Article Text

Download PDFPDF
Lung clearance index predicts pulmonary exacerbations in individuals with primary ciliary dyskinesia: a multicentre cohort study
  1. Florian Singer1,
  2. Anne Schlegtendal2,
  3. Sylvia Nyilas3,
  4. François Vermeulen4,
  5. Mieke Boon4,
  6. Cordula Koerner-Rettberg2,5
  1. 1 Division of Respiratory Medicine, Department of Paediatrics, Inselspital, Bern University Hospital, University of Bern, Bern, Switzerland
  2. 2 Department of Paediatric Pulmonology, University Children's Hospital of Ruhr University Bochum at St. Josef-Hospital, Bochum, Germany
  3. 3 Department of Diagnostic, Interventional, and Paediatric Radiology, Inselspital, Bern University Hospital, University of Bern, Bern, Switzerland
  4. 4 Paediatric Pulmonology, Department of Paediatrics, University Hospital Gasthuisberg, Leuven, Belgium
  5. 5 Children's Hospital, Research Institute, Marien Hospital Wesel, Wesel, Germany
  1. Correspondence to Dr Cordula Koerner-Rettberg, Children's Hospital and Research Institute, Marien Hospital Wesel, 46483 Wesel, Germany; Cordula.Koerner-Rettberg{at}prohomine.de

Abstract

Background Lung clearance index (LCI) is a promising lung function outcome in individuals with primary ciliary dyskinesia (PCD). The impact of events clinically important for individuals with PCD, such as pulmonary exacerbations, on LCI is unknown.

Methods We conducted an international, multicentre, observational cohort study to assess the association of LCI and risk of pulmonary exacerbation, specific changes in LCI during pulmonary exacerbation and global variability of LCI across four visits every 4 months. Ninety individuals with PCD, aged 3–41 years, underwent nitrogen multiple-breath washout (MBW) and spirometry measurements. The association of LCI and pulmonary exacerbations was assessed by Cox proportional hazards and random-effects regression models.

Results We obtained 430 MBW and 427 spirometry measurements. In total, 379 person-years at risk contributed to the analysis. Per one unit increase (deterioration) in LCI, the risk of future pulmonary exacerbation increased by 13%: HR (95% CI), 1.13 (1.04 to 1.23). If LCI changed from a range of values considered normal to abnormal, the risk of future pulmonary exacerbations increased by 87%: 1.87 (1.08 to 3.23). During pulmonary exacerbations, LCI increased by 1.22 units (14.5%). After pulmonary exacerbations, LCI tended to decline. Estimates of variability in LCI suggested lower variation within individuals compared with variation between individuals. Findings were comparable for forced expiratory volume in 1 s.

Conclusion On a visit-to-visit basis, LCI measurement may add to the prediction of pulmonary exacerbations, the assessment of lung function decline and the potential lung function response to treatment of pulmonary exacerbations.

  • primary ciliary dyskinesia
  • lung physiology
  • paediatric lung disease
  • respiratory infection

Data availability statement

Data are available upon reasonable request. Data are available upon reasonable request.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Key messages

What is the key question?

  • Can lung clearance index (LCI) predict future pulmonary exacerbation in individuals with primary ciliary dyskinesia?

What is the bottom line?

  • LCI increase from a range considered normal to abnormal increases the risk of future pulmonary exacerbations by 87%.

Why read on?

  • During pulmonary exacerbation, LCI increases by 1.2 units but does not fully recover 4 months after pulmonary exacerbation.

Introduction

Primary ciliary dyskinesia (PCD) is a rare genetic disease with heterogeneous structural and functional impairment of cilia, including airway epithelia.1 2 Chronic infection and inflammation lead to progressive and irreversible loss of lung function comparable to cystic fibrosis (CF), but generally less severe.3 4 In early PCD lung disease, pulmonary function impairment is mainly characterised by increased ventilation inhomogeneity arising in obstructed peripheral lung compartments.5–7 Multiple-breath washout (MBW)-derived lung clearance index (LCI) is a reliable method to measure ventilation inhomogeneity.8 9 MBW is standardised and feasible even in young children with PCD.10 11 LCI is increased (abnormal) in the majority of patients with PCD despite normal forced expiratory volume in 1 s (FEV1) from spirometry.7 12–14 LCI correlates with the severity of cilia ultrastructure alteration15 and structural airway pathology such as airway wall thickening and mucus plugging.6 12

