1 INTRODUCTION

Asthma is a common chronic airway disease defined by intermittent obstruction of the airways due to chronic inflammation.1 Clinical evaluation of patients suspected of asthma is based on airway symptoms, spirometry, and measures of airway hyperresponsiveness that are not always readily available in primary care nor feasible in preschool children. There is a pressing need for objective and readily available markers of airway inflammation in patients with asthma.

The etiology of asthma is complex and influenced by many factors, including genetic propensity, exposure to ubiquitous allergens, irritants, respiratory tract infections during early childhood, and altered airway microbiome.2-5 Surfactant Protein D (SP-D) is a pattern recognition molecule that belongs to the collectin family expressed in multiple human organ systems, including alveolar type II cells and Clara cells.6 SP-D plays an important role in the innate immune system by binding bacteria, viruses, fungi, and parasites for clearance via opsonization in phagocytes, as well as aiding in the removal of allergens.7-9 Several experimental studies have indicated that SP-D levels in broncho-alveolar lavage fluid (BAL) decrease and serum levels increase in patients suffering from asthma possibly due to both induced SP-D synthesis and decreased air–blood barrier integrity.10-13 Recent clinical studies have revealed a correlation of serum SP-D (sSP-D) with asthma severity and a decreased BAL/serum SP-D ratio supporting the hypothesis of leakage of degraded SP-D to the circulation,5, 14, 15 whereas other studies have failed to demonstrate an association.16, 17

The surfactant protein D-encoding gene (SFTPD) is located at the genomic position 10q22.2-23.1.18 Three structural single nucleotide polymorphisms (SNP), rs721917, rs2243639, and rs3088308 have previously been linked with the circulating levels of SP-D and pulmonary disease.19 In particular SNP in rs721917 has been shown to be associated with several chronic and acute respiratory diseases, although the data on the association to asthma have been conflicting.20-22

We intended to elucidate the clinical significance of SP-D in childhood asthma in a real-life setting. The aims of this study were to investigate if the serum level of SP-D and SP-D gene variants were associated with asthma in children, adolescents, and young adults, and whether serum SP-D measured at the time of asthma diagnosis was associated with persistent asthma. We hypothesized that serum SP-D was increased in subjects with persistent asthma and associated with asthma severity.

2 METHODS 2.1 Study design

Prospective follow-up including 449 adolescents and young adults (11–27 years) from a closed cohort of children diagnosed with asthma during 2003–2005 (0–16 years) at all four pediatric outpatient clinics in the Region of Southern Denmark.23 At follow-up during 2016–2017, an additional 314 healthy controls with no medical history of asthma or use of asthma medication, and within the same age range were recruited. Controls were recruited through a notice on the digital learning systems from the same schools as the origin of cases.

2.2 Baseline data collection at time of diagnosis

Baseline data were obtained from a structured database and biobank established prospectively during the baseline examination as previously described.23 The data were collected prospectively and consecutively by the treating staff and included a structured questionnaire-based interview (regarding predisposition, medical history, medication use, and environmental exposures), clinical examination by a pediatrician, blood sampling, skin prick test, spirometry with bronchodilator test and exercise test where appropriate.

2.3 Follow-up data collection

The follow-up examination was performed by the research team which included medical doctors and nurses with experience within the field of pediatric asthma. For each participant, a structured questionnaire-based interview based on the same items as the baseline questionnaire was completed. Subsequently, the subjects participated in a clinical examination, blood sampling, measurement of fractional exhaled nitric oxide (FeNO), spirometry with bronchodilator test, and mannitol provocation test. The subjects also completed questionnaires regarding asthma control (ACT, Asthma Control Test) and quality of life (Asthma Quality of Life Questionnaire).24, 25 Controls completed the same examination program as the cases except for mannitol provocation test and asthma questionnaires.

