Introduction

Cystic fibrosis (CF) is a genetic disease caused by mutations in the CF transmembrane conductance regulator (CFTR) gene.1 CF-causing mutations result in defects in the expression or function of CFTR protein, leading to chronic pulmonary infection, inflammation and progressive lung damage.1 Manifestations of CF are present at birth,2 and disease progression begins early, with structural lung abnormalities present in infancy and childhood.3 4 Lung function (as measured by forced expiratory volume in 1 s (FEV1)) progressively declines during adolescence and adulthood, while the risk of lung or liver disease and CF-related diabetes increases, ultimately leading to organ failure and transplant or premature death.2 5 While CF is a multisystem disorder,1 6 the primary cause of death is progressive lung disease.7 In recent decades, the life expectancy of people with CF in the USA has steadily increased with newborn screening and early treatment with comprehensive standardised care.7–9

Ivacaftor (IVA) is a CFTR potentiator used alone or in combination with CFTR correctors to increase CFTR protein function at the cell surface.1 10 11 IVA was first approved by the FDA in January 2012 for people with CF aged ≥6 years with the G551D mutation.12 The indication was subsequently expanded to include people with CF who are as young as 1-month old and have at least one mutation in the CFTR gene that is responsive to IVA based on clinical data, in vitro assay data, or both.13 EU regulatory approval of IVA was first received in 2012, and IVA is currently approved in over 30 countries across Europe, Australia and Latin America in children aged ≥4 months.14 Clinical trials and long-term real-world studies have demonstrated that IVA treatment is associated with sustained improvements in lung function and decreased pulmonary exacerbation (PEx) rates and mortality risk.15–17 In a parallel US real-world study examining the impact of IVA on long-term health outcomes of people with CF, IVA was associated with significantly lower rates of death, lung transplant, PEx and hospitalisation; better lung function and higher body mass index and body mass index z-scores over a maximum follow-up of nearly 8 years.18

Because lung structure abnormalities may be present in young children with CF, even in the absence of measurable functional changes,3 4 19 early intervention may be critical for maximising lung function preservation. However, evidence demonstrating the impact of age at IVA initiation on pulmonary outcomes is lacking. Therefore, the purpose of this study was to evaluate pulmonary outcomes in people with CF who initiated IVA at younger vs older ages.

MethodsData sources

This study used deidentified, retrospective data from the US Cystic Fibrosis Foundation Patient Registry (CFFPR),20 which included vital status (including death), anthropometrics, genetic mutations, medications, pulmonary function, respiratory microbiology, hospitalisations, PEx, pregnancy, other CF-related complications and organ transplant from January 2003 to December 2019. The US CFFPR protocol review committee provided feedback and approved the protocol prior to study start. The Advarra Institutional Review Board granted an exemption for this research project based on the scope of the research in July 2020.

Study design

In this longitudinal study, people with CF were grouped into cohorts based on their age at IVA initiation (6–10, 11–15, 16–20 and 21–25 years). Three separate analyses were performed to compare pulmonary outcomes in younger versus older IVA initiators (6–10 vs 11–15 years, 11–15 vs 16–20 years and 16–20 vs 21–25 years). Demographics and clinical characteristics were evaluated during the baseline period, defined as the 1-year period preceding the starting age of the younger IVA initiator group and prior to IVA initiation (eg, while aged 5 years for the 6–10 vs 11–15 comparison). The date of IVA initiation was defined as the encounter date prior to that on which IVA use was first identified in the US CFFPR database. If there was no prior encounter date within 6 months of the first identification of IVA use, then the date of the encounter on which IVA use was first noted in the US CFFPR database was used as the date of IVA treatment initiation. For each age group analysis, pulmonary outcomes were compared between younger and older IVA initiators during the specified 5-year outcome assessment period. This allowed assessment of pulmonary outcomes in both groups when they were in the same age range and receiving IVA. The outcome assessment periods comprised people with CF aged 11–15 years (for the 6–10 vs 11–15 years IVA initiator comparison), aged 16–20 years (for the 11–15 vs 16–20 years comparison) and aged 21–25 years (for the 16–20 vs 21–25 years comparison). The index date, or start of the outcome assessment period, for an individual in the younger initiator group in each comparison was defined as the date on which they reached the required age to begin contributing data to the 5-year outcome assessment period. The index date for the older IVA initiators was the date of IVA initiation because their age at IVA initiation was, by definition, within the 5-year age range of the outcome assessment period. Index dates were defined independently for each age group comparison. For example, individuals who initiated IVA aged 11–15 years would have had an index date corresponding to the date of IVA initiation when assessed as an older initiator in the 6–10 versus 11–15 analysis while when assessed as a younger initiator in the 11–15 versus 16–20 analysis, they would have had an index date corresponding to when they turned age 16 years. Individuals were followed from the index date and censored in the event of the following: first occurrence of IVA discontinuation, treatment with a CFTR modulator therapy other than IVA, pregnancy, lung transplant, death, end of the 5-year outcome assessment period or end of data availability. The study design is shown in figure 1.