However, current interpretation of LCI in individuals with PCD is limited due to the paucity of longitudinal data.14 16 One small study described global variability of LCI across three visits.16 The impact of events clinically important for patients, such as pulmonary exacerbation (PEX), on LCI remains unknown. We hypothesised that PEX is associated with increased ventilation inhomogeneity measured by LCI in young individuals with PCD. The primary objective of this study was to assess the association of LCI and PEX within the context of long-term global variability of LCI.

Methods

Study design

We conducted an international, multicentre, observational cohort study to assess (1) changes in LCI over time to estimate the risk of PEX, (2) specific changes in LCI during PEX and (3) natural, global changes in LCI, that is, variability during follow-up as far as 4 years. Patients were consecutively recruited from PCD outpatient clinics from tertiary care centres, named centre 1 and centre 2, at the Ruhr University Hospital of Bochum (Germany) and the University Hospital Gasthuisberg, Leuven (Belgium). Study participants underwent assessments in the following order: MBW, spirometry, and sputum or cough swab, followed by clinical visits in the outpatient clinics on the same day. Data were collected prospectively during routine outpatient visits between March 2011 and August 2016 and analysed later. Source data of the study were electronic patient files. Some baseline data were reported previously.12

Study population

Eligibility criteria were a confirmed or highly likely diagnosis of PCD,2 age ≥3 years and ability to perform MBW. The diagnostic algorithm for PCD was based on the European Respiratory Society guidelines2 and on the following five diagnostic items: (1) a clinical history of symptoms suggestive of PCD in combination with (2) low nasal nitric oxide, (3) abnormal ciliary beat frequency and pattern in high-speed video-microscopy analysis, (4) a hallmark ciliary defect in transmission electron microscopy17 and/or (5) pathogenic biallelic mutations (online supplemental table 1). Immunofluorescence analysis of ciliary proteins was performed in addition in centre 1. A highly likely diagnosis of PCD was defined as a clinical history of symptoms suggestive of PCD in combination with low nasal nitric oxide and/or abnormal ciliary beat frequency and pattern in respiratory cell cultures. We here refer to study participants as individuals with PCD. We obtained written informed consent from parents or caregivers and adult patients, and assent from study participants if older than 12 years (Bochum) or 14 years (Leuven).

Supplemental material

Clinical outcomes

We defined PEX by the treating clinician’s decision to start or change antibiotic treatment because of perceived pulmonary symptoms and in the presence of one or more of the following five items: (1) increased cough, or change in sputum volume and/or colour, (2) increased shortness of breath, (3) tiredness, (4) temperature >38°C and (5) loss of appetite or failure to thrive. In this study, the definition of PEX was similar to the expert consensus definition published later.18

Lung function results from spirometry but not MBW were routinely disclosed to treating clinicians. Culture-based microbiology from sputum or oropharyngeal swabs was obtained at every visit and routinely disclosed to the treating clinicians.

Multiple-breath washout

MBW was performed using an available setup (Exhalyzer D; Eco Medics AG, Dürnten, Switzerland) according to current recommendations.8 9 Both centres used the same equipment and protocol. LCI was calculated as recommended.8 9 LCI values are dimensionless, and we refer to LCI units. Further details are provided in online supplemental file 1.

Spirometry

Spirometry was performed after MBW and according to the American Thoracic Society/European Respiratory Society standards.19 Both centres used the same equipment (Jaeger Masterscreen PFT; CareFusion, Hoechberg, Germany). We reported FEV1.

Variables

We standardised FEV1 for height, age, sex and ethnicity to z-score using the Global Lung Function Initiative reference equations.20 No such reference equations for LCI yet exist. However, for descriptive analysis of abnormal lung function in the current study population, we transformed LCI into z-score, that is, measured LCI−predicted LCI (7.04)/SD (0.45) from a reference population published recently.21 FEV1 z-score ≤−1.96 and LCI z-score ≥1.96 were defined as lower and upper limits of normal. We used raw values of LCI units for statistical analyses, as LCI z-score introduces heteroscedasticity, that is, unbalanced variance in the distribution of higher (abnormal) versus lower LCI values. To assess lung function variability specific to PEX and global lung function variability, we calculated the absolute change in FEV1 (z-score) or LCI (units) and the relative change (%) between consecutive measurements. These changes were derived from a visit-by-visit basis and therefore contained information from measurements before and during PEX.