2.4 Serum surfactant protein D

Serum levels of SP-D were measured by immunoassay as previously described.26 Serum SP-D was analyzed on samples obtained at baseline as well as samples obtained at follow-up.

2.5 SFTPD genotyping

DNA purification and genotyping were carried out by PentaBase Aps using Maxwell® 16 Blood DNA Purification Kit (Promega, AS1010) and nuclease resistant probes, EasyBeacons™, developed by PentaBase,27 in a 2-step Real-Time PCR, followed by melt analysis. SP-D genotyping was performed for three single-nucleotide variants conferring amino acid substitutions in the mature protein (rs721917, rs2243639, and rs3088308). Information on the SFTPD specific primers and probes (Table E1) and a detailed description of the method are provided in the online repository.

2.6 Definition of asthma

Current asthma at follow-up was defined by recurrence or persistence of at least two of three symptoms: cough, wheeze, and shortness of breath (not triggered only by infection) within the last 12 months and at least one of the following: positive bronchodilator reversibility test (salbutamol/terbutaline) and/or positive mannitol test. Subjects were defined as having asthma if they had a history of asthma symptoms and daily use of inhaled corticosteroids (ICS), fixed combination of ICS, and long-acting beta-2-adrenoreceptor agonists (LABA) or classical exercise-induced asthma symptoms and clinical effect of inhaled beta-2-adrenoreceptor agonists. Allergic asthma was defined as having current asthma and concurrent allergic sensitization to ≥1 inhalant allergens. Severe asthma was defined as poor control despite maximal maintenance therapy (GINA step 4).28

2.7 Allergic sensitization

Specific immunoglobulin E (sIgE) at follow-up was measured quantitatively using Single ImmunoCAP™ after an initial qualitative screening with ImmunoCap™ Phadiatop™ (Thermo Fisher Diagnostics Aps). The samples were tested for the presence of sIgE for 10 inhalant allergens (birch, timothy grass, mugwort, horse, dog, cat, Dermatophagoides pteronyssinus, Dermatophagoides farinae, Alternaria, and Cladossporium). Sensitization was defined as having sIgE ≥ 0.35 kU/l.

2.8 Clinical measures

Lung function was measured by spirometry (SpiroUSB, Carefusion Ltd), bronchodilator test was performed with salbutamol (Buventol Easyhaler™), and mannitol provocation testing was performed using the commercially available test kit (Osmohale™, Pharmaxis Ltd.). All measurements were performed according to generally accepted methods and criteria.29-31 Parameters obtained for statistical analysis were forced expiratory flow rate at one second (FEV1), forced vital capacity (FVC), FEV1/FVC, and forced expiratory flow 25%–75% (FEF25–75). We used the multi-ethnic reference intervals for spirometry developed by the European Respiratory Society Task Force.32

Classification of the response to mannitol was performed in subjects with a positive test according to the cumulative dose of mannitol required to induce a 15% reduction in FEV1.33 Fractional exhaled nitric oxide (FeNO) was measured before spirometry using NIOX VERO™.

2.9 Statistical analysis

The distribution of data was evaluated visually using histograms and normal quantile-quantile (QQ) plots. Because of a non-normal distribution of observations of sSP-D, approximation to normal distribution was achieved by logarithmic transformation (ln, the natural logarithm) before further statistical analysis. All reported values of sSP-D and 95% confidence intervals were ln-transformed.

The subjects were classified according to the clinical assessment at follow-up: current asthma, no current asthma, and controls and subgroups of allergic phenotype and response to mannitol.

A regression-based comparison using linear regression was used to test differences between ln sSP-D as outcome and classification as exposure with an initial F-test to test the overall difference between groups.

The relationships between ln(sSP-D) and continuous clinical measures were assessed by estimating kernel-weighted local polynomial regression. The association between sSP-D as outcome and relevant exposure variables was estimated by linear regression models and presented by back-transformed coefficients (eln(β)) and confidence intervals (95%) with associated p-values. The coefficients are interpreted as the fold increase per unit change in a continuous variable or the fold difference between levels of a binary explanatory variable.34 The longitudinal association of baseline sSP-D with current asthma at follow-up was estimated using logistic regression models.