Figure 1

Schematic illustration of the outcome assessment period in the comparison of younger (ages 6–10 years) vs older (ages 11–15 years) initiators of IVA. In the comparisons of IVA initiators, the baseline period was the 1-year period prior to IVA initiation for the younger initiator group and the outcome assessment period was when both groups were in the same age range and receiving IVA. Here we depict a participant in the age 6–10 years vs 11–15 years comparison, the patient in the younger initiator group initiated IVA at age 8 and whose baseline period began at age 5 years (younger initiator; top) and the patient in the older initiator group initiated IVA at age 11 years and whose baseline period also began at age 5 years (older initiator; bottom). The outcome assessment period for this comparison analysis was during ages 11–15 years, which began on the date the participant reached age 11 years and was receiving IVA. This figure highlights that a participant who initiated IVA at age 6 years (younger initiator; top) only began contributing data to the outcome assessment period 5 years later, whereas one who initiated IVA at age 11 years (older initiator; bottom) immediately began contributing data to the outcomes assessment period. * The end of the outcome assessment period was the earliest date of death, end of data availability, first occurrence of treatment with a CFTR modulator therapy other than IVA, death, pregnancy, IVA discontinuation, or lung transplant.CF, cystic fibrosis; IVA, ivacaftor; y, years.

Study population

A flow diagram of cohort selection is shown in figure 2, along with the three age group comparisons, corresponding baseline and outcome assessment periods. Inclusion criteria were clinical diagnosis of CF and presence of a CFTR-gating mutation (G551D, G1244E, G1349D, G178R, G551S, S1251N, S1255P, S549N or S549R); aged ≥6 years at IVA initiation; IVA initiation on or after 31 January 2012, through 31 December 2018; and data available during the 1-year baseline period. Exclusion criteria were lung transplant or evidence of use of CFTR modulator therapy that had not yet received FDA approval prior to IVA initiation; evidence of pregnancy in the calendar year prior to or in the same year as IVA initiation and presence of R117H or residual function CFTR mutation.

Figure 2

Study population flowchart. At this step, 441 people were dropped for the following reasons: 148 did not have any indication of IVA use, 266 had a date of IVA treatment initiation that was prior to the approval date of IVA (ie, January 31, 2012), 13 had their first indication of IVA use prior to the approval date of IVA, 13 had a date of IVA treatment initiation in 2019, and one had insufficient encounter data so a date of IVA treatment initiation could not be defined.†Among the 602 people with CF included in the comparative analyses, 502 (83.4%) had ≥ 1 copy of G551D mutation.‡ As pregnancy and lung transplant events were only included in the annual data, the date of these events was imputed as January 1. The end of data availability was imputed as December 31 of the last year for which annual data were available.§People included in each of the age group comparisons had to have ≥1 year of annual data available for the specific baseline period and the outcome assessment period. For the outcome assessment period of 11–15 years, the younger and older IVA initiators comprised people who initiated IVA between ages 6–10 and 11–15 years, respectively; for the outcome assessment period of 16–20 years, the younger and older IVA initiators comprised people who initiated IVA between ages 11–15 and 16–20 years, respectively; and for the outcome assessment period of 21–25 years, the younger and older IVA initiators comprised people who initiated IVA between ages 16–20 and 20–25 years, respectively. CF, cystic fibrosis; CFFPR, US Cystic Fibrosis Foundation Patient Registry; CFTR, CF transmembrane conductance regulator; IVA, ivacaftor; RF, residual function; US, United States; y, years.

