Evaluation of the Pharmacokinetics and Pharmacodynamics of Liposomal Amikacin for Inhalation in Cystic Fibrosis Patients with Chronic Pseudomonal Infections Using Data from Two Phase 2 Clinical Studies

Abstract

The pharmacokinetic-pharmacodynamic (PK-PD) relationships between serum exposure measures of liposomal amikacin for inhalation (LAI) and the change in pulmonary function test (PFT) measures and number of CFU from baseline were evaluated in cystic fibrosis (CF) patients chronically infected with Pseudomonas aeruginosa. A dose of 70, 140, 280, or 560 mg of LAI or placebo was administered to CF patients once daily for 28 days. PFTs and sputum samples for microbiology were assessed on days 7, 14, 21, 28, 35 (for log₁₀ CFU), and 56 (for PFTs). Serum, urine, and sputum samples were collected for PK evaluation. The relationships between efficacy endpoints (relative change in forced expiratory volume in 1 s [FEV₁ {expressed in liters}] and FEV₁% predicted and the absolute change in log₁₀ CFU of P. aeruginosa from baseline) and exposure measures (dose, day 1 area under the curve [AUC], dose/MIC ratio, and day 1 AUC/MIC ratio) and baseline MIC value were assessed. The serum and urine PK data were best fit by a 3-compartment model (lung, serum, and urine) with linear clearance and interoccasional variation on total and renal clearance. Significant univariable relationships between dose or day 1 AUC and the relative change in PFT measures (P ≤ 0.017) or the absolute change in log₁₀ CFU from baseline (P ≤ 0.037) on the study days were identified. Repeated-measures mixed-effects models, which showed dose- and AUC-related improvements for each efficacy endpoint (P ≤ 0.041), predicted the observed data well. The increases in the relative change in FEV₁ and FEV₁% predicted of 11% and 9.9%, respectively, and a 1.23-log₁₀ CFU reduction per 560 mg of LAI estimated on day 7 were comparable to the observed increases of 10.7% and 10.3%, respectively, and a 1.24-log₁₀ CFU reduction on the same day. The model-estimated PFT effects were predicted to be sustained to day 28. An additional 0.451-log₁₀ CFU reduction (P = 0.022) was estimated on day 14 relative to day 7, with a persistence of effect predicted to day 35.

INTRODUCTION

Cystic fibrosis (CF) is chronic genetic disease caused by a mutation in the cystic fibrosis transmembrane conductance regulator (CFTR) gene. This gene encodes the CFTR protein (1), the absence of which can result in the production of thick viscous mucus within the lungs, liver, intestines, and several endocrine glands, such as the pancreas and testes. In the respiratory system, the production of viscous mucus leads to the impairment of the mucociliary defense mechanisms and lung function, often resulting in chronic life-threatening infections by bacteria, such as Pseudomonas aeruginosa (2).

Given that lung infections in CF patients arising from P. aeruginosa are associated with an increased decline in pulmonary function and with significant morbidity and mortality (1, 3–5), the eradication of this pathogen from the lungs of infected patients is an important treatment endpoint. Antibacterial agents developed to meet this need include inhalational therapies, such as tobramycin solution for inhalation (6) and lyophilized aztreonam for inhalation (7), both of which are approved in the United States for the management of CF patients with P. aeruginosa infections.

Inhalational therapies for the treatment of CF patients became more mainstream with the introduction of tobramycin inhalation solution in the United States. Given that tobramycin inhalation solution is administered twice daily and that compliance is poor, as measured by a treatment diary (8), the development of a unique liposomal inhalational formulation that can be administered once daily and that potentially improves compliance with minimal toxicity is of great interest. Liposomal amikacin for inhalation (LAI), at a concentration of 70 mg/ml, is currently being developed for the treatment of CF patients chronically infected with P. aeruginosa. The development of this formulation was preceded by the development of an older formulation of LAI with a concentration of 50 mg/ml, which was evaluated in two open-label phase 1b/2a studies (9).

Using data from the above-described two phase 1b/2a studies in which CF patients received a 500-mg dose of LAI using a Pari LC Star jet nebulizer once daily for 14 days, we previously reported that the pharmacokinetics (PK) of LAI was best described using a three-compartment model (lung, central, and urine) and linear elimination (9). Using individual fitted exposures based on this model, significant relationships between the day 1 serum area under the curve (AUC)-to-MIC (AUC/MIC) ratio and the absolute change in P. aeruginosa log₁₀ CFU from baseline on days 7 and 14 were identified. Significant relationships between the absolute change in P. aeruginosa CFU from baseline and the absolute change in forced expiratory volume in 1 s (FEV₁ [expressed in liters]) or FEV₁% predicted from baseline on days 7 and 14 were also identified. However, this analysis was limited by a small sample size (n = 24) and the fact that only one dose (500 mg) was studied. Using data from two dose-ranging placebo-comparator studies, one which was a phase 1b/2a study and the other which was a phase 2a study, and which included a larger number of patients and were conducted using a wider range of doses of the new formulation that were delivered using a more efficient device (10), PK and PK-pharmacodynamics (PD) analyses were undertaken.

The first objective of these analyses was to characterize the disposition of amikacin in serum, urine, and sputum samples after the administration of LAI. Using serum exposures derived from the above-described analysis, the second objective was to evaluate the PK-PD relationships between various efficacy endpoints and amikacin exposure measures or baseline MIC value. The efficacy endpoints included the change from baseline in two pulmonary function tests (PFTs), FEV₁ and predicted FEV₁% predicted, and the change from baseline in the bacterial burden of P. aeruginosa in sputum.

MATERIALS AND METHODS

Patient population.

The study population for these analyses, which consisted of CF patients with mild to moderate obstructive lung disease (FEV₁% predicted, ≥40%) who were chronically infected with P. aeruginosa, was obtained from two multiple-dose parallel-group dose escalation studies conducted in eastern Europe (study 1) and the United States (study 2). The details of both studies are reported by Clancy et al. (10). The patients enrolled in these studies were ≥6 years of age and had a confirmed diagnosis of CF, a history of P. aeruginosa infection (defined as three positive cultures in the last 2 years, with one positive culture in the last 3 months), FEV₁% predicted of ≥40%, and arterial oxygen saturation (SaO₂) of ≥90%. Patients were excluded if they met any of the following criteria: use of any antipseudomonal antibiotics or initiation of chronic CF therapy (Tobi, recombinant human DNase [rhDNase], macrolide antibiotics, or high-dose ibuprofen) within 28 days prior to screening, history of sputum or throat culture positive for Burkholderia cepacia within 2 years of screening, history of mycobacterial or Aspergillus infections, history of biliary cirrhosis with portal hypertension or splenomegaly, absolute neutrophil counts of <1,000/mm³, liver enzymes ≥3 times the upper limit of normal at screening, serum creatinine >1.5 times the normal range at screening, a history of daily continuous oxygen supplementation or a requirement of >2 liters/min of oxygen at night, or changes in chest X ray at screening (or within 3 months prior) with new onset infiltrates or changes that would compromise the safety of the patients or the quality of the data.

