Long-term effectiveness of dual CFTR modulator treatment of cystic fibrosis

Discussion

This study provided real-world data of the long-term effectiveness of LUM/IVA and TEZ/IVA on important pulmonary outcomes and nutritional status, covering almost 4000 patient-years of observation in pwCF homozygous for F508del, up to 3 years after the introduction of these dual CFTR modulating therapies. Although the pivotal RCTs and open-label extension trials demonstrated a clear efficacy of LUM/IVA and TEZ/IVA on several clinical end-points in pwCF with a baseline FEV₁ between 40% and 90% pred [4–7], our results emphasised that real-world effectiveness is less pronounced, with considerable differences in long-term trends among pwCF and a FEV₁ below 40% pred or above 90% pred upon CFTR modulator initiation.

Real-world improvement of annual FEV₁ decline was slightly lower than the 1% pred change in FEV₁ decline estimated by the long-term open-label extension trial data [6, 7]. This was demonstrated by a mean change of 0.81% pred and 0.88% pred per year after CFTR modulator initiation in both the subgroup with baseline FEV₁ 40%–90% pred and in the entire cohort, respectively. In contrast with the short-term trials [4, 5], the acute change of FEV₁ after modulator initiation was limited in the entire cohort. However, we did observe an acute improvement of 2.59% pred in the subgroup of participants with a baseline FEV₁ between 40% and 90% pred that approximated the original trial results [4–7].

Interestingly, the mean acute improvement of FEV₁ in participants with a baseline FEV₁ <40% pred was not significantly different from the group with a pre-modulator FEV₁ 40%–90% pred. Moreover, the improvement of FEV₁ decline was even higher in those with FEV₁ <40% pred before CFTR modulator initiation. Similar short-term improvements in pwCF and severe lung disease were already reported in subgroup analyses of clinical trials and in several case series [11], but the long-term benefits in this subgroup have not yet been demonstrated before.

In addition, long-term changes in BMI and BMI Z-score in this study were moderate compared to previous trials [6, 7], and despite the acute decrease in the duration of i.v. antibiotic use in the first year after modulator initiation, the mean duration of i.v. antibiotic treatment continued to increase again in the subsequent years.

Taken together, the results of this study emphasise that translation of clinical trial results into daily clinical practice can be difficult, especially in chronic diseases like CF. Most of the discrepancies are probably explained by the different populations, design and settings of traditional trials compared to observational real-world studies. This could be related to the relatively short follow-up of RCTs, as well as to the stringent selection criteria which usually exclude people with, for example, severe or limited lung disease (FEV₁ <40% pred and >90% pred) or people with CF-related comorbidities such as diabetes and liver disease. In addition, clinical trial conditions regarding co-medication and treatment adherence are strictly controlled, whereas temporary or permanent treatment discontinuation is more likely to occur in practice [8]. Real-world studies with a long-term follow-up are therefore important to provide additional post-approval data of the impact and sustainability of treatments on the entire heterogeneous population [12].

So far, seven studies have been published that assessed the effectiveness of LUM/IVA in a real-world setting [13–19]. Most of these studies were conducted in small populations, examining different subgroups and outcomes with a follow-up period of 1 year after LUM/IVA initiation and a limited observation period, not exceeding 845 patient-years.

The present study substantially contributes to the existing real-world evidence, because the follow-up period covered on average 7 years before CFTR modulator treatment and up to 3 years after modulator initiation. Moreover, this study included 3844 patient-years of observation of a relatively large and heterogeneous population of F508del-homozygous pwCF aged 12 years and older at different disease stages, which reflects daily clinical practice. In addition, we adjusted for the confounding effect of age, which is known to be associated with rate of lung function decline [20].

Overall, our results were consistent with previous studies that suggested real-world effectiveness to be less pronounced compared to the initial trials. Most studies reported a moderate change from baseline FEV₁ % pred [13, 14] and a moderate change in FEV₁ % pred decline after 1 year [16, 17] or 2 years [18] of follow-up. The discrepancy with a different recently published study that focused on predictors of long-term clinical outcomes using encounter-based FEV₁ measurements [19] might be explained by the inclusion of annual best FEV₁ measurements in the NCFR. Annual best measurements may provide a better estimation of long-term trends, as this reduces the impact of measurement variability over time compared to multiple repeated measurements. Given the strong (nonlinear) association of lung function decline with age [20, 21], trends were adjusted for age at baseline in this study. The short- and long-term improvement of BMI and nutritional status could be interpreted as modest and was more profound in adolescents [13, 14, 18]. The use of the different reference values for adults and adolescents limits a direct comparison of BMI and BMI Z-score trends between age groups, which has also not been assessed in other real-world studies. Nevertheless, similar differences were reported in the PROGRESS trial, showing an increasing BMI trend in treated pwCF and matched registry controls, whereas BMI Z-score and weight-for-age trend improved after LUM/IVA initiation compared to a decline in matched registry controls [6]. Moreover, LUM/IVA and TEZ/IVA might induce a short-term improvement of pulmonary exacerbations [13, 14] and reduce the use of i.v. antibiotics in the first year after treatment initiation in pwCF above 12 years of age, but this improvement was not sustained in the subsequent years [18, 19]. This could indicate that the benefit of dual CFTR modulators on severe pulmonary exacerbations diminishes in the long-term, but it could also be related to a decreasing long-term adherence to modulators or to a reduced prescription or adherence to other co-medication such as dornase α, hypertonic saline and inhaled antibiotics in a real-life setting.

The contrast between short- and long-term changes in this study also illustrates that traditional short-term clinical end-points such as FEV₁ % pred might not always be the best measures to capture treatment benefits, especially when effect sizes are limited, populations are heterogeneous and sample sizes are small, which frequently occurs in rare diseases such as CF. Long-term trials or observational real-world studies might partially overcome this problem because they could reveal an inhibition of disease progression, but alternative approaches will be needed since long-term studies are not always feasible and require sufficient short-term evidence first.

An important limitation of this study was the relatively large proportion of missing data in i.v. antibiotic treatment duration, which was not consistently collected in the NCFR throughout the entire study period, particularly in the years before CFTR modulator initiation (2010–2014). Although we used appropriate statistical models to adjust for missing data, we cannot rule out that this might have influenced the results. Even though we did adjust for the most important confounders age and sex, we were not able to include data regarding either treatment discontinuation and side-effects or concomitant treatments such as hydrators, dornase α, azithromycin or other inhaled or oral antibiotics, which might have respectively underestimated or overestimated the reported effectiveness. Due to the transition from LUM/IVA to TEZ/IVA during the observation period, this study provides combined results about the effectiveness of both dual CFTR modulators. Based on the additional subgroup analyses that compared the groups who did and did not switch to TEZ/IVA, the influence of transition was considered as limited.

In conclusion, this real-world study showed that long-term FEV₁ decline improved up to 3 years after the introduction of LUM/IVA and TEZ/IVA, which was also observed for BMI Z-score in children, but not for BMI in adults. Intravenous antibiotic treatment duration was reduced in the first year after modulator initiation, but this duration increased in the subsequent years. Compared to the efficacy reported in previous clinical trials, real-world effectiveness of the dual CFTR modulators is less pronounced and varies considerably between pwCF and different baseline FEV₁ levels.