Systematic review of the dry powder inhalers colistimethate sodium and tobramycin in cystic fibrosis

Quantity of research available

The search retrieved 866 potentially relevant citations; 857 from searches of electronic databases, and nine from secondary searches of relevant reviews, articles and the sponsor’s submissions. Of these, 841 were excluded at the title and abstract stage, leaving 25 potentially includable citations.

The full texts of the 25 articles were obtained for investigation; of these, 20 did not meet the inclusion criteria and were excluded (fig. 1)[12]. Three studies comparing colistimethate sodium DPI or tobramycin DPI with a nebulised antibiotic were included in the review. One study (the EAGER trial) investigated tobramycin DPI in combination with the TOBI Podhaler, and two studies (COLO/DPI/02/06 and COLO/DPI/02/05) investigated colistimethate sodium DPI in combination with the Turbospin device. Information about the three trials included in the systematic review was initially available from five sources; these comprised one published journal article [13], two conference abstracts [14, 15] and two manufacturers’ submissions to NICE [16, 17] (with subsequent clarifications from the manufacturers). Additional sources became available during the appraisal process comprising two European Public Assessment Reports (EPARs), which were published by the European Medicines Agency (EMA) (for TOBI Podhaler in 2011, updated in 2012 [18], and for Colobreathe in 2012 [1]), and a journal article relating to the COLO/DPI/02/06 study [19]. Whilst EPAR sources are not peer reviewed, they are scrutinised through the EMA licensing process. Data received after seeking clarifications from both manufacturers, but which had not been made publicly available, are not reported here.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Flow diagram of study inclusion. CF: cystic fibrosis.

Study characteristics

All studies were open label, multicentre studies, two of which were multinational studies. A summary of characteristics of the included studies can be found in table 1. The EAGER trial [13, 16, 18] was a large trial (n=533) performed between 2006 and 2009 that compared tobramycin DPI and nebulised tobramycin. The COLO/DPI/02/06 trial [1, 17] was a slightly smaller trial (n=380) that finished data collection in 2007 and compared colistimethate sodium DPI with nebulised tobramycin. Both of these trials were powered to detect clinically relevant changes in FEV₁. The COLO/DPI/02/05 trial was much smaller (n=16), and the dates of the study were not reported; compared colistimethate sodium DPI with nebulised colistimethate sodium. The EAGER trial and COLO/DPI/02/06 studies were both 24 weeks in duration, whilst COLO/DPI/02/05 was a crossover trial that reported outcome data at 4 weeks (before crossover) and 8 weeks only (after crossover).

Inclusion and exclusion

Criteria were largely comparable between the two major trials (EAGER and COLO/DPI/02/06). Patients in all three trials had a chronic P. aeruginosa infection and were clinically stable. They were aged >6 years in the EAGER and COLO/DPI/02/06 trials, and >8 years in the COLO/DPI/02/05 trial. Patients in all trials continued with usual CF treatments (except other routine anti-pseudomonal treatments). The criteria used to define a chronic infection did not meet with EMA recommendations in any trial [10]. As such, it is the comparability of the degree of infection that is unclear between the included patients in the trials.

The COLO/DPI/02/06 trial had a run-in period whereby participants were required to have received 16 weeks (two cycles) of nebulised tobramycin prior to beginning the trial in order to eliminate the short-term peak in FEV₁ % predicted efficacy, which has been documented in patients receiving tobramycin, and which is not sustained over time [20]. Conversely, the EAGER trial had a proportion of patients (6.2%) who had not used tobramycin or any other anti-pseudomonal antibiotic before, and a substantial proportion (67.6%) who had not used anti-pseudomonal antibiotics immediately prior to the 28-day wash-out period. Some or all of these patients may have experienced an initial peak in efficacy during the trial, and may also be more likely to experience adverse events associated with tobramycin than patients in the COLO/DPI/02/06 trial.

Patient characteristics

Baseline characteristics of patients are presented in table 2. The patients in the COLO/DPI/02/06 trial had a lower mean age than those in the EAGER trial. Mean age was not reported for the COLO/DPI/02/05 trial. The baseline FEV₁ % pred values were similar between these two trials, with the FEV₁ % pred in COLO/DPI/02/06 being slightly lower, indicating that patients had slightly worse lung health to start with overall in comparison to the patients in the EAGER trial. In both trials, the lack of consistency in inclusion criteria and conformity with EMA guidelines [10] affects the comparability of the trials to one another.