To ease interpretation, we generated ‘worst–moderate–best’ lung function categories (tertiles) balanced for the number of observations.22 The LCI category ‘best’ corresponded to a range of LCI values considered normal or mildly elevated. The LCI categories ‘moderate’ and ‘worst’ comprised abnormal LCI values. The FEV1 categories ‘best’ and ‘moderate’ contained normal FEV1 values; the FEV1 category ‘worst’ contained abnormal FEV1 values (online supplemental table 2). We categorised Pseudomonas aeruginosa (Pseudomonas) infection burden as (1) free, (2) intermittent (≤50% of cultures positive for Pseudomonas) and (3) chronic colonisation (>50% of cultures positive for Pseudomonas).23 24

Statistics

The sample size was a convenience sample of eligible patients in each centre. For the characteristics of the study population, summary statistics (median and proportions) were used and compared between (1) individuals experiencing PEX and those who did not, and (2) between centres for sensitivity analysis by two-sample Wilcoxon rank-sum or signed-rank test and Fisher’s exact test, as appropriate. We adjusted univariable and multivariable models for centre effects and multiple observations within individuals. We additionally adjusted multivariable models for the following potential confounders selected a priori: age, time between visits, number of visits and evidence of Pseudomonas infection in respiratory specimens.25 26 We also explored the influence of gender, body mass index and type of current treatment and found no relevant change in model estimates (data not shown).27 To assess the association of lung function and risk of PEX, we fitted univariable and multivariable Cox proportional HR models, allowing for multiple observations within individuals (clustering) and multiple failure data (PEX), and computed Kaplan-Meier curves. We calculated the diagnostic performance for future PEX by assessing various LCI cut-offs. To evaluate the predictive value of LCI, we calculated concordance from the Cox models. Gönen and Heller’s concordance coefficient K is the proportion of all usable subject pairs in which the predictions and outcomes are concordant, or simply stated, the count of the number of correct predictions. This estimator does not depend on observed time directly and is robust to the degree of censoring. Values of 0 or 1.0 indicate perfect separation of subjects with different outcomes.28 For sensitivity analysis, we fitted Cox proportional HR models to baseline LCI values only and restricted analysis to 1 year of observation.

We calculated differences in lung function values on the individual and population level: On the individual level, we assessed lung function differences between visits in the group of individuals experiencing no PEX and in the group of individuals experiencing at least one PEX. On the population level, we assessed differences between all stable visits and all visits during PEX across the whole population. To assess the association of these lung function changes and PEX, we applied univariable and multivariable random-effects multilevel regression models. To assess global lung function variability during the study, we calculated three standard estimates of variability and plotted differences in lung function measurements versus the mean of pairwise measurements.29 30 The coefficient of variation (CV) was calculated as within-subject SD×100/within-subject mean. Limits of agreement were calculated as mean difference±1.96×SD of the lung function differences. The coefficient of repeatability (CR) was calculated as 1.96×SD of the lung function differences. The intraclass correlation coefficient (ICC) estimates clustering or tracking of measurements over time within individuals from a random-effects model.30 The ICC is an estimate of the variance between individuals in relation to the total variance, which is a combination of the variances between and within individuals. Missing data including loss to follow-up were assumed to occur at random and not specifically addressed. P values <0.05 were considered statistically significant, and all analyses were performed using Stata V.14.2 (Stata, College Station, Texas, USA).

