Mannose-binding lectin defi ciency and disease severity in non-cystic fi brosis bronchiectasis : a prospective study

BACKGROUND
Mannose-binding lectin (MBL) is a key component of innate immunity. MBL deficiency is common (10-30% of the general population depending on the definition used) and has been associated with disease progression in cystic fibrosis. We aimed to assess the effect of MBL deficiency on disease severity in non-cystic fibrosis bronchiectasis.


METHODS
We recruited patients with non-cystic fibrosis bronchiectasis and age-matched and sex-matched controls at a specialist bronchiectasis clinic in Edinburgh, UK. We assessed MBL function with genotyping (low-expressing genotype [deficiency] defined as homozygosity for exon 1 mutations [YO/YO] or compound heterozygosity [XA/YO]; YA/YO and XA/XA genotypes were defined as intermediate-expressing with all other genotypes defined as high-expressing) and serum measurements (deficiency defined with two parameters: <500 ng/mL or <200 ng/mL). We assessed rates of exacerbation, chronic bacterial colonisation, and lung function during 4 years of follow-up.


FINDINGS
We included 470 patients with bronchiectasis and 414 controls. MBL genotype frequencies and MBL serum concentrations did not differ between patients and controls. 55 (12%) patients with bronchiectasis had low-expressing genotypes. These patients had a mean of 2·7 exacerbations per year (SD 1·8), compared with 1·9 per year (1·2) for 135 patients with intermediate-expressing genotypes and 1·9 per year (1·3) for 280 patients with high-expressing genotypes (p<0·0001). Chronic colonisation with bacteria was most frequent in patients with low-expressing genotypes (47 [85%] patients vs 82 [61%] patients with intermediate-expressing genotypes and 183 [65%] patients with high-expressing genotypes; p=0·0041); especially P aeruginosa colonisation (19 [35%] patients vs 13 [10%] patients and 36 [13%] patients; p<0·0001). Patients with low-expressing genotypes were more likely to be admitted to hospital for severe exacerbations during follow-up (27 [49%] patients vs 42 [31%] patients and 87 [31%] patients; p=0·032). Patients with low-expressing genotypes also had increased scores for radiological severity and worse quality of life compared with the other two groups. MBL serum deficiency (<200 ng/mL) was associated with increased exacerbations, hospital admissions, and radiological severity. When <500 ng/mL was used as the definition of deficiency, the associations with exacerbation frequency and radiological severity were no longer significant.


INTERPRETATION
MBL might be an important modifier of disease severity in non-CF bronchiectasis.


FUNDING
UK Medical Research Council, UK Chief Scientists Office.


Introduction
Bronchiectasis is a chronic infl ammatory lung disease characterised by permanent dilatation of the bronchi. 1 Patients with the disease have daily cough, sputum production, and recurrent respiratory infections. 1 Central to the pathogenesis of bronchiectasis is a cycle of failed bacterial clearance, airway infl ammation, and airway structural damage. 2 Patients become chronically colonised with pathogens owing to a failure of host immune defences. 3 The cause of adult bronchiectasis for most patients is unknown. 4 Mannose-binding lectin (MBL) is a soluble patternrecognition molecule of the innate immune system. 5 MBL binds to glycoconjugates containing mannose, fucose or N-acetylglucosamine on the surface of a wide range of clinically important bacteria, viruses and fungi, activating the lectin pathway of com plement. 6,7 Through complement activation, MBL promotes phagocytosis and leucocyte chemotaxis and activation. 8,9 MBL might also have a role in the resolution of infl ammation in the lung through the clearance of apoptotic cells and in suppression of proinfl ammatory cytokine secretion. 10,11 MBL defi ciency is one of the most common immune defects, aff ecting 10-30% of people depending on the defi nition of defi ciency used. 12 MBL defi ciency has been associated with recurrent respiratory infections, and is associated with the presence of bronchiectasis in patients with common variable immunodefi ciency. 13 Although MBL defi ciency is not the cause of cystic fi brosis, several studies have shown that such a status modifi es the course of disease in cystic fi brosis, leading to a more rapid decline in forced expiratory volume in 1 s (FEV 1 ), early acquisition of Pseudomonas aeruginosa, infection with Burkholderia cepacia, and death. [14][15][16] In this study, we aimed to assess whether mannose-binding lectin defi ciency was also associated with disease severity and clinical outcomes in adults with non-cystic fi brosis bronchiectasis.

