Skip to main content
NCBI home page
European Journal of Human Genetics logoLink to European Journal of Human Genetics
. 2010 Jan 13;18(6):680–684. doi: 10.1038/ejhg.2009.226

Use of a modeling framework to evaluate the effect of a modifier gene (MBL2) on variation in cystic fibrosis

Kathryn E McDougal 1, Deanna M Green 1, Lori L Vanscoy 2, M Daniele Fallin 3, Michael Grow 4, Suzanne Cheng 4, Scott M Blackman 1, J Michael Collaco 1, Lindsay B Henderson 1, Kathleen Naughton 1, Garry R Cutting 1,*
PMCID: PMC2874654  NIHMSID: NIHMS174670  PMID: 20068595

Abstract

Variants in mannose-binding lectin (MBL2; protein MBL) have shown association with different aspects (eg, lung function, infection, survival) of cystic fibrosis (CF) in some studies but not others. Inconsistent results may be due to confounding among disease variables that were not fully accounted for in each study. To account for these relationships, we derived a modeling framework incorporating CFTR genotype, age, Pseudomonas aeruginosa (Pa) infection, and lung function from 788 patients in the US CF Twin and Sibling Study. This framework was then used to identify confounding variables when testing the effect of MBL2 variation on specific CF traits. MBL2 genotypes corresponding to low levels of MBL associated with Pa infection 1.94 years earlier than did MBL2 genotypes corresponding to high levels of MBL (P=0.0034). In addition, Pa-infected patients with MBL2 genotypes corresponding to low levels of MBL underwent conversion to mucoid Pa 2.72 years earlier than did patients with genotypes corresponding to high levels of MBL (P=0.0003). MBL2 was not associated with the time to transition from infection to conversion or with lung function. Thus, use of a modeling framework that identified confounding among disease variables revealed that variation in MBL2 associates with age at infection with Pa and age at conversion to mucoid Pa in CF.

Keywords: infection, confounding, immunity

Introduction

Studies of twins and siblings affected with cystic fibrosis (CF) show that genetic modifiers have a major role in variation in CF pulmonary disease.1 One of the first genes proposed to modify CF was mannose-binding lectin 2 (MBL2).2 MBL2 encodes mannose-binding lectin (MBL), a collectin with a central role in the innate immune response. MBL is responsible for pathogen recognition, oposonization, and activation of complement.3 Low MBL levels are associated with increased rates of infection, especially in susceptible populations.4 Infection with a variety of bacteria and fungi, in particular Pseudomonas aeruginosa (Pa), has an important role in the course of CF pulmonary disease.5 Thus, modulation of lung infection owing to variation in MBL levels provides an attractive mechanism for a genetic modifier of CF. Variation in the MBL2 gene was initially associated with lung function measures (forced expiratory volume in 1 s; FEV1) in Pa-infected CF patients and with survival of CF patients.2 Later studies associated variation in the MBL2 gene with infection with Pa and Burkholderia cepacia, as well as with age of Pa infection.6, 7, 8, 9, 10, 11, 12, 13, 14 However, numerous other studies failed to find an association with one or more of these attributes.2, 6, 7, 9, 10, 11, 12, 13, 14, 15 Thus, the role of MBL2 variation as a genetic modifier of CF has been of some debate.16

Lack of consistency among association studies can have many sources including population stratification, insufficient power, and variation in environmental factors.17 Another reason for inconsistency may be the presence of unaccounted confounding among different clinical variables. If the patient groups being compared differ by a variable that affects the exposure and the outcome, that variable may be a confounder of the relationship between the exposure and the outcome.18 Therefore, in a disease state with many complex interrelated variables such as CF pulmonary disease, careful consideration of how these variables relate to each other is essential. Taking into account relationships between the exposures and outcomes through a modeling framework can facilitate recognition of and adjustment for confounding variables. We therefore developed modeling frameworks that incorporated variables commonly used in previous MBL2 analyses and tested for association between each variable and variation in the MBL2 gene.