Drug dosage and administration.

The patients enrolled in each study were randomized to receive 70, 140, 280, or 560 mg of LAI or placebo once daily for a period of 28 days via a Pari eFlow nebulizer for 2 to 5 min per 140 mg of LAI.

Sample collection and drug assay.

The serum, sputum, and urine samples to be assayed for amikacin were collected on days 1, 14, and 28. On each of these days, the serum samples were collected prior to dosing and at 0 to 1, 4, and 6 to 8 h postdose. The sputum samples were also collected prior to dosing and at 0 to 1 h postdose on days 1, 14, and 28. The urine samples were collected at 12-h intervals for up to 24 h on the same days as the other samples. The concentration of amikacin in each urine sample was multiplied by the corresponding urine volume in order to compute the amount of amikacin excreted into the urine.

The serum, sputum, and urine samples were assayed for amikacin using liquid chromatography-tandem mass spectrometry, with lower limits of quantification of 0.15, 0.5, and 0.1 μg/ml, respectively. The % coefficient of variation (%CV) of the assay was <4.8% for serum, <11% for sputum, and <5.5% for urine (data on file with Insmed, Inc., and KCAS, unpublished data).

Pulmonary function tests.

Using a standardized spirometer to evaluate FEV₁, FEV₁% predicted, forced vital capacity, forced expiratory flow between 25% and 75% of forced vital capacity (FVC), and vital capacity, spirometry was conducted at screening, predose on days 1 (baseline), 2, 7, 14, 15, 21, 28, and 29, and during follow-up on days 35, 42, 49, and 56.

Sputum microbiology.

Separate sputum samples were collected for microbiological analysis at screening (days −14 to 0), predose on days 1 (baseline), 14, 21, and 28, and during follow-up on day 35 (i.e., 7 days after the completion of dosing in the study). The sputum samples were processed at a central microbiology laboratory using standard biochemical techniques in order to identify P. aeruginosa. The density of P. aeruginosa in the sputum was considered to be the sum of all the morphologically distinct isolates (morphotypes) per g of sputum. The culture and susceptibility profile analyses of each morphotype of P. aeruginosa were conducted using the Etest method, according to the guidelines recommended by the manufacturer (AB Biodisk NA, Inc., Culver City, CA). In a given sample, the highest amikacin MIC associated with the above-described distinct morphotypes of P. aeruginosa was selected as the final value to represent the baseline MIC value.

Pharmacokinetic analysis.

Candidate PK models, informed by the results of our previous PK analysis (9), were fit to the serum concentration and urine amount data simultaneously using the Monte Carlo parametric expectation maximization (MC-PEM), as implemented in the open-source software S-ADAPT 1.56 (11) (pmethod = 8). The precision of the mean PK parameter estimates involved calculating the standard error of the estimate (SEE) using the full second-derivative matrix along with the third and fourth central moments (11) (pop_err = 4). For serum concentrations observed after the first dose and flagged as being below the limit of quantification (BLQ), the population analysis program (S-ADAPT 1.56) fit the BLQ value using the Beal method 3, which considers the BLQ value to be part of a normal distribution, with the true value being between negative infinity and the lower limit of quantification (12). The weighting of the serum concentrations and urine amounts was based on the reciprocal of the estimated observation variance. Separate additive error models were used to describe residual variability for both the serum and urine data. Standard model selection techniques were used to discriminate among the candidate PK models. The serum AUC values for each patient on each of the PK sampling days (days 1, 14, and 28) were calculated by dividing the post hoc estimates of total clearance based on the final population PK model by the administered LAI dose. The sputum concentration values obtained at each of the scheduled nominal time points were evaluated and summarized by dose.

Pharmacokinetic-pharmacodynamic analyses.

Univariable and multivariable PK-PD analyses for efficacy, as described below, were conducted using data from all patients, including those who received a placebo.

Univariable analyses.

The independent variables considered for the univariable PK-PD analyses for efficacy included amikacin exposure measures and the baseline MIC value for P. aeruginosa. The amikacin exposure measures included the following: dose, day 1 AUC, dose/MIC ratio, and the day 1 AUC/MIC ratio.

The PFT efficacy endpoints representing the dependent variables evaluated in these analyses included the relative change in FEV₁ and FEV₁% predicted from baseline on days 7, 14, 21, 28, and 56. Relative changes in the PFT efficacy endpoints from baseline were calculated by subtracting the PFT value on the study day of interest from the baseline value on day 1, dividing this difference by the baseline value, and multiplying the result by 100.

The relationships between the relative change in each of the PFT efficacy endpoints from baseline on days 7, 14, 21, 28, or 56 and each of the exposure measures or baseline MIC were evaluated. The direction and strength of each relationship, based on data for each of the above-described study days, were assessed using the Spearman's rank correlation test (13).

The relationships between the absolute change in log₁₀ CFU of P. aeruginosa from baseline on days 7, 14, 21, 28, and 35 and each of the above-described exposure measures or baseline MIC were also evaluated in the same manner as for the PFT efficacy endpoints. The absolute change in log₁₀ CFU of P. aeruginosa was calculated by subtracting the log₁₀ CFU value on the study day of evaluation from the predose baseline value on day 1. The relationships between each of the above-described PFT efficacy endpoints and the absolute change in log₁₀ CFU of P. aeruginosa from baseline on days 7, 14, 21, and 28 were also assessed.

Multivariable analyses.

Multivariable repeated-measures mixed effects (RMME) analyses were conducted to evaluate the relationships between each of the efficacy endpoints (relative change in FEV₁ and FEV₁% predicted from baseline and absolute change in log₁₀ CFU from baseline) and amikacin exposure measures (dose, day 1 AUC, dose/MIC ratio, and day 1 AUC/MIC ratio). Random effect intercepts were estimated for each patient to account for an anticipated correlation among repeated measures within patients. Age, body mass index (BMI), baseline MIC of P. aeruginosa, baseline FEV₁% predicted, and study day were included in each model as covariates. MIC was included only in models for which dose or day 1 AUC was assessed. Study day was modeled as a categorical variable to allow for nonlinearity in the relationship between the efficacy endpoints and study day, and the change from baseline in the efficacy endpoints on days 14, 21, 28, and 56 was assessed relative to the change from baseline on day 7. The evaluation of the study day in this manner allowed for an assessment of whether changes in the efficacy endpoints were maintained over the period of study. The interaction between study day and the exposure measure evaluated, which implies that the mean change from baseline for a given efficacy endpoint over time beyond day 7 was dependent upon exposure, was assessed for each model using the likelihood ratio test. Interactions failing to reach statistical significance were removed from each model. The model-estimated effects for the relevant exposure measure values for the average patient (i.e., using the mean age, mean BMI, and baseline FEV₁% predicted for all patients studied) were calculated for RMME models for which the relationship between the exposure measure and efficacy endpoint was significant.