All models were estimated using both a univariate model and a multivariate model to accommodate the uncertainty of potential confounders. In the multivariate models, we included age, sex, body mass index (BMI) (numeric), smoking status, and ethnicity, all factors previously reported to affect the constitutional levels of SP-D.35, 36

The multivariate models assessing the association between sSP-D and clinical measures of asthma were estimated on all study subjects and included asthma classification at follow-up as an interaction term. Furthermore, all models were re-estimated including smoking status, gender, and genotype in separate models as potential effect modifiers. Smoking status at follow-up was defined as current active smoking whereas smoking status at baseline was defined as current parental smoking.

Analysis of genotype association was performed by initial testing of Hardy–Weinberg equilibrium of each SNP among the groups. Logistic regression models were applied to assess the association of asthma classification with genotypes.

To evaluate the importance of missing values on sSP-D at follow-up, subjects with a missing sSP-D were assigned both the highest and the lowest measured value in the study population, and the main analysis was re-estimated to assess whether conclusions changed.

All statistical analysis was performed using Stata 15 (StataCorp LCC) with the level of significance set to 0.05. Due to the risk of type one error caused by multiple testing findings should be interpreted as exploratory.

2.10 Ethics

The study was conducted in accordance with the Declaration of Helsinki approved by the Regional Scientific Ethical Committee of Southern Denmark (S-20120093) and the Danish Data Protection Agency (95-50819). Before enrollment, informed consent was obtained from each participant and from the parents of participants below 18 years of age.

3 RESULTS 3.1 Study population

The asthma diagnosis was confirmed in a total of 1014 children at baseline during 2003–2005 of whom 449/1014 subjects participated in the follow-up examination during 2016–2017 (follow-up rate, 0.443). 196 classified as having current asthma and 253 as having no current asthma. A comparison of the included subjects with the subjects lost to follow-up is available in the online repository (Table E2). At follow-up, 314 healthy controls were recruited. The basic and clinical characteristics of the study population stratified by classification at follow-up are presented in Table 1. The groups had significant differences according to age, sex, BMI, ethnicity, and current smoking.

Table 1. Characteristics of the study population according to classification at follow-up: controls, no current asthma, and current asthma Controls No current asthma Current asthma (n = 314) % (n = 253) % (n = 196) % (cases/total) (cases/total) (cases/total) p-value Basic characteristics Gender - female 61.8 (194/314) 39.9 (101/253) 46.4 (91/196) <.001 Age - mean (SD) 18.4 (4.5) 17.6 (4.4) 18.7 (4.1) .030 BMI - mean (SD) 22.2 (3.8) 22.9 (4.7) 23.2 (5.2) .031 Ethnicity - Caucasian 98.1 (308/314) 96.4 (244/253) 95.9 (188/196) .312 Parental asthmaa 10.5 (33/314) 29.6 (75/253) 33.2 (65/196) <.001 Siblings - median (IQR) 1 (1–2) 1 (1–2) 1 (1–2) .480 Active smokers 5.4 (17/314) 8.7 (22/253) 10.7 (21/196) .080 Symptom score (ACT)b Well-controlled (score > 19) – 94.0 (236/251) 77.4 (151/195) Poor controlled (score ≤ 19) – 6.0 (15/251) 22.6 (44/195) Patient-reported symptoms Hayfever 2.9 (9/314) 44.7 (113/253) 74.5 (146/196) <.001 Eczema 2.2 (7/314) 11.1 (28/253) 27.7 (54/195) <.001 Food allergy 1.0 (3/314) 5.1 (13/253) 10.2 (20/196) <.001 Urticaria 0.6 (2/314) 4.3 (11/253) 13.8 (27/196) <.001 Current medicationc No treatment – 87.0 (220/253) 11.7 (23/196) SABA only – 13.0 (33/253) 19.9 (39/196) ICS low dose – – 29.6 (58/196) ICS moderate or ICS low dose + LABA – – 29.6 (58/196) ICS high or ICS moderate dose + LABA – – 9.2 (18/196) Add on (thiotropium/biologicals) – – – Allergic sensitization Inhalant allergens 18.8 (58/309) 54.5 (134/246) 78.1 (153/196) <.001 Food allergens 2.6 (8/309) 10.6 (26/246) 16.3 (32/196) <.001 Abbreviations: ACT, Asthma Control Test; ICS, inhalant corticosteroids; LTRA, leukotriene receptor antagonists; SD, standard deviation; SABA, short-acting beta2 agonists. aMedical history of asthma in ≥1 parent. bNot filled in by controls. cSelf-reported symptoms. 3.2 Association of sSP-D with potential confounders

