Mannose-binding lectin gene polymorphic variants predispose to the development of bronchopulmonary complications but have no influence on other clinical and laboratory symptoms or signs of common variable immunodeficiency

Abstract

Mannose-binding lectin (MBL), activating protein of the lectin pathway of the complement system, is an important component of the non-specific immune response. MBL2 gene polymorphisms, both in the coding and promoter regions, lead to low or deficient serum MBL levels. Low serum MBL levels were shown to be associated with serious infectious complications, mainly in patients in whom other non-specific immune system barriers were disturbed (granulocytopenia, cystic fibrosis). We have analysed two promoter (−550 and −221) and three exon (codons 52, 54 and 57) MBL2 polymorphisms in a total of 94 patients with common variable immunodeficiency (CVID) from two immunodeficiency centres. Low-producing genotypes were associated with the presence of bronchiectasis (P = 0·009), lung fibrosis (P = 0·037) and also with respiratory insufficiency (P = 0·029). We could not demonstrate any association of MBL deficiency with age at onset of clinical symptoms, age at diagnosis, the number of pneumonias before diagnosis or serum immunoglobulin (Ig)G, IgA and IgM levels before initiation of Ig treatment. No association with emphysema development was observed, such as with lung function test abnormalities. No effect of MBL2 genotypes on the presence of diarrhoea, granuloma formation, lymphadenopathy, splenomegaly, frequency of respiratory tract infection or the number of antibiotic courses of the patients was observed. Our study suggests that low MBL-producing genotypes predispose to bronchiectasis formation, and also fibrosis and respiratory insufficiency development, but have no effect on other complications in CVID patients.

Introduction

Common variable immunodeficiency (CVID) is a primary hypogammaglobulinaemia affecting both sexes with clinical manifestation beginning at any age over 2 years [1]. Besides frequent and complicated respiratory tract infections (RTI), the patients suffer frequently from other symptoms – diarrhoea, autoimmune diseases, splenomegaly, lymphadenopathy and granuloma formation [2]. Although mutations in genes coding for inducible co-stimulator (ICOS), CD19, B cell-activating factor of the tumour necrosis factor (TNF) family receptor (BAFF-R), and possibly transmembrane activator and calcium-modulating cyclophilin ligand interactor (TACI), were documented in some patients [3], in the majority of affected people the genetic background is unknown. Factors influencing the clinical course and various laboratory abnormalities of CVID are still unknown in general. Published studies of genetic polymorphisms have shown that the vitamin D receptor and interleukin (IL)-6 allelic polymorphisms were associated with immunophenotypic abnormalities in CVID patients, and particular variants of TNF and IL-10 alleles conferred susceptibility to granulomatous forms of CVID [4,5], but probably other disease-modifying genes are also involved.

Mannose-binding lectin (MBL) is an important component of the innate humoral immune response. It binds to polysaccharide groups on the surface of various microbes activating the lectin complement pathway, which is independent of previous antigen–antibody interaction. The gene coding for MBL, designated as MBL2, is located on chromosome 10. Various variants on exon 1 influencing serum MBL levels have been described. A single nucleotide mutation in codon 54 leads to Gly → Asp substitution (variant B), in codon 57 Gly → Glu (variant C) and in codon 52 Arg → Cys (variant D), while the normal, non-mutated allele is designated as A[6]. Homozygosity in the A allele leads to normal serum levels of MBL, while individuals heterozygous for one of the polymorphic alleles have decreased levels of MBL, reaching approximately one-tenth of the normal levels. Homozygotes or compound heterozygotes for mutated alleles have very low serum MBL levels, hardly detectable by conventional enzyme-linked immunosorbent assay [7], although marked interindividual variation can be documented [8]. Also, polymorphisms in the promoter were documented to lead to alteration of serum MBL levels. H/L, Y/X and P/Q polymorphisms at positions −550, −221 and +4 were described. When in cis position with the wild allele, HYA, LYA and LXA haplotypes are associated with high, low and deficient serum MBL levels respectively [9]. In summary, in healthy individuals various combinations of structural and promoter polymorphisms lead to a marked variation of up to 1000-fold in MBL concentrations [6].