Study design and patients
We recruited patients with non-cystic fi brosis bronchiectasis from a regional specialist bronchiectasis clinic at the Royal Infi rmary of Edinburgh (Edinburgh, UK). Bronchiectasis was defi ned as presence of bronchial dilatation on high-resolution CT scanning with a compatible clinical history of daily cough with sputum production and recurrent respiratory infections.
We excluded patients with primary immunodefi ciency (eg, common variable immunodefi ciency), active malignant disease, cystic fi brosis, active allergic bronchopulmonary aspergillosis (as defi ned elsewhere; 4 current treatment with corticosteroids or itraconazole was also an exclusion), interstitial lung disease, active mycobacterial disease, current smoking (within 2 years), HIV infection, or current chronic liver disease.
We recruited age-matched and sex-matched healthy volunteers from the spouses and partners of patients attending the outpatient clinics at the Royal Infi rmary of Edinburgh.
The study was approved by the local research ethics committee and all participants provided written informed consent.

Procedures
At the time of inclusion, all patients were clinically stable with no antibiotic use in the preceding 4 weeks. We followed up patients for 4 years with review every 6 months. Patients provided blood samples for genomic DNA extractions, serum for measurement of MBL, and spontaneous sputum samples for bacteriological analysis and markers of airway infl ammation.
We assessed severity of bronchiectasis by scoring highresolution CT scans with a modifi ed Reiff score. 4,17 At every visit, all patients underwent clinical assessments including spirometry FEV 1 , forced vital capacity with the highest of three technically satisfactory measurements re corded), 18 and chest radio graphy. Patients completed the St George's respiratory questionnaire (minimum clin ically important diff erence 4 units) 19 and the Leicester cough questionnaire (minimum clinically important diff erence 1·3 units) 20 as measures of quality of life and cough severity.
We recorded unscheduled hospital admissions in the previous year for severe exacerbations from patient histories and verifi ed reports by use of an administrative database that recorded all regional hospital admissions. We quantifi ed outpatient antibiotic use for exacerbations of bronchiectasis from patient histories and verifi ed fi ndings against primary-care prescription records. Such clinical databases are widely used in clinical research in the UK. 21 We classifi ed patients as chronically colonised if they isolated in sputum culture a potentially pathogenic microorganism on two occasions at least 3 months apart in 1 year while clinically stable.
We undertook quantitative and qualitative bacteriological analysis as previously described. 22 For measure-ment of markers of airway infl ammation, sputum was ultracentrifuged at 50 000 × g for 90 min at 4°C. The sol phase was removed and immediately frozen at -70°C. We measured markers of airway infl ammation as previously described. 23 We did serum measurement of MBL as described elsewhere. 24 Briefl y, diluted sera were incubated in mannan-coated ELISA plates and bound MBL was detected with a specifi c monoclonal antibody to MBL (HYB131-01) followed by anti-mouse immunoglobulin conjugated to alkaline phosphatase with p-nitrophenyl phosphate as a substrate. Previous validation of this   assay showed an intra-assay coeffi cient of variation of 3·5% and inter-assay coeffi cient of variation of 7%. 24 To assess the eff ects of changes in MBL on disease severity over time, we measured serum MBL at study baseline, midpoint, and at the end of the study in all patients. In addition, to determine the eff ect of exacerbations on serum MBL, we recruited 68 patients attending the Royal Infi rmary of Edinburgh bronchiectasis service for treatment of exacerbations. 23 Serum samples were obtained at day 1 (start of exacerbation) and day 14 (end of exacerbation). We treated patients with intravenous antibiotic therapy on the basis of their previous sputum microbiological results for 14 days. Repeat measurements were then made at least 3 months after exacerbation to determine return of MBL concentrations to baseline levels.
Genomic DNA was isolated from EDTA (edetic acid)anticoagulated whole-blood samples with the Nucleon BACC-3 kit (Gen-Probe, MA, USA). Isolated DNA was quality-tested and genotyping was done at the Wellcome Trust Clinical Research Facility Genetics Core (Edinburgh, UK). Validated Taqman allele specifi c PCR primers were purchased from Applied Biosystems (CA, USA) and PCR done on the Applied Biosystems 7900HT according the manufacturer's instructions. We studied six single-nucleotide polymorphisms (SNPs) known to have the greatest eff ect on MBL serum concentrations: the exon-1 polymorphisms B-rs1800450, C-rs1800451, and D-rs5030737, the promoter polymorphisms H/L, rs11003125, X/Y, and rs7096206, and the 5′-untranslated region SNP P/Q, rs7095891. These SNPs comprise seven well-characterised "secretor haplotypes" (HYPA, LYPA, LYQA, LXPA, HYPD, LYPB, and LYQC) which strongly infl uence circulating MBL concentrations 24 (table 1).
No universally agreed upon defi nition of MBL defi ciency exists. MBL function can be assessed by genotype, serum concentrations, or functional activity in complement activation assays. Although these assessments are strongly correlated, they do not provide identical results. To account for this, we present data with three defi nitions of MBL defi ciency determined a priori as genotypes associated with MBL defi ciency, serum concentrations of less than 500 ng/mL, and serum concentrations of less than 200 ng/mL.

