf}}}\right]}^{{\theta }_n}+{\theta }_2*\hspace{0.17em}\text{AUC}& \text{Continuous}\end{array}$$

where the probability of an event for individual i is given by p_i, λ_i is the logit, which is the natural log of the odds ratio, θ₁ is the baseline probability of success, θ₂ is the effect of drug exposure on the probability of success, and θ_n is the selected covariate (Cov) effect on the baseline probability of success independent of drug effect.

Model selection was based on goodness-of-fit criteria, including diagnostic plots, convergence with at least 2 significant digits, precision of parameter estimates, and the objective function value (OFV). Estimates of parameter precision were obtained from the asymptotic variance-covariance matrix of the estimates and are described as a percent relative standard error (% RSE). No further model evaluation methods were implemented.

RESULTS

Data for analysis.

The anidulafungin plasma concentration data were available from 262 patients (819 observations) and used for the population PK analysis. There were 262 patients with exposure-and-efficacy data pairs and 235 patients with exposure-and-safety data pairs. Specific data pairs used for exposure-efficacy analysis in each study are presented in Table 1. The demographic characteristics of these patients are summarized in Table 2. In this analysis, 68% (178/262) of patients received the 100/50-mg dosing regimen. AUC values in the database ranged from 25.4 to 202 mg · h/liter, and three-quarters of the patients had AUC values below 72 mg · h/liter. Most of the isolates were Candida albicans (75%; 137/183), followed by Candida glabrata (11%; 21/183). The baseline MIC values of the majority of Candida species were low (0.03 to 0.5 mg/liter), and there were few relatively “high” MIC organisms (3 isolates with MICs of 4 mg/liter). The distributions of AUC, MIC, and AUC/MIC in studies 1 and 2 are presented in Fig. 1 and were used as the basis to set the bins for continuous covariates (an attempt was made to balance the number of observations in each bin).

Table 2.

Summary of subject demographics

Characteristic	Median value (range) or count for study
	Pooled	1	2	3	4
No. of subjects (no. of concns)	262 (819)	129 (350)	87 (211)	19 (180)	27 (78)
Wt (kg)	61 (31–154)	52 (31–122)	72 (36–154)	59 (49–92)	70 (46–111)
Age (yr)	43 (18–88)	38 (20–65)	57 (18–88)	40 (30–53)	61 (21–79)
Gender (male/female)	131/131	60/69	40/47	12/7	19/8
Race\ethnicity
Caucasian	118	29	56	8	25
Black	62	26	27	8	1
Asian	41	41	0	0	0
Other	41	33	4	3	1

Fig 1.

Distribution of AUC, MIC, and AUC/MIC in study 1 (upper panels, all treated EC patients) and study 2 (lower panels, all treated ICC patients).

Population PK analysis.

The additional data collected from studies 3 and 4 were consistent with the original data used in the previous population PK analysis. There was little difference in parameter estimates among the two analyses with different PK data sets (Table 3). All parameter estimates were within 20% of the original values. This analysis further validated the previously defined population PK model for anidulafungin.

Table 3.

Comparison of anidulafungin parameter estimates from population PK model at two stages (original data versus final data, FOCEI method)^a

Parameter	Modeling approach	Typical value (% RSEb)		Interpatient % CV (% RSEb)
		Original data	Final data	Original data	Final data
CL (liters/h)	CL = θ1 + (wt − Mwt) × θ5 + gender × θ6 + study × θ7			28.0 (17.6)	27.2 (16.2)
θ1		0.768 (3.80)	0.777 (3.31)
θ5		0.00417 (26.9)	0.00461 (21.6)
θ6		0.166 (25.4)	0.183 (20.9)
θ7		0.278 (20.8)	0.256 (20.7)
V1 (liters)	V1 = θ2 × wt			NS	NS
θ2		0.215 (20.3)	0.170 (24.0)
Q (liters/h)	Q = θ3			NS	NS
θ3		20.3 (16.7)	21.6 (10.5)
V2 (liters)	V2 = θ4			NS	NS
θ4		19.6 (15.1)	23.5 (11.0)
Vss (liters)		33.4c	34.6c	14.3d	10.0d
t1/2 (h)		25.6c	26.5c	29.1d	27.3d
Residual error parameter
σ21prop (%)		24.0 (9.69)	24.0 (9.79)	NA	NA

^aAbbreviations: FOCEI, first-order conditional estimation with interaction; CL, clearance; V₁, central volume of distribution; Q, intercompartmental clearance; V₂, peripheral volume of distribution; V_ss, volume of distribution at steady state; t_1/2, terminal-phase half-life; σ²_1prop, proportional component of the residual error model; NS, not supported in model; NA, not applicable; wt, weight (kg); Mwt, 60 kg; gender, 1 for males and 0 for females; study, 1 for study 2 (ICC study) and 0 for all other studies.