RESULTS

Patient population.

A total of 105 patients (47 males and 58 females) in both studies received either active treatment or a placebo. In study 1, 64 patients (26 males and 38 females) received either active treatment (n = 42) or a placebo (n = 22). In study 2, 41 patients (21 males and 20 females) received either active treatment (n = 27) or a placebo (n = 14). The summary statistics of the baseline demographic characteristics and the baseline PFT and microbiological data for the analysis population and patients stratified by study and dose group are provided in Table 1. Baseline CFU and MIC data were available for 65 patients in study 1 and 36 patients in study 2. For the microbiology assessments, there were 22 patients (21.8%) with MIC values of ≥256 μg/ml, of which 11 were enrolled in study 1 (16.9%) and 11 were enrolled in study 2 (30.6%).

TABLE 1.

Summary statistics of the baseline demographic characteristics, PFTs, and microbiological data for the analysis population and patients stratified by study and dose group

Variablea	Median (minimum, maximum) by dose and patient group (n)
	All patients (105)	Study 1 (64)			Study 2 (41)
		Placebo (22)	280 mg (21)	560 mg (21)	Placebo (14)	70 mg (7)	140 mg (5)	560 mg (15)
Age (yr)	20 (6, 68)	16.5 (6, 29)	15 (9, 28)	18 (7, 29)	26 (13, 37)	37 (20, 47)	33 (28, 43)	28 (9, 68)
Wt (kg)	55 (17, 94.8)	46.7 (18, 70.5)	47.6 (22, 66)	51.3 (17, 78)	59.1 (39, 82.6)	59.9 (51, 67.9)	80.4 (72.2, 94.8)	64.3 (28.4, 94.1)
Ht (cm)	163 (108, 194)	158 (118, 186)	156 (129, 182)	163 (108, 180)	166 (148, 194)	161 (159, 164)	176 (168, 184)	168 (132, 180)
Ideal body wt (kg)	54.8 (5.71, 92.7)	50.8 (14.6, 84.1)	48.6 (24.5, 80.3)	54.8 (5.71, 77.7)	57.7 (43.4, 92.7)	54.5 (51.1, 56.2)	72.7 (64.4, 81.7)	65.3 (26, 78.2)
BSA (m2)	1.59 (0.73, 2.19)	1.44 (0.78 1.91)	1.45 (0.898, 1.79)	1.55 (0.73, 1.98)	1.66 (1.28, 2.11)	1.65 (1.51, 1.78)	1.99 (1.85, 2.17)	1.76 (1.03, 2.19)
BMI (kg/m2)	20.2 (12.9, 30.8)	18.9 (12.9, 27.3)	18 (13.2, 22.1)	18 (13.2, 26.7)	21.9 (17.5, 26.1)	23 (20.2, 25.2)	26.1 (21.9, 30.8)	22.5 (16.3, 28.9)
Creatinine clearance (ml/min/1.73 m2)	116 (60.5, 347)	162 (60.5, 347)	174 (62.7, 273)	103 (60.7, 291)	114 (82.3, 259)	90.7 (86.6, 162)	90.1 (83, 103)	113 (67.8, 145)
Baseline FEV1 (liters)	2.16 (0.53, 4.22)	2.04 (0.84, 3.54)	2.1 (0.7, 3.81)	2.12 (0.53, 4.01)	2.26 (1.4, 4.22)	1.84 (1.38, 2.43)	2.88 (2.47, 3.41)	2.52 (1.18, 4.17)
Baseline FEV1% predicted (liters/s)	65 (38, 131)	61 (42, 131)	63 (38, 105)	58 (43, 98)	70 (44, 86)	65 (42, 72)	69 (55, 80)	67 (45, 102)
Baseline FEF25–75% (liters/s)	1.28 (0.11, 4.21)	1.31 (0.11, 4.21)	1.29 (0.31, 3.56)	1.33 (0.38, 3.61)	1.16 (0.66, 2.56)	0.76 (0.57, 2.01)	1.87 (1, 2.99)	1.53 (0.5, 3.86)
Baseline FVC (liters)	3.09 (0.53, 6.64)	2.52 (1.22, 4.52)	2.83 (1.23, 4.89)	2.81 (0.53, 5.13)	3.51 (2.14, 6.64)	2.69 (2.58, 4.23)	4.35 (3.66, 4.69)	3.75 (1.85, 5.14)
Baseline log10 CFUb	7.54 (0, 9.05)	7.24 (3.08, 8.3)	7.32 (3.59, 8)	7.26 (4.58, 8.99)	8.08 (0, 8.79)	8.53 (7.64, 9.04)	8.07 (6.3, 9.05)	8.04 (6.41, 8.94)
Baseline MIC (μg/ml)b	24 (0, ≥256)	24 (2.0, ≥256)	32 (3, 256)	16 (4, 256)	112 (8, 256)	32 (16, 256)	32 (1.5, 256)	64 (2, 256)

^aBSA, body surface area; FEF_25–75%, forced expiratory flow between 25% and 75% of forced vital capacity.

^bn for these measurements were 101 for all patients, 22 for placebo in study 1, 21 for 280 mg in study 1, and 22 for 560 mg in study 1, 11 for placebo in study 2, 7 for 70 mg in study 2, 5 for 140 mg in study 2, and 13 for 560 mg in study 2.

Pharmacokinetic analysis.

PK data were available for all patients receiving active treatment in studies 1 and 2. Thus, serum, urine, and sputum sample PK data were available for a total of 69 patients. The most robust fit to the serum concentration and urine amount data was obtained using a three-compartment population PK model (one absorption site, the lung, one central compartment, and one urine compartment) with zero-order drug input into the lungs, a first-order process from lungs to the central compartment, and linear elimination. Amikacin in the urine was modeled as accumulating in the urine compartment after clearance from the central compartment by a first-order renal process. The urine compartment was modeled as being emptied at the end of the collection intervals. Given that the predose samples on days 14 and 28 were undetectable, bioavailability relative to day 1 was estimated for days 14 and 28 as one component of interoccasional variability (IOV). In addition, random IOV was estimated for total and renal clearance (CL_R) for all study periods.

The parameter estimates and %SEE for the final population PK model are provided in Table 2. As evidenced by the relatively low %SEE values, the precision of the mean population PK parameters was high. With the exception of the apparent volume of the central compartment (V_c/F) and the first-order rate of drug absorption (ka), the degree of the interindividual variability (%CV) was moderate.

TABLE 2.