Significant relationships were found between sSP-D and age (exp , 0.97 [95% CI, 0.96–0.98]), female sex (exp , 0.76 [95% CI, 0.70–0.82]), BMI (exp , 0.97 [95% CI, 0.96–0.98]), and ethnicity (exp , 1.58 [95% CI, 1.26–1.97]) but no significant association to current smoking (exp , 0.95 [95% CI, 0.83–1.10]).

3.3 Association of sSP-D with asthma at follow-up

All observations stratified by classification at follow-up are presented in Figure 1A. No difference was found in mean levels of sSP-D among subjects with current asthma 6.72 (95% CI, 6.64–6.80), no current asthma 6.79 (95% CI, 6.71–6.86), and controls 6.70 (95% CI, 6.65–6.76). Based on the findings regarding potential confounders, a multivariate model was estimated which did not change the conclusion.

Boxplot presenting log-transformed serum levels of surfactant protein D (SP-D) (unadjusted) measured at follow-up stratified by (A) classification at follow-up, (B) allergic phenotype, and (C) asthma severity

When comparing mean levels of sSP-D in subgroups, no differences were found according to allergic phenotype, allergic asthma 6.75 (95% CI, 6.66–6.84), nonallergic asthma 6.61 (95% CI, 6.42–6.80) (Figure 1B), and severity, mild asthma 6.76 (95% CI, 6.62–6.90), moderate asthma 6.70 (95% CI, 6.60–6.81) and severe asthma 6.68 (95% CI, 6.35–7.02) (Figure 1C).

No associations were found between sSP-D and FEV1, FVC, FEF 25%–75% and response to SABA (Figure 2), nor to self-reported atopic symptoms other than asthma, ACT-score, FeNO, and allergic sensitization (data not shown). We found no association of sSP-D with the outcome of mannitol test nor to the severity of the response to mannitol (Figure 3). There was a weak association of sSP-D with FVC in the unadjusted model that diminished when age was included in the model.

Association of serum SP-D with parameters obtained from spirometry. Local kernel-weighted mean of ln(SP-D) on FEV1, FVC, FEF 25%–75% and response to SABA (% increase in FEV1). Shaded areas indicate 95% confidence bands. The linear slope coefficients are presented as back-transformed coefficients (eln(β)) with 95% confidence intervals from a univariate and multivariate model. The multivariate model includes adjustment for age, sex, BMI, active smoking, ethnicity, and asthma classification at follow-up. FEF 25%–75%, forced expiratory flow 25%–75%; FEV1, forced expiratory flow rate at 1 s; FVC, forced vital capacity; SP-D, surfactant protein D

Boxplot presenting log-transformed serum levels of surfactant protein D (SP-D) (unadjusted) measured at follow-up and stratified by response to mannitol. Mannitol test was only performed in subjects previously diagnosed with asthma: current asthma and no current asthma

No significant effect modification by gender, smoking status, genotype nor asthma classification at follow-up was found in any model.

