Section 1. Introduction

The earliest known description of cystic fibrosis (CF) was in 1938 by Dorothy Andersen of the New York Babies Hospital.¹ Since that time, survival from the disease has increased steadily. Only 3% of adults with CF born between 1947 and 1949 could expect to survive to 30 years of age.² The median predicted survival from the 2009 UK registry is now 34 years,³ and the previously predicted survival to >50 years of age for children born in 2000 is now looking realistic, even in the absence of effective therapy to correct the genetic defect.⁴

During the 1970s and 1980s, the main emphasis on CF treatment focused on antibiotics. Although intravenous (IV) access techniques improved and patients lived longer, many became chronically infected with Pseudomonas aeruginosa. In the UK, 36% of the CF population is chronically infected with P. aeruginosa and 15% with Staphylococcus aureus.³ In the US, the figures are 52%  and 51%, respectively.⁵

Reduced levels of chronic P. aeruginosa infection have been attributed by some to the aggressive use of nebulized antibiotics, regular microbiological monitoring, prompt antibiotic treatment of first isolates, and intensive use of IV antibiotics when inhaled antibiotics have failed.^{6, 7, and 8} One Belgian center achieved a rate of chronic P. aeruginosa of 20.7%, compared with a national average of 48%.⁶ In the UK, Lee and colleagues reported a decline in the number of patients with chronic P. aeruginosa infection, from 24.5% to 18.1% (P < 0.05), which was thought to be the result of these measures.⁷ An increase in representation among individuals with a relatively “mild” CF phenotype identified by molecular diagnostic techniques may also have contributed to the observed decrease in prevalence of chronic P. aeruginosa infection in the population.

Section 2. Continuous vs intermittent therapy: how did we get here?

Three antibiotics are commonly used for inhalation therapy in patients with CF: tobramycin, colistin, and aztreonam. Colistin has been in use in Europe since the 1980s, following studies by Littlewood and colleagues⁹ and Jensen and associates.¹⁰ Its use remains widespread, due largely to its tolerability profile and the fact that P. aeruginosa resistance is relatively rare.¹¹ Inhaled tobramycin has been the preferred chronic suppressive therapy against P. aeruginosa in North America since the landmark trial by Ramsey and collaborators in 1999.¹² Only one head-to-head comparison of inhaled colistin vs. tobramycin has been conducted to date,¹³ and although tobramycin appeared to be more efficacious, the dose of colistin used was 1 MU twice daily, which is only half the maximum recommended dose. Inhaled aztreonam is relatively new to the market, having been granted US Food and Drug Administration (FDA) approval for use in 2010.

The specific reasons for choice of inhaled antibiotic are beyond the scope of this article. A critical question in the selection of a therapeutic regimen of chronic suppressive aerosolised antibiotics for CF is whether to use continuous or intermittent therapy (usually 28-day on/off cycles). If the efficacy and safety of the two approaches were similar, then intermittent therapy would win hands down because of increased convenience and reduced cost. Two other potential advantages, although intuitively sensible, should be accepted only if robust evidence is present – namely, greater adherence to treatment and reduced antimicrobial resistance. Comparison of the two approaches to scheduling is difficult, because historically, a continuous regimen has been used with aerosolised colistin and an intermittent regimen with tobramycin (and lately with aztreonam).

The origin of the 28-day intermittent treatment cycle can be traced back to a trial of 3‐times-daily nebulized tobramycin, administered continuously for 3 months. This study was conducted in 22 patients (no control group) in the 1980s.¹⁴ Mean forced expiratory volume in 1 second (FEV₁) improved significantly in these patients at 28 days, but the improvement had diminished by the end of 3 months of treatment and was close to baseline approximately 1 month after treatment ceased. In addition, substantially more bacterial isolates with reduced susceptibility to tobramycin were observed after 3 months of nebulized tobramycin, although this proportion declined over the following year (off treatment). A subsequent randomized, blinded, placebo-controlled, crossover study by Ramsey and colleagues¹⁵ (Table 1) again reported significant improvements in FEV₁ and forced expiratory flow between 25% and 75% of forced vital capacity (FEF_25%–75%) during 28 days of treatment with inhaled tobramycin – improvements that diminished from days 28 to 56 of treatment. There was also an associated decrease in P. aeruginosa sputum density by a factor of 100 with tobramycin treatment during the first 28 days, but less of a decline following the first 28 days.¹⁵ Based on these two studies, it was postulated that continued administration of tobramycin beyond 28 days would not result in an increased treatment effect and was more likely to lead to selection for bacterial isolates resistant to tobramycin.¹⁴