Study withdrawals

In both trials, more patients withdrew due to adverse events than for any other single reason. Withdrawal of consent/patient request was the second most common reason. In the EAGER trial, adverse events accounted for proportionately more withdrawals in the tobramycin DPI arm (13%) than in the nebulised tobramycin arm (8.1%). Similarly, more patients withdrew consent from the trial in the DPI arm (8% versus 4.3%).

In the COLO/DPI/02/06 trial, the same pattern was seen with more patients withdrawing from the colistimethate sodium DPI arm (9.8%) than from the nebulised tobramycin arm (1.6%) due to adverse events, although withdrawals due to patient request were lower in the DPI arm (2.7% versus 5.8%).

Quality assessment

All three trials randomised patients to treatment. The method of randomisation was acceptable in the EAGER trial, but was not clearly described in the COLO/DPI/02/06 trial. The EAGER and COLO/DPI/02/06 studies adopted a method of allocation concealment using an interactive voice system. Participants were not blinded to the treatment arm in two of the studies (EAGER and COLO/DPI/02/06), although the Forest Laboratories UK Ltd submission states that FEV₁ data were collected by a blinded investigator. The COLO/DPI/02/05 study does not state whether any blinding was attempted. Due to the fact that part of the assessment was to establish patient convenience, blinding would have been inappropriate. However, this still represents a risk of bias for potential benefit and harm from both patients and investigators.

Two of the studies (COLO/DPI/02/06 and COLO/DPI/02/05) reported that relatively similar numbers of patients in the intervention and control groups dropped out of the study; however, data did not support this. The EAGER trial reported a somewhat higher attrition in the intervention group (26.9%) compared to the control group (18.2%) and, as data were not corrected for withdrawals, the study may therefore be at high risk of attrition bias. It would seem that more evidence was recorded than was reported for some FEV₁ % pred measurements and for body mass index (BMI) in the EAGER trial, which may indicate a degree of reporting bias. In addition, there was a substantial delay in publishing the trial results for the COLO/DPI/02/06 study, which finished collecting data in 2007. The results have recently been published in a peer-reviewed journal [19]. Delayed publication of clinical trial data is an area of concern for reporting the results of empirical work generally, but is a specific concern for trials in CF [21].

Lung function

The most commonly reported measure of lung function from the included studies was FEV₁. Where available, data were obtained at 4, 20 and 24 weeks. Whilst outcome data for FEV₁ at 24 weeks were provided in the data submissions from Forest Laboratories UK Ltd, only data at 20 weeks was available in full from any source for the EAGER trial. Tobramycin was administered for 28 days followed by 28 days off treatment, which results in a peak and trough in FEV₁ values. This has the potential to affect estimates of comparative efficacy, and it would seem appropriate to consider results at both the peak and trough of the efficacy cycle. As such, data from both 20 and 24 weeks are presented within this review where possible.

The presentation and analysis of data varied across the studies. The COLO/DPI/02/06 reported several analyses for FEV₁ data, including intention-to-treat (ITT) and per protocol populations with last observation carried forward (LOCF) imputation and without imputation. Tests specified a priori showed that the data were non-normal in distribution; therefore, additional non-parametric analyses and analyses using logarithmically transformed data were also performed by the manufacturer to correct for this. The EAGER trial data in the manufacturer’s submission were not transformed, and a non-parametric test was not performed, although no test of normality was apparently planned or performed. No imputation was performed on the Novartis data, and only limited data were presented at 24 weeks. Some adjusted comparative data were presented, with adjustments for main effects treatment, baseline FEV₁ % pred and pooled centre.

Non-inferiority was defined in the COLO/DPI/02/06 trial as having been satisfied if the lower limit of the 95% confidence interval was no lower than -3.0% for both the per protocol and ITT populations. It was also stated that the primary efficacy analysis was performed using a LOCF imputation and that the Shapiro-Wilk test would be used to check for non-normal distribution. Non-inferiority was reported in the study by Schuster et al. [12] for logarithmic analysis (table 3) but not for the original analysis on the absolute scale. Additional data from the industry submission indicates that the conclusion on non-inferiority in the logarithmic analysis is based upon patients with data at baseline and 24 weeks (completers), in the per protocol population. However, the LOCF analysis did not demonstrate non-inferiority. A further non-parametric analysis was conducted as tests indicated that the assumption of normality was not fully satisfied by the logarithmic transformation and non-inferiority was demonstrated in this analysis. Similar results were reported for data at 20 weeks.