3.4 Association of baseline sSP-D with current asthma at follow-up

To investigate the longitudinal relationship of sSP-D obtained at baseline with asthma status at follow-up multivariate logistic regression models were estimated with asthma classification at follow-up as response variable: current asthma vs no current asthma. The models included age, sex, BMI, ethnicity, and active parental smoking as covariates. No association of baseline sSP-D was found with current asthma at follow-up nor to FEV1, FVC, FEF 25%–75%, and response to mannitol.

3.5 Association of baseline sSP-D with clinical measures obtained at baseline

The association of baseline sSP-D with clinical measures obtained at baseline was evaluated and a weak association of baseline sSP-D with FEV1 at baseline (age 5–16 years) was found in the adjusted model (exp 0.99 [95% CI, 0.99–1.00], p = .011). We found no association between baseline sSP-D and FVC, FMEF, peak flow, atopic comorbidity, allergic sensitization, or blood eosinophils obtained at baseline.

3.6 Association STFPD gene variants with sSP-D, asthma, and lung function

The frequencies of SNP's in the total number of genotyped subjects (n = 749) are presented in Table E3 in the supporting information. All SNP's were in Hardy–Weinberg equilibrium A univariate model and a multivariate model including all three SNP's were estimated to evaluate the association between genotypes and asthma. No association of STFPD variants, rs721917, rs3088308, and rs2243639 with current asthma (Table 2) nor to allergic asthma, FEV1, FVC, and FEF 25%–75% was found (data not shown). We found a strong association of rs721917 genotype with constitutional levels of sSP-D both at baseline and follow-up (Table 3) and a high degree of tracking comparing the baseline levels of sSP-D with sSP-D at follow-up (Figure E4).

Table 2. Association of SFTPD polymorphisms and asthma at follow-up (2016–2017) “Current asthma” vs. “Controls” “Current asthma” vs. “No current asthma” (n = 502) (n = 442) SNP Genotype OR (95% CI) p-value OR (95% CI) p-value rs2243639 CC Ref. – Ref. – CT 1.06 (0.71–1.57) .787 0.90 (0.60–1.37) .633 TT 0.77 (0.45–1.33) .343 0.83 (0.46–1.49) .537 rs3088308 AA Ref. – Ref. – AT 0.83 (0.48–1.45) .518 0.93 (0.52–1.67) .801 TT 6.21 (0.69–56.05) .104 5.06 (0.56–45.68) .149 rs721917 AA Ref. – Ref. – AG 1.33 (0.89–2.00) .164 1.11 (0.73–1.69) .627 GG 1.11 (0.66–1.87) .687 1.23 (0.71–2.15) .459 Note: Unadjusted logistic regression models are presented with odds ratios (ORs), 95% confidence intervals (CIs), and related p-values. Table 3. Associations of sSPD with SFTPD variants rs2243639, rs3088308, and rs721917 Univariate model Multivariate model Variant Genotype expβ (95% CI) p-value expβ (95% CI) p-value Follow-up rs2243639 CC Ref. – Ref. – CT 1.04 (0.95–1.13) .404 0.86 (0.77–0.95) .003 TT 1.13 (1.01–1.28) .035 0.76 (0.65–0.88) <.001 rs3088308 AA Ref. – Ref. – AT 0.75 (0.66–0.84) <.001 0.83 (0.74–0.94) .003 TT 0.65 (0.42–1.00) .048 0.77 (0.50–1.18) .222 rs721917 AA Ref. – Ref. – AG 0.79 (0.73–0.86) <.001 0.74 (0.67–0.82) <.001 GG 0.68 (0.61–0.76) <.001 0.60 (0.52–0.70) <.001 Baseline rs2243639 CC Ref. – Ref. – CT 1.01 (0.90–1.13) .848 0.89 (0.78–1.02) .091 TT 1.08 (0.93–1.27)