The importance of MBL in anti-microbial defence has been documented by studies that showed increased occurrence of invasive infections caused by Streptococcus pneumoniae[10] and Neisseria meningitidis[11] in people with MBL deficiency. MBL deficiency increases the probability of attraction of human immunodeficiency virus infection [12,13], and possibly also shortens survival of patients in the acquired immune deficiency syndrome stage [12]. The frequency of complications of hepatitis B infection is also associated with MBL2 genotypes [14,15].

Data concerning the influence of the MBL2 genotype on the CVID phenotype are limited. Mullighan et al. [16] analysed MBL2 polymorphisms in 163 CVID patients and 100 controls. They found that low MBL-producing alleles were associated with earlier clinical manifestations of CVID. This was most significant in patients with the LXPA haplotype. They also found that the MBL2+4 Q allele was associated with autoimmune manifestation. Fevang et al.[17] found that serum MBL concentrations correlated negatively with the frequency of lower RTI and the presence of bronchiectasis. Andersen et al.[18] observed an increased frequency of severe RTI before the initiation of immunoglobulin (Ig) treatment in patients heterozygous for MBL2 exon 1 structural gene variants. MBL2 exon 1 polymorphic variants were found in 16 of 23 of the patients with various forms of primary hypogammaglobulinaemia with a proven mycoplasma infection compared with two-thirds in the general population, showing that MBL deficiency predisposes to mycoplasma infections in hypogammaglobulinaemic patients [19]. None of these results were confirmed by additional studies.

In this study we have analysed MBL2 exon 1 and promoter polymorphic markers in CVID patients from two immunodeficiency centres and correlated them with various clinical and laboratory parameters to assess to what extent the MBL2 gene could be regarded as a disease-modifying gene in CVID.

Patients and methods

Ninety-four patients with CVID were included into the study: 51 females and 43 males aged 12–82 years [mean 45·4, standard deviation (s.d.) = 14·7]. Fifty-four patients were from the Department of Clinical Immunology and Allergology in Brno and 40 from the Department of Rheumatology and Clinical Immunology in Freiburg. None of them was known to have ICOS (tested in 51 patients) or TNFRSF13C (coding for BAFF-R; tested in six patients) mutations, while the TNFRSF13B (coding for TACI) mutation was documented in 10 of 87 patients tested.

Fifty-two patients fulfilled the European Society for Immunodeficiencies diagnostic criteria for CVID [1]. In 42 patients, mainly those whose treatment was initiated before the mid-1990s (the introduction of relevant tests in our laboratories), diagnosis was made by low Ig levels, clinically significant immunodeficiency and exclusion of other causes of hypogammaglobulinaemia.

Three hundred and fifty-nine healthy donors of Czech origin were used as control subjects for assessing the frequency of MBL2 genotypes, as published previously [20].

The onset of the disease was defined as the age when the first episode of pneumonia or a marked increase in the frequency of RTI occurred. In patients without significant immunodeficiency symptoms, the date of statement of a CVID diagnosis was considered to be the beginning of the disease.

The presence of bronchiectasis and lung fibrosis was determined by high-resolution computerized tomography. Data were available in 66 patients. Respiratory functions were determined by spirometry; the data were available in 90 patients. Obstructive disease was graded as mild, moderate and severe if forced expiratory volume in 1 s was 60–79%, 45–59% and < 45% of the predicted value respectively. Restrictive lung disease was graded as mild, moderate and severe if vital capacity was 60–79%, 45–59% and < 45% of the predicted value, respectively. Splenomegaly was defined by the length of the spleen over 11 cm on ultrasonography.

Serum Ig levels were measured by radial immunodiffusion, turbidimetry or nephelometry, the method being dependent upon the year of diagnosis of the patients. In the case of ‘immeasurable’ Ig serum levels, a lower detection limit was used for calculation. B lymphocyte subpopulations were determined by flow cytometry, as published previously [21], and the patients were subdivided according to the ‘Freiburg classification’[22].