Non-inferiority was defined in the EAGER trial as having been satisfied if the lower limit of the 95% confidence interval was no lower than -6% in the ITT population. No imputation was carried out and no tests for normality were planned. Data were available at 20 weeks but not 24 weeks. The EAGER trial reported non-inferiority for tobramycin DPI at 20 weeks, supported by least squares mean±se difference relative change of 1.1±1.75%, which meets the criteria for non-inferiority. As noted previously, this analysis was performed with no imputation of data in the ITT population. A non-inferiority analysis was not presented for the data at 24 weeks, where FEV₁% measurements are expected to be lower for both study arms than at 20 weeks. It is unclear whether this would have affected the estimate of comparative efficacy between the two arms.

Both the EAGER trial and the COLO/DPI/02/06 trial conclude that tobramycin DPI and colistimethate sodium DPI, respectively, are non-inferior to nebulised tobramycin as they have met their respective pre-defined non-inferiority criteria. However, it is unclear whether the standards for non-inferiority are comparable between trials.

Microbiological outcomes: colony density and resistance

In the COLO/DPI/02/06 trial, resistance (4 mg·L⁻¹ or 8 mg·L⁻¹ breakpoints) to colistimethate sodium remained very low (≤1.1%) in the colistimethate sodium DPI arm, whilst resistance to tobramycin did not change substantially during the study. In the EAGER trial, resistance (8 mg·L⁻¹ breakpoint) to tobramycin started at around 20% and was lower at 24 weeks. It is unclear if this change was statistically significant.

Sputum density tests were conducted in the EAGER trial only. Mean change from baseline log₁₀ values showed numerically greater reductions in sputum density and were achieved with tobramycin DPI at 20 weeks in comparison to nebulised tobramycin. Results at 24 weeks were not reported. Statistical significance was not reported.

Exacerbations

Data on protocol defined acute exacerbations were not reported in a consistent way across the three included trials. The EAGER trial did not report acute exacerbations and did not have a protocol definition of an exacerbation. Instead they report “lung disorders” defined as “generally reported by the investigator as pulmonary or CF exacerbation” [13]. This is not an entirely specific measure and it is unclear what other events may also have been included in this outcome. However, the percentage of patients experiencing at least one episode of lung disorder was 33.8% (dry powder) versus 30.1% (nebulised).

The COLO/DPI/02/06 trial did report protocol defined acute exacerbations. Numerically, more patients on colistimethate sodium DPI experienced acute exacerbations compared to nebulised tobramycin in COLO/DPI/02/06 (31.1% versus 26.1%, respectively), although it is unclear if the same trend would be observed for data relating to the number of events. Acute exacerbations were not reported in the COLO/DPI/02/05 trial. It is not possible to draw a comparative conclusion as to the relative efficacy between trials in terms of exacerbations, given the difference in the way they have been reported, and the uncertainty about the comparability of the patient data populations and characteristics.

Body mass index

The EAGER trial reported baseline BMI but did not report data for time-points at the end of the trial at either 20 or 24 weeks [16]. The COLO/DPI/02/06 trial reported BMI data for the ITT population, which demonstrated very little change in BMI from baseline [17]. BMI data for the COLO/DPI/02/05 was not presented as it was not an outcome under investigation in that trial.

Health-related quality of life

Two trials investigated this outcome (COLO/DPI/02/06 and COLO/DPI/02/05). COLO/DPI/02/06 data are presented as the adjusted means of Cystic Fibrosis Questionnaire (CFQ) domains from baseline to week 24. Most of the scores tended to be in favour of the dry powder intervention although none of the differences are statistically significant. This quality of life measure has not been validated and is not preference based. The COLO/DPI/02/05 trial also used the CFQ measure but the data are not available in the public domain. There was no quality of life data reported in the EAGER trial. However, treatment satisfaction was measured using the Treatment Satisfaction Questionnaire for Medication.