Study measures

Pulmonary outcomes evaluated included percent predicted FEV1 (ppFEV1; as estimated by Global Lung Initiative equations) and PEx. ppFEV1 was calculated as the average of the best available ppFEV1 measurements from any clinic visit for two quarters over each 6-month period during the outcome assessment period. If ppFEV1 measurements were not available in one quarter, the highest ppFEV1 in each 6-month period was used. In contrast to taking an average of all available ppFEV1 measurements, this approach limited the effect of the frequent testing often observed during PEx periods, which may correspond to transitory declines in lung function. PEx, defined as receipt of home intravenous antibiotic or clinician-reported hospitalisation for PEx, were calculated as the rate per person-year during the outcome assessment period.

Statistical methods

Baseline characteristics were described using frequencies and percentages for categorical variables and means, SD, medians and IQRs for continuous variables. All analyses were performed using SAS Enterprise Guide V.7.1 (SAS Institute, Cary, North Carolina).

Standardised mortality/morbidity ratio weighting

For each age group comparison, standardised mortality/morbidity ratio (SMR) weighting was used to reweight the clinical and demographic characteristics of older initiators to resemble the younger initiators at baseline, which allowed comparison between the age groups during the outcome assessment period.21 22 SMR weights were generated based on propensity scores (PS) to adjust for potential confounding by balancing the distribution of baseline covariates between younger and older IVA initiators.23 The PS was defined as the probability of receiving IVA at a younger versus older age conditional on observed baseline covariates selected based on clinical significance and sample size and was calculated using a logistic regression model. The baseline covariates included sex, health insurance type, average of best available ppFEV1 in each quarter (categorical; for 11–15 vs 16–20 years and 16–20 vs 21–25 years age group comparisons only), change in ppFEV1 (categorical; for 11–15 vs 16–20 years and 16–20 vs 21–25 years age group comparisons only), prevalence of any PEx, prevalence of any CF-related complications, prevalence of any respiratory microorganisms, all-cause hospitalisations (categorical) and any medication use. To eliminate extreme PS, weight trimming at the 1st and 99th percentiles was performed.21 23

SMR weighting was used to estimate the average treatment effect in younger initiators (ie, the average difference that would be found if all people in the younger age group initiated treatment at younger ages vs if they initiated treatment at older ages) by assigning younger initiators a weight of 1, and older initiators a weight equal to the ratio of the estimated PS odds of being a younger initiator. Standardised differences were calculated to evaluate the comparability of baseline characteristics after SMR weighting, with a difference of <10% indicating that SMR weighting had appropriately balanced covariates between groups. Baseline characteristics that remained unbalanced after weighting were included as covariates in the regression analysis.

For each age group comparison, adjusted mean differences in ppFEV1 between younger and older IVA initiators were estimated using an SMR-weighted generalised estimating equation model that accounted for correlation in repeated ppFEV1 measures from the same person. Time was included in the model to account for differences in duration of follow-up. PEx rates (per person-year) were calculated by dividing the frequency of PEx by person-time, and an SMR-weighted generalised linear model with a negative binomial distribution was used to calculate adjusted rate ratios. A non-parametric bootstrap procedure with 999 replications was used to calculate 95% CI for all SMR-weighted analyses.