The MBL2 genotype determination was performed by multiplex polymerase chain reaction (PCR), as described previously [20]. Briefly, the promoter polymorphisms were detected using the double amplification refractory mutation system method. Three separate amplifications with sequence-specific sense and anti-sense primers were carried out to determine HY, LY and LX promoter haplotypes respectively. The first exon MBL2 gene mutations were identified using the multiplex PCR method with sequence-specific primers. One reaction with primers specific to B, C and D alleles, and another reaction specific to the A allele in codons 52, 54 and 57, were performed. A 4% MetaPhor agarose gel electrophoresis was used to discriminate PCR products of 128, 135 and 143 base pairs. All reactions included internal control of amplification. Assignment of haplotypes was based on the strong linkage disequilibrium between the promoter variants and the first exon alleles, and the existence of the frequent haplotypes HYA, LYA, HYD, LYB, LYC and LXA. All suspected LYD haplotypes were confirmed by a separate long-chain PCR reaction with sequence-specific primers.

HYA/HYA, HYA/LYA, HYA/LXA, LYA/LYA and LYA/LXA genotypes were considered to be associated with normal levels of serum MBL; the patients with this genotype were labelled as ‘normal’ (N). Patients with HYA/HYD, HYA/LYB, HYA/LYC, HYA/LYD, LYA/HYD, LYA/LYB, LYA/LYC, LYA/LYD and LXA/LXA genotypes were considered to have low serum MBL levels forming the group ‘low’ (L), while HYD/HYD, HYD/LYB, HYD/LYC, HYD/LYD, LYB/LYB, LYB/LYC, LYB/LYD, LYC/LYD, LYD/LYD, LXA/HYD, LXA/LYB, LXA/LYC and LXA/LYD genotype holders were considered to have deficient MBL levels (the ‘deficient’ group: D).

Statistical analysis

Testing the difference in two continuous variables was performed using either the two-tailed t-test or the Mann–Whitney test according to normality of data, which was assessed by the Kolmogorov–Smirnov test. In the case of more than two variables tested, the Kruskal–Wallis test was used. Categorical variables were analysed using the appropriate test for the contingency tables, i.e. using Spearman's test for two variables with more than two levels and Fisher's exact test in the case of two variables with binary outcome. A standard level of statistical significance α = 0·05 was used, i.e. a P-value < 0·05 was considered to be statistically significant. However, because of multiple hypotheses testing, standard Bonferroni correction was applied to the α-level resulting in the appropriate critical value. The statistical package statistica (StatSoft, Inc., Tulsa, OK, USA), version 7, was used.

This study was approved by the Ethics Commission of the Centre for Cardiovascular Surgery and Transplantation in Brno. All patients and donors gave their written consent prior to genetic analysis.

Results

Fifty-eight CVID patients had genotypes associated with normal MBL levels, 19 patients had genotypes associated with low MBL levels and 17 patients exhibited genotypes associated with deficient MBL levels.

The frequency of MBL2 genotype groups leading to normal, low and deficient MBL production in the general Czech population [20] and CVID patients did not show any significant differences (Spearman's test, data not shown), such as frequency of determined alleles or patients having at least one of the determined alleles (Fisher's exact test, data not shown).

State before CVID diagnosis

On comparing age at onset, age at diagnosis and the number of pneumonias before diagnosis, no significant differences were observed (see Table 1). There were no differences in serum IgG, IgA and IgM levels before the diagnosis comparing N versus L+D and N+L versus D groups (Mann–Whitney test, data not shown).

Table 1.

Comparison of age at onset, age at diagnosis, diagnostic delay and number of pneumonias in patients with genotypes associated with normal (N), low (L) and deficient (D) serum mannan-binding lectin levels.

	Normal (n = 58)	Low (n = 19)	Deficient (n = 17)	N versus L versus D P-value	N versus L+D P-value	N+L versus D P-value
Age at onset (years)	26·7 (13·7)	28·9 (17·2)	34·9 (15·8)	0·174*	0·122**	0·062**
Age at diagnosis (years)	32·9 (13·6)	35·9 (17·3)	38·4 (17·8)	0·460*	0·191**	0·240**
Diagnostic delay (years)	6·1 (7·6)	7·0 (8·0)	3·5 (4·4)	0·494*	0·866***	0·304***
Pneumonias before diagnosis	1·5 (2·7)	1·6 (3·4)	0·76 (1·0)	0·758*	0·889***	0·617***

The data are given as mean (standard deviation).