Results

We enrolled 90 individuals with PCD. At baseline, the median age was 12.8 years, and 68 (75.6%) individuals were younger than 18 years (table 1). The age range was 3.6–42.2 years. Patient characteristics were similar between centres (online supplemental tables 1 and 3). In total, 71 (78.9%) had abnormal LCI and 22 (24.4%) had abnormal FEV1 at baseline. As expected, LCI and FEV1 were correlated (R2=0.40; online supplemental figure 1). On average, individuals had four visits every 4 months across 1 year. We obtained 430 MBW and 427 spirometry measurements from 436 visits. Twenty-one (5.8%) out of 365 respiratory samples were Pseudomonas-positive. Thirty-nine visits in 23 individuals were classified as PEX. The prevalence (95% CI) of PEX was 9.5% (6.6% to 13.1%). Individuals with at least one PEX were followed more closely in outpatient clinics. These individuals had on average two more visits and 7 weeks shorter time intervals between visits compared with those individuals who did not experience any PEX during the study. The 23 individuals who experienced at least one PEX had similar demographic characteristics, medication use, lower baseline FEV1 and similar baseline LCI compared with the 67 who did not experience one PEX during the study (table 1).

Table 1

Study population characteristics

Changes in lung function during PEX on the individual level

On the individual level, LCI significantly increased and FEV1 declined with the occurrence of PEX. Changes in LCI and FEV1 between (1) stable visits, (2) stable visits prior to PEX, (3) symptomatic visits during PEX and (4) follow-up visits after PEX are presented in figure 1. LCI increased with PEX by a mean (95% CI) of 1.22 units (0.49 to 1.95), p=0.001. The relative change in LCI was 14.5% (2.2% to 26.9%), p=0.021. At the follow-up visit (n=28) after PEX, LCI declined relatively to the symptomatic visit, though not statistically significantly: −9.3% (−20.5% to 1.9%), p=0.10. FEV1 declined with PEX by −0.74 (−1.17 to −0.31) z-score, p=0.001; the relative change in FEV1 was −9.4% (−13.6% to −5.2%), p<0.001. At the follow-up visit (n=28) after PEX, FEV1 increased relatively to the visit during PEX: 11.4% (3.9% to 18.9%), p=0.003.

Figure 1

Specific changes in lung function during PEX. On the individual level, we calculated relative changes (%) in LCI and FEV1 on an observation-by-observation basis between consecutive visits. Changes are displayed separately for the 67 individuals experiencing no PEX (labelled as ‘stable, no PEX’) and the 23 individuals ever experiencing PEX: relative changes between stable visits (labelled as ‘stable’), stable visits preceding PEX and visits during PEX (labelled as ‘PEX’) and visits during PEX and visits following PEX (labelled as ‘follow-up’). Mean relative changes (diamond symbols) and 95% CI (vertical bars) were estimated from random-effects regression models to account for multiple measurements in the same individual. The analysis was based on 430 multiple-breath washout and 427 spirometry measurements from 436 visits, of which the 39 visits were classified as PEX. FEV1, forced expiratory volume in 1 s; LCI, lung clearance index; PEX, pulmonary exacerbation.

Changes in lung function during PEX on the population level

On the population level, LCI significantly increased during PEX by 12.6% and FEV1 declined by −10.3% (tables 2 and 3). Adjusting for age, time between visits, number of visits and Pseudomonas infection slightly increased the effect; LCI increased during PEX by 16.0% and FEV1 declined by −11.1% (online supplemental table 4). We additionally adjusted for treatment other than antibiotics, which had no relevant effect on the estimates (data not shown).

Table 2

Lung function during stable visits and PEX

Table 3

Changes in lung function between stable visits and PEX

Table 4

Lung function impairment and risk of pulmonary exacerbation

Lung function impairment and risk of PEX

In total, 379.3 person-years at risk contributed to the Cox proportional regression analysis. For LCI, after adjustment for multiple observations, the risk increase (HR) in PEX was 13% (p=0.003) per one unit increase in LCI and 87% (p=0.025, table 4) per one LCI category increase. Lung function categories were tertiles derived from 430 MBW and 427 spirometry measurements from 436 visits, of which 39 visits were classified as PEX (online supplemental table 2). After additional adjustment for age, number of visits and Pseudomonas infection, the risk increase in PEX was 16% (p=0.001) per one unit increase in LCI and 137% (p=0.004) per one LCI category increase (figure 2 and table 4). Sensitivity analysis restricted to baseline LCI as a predictor of PEX in 90 individuals with PCD revealed that per one unit increase in LCI, HR (95% CI) was 1.16 (1.03 to 1.30), p=0.014, for future PEX (n=23) across the whole study duration and 1.18 (1.03 to 1.35), p=0.019, for future PEX (n=15) across 1 year.