Population pharmacokinetic parameter estimates and the associated %SEE for amikacin

Parameterb	Population mean		Interindividual variability (%CV)
	Final estimate	%SEEa	Final estimate	%SEE
CLT/F (liters/h)	67.9	7.48	52.2	27.0
Vc/F (liters)	320	10.6	71.0	26.1
ka (h−1)	2.92	16.3	76.9	50.4
CLR (liters/h)	2.52	7.38	46.8	27.3
Relative F−1 on day 14	0.734	6.71	37.2	36.5
Relative F−1 on day 28	0.651	8.89	55.5	25.1
IOV on CLT/F	1.00		24.8	29.4
IOV on CLR	1.00		33.5	28.1
Residual error for serum	0.204	3.61
Residual error for urine	4.92	5.18

^a%SEE, standard error of the estimate.

^bCL_T/F, apparent total serum clearance (liters/h); V_c/F, apparent volume of distribution of the central compartment (liters); ka, first-order rate of drug absorption; CL_R, renal clearance of parent drug (liters/h); F⁻¹, inverse bioavailability relative to day 1; IOV, interoccasional variability.

The final model described the observed data with excellent precision, as evidenced by the r² of 0.965 for observed versus individual fitted serum concentrations and the r² of 0.958 for observed versus individual fitted urine amounts. The performance of the model was similar among the two studies, with r² values for the observed versus individual fitted serum concentrations of 0.960 and 0.976, and r² values for the observed versus individual fitted urine amounts of 0.896 and 0.978 for studies 1 and 2, respectively. The summary statistics for the serum amikacin AUC values on days 1, 14, and 28, stratified by dose, are provided in Table 3. The increase in bioavailability on days 14 (36%) and 28 (54%) relative to day 1 is reflected by the higher median AUC values observed by dose group on days 14 and 28 than those on day 1. The mean (%CV) elimination half-life of amikacin was 3.11 (37%) hours.

TABLE 3.

Summary statistics for serum amikacin AUC (mg · h/liter) values stratified by dose and study period

Dose (mg)	Median (min, max) AUC (mg · h/liter) on daya:
	1	14	28
70	1.04 (0.739, 1.5)	1.36 (0.993, 2.62)	1.51 (0.955, 2.35)
140	3.04 (1.65, 3.81)	4.07 (2.07, 6.99)	4.84 (1.96, 5.48)
280	4.34 (2.3, 11.1)	6.46 (1.92, 15.6)	6.9 (2.26, 27.2)
560	8.18 (3.32, 20.6)	10.8 (2.9, 54.3)	13.4 (1.82, 45.8)

^amin, minimum; max, maximum.

A large degree of variability was observed in the sputum amikacin concentrations both within and between patients. The summary statistics for sputum amikacin concentrations obtained predose and 1 h postdose, stratified by dose and study day, are presented in Table 4. An evaluation of the sputum amikacin concentrations at each sampling time revealed a dose-related increase in sputum concentration. No significant departures from dose proportionality were observed.

TABLE 4.

Summary statistics for sputum amikacin concentrations (μg/g), stratified by dose and study day

Day and time	Median (min, max) sputum amikacin concn (μg/g) by dose (mg)a:
	70	140	280	560
Day 1
1 h postdose	1,266 (110, 2,292)	1,626 (1,032, 2,550)	915 (15.2, 1,867)	2,286 (11.6, 11,220)
Day 14
Predose	29.8 (4.47, 900)	18.1 (11.4, 32.9)	19.3 (1.8, 106)	35.9 (2.17, 906)
1 h postdose	599 (85.8, 3,420)	1,086 (930, 3,228)	822 (58.4, 4,278)	2,187 (5.79, 13,014)
Day 28
Predose	12.8 (4.57, 241)	20.2 (9.36, 39.4)	14.9 (1.27, 302)	41.1 (3.29, 452.5)
1 h postdose	449 (207, 1,758)	1,050 (654, 3,828)	1,233 (68.4, 11,160)	1,758 (8.28, 15,109)

^amin, minimum; max, maximum.

Pharmacokinetic-pharmacodynamic analyses. (i) Exploratory evaluations.

The median (90% confidence interval [CI]) values for relative change in FEV₁ and FEV₁% predicted and the absolute change in log₁₀ CFU of P. aeruginosa from baseline by dose on the study days of evaluation are shown in Table 5. As assessed using the Wilcoxon signed-rank test, the changes in these efficacy endpoints were significantly different from baseline on all study days evaluated for patients receiving the 560-mg dose. The mean (95% CI) relative change in FEV₁ and FEV₁% predicted from baseline and the absolute change in log₁₀ CFU of P. aeruginosa from baseline by study day for each dose group are shown in Fig. 1. Consistent improvement in the relative change in FEV₁ and FEV₁% predicted and decreases in the absolute change in log₁₀ CFU of P. aeruginosa from baseline over time were most apparent for patients receiving the 560-mg dose compared to those receiving placebo.

TABLE 5.

Median (90% CI) values for the relative change in FEV₁ and FEV₁ % predicted from baseline and absolute change in log₁₀ CFU of P. aeruginosa from baseline and associated P value by dose on study days of evaluation

Dose (mg) for change from baseline type	Results by evaluation day:
	7		14		28		35 or 56 (follow-up)b
	Median (90% CI)a	P value	Median (90% CI)	P value	Median (90% CI)	P value	Median (90% CI)	P value
Relative change in FEV1
0	1.25 (−2.8, 4.84)	0.522	−1.55 (−5.57, 2.36)	0.363	1.17 (−2.49, 3.59)	0.451	−2.77 (−6.84, 1.24)	0.196
70	2.42 (−2.24, 9.64)	0.219	4.44 (−7.62, 17.1)	0.675	4.98 (−9.68, 11.7)	0.297	3.83 (−6.58, 7.33)	0.688
140	4.44 (−9.72, 4.86)	0.855	0.467 (−7.64, 5.87)	0.812	−1.17 (−18.4, 4.45)	0.812	0.229 (−16.3, 6.16)	1
280	10.9 (6.11, 16.7)	<0.001	8.49 (3.17, 15.4)	0.002	10.4 (3.9, 16.3)	0.002	2.07 (−2.23, 6.06)	0.294
560	10.7 (6.01, 16)	<0.001	10.7 (4.83, 17.5)	<0.001	7.33 (2.4, 12.7)	0.01	4.58 (0.617, 12.7)	0.025
Relative change in FEV1% predicted
0	0.476 (−3.46, 4.23)	0.831	−2.23 (−6.64, 1.54)	0.18	0.0542 (−3.56, 3.1)	0.963	−2.58 (−6.52, 1.32)	0.163
70	5.02 (−2.13, 10.8)	0.142	4.66 (−8.33, 18)	0.675	6.41 (−8.87, 12.2)	0.402	4.65 (−6.94, 8.19)	0.578
140	3.95 (−10.1, 5)	0.855	0 (−4.35, 4.94)	1	−0.725 (−17.7, 3.64)	0.812	−2.19 (−16.5, 6.25)	1
280	11.4 (6.47, 17.5)	<0.001	9.36 (2.87, 17.6)	0.005	9.27 (3.19, 15.8)	0.005	1.93 (−2.38, 6.12)	0.409
560	10.3 (5.33, 15.7)	<0.001	9.5 (3.7, 16.4)	0.005	5.75 (1.02, 10.8)	0.028	6.4 (0.481, 15.9)	0.032
Absolute change in log10 CFU
0	0.15 (−0.201, 0.542)	0.432	−0.399 (−0.77, −0.00653)	0.048	−0.325 (−0.835, −0.0457)	0.018	−0.076 (−0.666, 0.301)	0.654
70	−1.2 (−1.56, −0.533)	0.031	−0.938 (−3.67, 0.25)	0.063	−1.72 (−3.83, 0.0434)	0.063	−0.675 (−1.14, −0.267)	0.125
140	−0.605 (−1.41, 0.204)	0.125	−0.0827 (−1.77, 1.04)	0.812	−0.405 (−1.37, 0.286)	0.312	0.258 (−1.17, 1.78)	0.625
280	0.0567 (−0.659, 1.33)	0.896	−0.87 (−2.2, 0.0424)	0.070	−0.669 (−1.48, 0.632)	0.225	−0.18 (−0.913, 0.551)	0.596
560	−1.24 (−1.76, −0.658)	<0.001	−1.26 (−1.82, −0.676)	<0.001	−1.42 (−2.03, −0.839)	<0.001	−0.931 (−2.13, −0.298)	0.002