^b% RSE, percent relative standard error of the estimate = SE/parameter estimate × 100 (for variability terms, this is the % RSE of the variance estimate). For original data, n = 225; for final data, n = 262.

^cCalculated from individual parameter values.

^dCalculated as (standard deviation/mean) × 100.

Simulations with different demographics, including study effect.

The estimated exposure parameters and clearance at a 200/100-mg regimen in 12 scenarios are summarized in Table 4. The total exposure (AUC) in a typical 40-kg female was only approximately 13% higher than that in a typical 60-kg female patient, and the exposure in a typical 150-kg male was approximately 30% lower than that in a typical 60-kg male patient. If the exposure parameters between a typical 40-kg female patient and a typical 150-kg male patient were compared, a significant difference would exist (e.g., AUC of 146 versus 73 mg · h/liter, or 106 versus 61 mg · h/liter if the patient was from study 2).

Table 4.

Typical estimates of exposure parameters and clearance of anidulafungin in patients with different demographics following a 200/100-mg regimen at 1-mg/min infusion rate

Parameter	Value by patient gender and wt (kg)
	Male			Female
	40	60	150	40	60	150
Study 1 (EC)
Cmax, ss (mg/liter)	7.9	7.0	4.6	9.0	7.9	5.0
Cmin, ss (mg/liter)	3.4	3.1	2.2	4.5	4.0	2.6
AUCss (mg · h/liter)	115	104	73	146	129	84
CL (liters/h)	0.87	0.96	1.4	0.68	0.78	1.2
Study 2 (ICC)
Cmax, ss (mg/liter)	6.9	6.1	4.2	7.6	6.7	4.5
Cmin, ss (mg/liter)	2.3	2.2	1.7	3.0	2.8	2.0
AUCss (mg · h/liter)	89	82	61	106	97	69
CL (liters/h)	1.1	1.2	1.6	0.94	1.0	1.5

Simulations with recommended dosing regimens.

For the same patient population receiving a 100/50-mg or 200/100-mg once-daily regimen at an infusion rate of 1 mg/min, the estimated PK parameters are summarized in Table 5.

Table 5.

Estimated steady-state PK parameters of anidulafungin in patients with fungal infections receiving recommended regimens^a

PK parameterc	Mean value (% CV) for anidulafungin given i.v.
	Dosing regimen (LD/MD, mg)b		All
	100/50	200/100
Cmax, ss (mg/liter)	4.2 (22.4)	7.2 (23.3)
Cmin, ss (mg/liter)	1.6 (42.1)	3.3 (41.8)
AUCss (mg · h/liter)	55.2 (32.5)	110.3 (32.5)
CL (liters/h)			1.0 (33.5)
t1/2β (h)d			26.5 (28.5)

^an = 262.

^b% CV, percent coefficient of variation; LD/MD, loading dose/maintenance dose.

^cBased on estimated individual PK parameters from the final PK model with an infusion rate of 1 mg/min.

^dt_1/2β is the predominant elimination half-life that characterizes the majority of the concentration-time profile.

Exposure-efficacy analysis. (i) Study 1.

Based on diagnostic plots of the observed data presented in Fig. 2 and 3 (the first column), in all treated EC patients, there was a trend of positive association between AUC and the probability of a successful endoscopic response at the 2-week follow-up (ENDF) (n = 129), and AUC/MIC were also positively associated with the probabilities of a successful endoscopic response at the end of i.v. therapy (ENDE) and ENDF (n = 100). For clinically evaluable patients, AUC was positively associated with the probability of successful ENDE and ENDF (n = 102), but the signal for ENDE was small, probably due to an overall high success rate (>95%), while there was no association between AUC/MIC and the probabilities of successful ENDE/ENDF (n = 78) (data on file). The pairs with positive association were selected for further analysis using logistic regression. Age and gender were selected as potential covariates for evaluation.

Fig 2.