^*Kruskal–Wallis test was used for statistical evaluation;

^**t-test;

^***Mann–Whitney test.

B cell analysis

There were no differences in numbers of patients with B cells < 1% of peripheral lymphocytes (six in normal, four in low and two in deficient groups; P = 0·278, Spearman's test). In 76 patients in whom B cells were > 1% of peripheral lymphocytes the Freiburg classification was used, but no significant differences in the frequency of groups Ia, Ib and II [22] were observed (P = 0·894 for all groups, Spearman's test).

Lung abnormalities

The association of MBL2 genotype groups with the presence of bronchiectasis, lung fibrosis and emphysema is shown in Table 2; as can be seen, the presence of bronchiectasis and lung fibrosis was linked to defective MBL2 genotype groups. On assessing lung function tests (see Table 3), no relation between restrictive and obstructive disease and MBL2 genotype groups was observed, while respiratory insufficiency was associated mildly with the presence of defective MBL2 genotype groups.

Table 2.

Presence of lung abnormalities on computed tomography scan. The results were available for only 66 patients. Generalized bronchiectases were defined as bronchiectases in more than three lung lobes. The results of the extent of bronchiectasis, the extent of fibrosis and the extent of emphysema are given as no/localized/generalized.

	Normal (n = 43)	Low (n = 13)	Deficient (n = 10)	N versus L versus D P-value	N versus L+D P-value	N+L versus D P-value
Presence of bronchiectasis	15	10	5	0·009*	0·022**	0·999**
Extent of bronchiectasis	28/10/5	3/5/5	5/3/2	0·006*	0·013*	0·768*
Presence of fibrosis	12	8	3	0·037*	0·102**	0·999**
Extent of fibrosis	31/10/2	5/6/2	7/2/1	0·048*	0·091*	0·800*
Presence of emphysema	7	2	2	0·959*	0·999**	0·668**
Extent of emphysema	36/4/3	11/0/2	8/2/0	0·913*	0·893*	0·870*

^*Spearman's test;

^**Fisher's exact test. N, normal; L, low; D, deficient serum mannan-binding lectin levels.

Table 3.

Lung function abnormalities in patients with common variable immunodeficiency. The results are given as normal/mild/moderate/severe in the degree of obstructive disease and degree of restrictive disease lines (see Patients and methods; 90 patients evaluated) and no/partial/global in the respiratory insufficiency degree line (88 patients evaluated).

	Normal	Low	Deficient	N versus L versus D P-value	N versus L+D P-value	N+L versus D P-value
Obstructive disease	29	5	8	0·132*	0·277**	0·789**
Degree of obstructive disease	27/18/8/3	13/2/2/1	8/3/5/0	0·274*	0·408*	0·620*
Restrictive disease	8	3	2	0·859*	0·999**	0·999**
Degree of restrictive disease	48/7/1/0	15/2/1/0	14/2/0/0	0·869*	0·904*	0·783*
Respiratory insufficiency	3	4	3	0·029*	0·041**	0·380**
Degree of respiratory insufficiency	51/2/1	14/4/0	13/2/1	0·034*	0·033*	0·291*

^*Spearman's test;

^**Fisher's exact test. N, normal; L, low; D, deficient serum mannan-binding lectin levels.

Respiratory tract infections

The number of RTI and antibiotic courses in our patients during 1 year prior to inclusion into this study is given in Table 4; no differences in the frequency of RTI or of antibiotic courses in the subgroups of CVID patients were observed. There was no difference in the number of patients with X-ray-proven pneumonia after the initiation of Ig treatment (nine in the N group, four in the L group, five in the D group) in all three groups of patients (Spearman's test, P = 0·414), not even when the groups were merged (N versus L+D: P = 0·227, N+L versus D: P = 0·216, Fisher's exact test). There was also no difference in the number of patients who were on permanent or seasonal antibiotic prophylaxis (eight patients in group N, three patients in group L and one patient in group D; Spearman's test for all groups: P = 0·547, Fisher's exact test for N versus L+D: P = 0·760, N+L versus D: P = 0·688). No significant difference was observed comparing patients with more than five infections in 1 year prior to inclusion into the study (four of 49 patients in group N, three of 18 patients in group L, one of 15 patients in group D; Spearman's test for all groups P = 0·421, Fisher's exact test: N versus L+D: P = 0·708, N+L versus D: P = 0·999).

