Overall timelines

The first patient was screened on 6 June 2012 and the first dose was administered on 13 June 2012. The last patient was screened on 24 June 2013, the last dose was administered on 1 May 2014 and the last follow-up visit was performed on 30 May 2014.

Screening and recruitment

Following prescreening of clinic databases, 191 patients were considered likely to be suitable and agreed to a screening visit. Forty of these were not enrolled, as they failed to fulfil inclusion criteria (see Figure 15). Of the 151 who passed screening, 11 patients subsequently withdrew, either because of a change of mind or because of the development of an exclusion criterion. One hundred and forty patients were randomised: 62 (44%) to placebo and 78 (56%) to active treatment. Two patients in each group withdrew post randomisation and were not able to reattend for any further visits; therefore, this left 136 in the ITT cohort, predefined as being randomised and having any follow-up data available. The PP population was predefined as those patients receiving ≥ 9 doses and comprised 116 patients (placebo, n = 54; active treatment, n = 62). The reasons for 20 patients discontinuing are shown in Figure 15. Of note, the first small-molecule CFTR modulator, ivacaftor, was licensed and approved in UK countries during the trial for patients with the G551D mutation. Three of our subjects elected to leave the trial to enable them to receive this treatment; others were prepared to wait until the end of the trial and continued.

FIGURE 15

The CONSORT diagram showing screening, randomisation and patients dosed. The ITT and PP groups are identified.

Subject demographics

Subjects randomised to active-treatment and placebo groups were well matched at baseline for age (and proportion of paediatric patients aged < 18 years), sex, centre and CFTR gene mutation class (F508del/F508del vs. other) (Table 2). The two groups were also similar with regard to clinical characteristics including lung function (FEV₁%) and BMI, used here as a general measure of nutritional status.

TABLE 2

Baseline demographic data of subjects randomised to the active-treatment and placebo groups. Data are presented for both ITT and PP populations. Groups were well matched

Primary outcome: relative change in FEV₁ percentage

Out of 116 PP subjects, 114 (placebo, n = 54; active treatment, n = 60) had paired pre–post-treatment measurements of percentage predicted FEV₁. Two patients, despite fulfilling the PP definition of receiving ≥ 9 doses, were excluded from this analysis as they did not have spirometry performed at follow-up visits; in one case, this test was contraindicated owing to a recent surgically induced pneumothorax. The other patient had recently withdrawn from the trial because of time commitments and was unable to return for follow-up measurements. There was a significant (p = 0.046; Figure 16) TE in the primary outcome, percentage predicted FEV₁, with the active-treatment group having an ANCOVA-adjusted 3.7% (95% CI 0.1% to 7.3%) greater change in FEV₁ than the placebo group at 12 months’ follow-up. The effect was seen from month 1 onwards, with a sustained divergence of the two groups (see Figure 16).

FIGURE 16

Primary end point (relative change in percentage predicted FEV₁). Time course of the response of the primary outcome to either placebo (green) or active treatment (black). ‘Pre’ and ‘post’ indicate the mean of two measurements (more...)

Changes in FEV₁ over the course of the trial were variable between subjects. Figure 17 illustrates individual relative improvements from pretreatment to follow-up in a waterfall plot in which improvement is indicated by a positive value. Post hoc analysis showed that 21 subjects [placebo, n = 6 (11%); active treatment, n = 15 (25%)] demonstrated a change in percentage predicted FEV₁ of ≥ 5%.

FIGURE 17

The distribution of FEV₁ changes in individual patients, shown separately for the two groups. (a) Active treatment; and (b) placebo. Positive values indicate an improvement.

The TE in the ITT population with spirometry measurements both pre-dosing and within the protocol-defined window after their final dose (placebo, n = 56; active treatment, n = 65) was 3.6% (95% CI 0.2% to 7.0%; p = 0.039). The SAP prespecified an additional sensitivity analysis based on the area under the curve for the percentage predicted FEV₁. With this analysis, using the PP population, the estimated TE (in units of percentage predicted FEV₁, rather than the relative change) was 1.32% (95% CI –0.48% to 3.12%; p = 0.15), consistent with the relative TE observed in the primary analysis.