Observed and model predicted success probabilities of ENDF versus AUC in EC patients in study 1 (upper panels, all treated; lower panels, clinically evaluable). Stacked bar plot showing the probability of the efficacy endpoint, calculated as the ratio of the number of successful outcomes in a particular bin to the total number in that bin and presented as a percentage. In each bin, observed successful ENDF (all treated), 16/31, 36/59, and 26/39; observed successful ENDF (clinically evaluable), 15/22, 32/47, and 26/33.

Fig 3.

Observed and model predicted success probabilities of ENDF (upper panels) and ENDE (lower panels) versus AUC/MIC in all treated EC patients in study 1. Stacked bar plot showing the probability of the efficacy endpoint, calculated as the ratio of the number of successful outcomes in a particular bin to the total number in that bin and presented as a percentage. In each bin, observed successful ENDF (all treated), 18/34, 23/37, and 19/29; observed successful ENDE (all treated), 28/34, 35/37, and 27/29.

The key model-building steps for ENDF-AUC in both populations are summarized in Table 6, and the diagnostic plots for different models are presented in Fig. 2. Similar results were obtained in the two populations. The base model was described as an intercept model. Addition of anidulafungin exposure (AUC) as a linear function (the slope model) resulted in an improved goodness of fit on the conditional estimation of the probability of success (Fig. 2). However, the OFV did not decrease significantly, and the % RSE increased substantially compared with that for the intercept model. When age or gender was introduced into the model to describe the intercept, the % RSE on the intercept parameter was still more than 100%. Similar findings were observed in the analysis of ENDE and ENDF versus AUC/MIC for all treated patients (data on file), and the diagnostic plots for different models are presented in Fig. 3. Above all, adequate characterization of the correlation between anidulafungin exposure (including AUC/MIC) and response could not be established due to a very large % RSE on the intercept and slope parameters of the logit function.

Table 6.

Model-building steps (ENDF-AUC in study 1)

Patient group (na) and run no.	Description	OFV	Parameter estimate (% RSE)
			θ1	θ2	θ3
All treated EC patients (129)
1	θ1	173	0.425 (42)
2	θ1 + θ2 × AUC	172	−0.162 (433)	0.00937 (116)
3	θ1 × (age/38)θ3 + θ2 × AUC	171	−0.285 (122)	0.0121 (52)	−2 (80)
4	θ1 × θ3sex + θ2 × AUC	172	−0.126 (717)	0.00901 (138)	1.22 (366)
Clinical evaluable EC patients (102)
5	θ1	122	0.923 (24)
6	θ1 + θ2 × AUC	121	0.168 (490)	0.012 (106)
7	θ1 × (age/38)θ3 + θ2 × AUC	120	0.37 (162)	0.00887 (108)	1.21 (87)

^an, no. of patients.

(ii) Study 2.

As shown in Fig. 4 (first column), for all treated ICC patients, there was a trend of positive association between AUC and the probabilities of a successful global response at the end of i.v. therapy (GLBE) and at the 2-week follow-up (GLBF) (n = 87) but not with AUC/MIC (n = 71; data on file). No trend was observed for the clinically evaluable patients by graphical exploration (n = 50). This is not unexpected given the relatively small sample size. The diagnostic plots for different models on GLBE and GLBF versus AUC are presented in Fig. 4. Similar to findings for study 1, when anidulafungin exposure, age, or gender was introduced into the model, the OFV did not decrease significantly and the % RSE on the parameter estimates was very high (>100%). Adequate characterization of the effect of anidulafungin exposure on the response could not be established.

Fig 4.

Observed and model-predicted success probabilities of GLBF (upper panels) and GLBE (lower panels) versus AUC in all treated ICC patients in study 2. Stacked bar plot showing the probability of the efficacy endpoint, calculated as the ratio of the number of successful outcomes in a particular bin to the total number in that bin and presented as a percentage. In each bin, observed successful GLBF (all treated), 11/27, 14/32, and 18/28; observed successful GLBE (all treated), 16/27, 24/32, and 22/28.

(iii) Study 3.

In 19 patients with azole-refractory EC or oropharyngeal candidiasis, there was no clear separation on the anidulafungin AUC between patients with success and those with failure. With limited MIC data, no trend was observed between AUC/MIC and efficacy endpoints. There were four cases of recurrence and two deteriorations for the clinical response at the 2-week follow-up (CLNF). Excepting one deterioration case, all five other cases occurred in patients with AUC values less than 50 mg · h/liter. This may suggest that the likelihood of recurrence at the 2-week follow-up might be higher in patients with low exposure (e.g., AUC < 50 mg · h/liter) than in those with high exposure.