^aCI, confidence interval.

^bThe follow-up visit represented day 35 for absolute change in log₁₀ CFU from baseline and day 56 for relative change in FEV₁ or FEV₁% predicted from baseline.

FIG 1.

Mean (95% CI) relative change in FEV₁ (A) and FEV₁% predicted (B) from baseline and absolute change in log₁₀ CFU of P. aeruginosa from baseline (C) by study day for each dose group.

(ii) Univariable analyses.

Table 6 shows the Spearman's rank correlation coefficient (r_s) and associated P values for the relationships between the relative change in FEV₁ or FEV₁% predicted or the absolute change in log₁₀ CFU of P. aeruginosa from baseline and dose or day 1 AUC. As evidenced by r_s values of 0.211 to 0.419 and P values of ≤0.033 for the PFT endpoints and r_s values of −0.346 to −0.167 and P values of ≤0.12 for the absolute change from baseline in log₁₀ CFU, the direction of the relationships between all exposure measures and the relative change in FEV₁ or FEV₁% predicted and the absolute change in log₁₀ CFU from baseline for all study days was consistent; as a given exposure measure increased, the relative change in FEV₁ and FEV₁% predicted from baseline increased, and the absolute change in log₁₀ CFU from baseline decreased. Each of these relationships was statistically significant (r_s = −0.346 to −0.206; P ≤ 0.047), with the exception of the relationship between the absolute change in log₁₀ CFU from baseline and the day 1 AUC/MIC and dose/MIC ratio on day 35, the significance of which was only borderline (P = 0.09 and 0.12, respectively). It is important to note that the strength of the significant relationships between dose or day 1 AUC and the relative change in FEV₁ or FEV₁% predicted from baseline for all study days evaluated was modest.

TABLE 6.

Spearman's rank correlation coefficient and associated P values for the relationships between relative change in FEV₁ and FEV₁% predicted and absolute change in log₁₀ CFU of P. aeruginosa from baseline and dose or day 1 AUC

Day and value type	Relative change in FEV1		Relative change in FEV1% predicted		Absolute change (log10 CFU)
	Dose	Day 1 AUC	Dose	Day 1 AUC	Dose	Day 1 AUC
7
rs	0.351	0.357	0.332	0.342	−0.346	−0.32
P	<0.001	<0.001	<0.001	<0.001	<0.001	0.002
14
rs	0.388	0.362	0.343	0.321	−0.261	−0.263
P	<0.001	<0.001	<0.001	0.001	0.010	0.009
21
rs	0.294	0.282	0.263	0.255	−0.277	−0.25
P	0.003	0.005	0.009	0.011	0.006	0.014
28
rs	0.245	0.271	0.211	0.236	−0.264	−0.26
P	0.013	0.006	0.033	0.017	0.01	0.011
35 or 56a
rs	0.259	0.285	0.257	0.285	−0.218	−0.248
P	0.009	0.004	0.01	0.004	0.037	0.017

^aResults shown are for absolute change in log₁₀ CFU from baseline on day 35 and relative change in FEV₁ or FEV₁ % predicted from baseline on day 56.

Univariable assessment of the relationships between relative change in FEV₁ or FEV₁% predicted from baseline on days 7, 14, 21, 28, and 56 and the baseline MIC of P. aeruginosa demonstrated significant relationships only for those PFT efficacy endpoints assessed on day 21 (r_s = −0.233 and −0.252, respectively; P ≤ 0.0258). Univariable assessments of the relationships between the absolute change in log₁₀ CFU of P. aeruginosa from baseline on days 7, 14, 21, 28, and 35 and the baseline MIC failed to demonstrate any significant relationships.

The relationships between relative change in FEV₁ or FEV₁% predicted and absolute change in log₁₀ CFU on days 14, 21, and 28 relative to baseline were also statistically significant (P ≤ 0.048). The relative increases in each of these two PFT efficacy endpoints were associated with decreases in log₁₀ CFU relative to baseline on days 14 (r_s = −0.207 and −0.241; P ≤ 0.048), 21 (r_s = −0.308 and −0.347; P ≤ 0.003), and 28 (r_s = −0.255 and −0.236; P ≤ 0.025). Despite the statistical significance of these relationships, only a small amount of variability in the change in each of these two PFT efficacy endpoints was explained by an absolute change in log₁₀ CFU from baseline on any of the study days, as evidenced by the low r_s² values of 0.043 to 0.120.

(iii) Multivariable analyses.

The relationships between the efficacy endpoints and each of the exposure measures were modeled as linear functions using RMME models, since tests using a second-degree polynomial failed to produce evidence of nonlinearity. RMME models, in which dose or day 1 AUC was evaluated as an independent variable, demonstrated significant relationships with all of the efficacy endpoints evaluated (P, ≤0.007 and ≤0.041 for dose and day 1 AUC, respectively) compared to the models in which the dose/MIC ratio or day 1 AUC/MIC ratio were evaluated (P, ≤0.715 and ≤0.922 for dose/MIC ratio and day 1 AUC/MIC ratio, respectively). Thus, the presentation of results for the RMME models was limited to those containing dose or day 1 AUC data. As evidenced by the prediction of improvements in the relative change in FEV₁ and FEV₁% predicted from baseline of 10.7% and 9.94%, respectively, and a 1.23-log₁₀ CFU reduction for the 560-mg dose of LAI on day 7, compared to observed increases of 10.7% and 10.3%, respectively, and a 1.24-log₁₀ CFU reduction on the same day, the RMME models predicted the observed data well.

