Participants were enrolled between February 2004 and October 2007 at seven US CF centers: Nationwide Children's Hospital (Columbus, Ohio), Children's Hospital of Pittsburgh (Pittsburgh, Pennsylvania), Washington University (St. Louis, Missouri), Baylor College of Medicine (Houston, Texas), the University of Iowa (Iowa City, Iowa), University Hospital (Cincinnati, Ohio), Seattle Children's Hospital (Seattle, Washington), and the University of Washington Medical Center (Seattle, Washington). The IRB at each site approved the protocol and amendments. Participants provided written informed consent (and assent when applicable). Inclusion and exclusion criteria were as described in Supplementary Material
Sputum samples obtained at Visit 1 (Day -21 to -14) were cultured quantitatively.3
For each P. aeruginosa
morphotype present at ≥10% of the most prevalent, antibiotic susceptibilities were determined by semi-automated broth microdilution3
(Sensititre System, AccuMed, Westlake, Ohio) and biofilm susceptibility testing.11
Multiresistance was defined as resistance to all agents in at least two of three classes (aminoglycosides, fluoroquinolones, beta-lactams) by either conventional or biofilm testing, based on CLSI breakpoints.16
Conventional and biofilm regimens of two antibiotics were selected centrally using a published algorithm,17
with not more than one drug from each antibiotic class, as follows: aminoglycosides (amikacin, tobramycin); beta-lactams (ceftazidime, meropenem, piperacillin-tazobactam, ticarcillin-clavulanate); fluoroquinolones (ciprofloxacin); macrolides (azithromycin). For each bacterial morphotype, the algorithm calculated the conventional minimum inhibitory quotient (MIQ) and the biofilm minimum inhibitory quotient (BIQ) of each drug, defined as achievable serum concentration divided by conventional or biofilm minimum inhibitory concentration (MIC or BIC), respectively. For each subject's isolate set, MIQ and BIQ composite scores were used to rank antibiotics from most to least active. For each testing method, the two most active drugs from different antibiotic classes were selected. When the most active antibiotics had the same composite scores, a tiebreaker rule based on ease of administration, cost, and likelihood of adverse drug reaction was applied (hierarchy, most to least preferred: azithromycin, ciprofloxacin, ceftazidime, meropenem, ticarcillin-clavulanate, piperacillin-tazobactam, tobramycin, amikacin). To define drug allergies, the beta-lactams were subdivided into penicillin, cephalosporin, and carbapenem subclasses. If a participant had a history of allergic reaction, that class or subclass of antibiotics was excluded from potential assignment.
Participants were assigned to treatment groups using a block randomization procedure, stratified by study site. Participants and site personnel were informed of antibiotic assignments, but were not told the testing method (conventional or biofilm) on which this was based. The effectiveness of this blinding was not evaluated.
At Visit 2 (Day 0), spirometry was performed,18
and participants started on a 14-day course of their assigned antibiotic regimen; recommended dosing was as described in Supplementary Material
. Participants assigned to regimens that included an IV antibiotic had the option of home IV therapy (OptionCare, Buffalo Grove, Illinois) or admission to a General Clinical Research Center; all such participants who initiated treatment received the majority of the 14-day course at home. Visit 3 (Day 7 +/- 2 days) was an interim visit or telephone call to monitor adverse events and review medication use. Visit 4 (Day 14 +/- 2 days) was timed to occur 12–48 hours after the last antibiotic dose, a time at which residual concentrations in airway secretions were deemed unlikely to affect bacterial growth. A sputum sample was collected for quantitative culture and spirometry was performed.
The primary outcome measure was microbiological response, change in P. aeruginosa sputum density, calculated as log10 of end-of-treatment density minus log10 of screening density, in CFU/g. The secondary outcome measure was pulmonary response, change in forced expiratory volume in one second (FEV1), calculated as end-of-treatment FEV1 minus the baseline FEV1, in liters.
A CF Foundation Data Monitoring Committee conducted one interim review to assess study conduct and safety. A ≥10% rate of acute pulmonary exacerbation or need to change antibiotics in either treatment group was the pre-specified threshold for considering early study termination.
This pilot trial was designed to determine feasibility and estimate treatment effect and variability for planning a full-scale trial; owing to cost constraints, it was not powered to detect significant differences in microbiological or clinical efficacy between treatment groups. The target sample size was 20 randomized participants in each treatment group. Outcomes were analyzed using linear regression models that adjusted for baseline values, defined as sputum density measured at screening and FEV1 measured at initiation of antibiotic treatment. Analyses were performed using STATA™ (versions 9.2 and 10.0, College Station, TX) and S-plus (version 6.2, Insightful Corp., Seattle); modified intent-to-treat comparisons included all randomized participants who received at least one dose of their assigned regimen.