Adverse events

The percentage of patients experiencing any adverse event was high in the EAGER trial (DPI 90.3% versus nebulised 84.2%), the COLO/DPI/02/06 trial (DPI 93.6% versus nebulised 89.1%) and the COLO/DPI/02/05 trial (nebulised 100% versus DPI 60%), although this is to be expected in a patient population with CF who have a high level of baseline adverse events. The EAGER trial did not state how many events were severe, whilst both of the colistimethate sodium DPI trials reported more severe events in the intervention (DPI 39.0%) arm versus the control (nebulised 6.2%) arm. This can, in part, be explained by the fact that during the run-in period, patients were normalised to nebulised tobramycin. Patients experiencing adverse events during the run-in period may have withdrawn, or the events may have resolved before randomisation. As such, the population was subject to a degree of selection bias for patients tolerant to nebulised tobramycin. Serious adverse events (which are internationally defined as adverse events that cause death, are life threatening, require hospitalisation/prolong hospitalisation or result in disability or birth defect [22]) appear to occur approximately equally, but slightly less frequently, in the DPI treatments in both the EAGER (DPI 27.4% versus nebulised 29.2%) and COLO/DPI/02/06 (DPI 4.3% versus nebulised 6.2%) key trials.

In each of the three trials, the percentage of patients withdrawing from the study due to adverse events was higher in the dry powder intervention groups (EAGER 13.0%; COLO/DPI/02/06 11.8%; COLO/DPI/02/05 12.5%) than in the nebulised control groups (EAGER 8.1%; COLO/DPI/02/06 2.6%; COLO/DPI/02/05 0%). Patients in both the EAGER and COLO/DPI/02/06 trials were largely experienced with nebulised tobramycin and, as discussed earlier, it is likely that this difference in dropout rate is at least, in part, due to selection bias of patients tolerant to nebulised tobramycin and desensitisation to its adverse events through prior use.

table 4 documents the most common adverse events (≥5% in any group) occurring in any of the three trials. The data presented relate to the number of patients who experienced adverse events. Data on the actual number of events were available for the COLO/DPI/02/06 and COLO/DPI/02/05 studies but are not presented here. The most common adverse event in all three trials was cough. The percentage of patients experiencing cough was higher in the COLO/DPI/02/06 and COLO/DPI/02/05 studies than in the EAGER trial, although this may again represent a difference in the definition of cough used in the studies rather than an actual difference in incidence of cough, as the difference persists when comparing the nebulised tobramycin arms of each trial. Cough was more common in the DPI intervention group for all trials. Cough is a known side-effect of dry powder formulations and is thought to generally reduce over time with improved technique, and may be controlled with the use of bronchodilators, to some extent, in some patients.

Whilst no statistical comparisons have been made, other adverse events that appear to be worse in the DPI arm include chest discomfort and dysphonia in the tobramycin DPI arm in the EAGER trial, and throat irritation in the colistimethate sodium DPI arm in the COLO/DPI/02/06 trial. There are minor improvements in a number of some adverse events in the colistimethate sodium DPI arm (table 3).

Mortality

Three patients died in the tobramycin DPI group in the EAGER trial. Two patients died in the tobramycin nebulised group in the COLO/DPI/02/06 trial. None of the deaths are attributed to the study medication, and the studies were not powered for these rare events.

Compliance

Compliance with study medication was reported in both key trials, but it is not clear whether the definition of compliance, methods used to ascertain compliance and analyses provided are compatible between trials. In the COLO/DPI/02/06 trial, fewer patients were compliant with medication in the colistimethate sodium DPI arm than in the nebulised tobramycin arm (66.7% versus 70.7% respectively complied with >75% of doses). The EAGER trial did not define how compliance was judged, but simply states it was generally high with >90% compliance in both arms.

Conclusions

Both DPI formulations have been shown to be non-inferior to nebulised tobramycin as measured by FEV₁ %. However, the results of these trials should be interpreted with caution due to the means by which the results were analysed within the trials, the length of follow-up, and concerns about the ability of FEV₁ % to accurately represent changes in lung health. Clear and comparable data on acute exacerbations, a highly relevant outcome, are missing. However, within trials it would appear that both DPI formulations may increase the likelihood of an exacerbation. It appears that patients on DPI formulations experience more adverse events of cough and may be less likely to tolerate the treatment. As such, the advantages and non-inferiority of DPI treatments compared to nebulised tobramycin remain unclear when all relevant outcomes are considered. Full evaluation of the long-term efficacy of dry powder anti-pseudomonal antibiotics for chronic P. aeruginosa is not currently possible and, therefore, phase 4 follow-up trials are required to prove effectiveness for long-term adherence rather than short-term benefit, which will be major driver of drug/device development. Additionally, relevant data on compliance, convenience and HRQoL using a validated measurement are lacking.