Figure 2

Lung function and risk of PEX. Kaplan-Meier curve: lines reflect the proportion of individuals free of PEX within each lung function category (low–moderate–high) from a Cox proportional hazards regression model adjusted for age, number of visits, Pseudomonas-positive respiratory samples and multiple observations within individuals. Lung function categories were tertiles derived from 430 multiple-breath washout and 427 spirometry measurements from 436 visits, of which 39 visits were classified as PEX (online supplemental table 2). The LCI category ‘high’ corresponds to the FEV1 category ‘low’, meaning abnormal lung function. The LCI category ‘low’ corresponds to the FEV1 category ‘high’, meaning normal or only mildly impaired lung function. Analysis time=days in study. FEV1, forced expiratory volume in 1 s; LCI, lung clearance index; PEX, pulmonary exacerbation.

For FEV1, after adjustment for multiple observations, the risk increase in PEX was 37% (p<0.001) per one z-score decline in FEV1 and 63% (p<0.001, table 4) per one FEV1 category decline. After additional adjustment for age, number of visits and Pseudomonas infection, the risk increase in PEX was 41% (p<0.001) per one z-score decline in FEV1 and 69% (p<0.001) per one FEV1 category decline (figure 2 and table 4). We additionally adjusted for treatment other than antibiotics, which had no relevant effect on the estimates (data not shown). While negative predictive values to rule out PEX were good, positive predictive values to diagnose PEX were low (online supplemental table 5). Based on LCI measurement, survival times free of PEX for pairs of individuals were ordered correctly 61% of the time. This estimate improved after adjustments (table 4).

Table 5

Overall variability in lung function during the study

Overall variability in lung function

The high ICC values suggested that variation between individuals exceeded variation within individuals, indicating tracking of LCI and FEV1 within individuals (table 5). The CV of FEV1 and LCI was lower in individuals experiencing no PEX compared with those who experienced at least one PEX. The difference in the CV between the groups was statistically significant for FEV1 (table 5).

The magnitude of relative changes in lung function was not associated with the magnitude of lung function values, but variance, that is, the spread of lung function changes, increased with worse lung function (figure 3). There was no trend in LCI or FEV1 over time, and the magnitude of relative change in LCI was not systematically different from change in FEV1 (online supplemental file 1).

Figure 3

Lung function variability and association with lung disease severity. On the population level, the percentage change in lung function, that is, LCI or FEV1, between two visits was not associated with the magnitude of LCI or FEV1 values. The middle dashed lines are the mean difference; the upper and lower dashed lines are upper and lower limits of agreement.29 Closed black circles: the percentage change in lung function during PEX; open grey circles: the percentage change in lung function during stable visits. The analysis was based on 430 multiple-breath washout and 427 spirometry measurements from 436 visits, of which 39 visits were classified as PEX. FEV1, forced expiratory volume in 1 s; LCI, lung clearance index; PEX, pulmonary exacerbation.

Discussion

Summary

This is the first study to show the association between LCI and PEX in individuals with PCD. Individuals with abnormal LCI had an increased risk of future PEX, with a risk increment of 87% if LCI values changed from a normal range to abnormal. In contrast to LCI, FEV1 values were often normal. During PEX, LCI and FEV1 deteriorated by 14.5% and 9.4% in individuals with PCD. While FEV1 improved 4 months after PEX, LCI did not systematically recover. Within individuals, we found evidence of lung function tracking. Across 1 year on average, variability within individuals in LCI and FEV1 was low compared with relatively large variability between individuals.