Statistical analysis
Normally distributed data are presented as mean (SD) and non-normally distributed data are presented as median (IQR). We analysed deviation from the normal distribution with the D'Agostino and Pearson omnibus K² test. For comparisons of more than two groups of continuous data, we used one-way ANOVA or the Kruskal-Wallis test as appropriate. We used the χ² test to analyse more than two groups of categorical data. To adjust for confounders of the relation between MBL defi ciency and chronic colonisation, we used multivariable logistic regression analysis. To investigate the relation between MBL genotype and survival, we used the Cox's proportional hazard model to estimate survivor functions with survival during 4 years of followup as the dependent variable. We adjusted for variables associated with mortality (p<0·05) in univariate analysis. For all analyses, p<0·05 was regarded as signifi cant. Analyses were done with SPSS version 21 and Graphpad Prism software.

Role of the funding source
The sponsor of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had fi nal responsibility for the decision to submit for publication.

Results
We included 470 patients with non-cystic fi brosis bronchiectasis and 414 healthy controls (fi gure 1). No patients were lost to follow-up. Table 2 shows the characteristics of the study population. None of the patients included in the study was treated with longterm (>28 days) oral or nebulised antibiotics or oral corticosteroids. We noted no signifi cant diff erences in demographics or non-respiratory comorbid illnesses between patients with bronchiectasis and controls. In addition to a diagnosis of bronchiectasis, 55 patients had previously been diagnosed with asthma and 17 patients had chronic obstructive pulmonary disease.   (table 2). All studied SNPs were in Hardy-Weinberg equilibrium (p>0·05). Patients with high-expressing genotypes had a mean serum MBL concentration of 2300 ng/mL (SD 1300), compared with 950 ng/mL (SD 870) with inter mediate-expressing genotypes, and 130 ng/mL (SD 220) in patients with low-expressing genotypes (defi ned as MBL defi cient).
Genotype frequencies and serum MBL concentrations did not diff er between patients with bronchiectasis and controls ( We noted a higher frequency of bacterial colonisation for patients with low-expressing genotypes (table 4) and defi ciency for both serum concentrations (table 5). Rates of bacterial colonisation did not diff er between patients with intermediate-expressing geno types and those with high-expressing genotypes (p=0·36). For both serum concentration cutoff s, we noted a higher frequency of bacterial colonisation with H infl uenzae, and P aeruginosa in the MBL-defi cient group than in the non-defi cient group. However, this eff ect was not evident when patients with the low-expressing genotype were excluded, suggesting that this diff erence was wholly attributable to these patients. After adjustment for age, radiographic severity, and percentage predicted FEV 1 , the logistic regression analysis did not show a signifi cant eff ect of low-expressing genotypes on chronic colonisation (adjusted odds ratio 1·85, 95% CI 0·96-3·59; p=0·063). The relation between low-expressing genotypes and P aeruginosa was not signifi cant after adjustment for age, radiographic severity, and percentage predicted FEV 1 (1·25, 0·55-2·83; p=0·65).
Rates of colonisation with other bacterial species did not correlate with MBL defi ciency as defi ned by genotype, although the numbers of cases in each group were small. One patient was chronically colonised with B cepacia and this patient had YO/YO genotype.
MBL defi ciency defi ned by genotype or serum concentration was not related to percentage predicted FEV 1 or FVC (tables 4,5). Patients with the lowexpressing genotype had more severe radiological bronchiectasis with the modifi ed Reiff score (table 4), as did patients with serum defi ciency defi ned as less than 200 ng/mL (table 5). This eff ect was not evident when serum defi ciency was defi ned as less than 500 ng/mL.
Pul monary function did not diff er between patients with intermediate-expressing and high-expressing geno types.
Frequency of exacerbations in the year before the study was higher in the low-expressing genotype group than it    42 deaths occurred during follow-up, with 25 (60%) of these deaths deemed to be related to bronchiectasis. In an exploratory analysis, we noted that eight (15%) deaths occurred in 55 patients with low-expressing genotypes, ten (7%) deaths occurred in 135 patients with inter mediate-expressing genotypes, and 24 (9%) deaths occurred in 280 patients with high-expressing genotypes. Figure 3 shows the Kaplan-Meier analysis of all-cause mortality. Mortality did not diff er signifi cantly between groups (log-rank p=0·25). A Coxproportional hazard regression analysis adjusting for age, FEV 1 , bacterial colonisation, and radiological severity of bronchiectasis did not establish higher mortality in the low-expressing genotype group (hazard ratio 1·58, 95% CI 0·73-3·4; p=0·44). No increase in mortality was associated with serum MBL levels of less than 200 ng/mL (hazard ratio 1·43, 95% CI 0·65-3·12; p=0·37) and no increase in mortality associated with serum MBL levels of less than 500 ng/mL (1·07, 0·54-2·12; p=0·84).