Table 4.

Number of respiratory tract infections (RTI) and antibiotic (ATB) courses during 1 year prior to inclusion into the study in subgroups of common variable immunodeficiency patients. Sufficient data were available in 83 patients. The data are given as mean (standard deviation).

	Normal (n = 50)	Low (n = 18)	Deficient (n = 15)	N versus L versus D P-value*	N versus L+D P-value**	N+L versus D P-value**
No of RTI infections	2·4 (2·1)	3·3 (2·6)	2·6 (1·5)	0·541	0·576	0·810
No. of ATB courses	1·6 (1·8)	1·8 (2·6)	1·5 (1·3)	0·761	0·655	0·901

^*Kruskal–Wallis test;

^**Mann–Whitney test. N, normal; L, low; D, deficient serum mannan-binding lectin levels.

Other clinical indicators

The frequency of splenomegaly, lymphadenopathy, autoimmune phenomena, granuloma and chronic diarrhoea in CVID patients is given in Table 5. There were no significant differences between the groups studied.

Table 5.

Presence of splenomegaly, lymphadenopathy, autoimmune phenomena, granulomas and chronic diarrhoea in 94 patients with common variable immunodeficiency. N/L/D means positivity in the patients with genotypes associated with normal/low/deficient mannan-binding lectin levels respectively.

	N/L/D	N versus L versus D P-value	N versus L+D P-value	N+L versus D P-value
Splenomegaly	32/10/9	0·956*	0·999**	0·792**
Lymphadenopathy	45/11/16	0·561*	0·999**	0·106**
Autoimmune phenomena	15/5/6	0·658*	0·636**	0·551**
Granuloma	2/1/1	0·652*	0·635**	0·556**
Chronic diarrhoea	6/3/3	0·413*	0·629**	0·546**

^*Spearman's test;

^**Fisher's exact test.

Particular genetic variant analysis

We have analysed the relation of the polymorphic variants of MBL2 determined with the presence of subsequent clinical or laboratory data: presence of pneumonia after initiation of Ig treatment, chronic diarrhoea, presence of bronchiectasis, fibrosis, emphysema, respiratory insufficiency, obstructive lung disease, restrictive lung disease, chronic diarrhoea, granuloma formation, autoimmune phenomena, splenomegaly and lymphadenopathy; more than five infections in 1 year prior to inclusion into the study. Fisher's test was used for statistical analysis. Only the presence of autoimmune phenomena in patients negative for allele A (present in five of eight A- patients, compared with 21 of 86 A+ patients, Fisher's test: P = 0·035), and chronic diarrhoea for variant D (present in five of 17 D+ patients and in seven of 77 D− patients, P = 0·038) exceeded P < 0·05 but, because of multiple testing, only the resultant P-values < 0·002 should be considered statistically significant.

Discussion

The extensive variability of CVID stimulates searching not only for causative genes, but also for genes modifying the clinical course of the affected individual. One such disease-modifying gene in CVID might be MBL2. Previous studies have shown that MBL deficiency has only a minor, if any, influence on the morbidity or mortality of otherwise healthy people [23], but that it becomes symptomatic if other defence barrier(s) is/are disturbed, the best example being granulocytopenia during or after cytostatic treatment [24,25] or cystic fibrosis [26].

Much less clear is the association of MBL deficiency with various types of Ig production disturbances. A possible importance of MBL in patients with antibody deficiencies is supported by the observation that MBL2 non-A variants were associated with increased otitis media episodes at the age of 12–24 months, but not later [27]. The age span mentioned is the life period when maternally derived antibodies have waned, but adequate adaptive immunity is not yet developed. On the other hand, Aittoniemi et al.[28] could not document any influence of serum MBL levels on the clinical state of IgA-deficient individuals.