Lung function and PEX

We comprehensively assessed the association of lung function and PEX in mostly young individuals with relatively mild lung function impairment. LCI was sensitive to identify individuals with lung function abnormalities. The risk estimates (HRs) for future PEX per one category increase were larger for LCI compared with FEV1. However, the predictive accuracy based on the postestimation concordance coefficient of LCI was moderate and comparable with the coefficient of FEV1. The comparison of risk estimates per one original unit increase in LCI versus FEV1 z-score (SD) is influenced by the magnitude of units. Our findings are consistent with the literature suggesting that LCI is more sensitive than FEV1 in detecting lung function impairment in suppurative lung disease.7 22 31–33 LCI correlates to structural airway pathology detected in CT scans or MRI, which is not the case for FEV1.6 12 34

LCI increase may provide an early warning of PEX in individuals with PCD. Our data demonstrate that by an LCI increase from a normal range to abnormal, the crude risk increment for future PEX was ⁓90%. The time to the first PEX significantly decreased with increasing LCI values throughout the study. Our study also suggests that elevated LCI and infection with Pseudomonas are multiplicative exposure effects increasing the risk of PEX in individuals with PCD. This observation coincides with previous findings in individuals with CF.22 35 However, Pseudomonas abundance in respiratory specimens was low in our cohort and was possibly influenced by sampling techniques and missing data. Our study adds further evidence to few existing data, suggesting that increased LCI may predict PEX.22 In the study of Vermeulen et al 22 in 63 school-aged children with CF, an LCI increase of ⁓0.5 LCI units was associated with an increased PEX rate ratio of 12%. The latter estimate can be interpreted similarly to the HR reported in our study.

Our data suggest that LCI may add as objective measure to identify PEX and monitor treatment response to a course of antibiotics. From a clinical perspective, it is important to highlight that LCI may not recover after an exacerbation. LCI increased during PEX and declined 4 months later, though not systematically. The latter findings require careful interpretation as prevalence of the event of interest in our study (PEX) and data from follow-up after PEX were limited. Given the magnitude of changes discovered post hoc, future work should validate current findings in external cohorts. Antibiotic treatment may variably affect central and peripheral airway function. Indeed, maintenance of azithromycin for 6 months did not improve LCI in individuals with PCD, whereas the number of PEX declined.36 Several observational studies in school-aged children and adults with CF treated with antibiotics during PEX suggest that the change in LCI is heterogeneous, while LCI may systematically improve in preschool children.37 Despite antibiotic treatment, variable hyperinflation, shifting mucus plugging and reopening of poorly ventilated lung compartments may impair ventilation efficiency measured by the LCI.38 39 We and others have previously demonstrated that PCD lung disease substantially differs between individuals even while clinically stable.6 26 A multidomain approach covering the dynamics of lung imaging and function testing over time appears most suitable for comprehensive surveillance of PCD lung disease.

Overall variability in lung function

We found evidence of lung function tracking in PCD across 1 year of average observation time. FEV1 and LCI values strongly clustered within individuals. High ICC and limits of agreement values suggest relatively high consistency within individuals compared with the large variance between individuals. We did not observe trends in LCI or FEV1 over time and no association between the magnitude in lung function changes and lung function values. Yet, there was an inverse relationship between the magnitude of variance (spread of lung function changes) and lower lung function values. Our findings are in line with a study in 12 individuals with PCD14 and a larger study in children with CF.40 In individuals with PCD, FEV1 decline of −0.5% to −0.6% predicted per year across more than 1 year was reported.3 25 41 Heterogeneity in FEV1 trends over time has been described previously and relates to genotypes and diverse cilia defects in PCD.15 26

The ‘minimal clinically important difference’ (MCID) in LCI has not been established yet. However, individual tracking in the presence of large variance between individuals questions the use of distribution-based approaches for calculating the MCID in LCI. The CR assumes that the variance is proportional to the magnitude of measurement, which appeared to be not the case in our study (figure 3). Future studies are warranted to determine whether the individual patient’s variability by several longitudinal LCI measurements can be used to calculate what would be a significant change for that individual. Prospective studies on patient-reported outcomes and LCI seem more suitable for the estimation of cut-offs for the MCID.

Strengths and limitations

We conducted an international, multicentre, pragmatic study in a relatively large cohort of a rare chronic lung disease. Our findings appear generalisable: The study participants’ age range was broad, FEV1 was often normal and follow-up times varied. We adjusted regression models for follow-up times, multiple observations and multiple failure times, which are required when the same event (PEX) can occur in the same individual multiple times. We did not examine time to first event, as this may have dropped relevant information. We assume that the predictive power of LCI declined as time elapsed from the test. A change in LCI may be more predictive of PEX if LCI is measured in a shorter time frame. More stringent follow-up times may improve repeatability. Our study was not designed to assess as to how frequently LCI should be measured for optimum predictive power. The study population was not large enough to split the population into a training set and a test set. Therefore, we could not validate the predictive accuracy of LCI or calculate CIs for the differences between concordance coefficients.