Role of the funding source

The funder (Vertex Pharmaceuticals Incorporated) was involved in study design and data interpretation and reviewed and provided feedback during manuscript writing. All authors had appropriate access to study data, based on their role, for purposes of fully appraising results, and all authors had final responsibility for the decision to submit for publication.

ResultsBaseline demographics and clinical characteristics

The baseline characteristics of younger versus older IVA initiators for each age group comparison (ages 6–10 vs 11–15 years, 11–15 vs 16–20 years, and 16–20 vs 21–25 years) are shown in table 1. Baseline demographic and clinical characteristics were well balanced between younger and older IVA initiators in each age group comparison after SMR weighting. The proportion of women was broadly similar between younger and older initiators after SMR weighting except in the ages 11–15 versus 16–20 years comparison, where the proportion of women was slightly higher among younger versus older initiators (59.0% vs 52.7%; standardised difference, 12.8%); this variable was further adjusted for regression analyses. During the corresponding baseline periods, SMR-weighted mean (SD) ppFEV1 was similar between younger and older IVA initiators for those who initiated IVA during ages 11–15 versus 16–20 years (95.3 (17.3) vs 93.0 (15.2) at age 10) and for those who initiated IVA during ages 16–20 versus 21–25 years (88.7 (18.5) vs 86.1 (15.4) at age 15). Baseline ppFEV1 was not measured for those who were 5-year old during the baseline period because spirometry was not reliably performed at this age.

Table 1

Baseline demographic and clinical characteristics of younger initiators versus older initiators of ivacaftor

Pulmonary outcomes

The average follow-up during the three outcome assessment periods (11–15, 16–20 and 21–25 years) were as follows: 3.4 and 4.4 years for IVA initiators between ages 6–10 and 11–15 years, respectively; 3.7 and 4.4 years for IVA initiators between ages 11–15 and 16–20 years, respectively; and 3.5 and 3.9 years for IVA initiators between ages 16–20 and 21–25 years, respectively.

Figure 3 shows the association between age at IVA initiation and lung function during the outcome assessment period, comparing ppFEV1 between younger versus older initiators while both groups were in the same age range and receiving IVA. The mean number of ppFEV1 measurements per patient by age group of IVA initiation during the three outcome assessment periods were 7.14 and 8.96 for IVA initiators between ages 6–10 and 11–15 years, respectively; 7.63 and 9.02 for IVA initiators between ages 11–15 and 16–20 years, respectively; and 7.35 and 8.10 for IVA initiators between ages 16–20 and 21–25 years, respectively. Across all age group comparisons, mean ppFEV1 was higher during the outcome assessment period in those who initiated IVA at younger versus older ages, with the highest mean ppFEV1 in the youngest IVA initiator cohort (ages 6–10 years) (figure 3). The mean difference in ppFEV1 (95% CI) during the outcome assessment periods was 6.3 (2.6, 9.8) percentage points for ages 6–10 versus 11–15 years; 11.2 (7.0, 16.5) percentage points for ages 11–15 versus 16–20 years; and 5.9 (0.8, 11.7) percentage points for ages 16–20 versus 21–25 years. The greatest difference in lung function was observed among those who initiated IVA during early versus later adolescence (ages 11–15 vs 16–20 years).

Figure 3

Association between age at IVA initiation and ppFEV1. *Mean follow-up during the outcome assessment period was as follows: 3.4 and 4.4 years for IVA initiators between ages 6–10 and 11–15 years, respectively; 3.7 and 4.4 years for IVA initiators between ages 11–15 and 16–20 years, respectively; and 3.5 and 3.9 years for IVA initiators between ages 16–20 and 21–25 years, respectively. Two children who initiated IVA at ages 6–10 years were excluded because they did not have sufficient ppFEV1 measurements available during the outcome assessment period. † 95% CI did not include the null (0).‡ Mean number of ppFEV1 measurements per patient during the outcome assessment period were as follows: 7.14 and 8.96 for IVA initiators between ages 6-10 and 11-15 years, respectively; 7.63 and 9.02 for IVA initiators between ages 11–15 and 16–20 years, respectively; and 7.35 and 8.10 for IVA initiators between ages 16–20 and 21–25 years, respectively. CI: confidence interval; IVA, ivacaftor; ppFEV1, percent predicted forced expiratory volume in 1 second; y, years.