(iv) Study 4.

In 27 patients with IA, no association between the anidulafungin AUC and the efficacy endpoints (GLBE and GLBF) was identified, since there was a substantial overlap on exposure between success and failure cases. As noted earlier, amphotericin B was used concomitantly as the study treatment, which may affect this assessment.

Exposure-safety analysis.

Anidulafungin exposure (AUC), age, race, and gender were examined graphically as potential predictors for safety endpoints. Male and female patients were almost equally distributed (112/123). There were 57 patients with treatment-related AEs; of these, 34 patients had a single event. Only 4 patients had more than 3 treatment-related AEs (2 patients had 5 AEs, 1 had 6 AEs, and 1 had 12 AEs), and all these patients had AUC values between 61 to 68 mg · h/liter. No clear association between anidulafungin exposure and the treatment-related AEs, treatment-related GI AEs, or all-causality hepatic laboratory abnormalities was observed. Although females and patients over 60 years old (n = 44) had a slightly higher incidence of treatment-related AEs than males and younger patients, the magnitude was very small and deemed minimal.

DISCUSSION

Effect of weight on clinical outcome.

As stated above, the total exposure (AUC) in a typical 150-kg male could be approximately 30% lower than that in a typical 60-kg male patient. The efficacy data for the overweight patients from the four studies were evaluated. In total, there were 7 patients over 120 kg: 1 patient in study 1 (122 kg; AUC, 37.6 mg · h/liter) and 6 patients in study 2 (weights, 121, 134, 144, 145, 146, and 154 kg; AUC, 27.9 to 65.2 mg · h/liter). In study 1, this patient (no Candida sp. was isolated) had a response of failure for ENDE and ENDF, although the clinical response at the end of therapy (CLNE) and CLNF were assessed as success. The average anidulafungin total exposure in EC patients receiving the 100/50-mg regimen was 55 mg · h/liter (Table 5). The success rates at ENDE and ENDF in clinically evaluable patients in this study were 97.4% (225/231) and 45.5% (105/231), respectively (1). In study 2, five out of six patients had Candida infection (MIC, 0.03 to 0.5 mg/liter), and the patient without confirmed Candida infection was a 145-kg male with an AUC of 57 mg · h/liter (dose, 200/100 mg). All five overweight ICC patients had successful GLBE (100%), and only two of them had total exposures below 50 mg · h/liter: a 154-kg male (dose, 150/75 mg; AUC, 27.9 mg · h/liter; MIC, 0.12 mg/liter) and a 144-kg male (dose, 100/50 mg; AUC, 30.6 mg · h/liter; MIC, 0.5 mg/liter). The global success was achieved for overweight ICC patients even with relatively low exposure (e.g., AUC < 40 mg · h/liter). In study 2, the success rates of GLBE in the evaluable population were 84% (21/25), 90% (27/30), and 89% (25/28) for the 50-mg, 75-mg, and 100-mg groups, respectively (13).

In addition, the effect of weight on efficacy was assessed for ICC patients from the pivotal phase 3 study, where the 200/100-mg dosing regimen was evaluated (16). Of note, no anidulafungin concentration data were collected in this study. Since there were few patients weighing >120 kg, a body mass index (BMI) of ≥32 kg/m² was used as an indicator of obesity. The success rates of GLBE for the overweight patients and the normal-weight patients (BMI, 18.5 to 31.9 kg/m²) were not substantially different: 78.3% (18/23) and 72.5% (66/91) (data on file).

On the other hand, the total exposure in a typical 40-kg female is approximately 13% higher than that in a typical 60-kg female. The probability of a safety concern with a 40-kg female is expected to be low. Furthermore, a 260/130-mg dosing regimen was evaluated in 10 healthy subjects, and anidulafungin, which achieved higher exposures (mean AUC, 169 mg · h/liter) than that in a 40-kg female receiving a 200/100-mg regimen, was generally well tolerated (1). In the pivotal ICC study (16), all-causality and treatment-related AEs were compared between patients with low weight (<40 kg or <50 kg) and all other patients (≥40 kg or ≥50 kg). The number of low-weight patients was relatively small (<40 kg, n = 4; or <50 kg, n = 12), but they did not appear to have experienced a disproportionate number of AEs (data on file).