We do not assume that the ability of LCI to predict PEX differed between individuals with a confirmed diagnosis versus highly likely diagnosis of PCD. Recommended algorithms for making the diagnosis currently vary,1 2 and we cannot fully exclude misclassification. The definition of PEX in our study was comparable with a recent expert consensus.18 The association between FEV1 and PEX in our study should be interpreted cautiously. Clinicians were not blinded to FEV1. Clinical judgement of ‘severity’ of respiratory symptoms necessitating treatment may have been influenced by FEV1. Additional modification of clinical care other than antibiotic treatment may have contributed to observed changes in LCI and FEV1 such that treatment response should not exclusively be attributed to antibiotics. Controlled interventional trials should evaluate whether LCI is responsive to antibiotic treatment during PEX.

The lung function variability estimates reported in our study were comparable with previous findings in CF. The young age of many individuals in our study may have contributed to greater between-test variability of LCI compared with the between-test variability of LCI in older individuals with PCD.14 16 The lack of a gold standard to define lung function variability limited our ability to disentangle physiological fluctuation from subclinical deterioration. Some episodes with or without PEX may have been missed, and residual confounding may have increased estimates of variability. Studies including functional imaging may help to better characterise variability in LCI observed in apparently stable individuals with PCD.6

Clinical implications for monitoring in PCD

Our study suggests that, on a visit-to-visit basis, LCI measurements can provide additional information to predict the risk of future PEX. From a practical and physiological point of view, LCI measurement may be more helpful in individuals with normal or modestly elevated LCI compared with individuals with consistently very high LCI. LCI below the upper limit of normal (7.9) had a good negative predictive value (95%) for future PEX. Yet, positive predictive values were low in our study. Further studies are warranted to compare LCI-driven treatment to treatment based on standard monitoring and to establish the number needed for LCI-driven treatment to prevent one exacerbation.

Conclusion

LCI increase is associated with greater risk of future PEX. LCI increases during PEX, and LCI decrease indicates potential recovery after PEX. LCI tracks in individuals with PCD. Variability in LCI between individuals underlines the heterogeneity of PCD lung disease and questions the use of distribution-based estimates to estimate the MCID of LCI. This study supports the concept that LCI could serve as an adjunct to clinical and spirometric assessment of individuals with PCD.

Data availability statement

Data are available upon reasonable request. Data are available upon reasonable request.

Ethics statements

Ethics approval

The Ethics Committee of the Ruhr University Bochum (No. 5103) and the Ethics Committee of Leuven (ML5901/s51800) approved the study.

Acknowledgments

We thank Ben Spycher, Johanna Kurz and Philipp Latzin, University of Bern, for intellectual input. We thank Linda Boulanger, Nathalie Feyaerts and Marianne Schulte, who performed LCI measurements in Leuven. We thank all patients for their participation in the study.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • FS and AS contributed equally.

  • MB and CK-R contributed equally.

  • Contributors All authors drafted and approved the final version of the manuscript. FS, AS, MB and CKR were the lead authors with overall responsibility for the manuscript.

  • Funding MB and CKR participate in the Better Evidence to Advance Therapeutic Options for PCD (BEAT-PCD) network (COST action BM 1407). MB is supported by an unrestricted postdoctoral research grant from KOOR, University Hospital Leuven, Belgium. AS is supported by an unrestricted research grant from FoRUM, Ruhr-University Bochum.

  • Competing interests Relevant financial activities for this work: AS reports grant from Ruhr-University Bochum, payment to institution. MB reports postdoctoral grant KOOR from UZLeuven, payment to institution. Relevant financial activities outside the work: FS has grants/grants pending from LungenLiga Bern, CFCH, and payment for lectures from Vertex, Novartis. MB is a member of the European Reference Network for Rare Respiratory Diseases (ERN-LUNG) Project ID No 739546. She has grants/grants pending from Horizon 2020, MyCyFAPP.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles

  • Airwaves
    The Triumvirate