The frequency of PEx was also compared between younger and older IVA initiators (figure 4). IVA initiation during ages 6–10 years versus 11–15 years was significantly associated with a 52% lower PEx rate during the outcome assessment period of ages 11–15 years (rate ratio (95% CI) 0.48 (0.28 to 0.81)). Numerically lower PEx rates were also observed for younger versus older initiators during the outcome assessment periods across other age group comparisons, but the differences were not statistically significant (11–15 vs 16–20 years, rate ratio (95% CI) 0.79 (0.48 to 1.30); 16–20 vs 21–25 years, rate ratio (95% CI) 0.96 (0.56 to 1.71)).

Figure 4

Association between age at IVA initiation and PEx. *Mean follow-up during the outcome assessment period was as follows: 3.4 and 4.4 years for IVA initiators between ages 6–10 and 11–15 years, respectively; 3.7 and 4.4 years for IVA initiators between ages 11–15 and 16–20 years, respectively; and 3.5 and 3.9 years for IVA initiators between ages 16–20 and 21–25 years, respectively.† 95% CI did not include the null (1).‡ Mean number of PEx during the outcome assessment period were as follows: 100 and 244 for IVA initiators between ages 6-10 and 11-15 years, respectively; 208 and 322 for IVA initiators between ages 11–15 and 16–20 years, respectively; and 196 and 234 for IVA initiators between ages 16–20 and 21–25 years, respectively. § Mean number of person-years of follow-up during the outcome assessment period were as follows: 438 and 575 for IVA initiators between ages 6-10 and 11-15 years, respectively; 446 and 518 for IVA initiators between ages 11–15 and 16–20 years, respectively; and 395 and 448 for IVA initiators between ages 16–20 and 21–25 years, respectively. CI, confidence interval; IVA, ivacaftor; PEx, pulmonary exacerbation.

Discussion

In this study, we quantified the impact of earlier initiation of IVA on pulmonary outcomes by comparing lung function and PEx in people who initiated IVA at younger vs older ages.

Study results show that, when evaluated during the same age range, younger IVA initiators had higher mean ppFEV1 than older initiators. Additionally, initiating IVA treatment during ages 6–10 versus 11–15 years was associated with a 52% lower PEx rate during ages 11–15 years, with numerically lower PEx rates observed between younger versus older initiators at other ages. These findings demonstrate that earlier IVA initiation substantially improves pulmonary outcomes in people with CF and further suggest that, although improved lung function is likely to occur with IVA at any age,17 24 25 progressive lung damage cannot be completely reversed with later treatment. While successive birth cohorts in the US CFFPR (1992–2016) have shown increasing ppFEV1 with improvements in CF care,26 our results suggest that further gains in lung function may be anticipated for future birth cohorts with early use of IVA, and potentially with other CFTR modulator therapies.

Although lung function generally decreases with age in people with CF, the steepest decline is observed during adolescence.27 28 Consistent with this, we observed the largest difference in ppFEV1 between those who had initiated IVA during ages 11–15 versus 16–20 years when both groups were evaluated during ages 16–20 years. While IVA had a greater impact on lung function during adolescence, younger IVA initiators had higher mean ppFEV1 than older initiators across all age group comparisons.

A decline in lung function is associated with an increase in mean annual PEx rate in people with CF.29 PEx rate also generally increases with age, and repeated PEx lead to permanent loss of lung function, which in turn increases the risk of future PEx, resulting in a downward cycle that may lead to lung transplant or death.29–32 In this context, the significantly lower PEx rate observed later (during ages 11–15 years) among the youngest IVA initiators (initiation during ages 6–11 years) versus older IVA initiators (initiation during ages 11–15 years) further supports the potential for sustained benefits and mitigation of disease progression with early IVA treatment.