Overall, patients who were on the upper end of the weight spectrum did not appear to have worse efficacy outcomes due to low exposure, while patients on the lower end of the weight spectrum did not appear to have an excess number of AEs due to high exposure. Based on these findings, the same dosing regimen was approved for anidulafungin in all patients regardless of weight (1). Since very few data existed for patients weighing more than 150 kg or less than 40 kg in the clinical development program, the sponsor has committed to collecting more information on these patients in the postmarketing pharmacovigilance program since 2007. Until now, no obvious trend in the frequencies of serious AEs or lack of efficacy has been observed for patients with extremely low or high body weight. Nonetheless, caution should be taken when treating patients with extremely high body weight (e.g., >150 kg) due to the lack of sufficient data for this group.

PK-PD correlation.

An attempt was made here to investigate the relationship between anidulafungin exposure and efficacy and safety endpoints of interest. The correlation between drug exposure, isolate MICs, and the clinical response is difficult to interpret for antifungal compounds, including the echinocandin class, because of the complex nature of the patients treated in the clinical program and the lack of an established standard method to measure MICs at the time of the analysis (17, 18). It is known that in addition to drug exposure, other factors are also important determinants of successful clinical outcomes, such as a patient's underlying condition, disease severity, and duration of the treatment. The current exposure-response analysis was performed without taking into account these components due to limitations on modeling capacity. The baseline MICs of anidulafungin were determined by the NCCLS M27-A broth microdilution method (19). This was considered an acceptable method when the studies were conducted, although it differed from the currently recommended M27 method.

In this analysis, only AUC/MIC was selected as the PK-PD index for evaluation, since AUC and C_max are highly correlated and AUC provided more reliable information on exposure than C_max in this situation (sparse sampling). The time above the MIC was 100% for most of the patients, which would not provide valuable information on predicting the probability of clinical success. Of note, since the protein binding of anidulafungin is >99% (1), it is difficult to accurately estimate the free fraction of anidulafungin in the plasma and no free-AUC/MIC was evaluated. In addition, the majority of the isolates obtained in our studies were susceptible to anidulafungin with very low MIC values, and the variability in MIC values was considered low. Hence, AUC values alone were also assessed to allow more patients to be included for exposure-efficacy evaluation.

In study 1, a positive association between AUC/MIC and response (ENDE/ENDF) was observed for all treated EC patients (n = 100), but the association was not evident when the clinically evaluable patient population (n = 78) was tested. Caution should be used when interpreting these results. It is possible that factors like the overall high success rate in the EC patients and low MIC values may have limited the assessment of the correlation between AUC/MIC and efficacy in the EC patients.

In study 2, a positive association between AUC and response (GLBE/GLBF) was observed for all treated ICC patients (n = 87) but not with AUC/MIC (n = 71), while no trend was observed for the clinically evaluable patients (n = 50). These findings are not unexpected given the relatively small sample size. Although this study evaluated 3 different dosing regimens, the lowest dose (100/50 mg) still demonstrated an acceptable global success rate (84%), which may indicate that this lowest dose is able to achieve the exposure close to the upper end of the exposure-response curve. The MIC values were also quite low except for a few isolates. It was noted that two out of three patients with a MIC value of 4 mg/liter were clinically evaluable subjects (AUC values of 35.6 and 59.2 mg · h/liter), and both of them achieved global success.

Given the available data, no clear threshold value could be established, since patients with low exposure also achieved successful outcomes (e.g., AUC < 40 mg · h/liter in ICC patients). The currently approved dosing regimen for ICC treatment in adults is 200/100 mg once daily. At this dose, approximately 95% of patients would achieve the total exposure (AUC) over 40 mg · h/liter.

In summary, the same dosing regimen of anidulafungin can be administered to all patients regardless of body weight. Nonetheless, caution should be taken for patients with extremely high body weights (e.g., >150 kg). There was a trend of positive association between anidulafungin exposure and efficacy for EC and ICC patients; however, adequate characterization of the effect of anidulafungin exposure on the response could not be established due to the relatively small sample size. Since there was no clear separation of exposure between patients with success and failure, no threshold value for exposure could be determined. Additionally, there was no association between anidulafungin exposure and the treatment-related AEs or all-causality hepatic laboratory abnormalities. The correlation between anidulafungin exposure and clinical outcomes in ICC patients will be further investigated when more data are available from other phase 3/4 studies.