Major secondary outcomes

Physiology

There was also a significant TE in FVC (p = 0.031; Figure 18; see also Figures 25 and 26). No difference was observed in the LCI at the follow-up period, although the serial time point graph (Figure 19) demonstrates that the active-treatment group appear to have a better-preserved LCI in the earlier stages of the trial.

FIGURE 18

Forced vital capacity. Time course of the response of FVC to either placebo or active treatment. ‘Pre’ and ‘post’ indicate the mean of two measurements at the respective time points. Error bars indicate SEM. There was a significant (more...)

FIGURE 25

Secondary outcome measures. Forest plot showing the responses of secondary outcome measures to placebo or active treatment. To allow results from different end points to be plotted on a common scale, the estimated TEs were standardised to be presented (more...)

FIGURE 26

Responses of active and placebo groups separately for secondary outcomes. Forest plot showing the responses of the placebo or active-treatment arms when assessed by the predefined subgroup values shown. To allow results from different endpoints to be plotted (more...)

FIGURE 19

Change in LCI. There was no significant TE in LCI, both groups increasing (worsening) slightly over the year of the trial.

Computed tomography scans

Paired pre and post follow-up high-resolution CT scans were available for 115 patients; the one PP patient missing is the same patient also missing from the analysis of the primary end point as she had been lost to follow-up. The other patient missing from the primary outcome is represented here, as CT scanning did not pose the same risks following her pneumothorax as a forced expiratory manoeuvre would have done.

Bronchiectasis is defined as dilatation and thickening of the airways. It is considered irreversible. An intervention applied for this period of time would not therefore be expected to have any impact on the extent of bronchiectasis and this was the case in this cohort (Figure 20). Although it was not statistically significant, it was interesting to note a trend for less worsening in the severity of bronchiectasis in the active-treatment group than in the placebo.

FIGURE 20

Change in CT scores for bronchiectasis (a) extent; and (b) severity.

Gas trapping is a feature visible on expiratory films when areas of lung fail to empty properly because of airway obstruction. There were increases in the percentage of gas trapping in the placebo group that were not demonstrated in the active-treatment group and which led to a significant TE (p = 0.048; Figure 21).

FIGURE 21

Gas trapping on CT scan. Over the course of the study, gas trapping indicative of small airways disease, increased in the placebo group (worsened) but did not in the active-treatment group leading to a statistically significant TE (p = 0.048). (more...)

Scores of mucus plugging (large or small; Figure 22) and airway wall thickness (Figure 23) did not differ significantly between the two groups, although, for every parameter, the change was smaller in and, therefore, favoured the active-treatment group.

FIGURE 22

Mucus plugging on CT scan. (a) Large mucus plugging; and (b) small mucus plugging (sometimes termed ‘tree in bud’).

FIGURE 23

Airway wall thickness on CT scan.

Quality-of-life scores

Baseline quality-of-life scores for the domains of major interest in a trial of respiratory treatment (respiratory and physical) were high for the group as a whole (median 87.5%). There were no statistically significant TEs at the end of the trial, although, again, the TE appeared to favour the active-treatment group (Figure 24).

FIGURE 24

Quality-of-life (QoL) scores. Change in the (a) respiratory and (b) physical domains on the validated CF quality-of-life questionnaire CFQ-R.

Influence of baseline parameters on primary outcome

We had predefined that the influence of baseline parameters on TE of the primary outcome would be explored in an attempt to define a ‘responder’ group. We reasoned that differences between responders and non-responders could be based on any of the following:

• Severity of underlying lung disease: patients with milder disease could respond more because the inhaled medication reached further into the airways, or less because they were already so well. We examined this using FEV₁, LCI and CT parameters.
• Age.
• Sex.
• CFTR mutation class: although we expect gene transfer to be mutation independent, this seemed necessary to confirm.
• The presence or absence of P. aeruginosa: gene transfer could be affected by the presence of this organism either because of the greater levels of inflammation likely associated with this chronic infection or because of an interaction with the wide array of exoproducts it produces. We did not consider that we had sufficient power to break this down further on the basis of other infections.
• Concomitant medications. Those considered to be particularly important to examine were:
  • the anti-inflammatory agents, azithromycin and corticosteroids, either of which could affect the host response to the medication
  • DNase, which could degrade the DNA in the trial product (of note, patients receiving this drug were asked to withhold it for 24 hours before and after each dose)
  • hypertonic saline, which by influencing mucus clearance could potentially reduce contact time.
• Side effects caused by the trial medication: we considered the hypotheses that those patients who developed either lower respiratory or systemic side effects acutely after dosing could have either a higher (side effects resulting from successful delivery) or smaller (host response limited success of transfection) chance of being a responder. To examine this, we classified patients on the basis of having (1) lower respiratory or (2) flu-like reactions within 48 hours of at least four doses. There were no subjects in the latter group and, therefore, only the lower airway symptoms were examined in this analysis.