Intermittent therapy has been accepted as “standard of care” by such regulatory agencies as the FDA. It is likely that pharmaceutical companies undertaking clinical trials of new formulations of inhaled antibiotics will be expected to compare their product with intermittent nebulized tobramycin and presumably will adopt an intermittent regimen for the new product, as well for current ongoing trials.

Section 3. Comparisons of continuous vs. intermittent therapy

Only one trial (Nikolaizik et al., 2008²¹) compared continuous vs. intermittent therapy. Different doses of tobramycin were used in the two treatment arms, rendering comparison difficult. No other head-to-head comparisons of the same dose of inhaled antibiotic administered either continuously or in 28-day on/off cycles have been conducted to date, thus a direct comparison of safety and efficacy (or adherence and antimicrobial resistance) is not possible.

So can we make indirect comparisons? The randomized, controlled trials (RCTs) that have evaluated long-term aerosolised antibiotic therapy in patients with CF have been compared in an exhaustive systematic review, conducted by Ryan and coworkers.²² These 19 trials (with 1724 participants) found that lung function (as measured by FEV₁) improved and exacerbations of respiratory symptoms were less frequent in the antibiotic-treated group vs. the placebo-treated group.

In this article, we have categorized the RCTs of nebulized antibiotics for chronic suppressive therapy in CF conducted to date according to whether the regimen was intermittent (Table 1) or continuous (Table 2). Table 3 lists trials in which treatment was administered for 28 days only, which cannot be said to be intermittent or continuous, and have been included for comparison. As in the systematic review, RCTs of poor methodological quality or <4 weeks' duration have been omitted and six new studies have been added.^{13, 18, 19, 30, 31, and 32} We have omitted trials in which the intention of therapy was eradication of P. aeruginosa (not trials of chronic therapy). Of note, the study of intermittent aztreonam lysine by Oermann and colleagues¹⁸ is a continuation trial of two previous trials: AIR-CF1¹⁹ and AIR-CF2,²⁰ respectively. The Nikolaizik trial²¹ is not included in any of the tables, as it compares continuous and intermittent regimens, and a trial that considered only patients with Burkholderia cepacia complex has been omitted as well.³⁴

Section 4. Trials of intermittent therapy

Table 1 illustrates trials of intermittent therapy. These 5 trials have enrolled 1293 participants, with a median treatment duration of 20 weeks (range, 12 to 72 weeks). Whenever FEV₁ was a recorded outcome (not always the primary outcome), these data have been included in the table. Although FEV₁ is not necessarily the most clinically relevant outcome (or the most important to patients), it is objective, has been reported most consistently, and is closely related to prognosis.³⁵ Hence, it is included as the outcome measure in Table 1, in order to allow comparison between trials. In 3 of the 5 trials, the FEV₁ improved to a significantly greater degree in the active vs. the comparator group. Of the remaining 2 trials, Oermann and associates¹⁸ did not report a statistical test, although the authors did comment that generally there was greater improvement in FEV₁ observed in the 3‐times-daily compared with the twice-daily treatment group. The remaining study by Murphy and coworkers¹⁶ was terminated early and did not detect a significant improvement in FEV₁, although the trend appeared to favor active treatment.