Discussion
To our knowledge, this study is the fi rst to describe a genetic modifi er of disease severity in non-cystic fi brosis bronchiectasis. We showed that MBL defi ciency (defi ned by genotype) was related to severity of disease, including quality of life and frequency of exacerbations and admission to hospital (panel).
Defi ciency of MBL arises from variants in exon-1 and in the promoter region of the MBL2 gene. Exon-1 alleles designated B, C, and D have a profound dominant eff ect on MBL serum concentrations. Patients homozygous for exon-1 mutations or compound heterozygous for an exon-1 mutation and the X/Y promoter polymorphism have very low serum MBL concentrations, typically less than 200 ng/mL. 5,12,16 Our study showed that these patients, with low serum MBL concentrations, have an increased frequency of chronic bacterial colonisation and an increased incidence of H infl uenzae and P aeruginosa compared with patients with higher serum MBL concentrations.
Compared with patients with non-defi cient MBL expression, patients with low-expressing genotypes had severe bronchiectasis as assessed by radiological scoring, and an increased frequency of outpatient exacerbations and hospital admissions for severe exacerbations during 4 years of follow-up. These patients also had increased impairment in quality of life and worse cough severity as assessed by the Leicester cough questionnaire. Patients with lowexpressing genotypes had signifi cantly higher measures of neutrophil mediated airway infl ammation, primarily related to higher bacterial loads. Our study was not powered to assess mortality and larger multicentre studies are needed to address this outcome.
Despite our fi nding of increased severity of disease in patients with MBL defi ciency, measurements of FEV 1 and FVC did not diff er between MBL groups. This fi nding might seem inconsistent, but adds to a growing body of   When we used defi nitions of serum MBL defi ciency determined a priori (serum concentration <500 ng/mL or <200 ng/mL), we noted a signifi cant association between serum MBL and bacterial colonisation, hospital admis sion, exacerbations, and radiological severity. These diff erences were, however, wholly attributable to the presence of patients with the most severe MBL defi ciency in this group, because exclusion of patients with low-expressing MBL genotypes removed the association. We therefore conclude that patients with intermediate-expressing MBL genotypes are not at increased risk of severe disease. Our study suggests that genotype is the most useful method of defi ning MBL defi ciency, rather than serum levels, for future studies of MBL in bronchiectasis. Although serum levels are genetically determined, they can also be infl uenced by hormones, drugs, and the acute phase response. 28 We previously reported 29 no diff erences between patients with bronchiectasis and controls in serum MBL concentrations and have now confi rmed these fi ndings in a larger cohort. One other study 30 investigated the eff ect of serum MBL on disease severity in 133 patients with non-cystic fi brosis bronchiectasis. The study measured serum MBL concentrations but did not do genotyping, and reported no diff erence in disease severity between patients with serum concentrations of less than 600 ng/mL compared with those with more than 600 ng/mL. Although the study 30 was underpowered to show signifi cant diff erences, a subgroup analysis of 13 patients suggested that serum concentrations of less than 100 ng/mL were associated with severe disease with more P aeruginosa and H infl uenzae infections and a higher frequency of exacerbations. These results seem therefore to support the fi ndings of our present study, showing that only patients with severe MBL defi ciency have a worse phenotype. However, several diff erences existed between MacFarlane and colleagues' study 30 and our own, including that their study obtained MBL measure ments for clinical reasons and therefore included a much higher frequency of patients colonised with P aeruginosa than is typically reported in the literature.
Several large studies have now confi rmed a relation between low-expressing MBL genotypes and disease severity in cystic fi brosis. [14][15][16] Meta-analysis of studies 16 in adults with cystic fi brosis showed that MBL defi ciency was associated with early acquisition of P aeruginosa, increased infections with B cepacia, and increased mortality. Notably, the largest studies suggest that the poor prognosis associated with variant MBL alleles is only evident in those with low-expressing MBL genotypes (YO/YO and XA/YO). 14 The meta-analysis also reported no signifi cant eff ect on lung function or markers of cystic fi brosis severity with the intermediateexpression genotypes. 16 This fi nding sup ports the results of the present study that suggested poor prognosis is associated with the low-expressing geno types but that intermediate MBL expression is not associated with greater disease severity.
Evidence from studies of chronic obstructive pulmonary disease suggests that MBL defi ciency might predispose patients to exacerbations. 27 Our study recruited patients from a specialist bronchiectasis clinic and therefore few patients in our cohort had a diagnosis of chronic obstructive pulmonary disease.