For each of the continuous variables, the whole PP group was divided based on median values for that measure. The other variables were divided into two dichotomous groups.

Figure 27 illustrates that TE was independent of age, sex, mutation class, concomitant treatments and response to dosing. However, stratification by baseline disease severity on the basis of percentage predicted FEV₁ suggested a difference in TE between the lower half (49.6–69.2% predicted FEV₁) who had a TE of 6.4% (95% CI 0.8% to 12.1%) and the upper half (69.6–89.9% predicted) of TE 0.2% (95% CI –4.6% to 4.9%) (interaction analysis p = 0.065).

FIGURE 27

Stratification of primary outcome measure. Forest plot showing stratification of the primary outcome response by prespecified variables. To allow results from different end points to be plotted on a common scale, the estimated TEs were standardised to (more...)

This is further illustrated in Figure 28, which shows the evolution of FEV₁ over time during the trial in (a) the half of patients with more severe disease and (b) the half with milder disease. The greater TE was not because of the use of ‘relative change’ magnifying an effect, as this was also apparent when absolute change in FEV₁ percentage was examined (data not shown).

FIGURE 28

Primary outcome stratified by baseline FEV₁. Time course of the primary outcome response stratified by baseline FEV₁ at trial entry. (a) Severe group, baseline FEV₁ 49.6–69.2% predicted; and (b) milder group, baseline FEV₁ 69.6–89.9% predicted. (more...)

Influence of baseline FEV₁ on other outcomes

Having observed that baseline FEV₁ appeared to have an important influence in the magnitude of the TE based on the primary outcome, we next examined other outcome measures for this effect. Subjects were divided into upper and lower halves based on the median value for the entire group, as previously, and TEs were compared. Figure 29 shows that many of the assays mirrored the effect which had been seen for the primary outcome, namely the more severe group of patients showing an approximate doubling of the TE compared with the values in the whole, unstratified group. This resulted in an absolute improvement (vs. stabilisation) in many of the assays. Of note, in those with less severe FEV₁ at trial entry, biomarkers associated with smaller airways, in particular LCI, still showed a TE favouring the active-treatment group.

FIGURE 29

Stratification of secondary outcomes. Forest plot showing stratification of secondary outcomes by the severity of baseline FEV₁ at trial entry. To allow results from different end points to be plotted on a common scale, the estimated TEs were standardised (more...)

We confirmed that these benefits seen in the more severe subgroup were not related to an increased number of antibiotic courses during the trial. Both active-treatment and placebo groups received a median of three courses of oral or intravenous antibiotics. In the half with more severe disease, stratified by FEV₁, both placebo and active-treatment groups received three courses, while among those with less severe disease, the placebo group received three courses and the active-treatment group received two. Thus, the observed TEs were independent of concurrent antibiotic courses.

Summary of efficacy outcomes

The trial achieved its primary outcome confirming a statistically significant TE favouring the active-treatment group on relative change in FEV₁. This was underscored by statistically significant TEs in the secondary outcomes, FVC and gas trapping on CT. Furthermore, for all other secondary outcomes, although not statistically significant, the standardised TE analysis favoured active treatment. In general, these improvements were seen across the group and were independent of sex, age and mutation class. In contrast, the magnitude of TE was influenced by the severity of baseline lung disease assessed by FEV₁: patients in the more severe half at baseline experienced larger improvements not only in FEV₁ but also in several other outcomes. Changes in the small airway measure, LCI, were still seen in patients with less severe baseline FEV₁, suggesting that these patients can still benefit.