The RMME models evaluating the relationships between the relative change in FEV₁ or FEV₁% predicted from baseline and dose are presented in Tables 7 and 8, respectively. The RMME models evaluating the relationships between the relative change in FEV₁ or FEV₁% predicted from baseline and day 1 AUC are presented in Tables S1 and S2 in the supplemental material, respectively. An increase in dose and an increase in day 1 AUC were both significantly associated with an increase in relative change in FEV₁ and FEV₁% predicted from baseline on day 7 (P < 0.001). While BMI was not significant, baseline FEV₁% predicted and the study day were significantly associated with both PFT efficacy endpoints in each of the models (P, ≤0.003 and ≤0.047, respectively). Patient age was significant for the RMME models for FEV₁ (P ≤ 0.048). It is important to note that the model effects for days 14, 21, and 28 relative to the previous study day were not significant for the changes in FEV₁ (P, ≥0.284 for dose and ≥0.287 for day 1 AUC) or FEV₁% predicted (P, ≥0.068 for dose and ≥0.069 for day 1 AUC), suggesting that the changes on day 7 relative to baseline in these PFT efficacy endpoints were sustained to day 28. In the models for change from baseline of either FEV₁ or FEV₁% predicted, the interaction between the study day and dose or AUC was not statistically significant (P ≥ 0.08 for each PFT endpoint) and thus was not retained in the models. The lack of significance of the interaction between study day and dose or AUC indicates that the persistence of effect from day 7 to 28 was consistent across the range of doses or AUC values and that the substantial impact of exposure on these PFT efficacy endpoints occurred on day 7 relative to baseline.

TABLE 7.

RMME model evaluating the relationship between relative change in FEV₁ from baseline and dose

Variable	RMME model			RMME model effect by dose
	Model-estimated effect	95% CIa	P value	Dose (mg)	Model-estimated effect on day 7 for the avg patientb	95% CI
Dose (per 280 mg)	5.37	2.97, 7.78	<0.001	Placebo	1.57	−1.68, 4.83
Age (per yr)	−0.285	−0.571, 0.002	0.048	70	2.92	0.028, 5.81
BMI (per 5 kg/m2)	−0.319	−4.27, 3.63	0.872	140	4.26	1.65, 6.88
Baseline FEV1% predicted	−0.189	−0.314, −0.068	0.003	280	6.95	4.52, 9.38
Study day	—c	—c	0.001	560	12.3	8.78, 15.9

^aCI, confidence interval.

^bThe model-estimated effect for the average patient for each dose was calculated using age and BMI equal to the mean for all patients (i.e., a 22-year-old with a BMI of 20 kg/m² and a baseline FEV₁% predicted of 66%).

^c—, study day was evaluated as a categorical variable with values of 7, 14, 21, 28, and 56 days, yielding 4 effect estimates relative to day 7.

TABLE 8.

RMME model evaluating the relationship between relative change in FEV₁% predicted from baseline and dose

Variable	RMME model			RMME model effect by dose
	Model-estimated effect	95% CIa	P value	Dose (mg)	Model-estimated effect on day 7 for the avg patientb	95% CI
Dose (per 280 mg)	4.97	2.56, 7.38	<0.001	Placebo	1.16	−2.11, 4.43
Age (per yr)	−0.232	−0.518, 0.054	0.108	70	2.40	−0.509, 5.31
BMI (per 5 kg/m2)	−0.595	−4.54, 3.35	0.764	140	3.64	1.01, 6.28
Baseline FEV1% predicted	−0.219	−0.343, −0.094	<0.001	280	6.13	3.68, 8.58
Study day	—c	—c	0.0432	560	11.1	7.54, 14.7

^aCI, confidence interval.

^bThe model-estimated effect for the average patient for each dose was calculated using age and BMI equal to the mean for all patients (i.e., a 22-year-old with a BMI of 20 kg/m² and a baseline FEV₁% predicted of 66%).

^c—, study day was evaluated as a categorical variable with values of 7, 14, 21, 28, and 56 days, yielding 4 effect estimates relative to day 7.

As shown in Tables 7 and 8, the model-estimated increases in relative change in FEV₁ and FEV₁% predicted from baseline on day 7 of 5.37% and 4.97%, respectively, were associated with every 280-mg dose of LAI. These model-estimated increases were not significantly different than those on days 14, 21, and 28 relative to day 7, suggesting that the model-estimated increases on day 7 were sustained to day 28. For a 560-mg dose of LAI, improvements of 10.7% and 9.94% in relative change in the FEV₁ and FEV₁% predicted from baseline, respectively, would be expected on day 7, and as with the 280-mg dose, these effects would be predicted to be sustained to day 28. However, after treatment ended on day 28, model-estimated decreases (95% CI) of 2.55% (0.43, 4.68) (P = 0.018) and 1.31% (−0.88, 3.51) (P = 0.24) were estimated on day 56 relative to day 28 for each of these PFT efficacy endpoints, respectively. For the average patient (i.e., a 22-year-old with a BMI of 20 kg/m² and a baseline FEV₁% predicted of 66%), the model-estimated improvements on day 7 of 6.95% and 12.3% in the relative change in FEV₁ from baseline and 6.13% and 11.1% in the relative change in FEV₁ % predicted from baseline were associated with the 280- and 560-mg doses, respectively, with the persistence of effect for each dose predicted to day 28.

Similarly, as shown in Tables S1 and S2 in the supplemental material, the model-estimated increases in the relative change in FEV₁ and FEV₁% predicted from baseline on day 7 of 0.941% and 0.899%, respectively, were associated with every 1 mg · h/liter of day 1 AUC. These model-estimated increases on day 7 relative to baseline were not significantly different than those on days 14, 21, and 28 relative to day 7. The model-estimated improvements on day 7 for the average patient of 6.18% and 9.94% in the relative change in FEV₁ from baseline and 5.40% and 9.00% in the relative change in FEV₁% predicted from baseline were associated with day 1 AUC values of 4 and 8 mg · h/liter, respectively (which represented the median day 1 AUC values associated with 280 and 560 mg, respectively), with the persistence of effect predicted to day 28.