Our study confirmed the previous observation [17] that in CVID patients low MBL levels were associated with the presence of bronchiectasis; also, the presence of fibrosis was associated with the presence of defective genotypes. Interestingly, observations in CVID patients are, to our knowledge, the first described associations of MBL deficiency with bronchiectasis development. Although the numbers of patients in the evaluated groups were too low to draw any unequivocal conclusions, it seems that it is predominantly the decrease in serum MBL level (in both patients from the L and D groups) that predisposed to bronchiectasis or fibrosis development. On the other hand, patients with MBL-deficient genotypes did not have higher proneness to the mentioned complications than the patients with low MBL-producing genotypes.

The association of MBL2 genotype groups with respiratory insufficiency was also observed in our study, but this result could be questioned because of the low number of patients in whom respiratory insufficiency was present. On the other hand, we could not prove any influence of MBL status on the frequency of infections of patients under Ig treatment documented previously by others [17], or the frequency of antibiotic courses in CVID patients. Comparing our study and the above-mentioned studies, we have recorded all RTI in our patients, while in the above-mentioned study only lower RTI were documented [17]. As many of our patients were treated many years ago, we could not evaluate the number of lower RTI prior to Ig treatment, which was shown to be increased in patients heterozygous for structural polymorphisms associated with low MBL production [18]. However, when evaluating the number of pneumonias before making a CVID diagnosis, we could not document any difference among patients from different MBL2 genotype groups.

Unlike the study by Mullighan et al.[16], we could not confirm the earlier clinical manifestation of CVID in patients with defective MBL2 genotypes. Another study from Norway also showed no effect of MBL levels on the age of clinical manifestation of CVID[17]. Surprisingly, our data showed an even later (although not significant) manifestation of CVID in patients with low and mainly deficient MBL-producing genotypes. It is necessary to mention that the retrospective determinations of the onset of immunodeficiency symptoms, even when conducted by experienced physicians, are highly inaccurate in many cases. Also the fact that currently many patients are diagnosed much earlier than previously, even with mild clinical symptoms, should be taken into account as a possible difference from the above-mentioned study [16] published 8 years ago.

Several studies showed that MBL deficiency might be associated with autoimmune diseases such as systemic lupus erythematosus [29] or rheumatoid arthritis [30]. Mullighan et al.[16] found an association of autoimmune phenomena with the presence of the MBL2+4 Q polymorphism. Unfortunately, the polymorphism MBL2+4 was not determined in our study, as this polymorphism has only a minor impact on serum MBL levels [31]. Our study did not find any association of low or deficient MBL2 genotype groups or the particular polymorphic variants evaluated with autoimmune phenomena in our CVID patients.

The MBL status in this study was determined only by MBL2 genotyping, while the serum MBL level was not determined. This is a relatively common approach, as it allows simplification of the complex situation when the actual MBL level in a specific person is influenced not only by genetic background, but also by actual inflammatory status, as MBL reacts as an acute-phase protein [32]; thyroid hormones and the growth hormone were also shown to influence the production of MBL by hepatocytes [33]. Serum MBL in CVID patients may be influenced mainly by acute or chronic inflammation, which is common in these patients.

The observation about the influence of MBL deficiency on bronchiectasis and fibrosis development raises the question of whether, in patients with MBL deficiency (or holders of MBL2 genotypes associated with abnormal serum MBL levels), a more intensive Ig regimen should be applied compared with patients with normal MBL levels. In our opinion the results of our study do not support this approach strongly. Although the results of MBL determination might be taken into account in such considerations, we still do not have clear evidence that the intensity of Ig treatment has a protective effect on bronchiectasis development in general, still less so in the case of MBL deficiency. Only a prospective large-scale study would be able to answer this question.

In general, our study showed that the presence of low or deficient MBL-producing genotypes in patients with CVID is associated with chronic changes of the bronchi and the lungs: bronchiectasis, fibrosis development and respiratory insufficiency. On the other hand, we could not document any influence of MBL2 genotypes on the frequency of acute RTI, extrapulmonary manifestation or various laboratory parameters. It is supposed that various other disease-modifying genes and their mutual interactions as well as interactions with environmental factors must be involved in the variability of the disease.