Recent trials have used a primary outcome measure that is more relevant to patients. McCoy and coworkers²⁰ used time to IV antipseudomonal antibiotic use (21 days longer in the group receiving nebulized aztreonam lysine; P = 0.007). The primary efficacy endpoint in the study by Retsch-Bogart and colleagues¹⁹ was change in patient-reported respiratory symptoms on the Respiratory Scale of the CF Questionnaire-Revised³⁶ (9.7-point improvement; P = 0.001; a minimum difference of 5 points was set a priori).

Few side effects have been reported in these trials, although tinnitus (a recognized adverse effect associated with tobramycin use) occurred more often in the active group in the largest trial of inhaled tobramycin.¹² Clearly, nebulized antibiotics have the potential to do harm. Along with tinnitus, acute kidney injury has been reported with the use of nebulized tobramycin,³⁷ although the occurrence is rare. The use of 28‐day on/off cycles will reduce lifetime drug exposure, and, hence, the risk for renal toxicity and ototoxicity with nebulized tobramycin use. Furthermore, in animal models, the half-life of aminoglycoside antibiotics in the hair cells of the inner ear is measured in months,³⁸ so an alternate monthly regimen should allow for improved drug clearance.

So what about antimicrobial resistance? Burns and associates³⁹ reported the antimicrobial resistance data from the pivotal trial of nebulized tobramycin,¹² using the accepted minimum inhibitory concentration (MIC) breakpoints for parenteral therapy. The percentage of patients with a strain of P. aeruginosa having an MIC above the parenteral breakpoint (>16 μg/mL) increased from 13% to 23% among individuals receiving inhaled tobramycin over the 24-week trial (vs. a decrease from 10% to 8% among controls). The parenteral MIC breakpoint used to define resistance may not be relevant in these circumstances, however, as much higher concentrations of antibiotics are achieved by the inhaled route. Hodson and collaborators¹³ reported “a small increase” in MIC following 28 days of treatment in tobramycin-treated patients but not in colistin-treated patients, but they presented no data. The study of long-term nebulized aztreonam lysine¹⁸ reported no increase in antimicrobial resistance in either arm, although 3 patients in the times-daily group experienced first isolation of B. cepacia after commencing the study therapy. Neither the Burns paper³⁹ nor the 28-day duration aztreonam lysine trials^{19 and 20} demonstrated an increase in such other pathogens as B. cepacia complex, Stenotrophomonas maltophilia, and Achromobacter xylosoxidans.

Adherence to (or compliance with) treatment is usually better in clinical trials than in routine practice. Briesacher and colleagues⁴⁰ identified 804 patients with CF and determined their adherence to treatment by means of claims made through occupational health insurance plans. The authors concluded that only 7% of patients received 4 cycles of inhaled tobramycin per year (“high adherence”), compared with the 6 cycles prescribed. High adherence was associated with a reduced risk for hospitalization (odds ratio, 0.4; 95% confidence interval, 0.19 to 0.84) compared with low adherence (2 cycles per year). This is very different from the adherence reported in clinical trials. In the studies shown in Table 1, in which adherence has been defined, the definition of satisfactory adherence varies from participants taking 66%²⁰ to 80%¹³ of prescribed doses. With both definitions, adherence was >90%. A Canadian study estimated that about half of the cost of nebulized tobramycin might be recouped because of a reduced requirement for hospital-based and home IV antibiotic treatment.⁴¹ However, poor adherence in routine clinical practice may diminish considerably the economic benefits claimed. In addition, the demand on time is substantial when patients are receiving inhaled antibiotics. Administration time with nebulized tobramycin solution is 15 to 20 minutes, excluding the time required for cleaning and sterilization of the delivery device⁴²; this equates to approximately 14 to 18 hours per month. An on/off regimen would save a considerable amount of time every other month; however, whether this would result in improved adherence is not known.

Section 5. Trials of continuous therapy

In contrast to trials of intermittent therapy, trials of continuous nebulized antibiotics enrolled fewer participants (206 in total; see Table 2). Where FEV₁ was measured, most patients showed an improvement, with the exception of the study by Nolan and associates,²⁴ in which a slightly larger decrease in FEV₁ % predicted was reported in the active treatment group. However, the authors noted that this did not reach statistical significance.