The RMME models evaluating the relationship between the absolute change in log₁₀ CFU of P. aeruginosa from baseline and dose or day 1 AUC are presented in Table 9 for dose and Table S3 in the supplemental material for day 1 AUC. The RMME model results demonstrated that increased dose and day 1 AUC were each significantly associated with a decline in the absolute change in log₁₀ CFU from baseline on day 7 (P ≤ 0.002). The model-estimated reductions in bacterial burden from baseline on day 7 of 0.615 log₁₀ CFU and 0.113 log₁₀ CFU were associated with every 280-mg dose of LAI and 1 mg · h/liter of day 1 AUC, respectively (P = 0.002). While age, BMI, and baseline FEV₁% predicted were not significant, study day was significantly associated with an absolute change in log₁₀ CFU from baseline in each of these models (P = 0.032 and 0.041 for dose and day 1 AUC, respectively), with an additional significant model-estimated reduction (95% CI) of 0.451 log₁₀ CFU (0.064, 0.837) (P = 0.022) and 0.458 log₁₀ CFU (0.072, 0.845) (P = 0.020) on day 14 relative to day 7, which was independent of the magnitude of the dose or day 1 AUC.

TABLE 9.

RMME model evaluating the relationship between absolute change in log₁₀ CFU of P. aeruginosa from baseline and dose

Variable	RMME model			RMME model effect by dose
	Model-estimated effect	95% CIa	P value	Dose (mg)	Model-estimated effect on day 7 for the avg patientb	95% CI
Dose (per 280 mg)	−0.615	−1.01, −0.224	0.002	Placebo	0.174	−0.369, 0.717
Age (per yr)	−0.016	−0.060, 0.029	0.497	70	0.020	−0.464, 0.504
BMI (per 5 kg/m2)	0.300	−0.325, 0.924	0.340	140	−0.133	−0.572, 0.305
Baseline FEV1% predicted	−0.009	−0.029, 0.011	0.362	280	−0.441	−0.846, −0.036
Study day	—c	—c	0.032	560	−1.06	−1.63, −0.479

^aCI, confidence interval.

^bThe model-estimated effect for the average patient for each dose was calculated using age and BMI equal to the mean for all patients (i.e., a 22-year-old patient with a BMI of 20 kg/m² and a baseline FEV₁% predicted of 66%).

^c—, study day was evaluated as a categorical variable with values of 7, 14, 21, 28, and 35 days, yielding 4 effect estimates relative to day 7.

The model-estimated effects for absolute change in log₁₀ CFU from baseline on days 21, 28, and 35 did not demonstrate any differences relative to those on day 14 (P ≥ 0.362 for both RMME models containing dose and day 1 AUC), suggesting that the model-estimated effects reached a plateau on day 14 and were predicted to be sustained to day 35. In the models for the change in log₁₀ CFU from baseline, the interactions between study day and either dose or day 1 AUC were not statistically significant (P = 0.861 and 0.565, respectively), suggesting that the differences in the mean change in log₁₀ CFU from baseline from day 7 to 14 and the subsequent persistence of effect achieved from day 14 to 35 were consistent across the range of exposures evaluated. Thus, the interactions were not retained in each of these models.

For a 560-mg dose of LAI, the above-described model-estimated effects for absolute change in log₁₀ CFU from baseline per 280-mg dose translated to a reduction of 1.23 log₁₀ CFU from baseline on day 7, with an additional reduction of 0.451 log₁₀ CFU on day 14 relative to day 7. As mentioned earlier, the additional reduction on day 14 relative to day 7 was independent of the dose. Thus, a total 1.68-log₁₀ CFU reduction in P. aeruginosa from baseline on day 14 was associated with a 560-mg dose of LAI, an effect that was predicted to be sustained through day 35. As shown in Table 9, the model-estimated reductions in bacterial burden from baseline on day 7 of 0.441 and 1.06 log₁₀ CFU for the average patient were associated with the 280- and 560-mg doses, respectively. An additional reduction of 0.451 log₁₀ CFU was estimated on day 14 relative to day 7, an effect that was independent of the magnitude of the dose. Thus, the total model-estimated reductions in bacterial burden from baseline on day 14 of 0.892 and 1.51 log₁₀ CFU were associated with each of these doses, respectively, and these effects were predicted to be sustained to day 35.

Similarly, as shown in Table S3 in the supplemental material, the model-estimated reductions in bacterial burden from baseline on day 7 of 0.350 and 0.802 log₁₀ CFU for the average patient were associated with a day 1 AUC of 4 and 8 mg · h/liter of day 1 AUC, respectively. An additional reduction of 0.458 log₁₀ CFU was estimated on day 14 relative to day 7, an effect that was independent of the magnitude of the day 1 AUC. Thus, the total model-estimated reductions in bacterial burden from baseline on day 14 of 0.808 and 1.26 log₁₀ CFU were associated with each of these exposures, respectively, with the persistence of effect predicted to day 35.

DISCUSSION

The first objective of these analyses was to characterize the disposition of amikacin in serum, urine, and sputum samples after the administration of LAI. Using the serum exposures characterized from this analysis, the second objective was to characterize the PK-PD relationships between various efficacy endpoints and amikacin exposure measures (dose, dose/MIC ratio, day 1 AUC, and day 1 AUC/MIC ratio) or baseline MIC value. The efficacy endpoints evaluated included the relative change in the PFT efficacy endpoints (FEV₁ and FEV₁% predicted) from baseline and the absolute change in the bacterial burden of P. aeruginosa in sputum from baseline.

The serum concentration and urine amount PK data obtained following the administration of LAI were best described by a three-compartment model (lungs, serum, and urine) with a first-order linear absorption process from the lungs and a first-order elimination process. The structural PK model and population PK parameter estimates agreed well with those obtained from our previous analysis, including CL_R (42 ml/min based on these analyses compared to 56.7 ml/min based on the previous analysis) (9). The observation of a lower CL_R estimate relative to the mean creatinine clearance of 137 ml/min/1.73 m², similar to what was observed in our previous analysis (9), suggests the possible reabsorption of urinary amikacin at the proximal renal tubules at low systemic exposure. It is possible that this phenomenon is common with other aminoglycosides. However, due to saturation at the proximal renal tubules, reabsorption may be negligible after the administration of standard systemic doses. This hypothesis warrants further exploration.

Due to the limited number of sputum samples collected (two samples) in each sampling period on days 1, 14, and 28 and the large variability observed with drug concentration in sputum, the sputum amikacin concentrations were summarized statistically rather than modeled using compartmental or noncompartmental methods. The sputum amikacin concentrations were much higher than the serum amikacin concentrations, suggesting that high concentrations of amikacin are found in the lungs, despite the very low systemic exposure.