Interpretation
Our study shows that MBL defi ciency defi ned by genotype is associated with increased severity of bronchiectasis as defi ned by exacerbation frequency, hospital admission, radiological fi ndings, and health-related quality of life. Patients with an MBL defi ciency also had a higher frequency of colonisation with bacteria including Haemophilus infl uenzae and P aeruginosa. After 4 years of follow-up, patients had more exacerbations and hospital admissions without signifi cant association noted for survival or decline in lung function over time. MBL might be an important modifi er of disease severity in non-CF bronchiectasis. This postulation should be confi rmed in an independent cohort. Recent evidence that the antibiotic azithromycin, which is eff ective in bronchiectasis, 26 can reverse the basic defect associated with bronchiectasis 27 might lead to a new pharmacogenetic approach to treatment of this disease.
Bronchiectasis occurs commonly in patients with chronic obstructive pulmonary disease, 31 so whether MBL defi ciency also plays a part in chronic obstructive pulmonary disease and associated bronchiectasis will be important to understand.
The mechanism through which MBL predisposes to severe disease is not clear, although reports suggest that MBL binds P aeruginosa, S aureus, and B cepacia and other clinically relevant bacteria, leading to complement activation and enhanced clearance. 6,7 MBL defi ciency might therefore lead to defi cient opsonophagocytosis; however, other mechanisms have also been proposed. MBL seems to be important in clearance of apoptotic cells, a key mechanism for the resolution of infl ammation. Hodge and colleagues reported low concentrations of MBL in the airway of patients with chronic obstructive pulmonary disease and importantly, these low concentrations correlated with reduced apoptotic cell clearance. 32 To our knowledge, our report is the largest study to examine genetic modifi ers of disease severity in noncystic fi brosis bronchiectasis. 33,34 Strengths of the study included the relative absence of confounders such as long-term antibiotic and steroid use and current smoking. We included a broad spectrum of patients from individuals with very mild disease to those with very severe disease who had verifi ed hospital admissions and exacerbations using electronic data to reduce bias. Our data are from a single centre in the UK and are not necessarily generalisable to other health-care settings. Ideally our fi ndings should now be replicated in a large independent, multicentre cohort.
MBL defi ciency is a recognised immunodefi ciency, with clinical testing available in many centres and guideline recommendations suggesting such testing should be done for patients with suspected primary immunodefi ciency. 35 Identifi cation of a group of patients at increased risk of exacerbations and poor outcome could be very useful to clinicians and we hope will stimulate further research to improve outcomes in bronchiectasis. Presently, few evidence-based treatments exist for bronchiectasis. 1 MBL-replacement therapy has been developed and might form a new therapeutic avenue for diseases associated with MBL defi ciency. 36 Of more immediate clinical relevance to bronchiectasis, Hodge and colleagues showed that azithromycin could restore the failure of apoptotic-cell clearance associated with MBL defi ciency, which raises the question of whether future trials should specifi cally target long-term macrolide treatment for patients with MBL defi ciency to improve outcome. 25,32