Although they enrolled fewer participants, these studies were of longer duration (median, 9 months; range, 3 to 33 months) and provided a total of 283 person-years of drug exposure to yield data on adverse effects. The more recent studies of intermittent therapy, however, may include open-label continuation phases to assess long-term safety. Studies of continuous therapy are generally older (1981 to 1989 vs. 1993 to 2010 for studies of intermittent aerosolized antibiotic therapy). Trials in the 1980s were not subject to the same rigorous regulations on reporting adverse events that have applied over the last 2 decades. Nevertheless, in trials of continuous therapy, adverse effects were no more common in the active treatment group than in the comparator group.

In terms of antibiotic resistance, 2 studies did not report on this,^{26 and 28} whereas 4 reported no significant difference.^{10, 23, 24, and 25} Of the remaining 2 trials, MacLusky and colleagues²⁹ reported the development of tobramycin-resistant P. aeruginosa in 4 of the 12 active treatment arm patients who were originally infected with sensitive strains, and Stead and coworkers²⁷ reported the development of P. aeruginosa with partial resistance in 3 patients (1 to ceftazidime and 2 to carbenicillin), who regained full sensitivity 1 to 2 months following completion of the trial. Although it may well be that continuous use of some classes of inhaled antibiotics can result in persons with CF becoming refractory to their beneficial effects, in vitro antibiotic susceptibility testing is a notoriously poor method for predicting the efficacy of antibiotics delivered by any route in patients with CF.^{43, 44, and 45} For this reason, use of in vitro antibiotic susceptibility tests to predict the relative long-term efficacy of continuous vs. intermittent inhaled antibiotic therapy is problematic.

Section 6. Single-agent therapy vs rotation of antibiotics

We identified no clinical trials that directly compared a single inhaled antibiotic vs the rotation of antibiotics in patients with CF.

Section 7. The way forward

From a thorough review of the literature, it is apparent that both intermittent and continuous regimens of aerosolised antibiotics are effective in maintaining lung function and are associated with few adverse effects. Recent studies have shown that aerosolised antibiotics may reduce respiratory symptoms and defer the need for IV antibiotics. A 28‐day on/off regimen of nebulized tobramycin is associated with an increase in antimicrobial resistance, but in general, emergence of other pathogens is not seen with either intermittent or continuous therapy. A major confounding factor is the fact that trials of continuous therapy were conducted almost a decade earlier than trials of intermittent therapy. These trials used different antibiotics (gentamicin and beta-lactams, as well as tobramycin) and enrolled far smaller numbers of participants.

There are planned and ongoing studies (Table 4) of other antibiotics, but these generally follow the precedent of intermittent therapy. Clearly what is needed are well-designed and adequately powered RCTs of intermittent vs continuous therapy, using the same dose of the same antibiotic in each arm. The challenges and limitations associated with the designs of such trials are addressed in the article by VanDevanter et al. in this supplement.⁴⁶

Section 8. Conclusion

Both intermittent and continuous inhaled antibiotics work, although direct comparisons of their efficacy are difficult. The administration of intermittent antibiotics is less time-consuming to patients; however, whether this necessarily means improved adherence is not yet known. It would be easy to assume that adherence would be better with intermittent regimens, although the opposite may also be true if patients do not experience any appreciable deterioration in health during “off” months and therefore feel reluctant to resume treatment at the start of “on” months. The cost of intermittent therapy might be less expensive than that of continuous therapy, although this would be true only when the same drugs (in the same dosage) are compared head to head with both regimens, and we have not identified any trials in which this has been performed. Some evidence suggests that long-term use of inhaled tobramycin is associated with increased resistance; however, given that deterioration in lung function is the main predictor of mortality among individuals with CF, this factor must be taken in the context of the patient's well-being as a whole. In conclusion, additional trials are required before we can base our treatment decisions, at least in terms of scheduling strategy, on good evidence.