As evidenced by the low r_s values of 0.211 to 0.419, modest but statistically significant univariable relationships (P ≤ 0.033) were identified between the relative change in FEV₁ or FEV₁% predicted from baseline on all evaluated study days and the amikacin exposure measures (dose, day 1 AUC, dose/MIC ratio, and day 1 AUC/MIC ratio). In addition, statistically significant univariable relationships (P ≤ 0.048) were observed between the relative change in FEV₁ or FEV₁% predicted from baseline and the absolute change in log₁₀ CFU from baseline on days 14, 21, and 28. Given the lack of significant relationships between the relative change in the PFT efficacy endpoints from baseline on days 7, 14, 28, and 56 and the baseline MIC of P. aeruginosa, as well as between the absolute change in log₁₀ CFU from baseline on any of the study days and the baseline MIC, baseline MIC was not associated with the efficacy of LAI. Since 22.7% of the reported MIC values on day 1 were right censored (i.e., ≥256 μg/ml), the lack of association between the absolute change in log₁₀ CFU or PFT efficacy endpoints and baseline MIC is not surprising.

The relationships between the absolute change in log₁₀ CFU and day 1 AUC/MIC ratio and between the absolute change in log₁₀ CFU and both the relative change in FEV₁ and FEV₁% predicted are consistent with the results of our previous analysis (9). Similar to what was observed in the previous analysis, the low r_s² values associated with the relationships between the efficacy endpoints and amikacin exposure measures suggest that other independent variables that were not adequately captured are also influencing responses. Factors, such as previous antibacterial therapy, number of courses of antibacterial agents per year, and number of exacerbations per year, may also modify responses. Future analyses based on a larger sample size would allow for the relationships described herein to be examined in the context of these other response modifiers.

While the results of the Spearman's rank correlation test allowed for the evaluation of univariable relationships between the amikacin exposure measures and efficacy endpoints assessed on different study days, RMME analyses were conducted to evaluate the same relationships in a multivariable manner. Such analyses allow for the impact of independent variables of interest other than amikacin exposure (patient age, BMI, baseline FEV₁% predicted, and study day) to be evaluated. The study day analysis, which was based on all the data, allowed for an assessment of the persistence of drug effect for a given study day.

The RMME models that evaluated dose (P ≤ 0.007) and day 1 AUC (P ≤ 0.041) as independent variables rather than the dose/MIC ratio (P ≤ 0.715) or day 1 AUC/MIC ratio (P ≤ 0.922) were found to be more significantly associated with the efficacy endpoints evaluated. In addition, the RMME models showed that the model-estimated improvements on day 7 were not significantly different than those on days 14 and 28 for the relative change in FEV₁ and FEV₁% predicted from baseline and on days 14, 28, and 35 for the absolute change in log₁₀ CFU from baseline.

The RMME models demonstrated model-estimated improvements on day 7 in the relative change in FEV₁ and FEV₁% predicted from baseline of 12.3% and 11.1%, respectively, for the average patient administered 560 mg of LAI. These effects on day 7 were predicted to be sustained to day 28. The RMME model for absolute change in log₁₀ CFU also demonstrated a model-estimated reduction in bacterial burden from a baseline of 1.51 log₁₀ CFU on day 14 for the average patient administered 560 mg of LAI, an effect that was also predicted to be sustained to day 35.

To evaluate whether the magnitude of the model-predicted improvement in FEV₁% predicted due to LAI was consistent with the magnitude of effect observed for patients with CF receiving other antibacterial agents, the results of four clinical studies in which patients with CF received treatment with either 112 mg of tobramycin inhalation powder twice daily or 75 mg of inhaled aztreonam lysine thrice daily were assessed (14–17). Given that the baseline characteristics for the average patient who received treatment with LAI (i.e., 22 years of age with a BMI of 20 kg/m² and a baseline FEV₁% predicted of 66%) were similar to those described in the other four studies (i.e., patients who were 25 to 29 years of age with a BMI of 20 to 21 kg/m² and a baseline FEV₁% predicted of 52% to 55%), such comparisons for improvement in the FEV₁% predicted among agents were considered appropriate.

The model-estimated improvements in the relative change in FEV₁% predicted from baseline on day 7 associated with treatment with 560 mg of LAI for 28 days of 9.94%, which did not significantly change on days 14 and 28, were higher than that reported in patients with CF who received treatment with tobramycin inhalation powder or tobramycin inhalation solution. For the group of patients receiving tobramycin, the improvement in relative change in FEV₁% predicted from baseline at week 4 was approximately 3% for patients that received either tobramycin inhalation powder or tobramycin inhalation solution (14). The model-predicted improvement for treatment with 560 mg of LAI was, however, comparable to that reported in patients who received treatment with inhaled aztreonam lysine. An improvement of approximately 8% to 10% (15–17) was evident for the aztreonam lysine-treated patients. It is important to note that a statistically significant median (95% CI) increase in FEV₁% predicted relative to baseline of 6.4% (0.481, 15.9) on day 56 (P = 0.032) was observed for patients who received treatment with 560 mg of LAI (i.e., 28 days after the end of therapy). Sustained improvement effects after 28 days off therapy were not observed for patients who received treatment with inhaled tobramycin powder (14, 16), tobramycin inhalation solution (14), or inhaled aztreonam lysine (16, 17).

Similar to the above-described comparisons for the relative change in FEV₁% predicted, the model-estimated reductions in bacterial burden from baseline on day 14 associated with the administration of 560 mg of LAI of 1.68 log₁₀ CFU, which did not significantly change on days 21, 28, and 35, were comparable to those reported for patients with CF who received treatment with either tobramycin inhalation powder or inhaled aztreonam lysine. Reductions in the bacterial burden of P. aeruginosa from baseline on day 28 of approximately 1.7 and 1.3 log₁₀ CFU were evident for patients who received treatment with tobramycin inhalation powder and tobramycin inhalation solution, respectively (14). Reductions in the bacterial burden from baseline of 0.6 to 1.5 log₁₀ CFU at the same time point were evident for patients who received treatment with inhaled aztreonam lysine (15–17). While an improvement in the log₁₀ CFU on days 14 and 28 of 0.451 log₁₀ CFU was evident, independent of dose, the predicted additional 1.23-log₁₀ CFU decrease for the absolute change in log₁₀ CFU due to 560 mg LAI alone was comparable to that observed with tobramycin inhalation solution or inhaled aztreonam lysine. An assessment of the reduction in bacterial burden after the end of the first 28 days of treatment could not be adequately made since the sputum samples for microbiological analysis were collected on day 35 for patients receiving treatment with LAI and on day 42 or 56 for patients receiving treatment with the other two agents.

In conclusion, a population PK model was developed based on data from CF patients with chronic infections due to P. aeruginosa, which adequately described the PK of LAI. Univariable PK-PD relationships were observed between all efficacy endpoints and the amikacin exposure measures evaluated. Based on the RMME models, a LAI dosing regimen of 560 mg once daily was associated with statistically significant improvements in FEV₁, FEV₁% predicted, and a reduction in log₁₀ CFU from baseline. As doses of >560 mg were not studied in these analyses, extrapolations of drug effects beyond this dose level should be made with caution. These data serve to support the efficacy of a once-daily dose of 560 mg LAI for the treatment of CF patients with chronic infections due to P. aeruginosa.