In clinical and real-world studies, IVA promotes improvements in lung function (ppFEV1) and reductions in PEx in children aged as young as 6 years17 24 25 and improvements in lung structure abnormalities in children aged as young as 2 years.33 Furthermore, IVA use has demonstrated a substantial reduction in the rate of lung function decline in people with CF aged ≥6 years.34 Our findings are consistent with this limited available evidence and show that early IVA use positively impacts respiratory24 25 34 outcomes in children. A recent interim analysis using US and UK registry data found that children with CF aged 2 to <6 years receiving IVA improved across multiple outcomes (eg, nutritional parameters, PEx, hospitalisations, Pseudomonas aeruginosa prevalence) versus a concurrent untreated cohort.35 Furthermore, clinical trials have reported improvements in markers of pancreatic function in children aged 4 to <24 months.36 37 Although these studies support treatment initiation early in life, our study is novel in quantifying the impact of early IVA initiation (as early as age 6 years) on pulmonary outcomes using robust methodology. The present findings provide real-world evidence of the benefit of initiating IVA earlier versus later in life, which can alter the course of CF by preserving lung function and minimising PEx, potentially improving long-term outcomes and survival.

To ensure sufficient follow-up, we focused on a small subgroup of people with CF who had CFTR-gating mutations and were eligible for IVA during the study period. However, the population of people with CF who may benefit from CFTR modulator treatment has expanded with the development of new therapies combining IVA and CFTR modulators that correct defects in protein expression and trafficking, including the triple combination of elexacaftor, tezacaftor and IVA (ELX/TEZ/IVA).1 Although long-term real-world data on ELX/TEZ/IVA are still emerging,38 the similarities to IVA monotherapy in mechanism and clinical trial efficacy25 39 40 suggest that early administration of ELX/TEZ/IVA may similarly improve respiratory outcomes and, in turn, increase the life expectancy of people with CF.

This study had several strengths. First, in each age group comparison, we used SMR weighting rather than matching to balance baseline characteristics between younger and older IVA initiators. This ensured the inclusion of all people treated with IVA in each age group, thereby maximising sample size and statistical power. SMR weighting also ensured balanced groups at baseline. Second, the use of US CFFPR data allowed for a longitudinal analysis of people with CF in the USA, and, as >80% of this population is included in the US CFFPR,41 our findings are highly generalisable within the USA. Third, following FDA approval, there was broad and rapid uptake of IVA among people with CF, with use in 64% of the eligible population by June 2012 and in 81% by December 2012.42 Thus, it is less likely that disease severity or socioeconomic status impacted IVA initiation (figure 2).

This study also had certain limitations. First, as with any observational study, only measured covariates could be controlled for. Therefore, although SMR weighting minimised differences between age groups, there may be residual confounding. In particular, we were not able to include ppFEV1 in the SMR weighting of the youngest age group comparisons due to the lack of ppFEV1 measures at age 5. However, the proportion of children aged 5 experiencing PEx, as a measure associated with lung function, was included in the SMR weighting. Other limitations inherent to observational data, such as lack of standardised assessments and regular clinic visits or missing data, may also introduce variability. While the CFFPR is a robust database, accurate treatment and outcome data, such as ppFEV1 measurements, may be more limited for individuals without adequate follow-up. Additionally, informative censoring may have occurred if people who were censored differed from those who remained, in terms of risk factors for the outcome of interest. Moreover, older initiators must have survived until a later age to initiate IVA and thus be included, which may have resulted in older initiators being healthier than younger initiators at baseline. However, because both groups were balanced on baseline characteristics after SMR weighting and were evaluated at the same age ranges, this impact should be minimal. Furthermore, individuals were grouped by age range to ensure adequate sample size but may have had heterogeneous disease severity. Also, we limited this analysis to 3 age group comparisons based on IVA approval timing and data availability. Finally, changes over time in recording practices, standard of care and other information not captured in the US CFFPR may have introduced additional confounding.