Patients and Methods

Subjects and clinical data

Pulmonary function, growth parameters, infection, and ethnicity/race data were derived from the CF Twin and Sibling Study19 supplemented using data from the Cystic Fibrosis Foundation Data Registry (Bethesda, MD, USA). Informed consent was received from all participants and/or guardians. All patients had DNA available for genotyping. Complete infection information (a recorded positive or negative culture and age of first positive culture, if applicable) for both Pa and mucoid Pa was available for all patients. In CF patients, Pa infection rates are relatively constant within the first 4 years of life, but rise quickly thereafter with increasing age.20 Therefore, patients were excluded from the infection analysis if they were diagnosed ≥4 years of age and had a positive Pa culture within 1 year of diagnosis. Respiratory cultures were obtained by throat swab, expectorated or induced sputum, or bronchoalveolar lavage. There was no difference in the number of clinic visits in the most recent year by MBL-sufficient/deficient status in uninfected patients. Over 30 000 cultures were available for the patients in this study (ranging from 2 to 112 per subject) and there was no difference in the mean number of cultures per subject by MBL-sufficient/deficient status. Age of patient at last pulmonary function test (PFT) was used as the age variable and only infection data collected before that age were used. Any data collected after lung transplant were dropped.

Pulmonary function was an average of the best PFT from each quarter of the most recent year of data as previously described.1 No PFT data were used from patients who were less than 6 years of age or who had received a lung transplant. The average was available in two forms: FEV1% predicted21 and a CF-specific percentile for FEV1.22 As CF-specific percentiles account for age as well as sex and height, the longitudinal decline in lung function seen with traditional predictive equations is corrected for and patients of different ages can be easily compared.1 In all analyses of lung function that required age correction to prevent potential confounding, the CF-specific percentile for FEV1 was used.22 If age correction was not necessary, the FEV1% predicted was used to enable comparison with previous studies. Lung function data were used for patients 6 years of age up to 40 years of age to account for the cohort used in deriving the CF-specific percentile.

Genotyping

Genetic variation in MBL2, including three amino-acid substitutions in exon 1 that prevent formation of the active MBL oligomer and promoter mutations that reduce transcription levels, tends to lower levels of active circulating MBL.23, 24, 25, 26 MBL2 alleles ‘B' (rs1800450), ‘C' (rs1800451), ‘D' (rs5030737), and promoter variation ‘X/Y' (rs7096206) were genotyped by hybridization of PCR-amplified DNA fragments encompassing each SNP to sequence-specific oligonucleotides in linear arrays as previously described.27, 28 An allele that does not contain a B, C, or D coding variant was designated as ‘A,' whereas the presence of any variant was classified collectively as allele ‘O'. The range of MBL levels corresponding to MBL2 diplotype in CF patients has been well described.2, 7, 11, 14 YO/YO and XA/YO individuals can be considered to have a ‘deficiency' of MBL, whereas YA/YO, XA/XA, XA/YA, and YA/YA individuals have a high enough level of MBL to be considered ‘sufficient'.2, 7, 14

Multiplex PCR with biotinylated primers was performed as previously reported.27, 28 Arrays were hybridized and washed using a BeeBlot machine (product number 1003; Bee Robotics Ltd, UK) with BeeBlot programmable software version 2.1b. Assay conditions were as follows: hybridization at 53°C for 10 min, wash at 25°C for 1 min, conjugation at 53°C for 10 min, wash at 25°C for 1 min, wash at 25°C for 10 min; citrate at 25°C for 5 min, substrate at 25°C for 10 min, deionized water at 25°C for 1 min (3 × ). Genotypes were determined using image analysis software (StripScan 5.8.0) provided by Roche Molecular Systems Inc (Pleasanton, CA, USA). Haplotype and diplotypes were constructed on the basis of previously reported haplotype data and by family data. TGFB1 genotyping for codon 10 was performed as previously reported.29 All SNPs meet the Hardy–Weinberg equilibrium requirements in parents. For MBL2 allele frequencies (Table 1).

Table 1. MBL2 genotype frequencies and breakdown of MBL2 analysis variables in 788 Caucasian CF patients.

Infection and pulmonary analyses MBL2 genotypes Number of patients Percent of patients
MBL sufficient YA/YA 235 29.8
  YA/XA 183 23.2
  XA/XA 29 3.7
  YA/YB 121 15.4
  YA/YD 64 8.1
  YA/YC 25 3.2
MBL deficient XA/YB 40 5.1
  XA/YD 30 3.8
  XA/YC 6 0.8
  YB/YB 19 2.0
  YB/YD 20 2.5
  YB/YC 4 0.5
  YD/YD 5 0.6
  YD/YC 6 0.8
  YC/YC 1 0.1
Open in a new tab

This chart contains individuals who are related to each other. For monozygous twins, one individual per set was used. Family relatedness was corrected for in all statistical analyses.

Statistical analysis

All analyses were performed in Intercooled Stata 8.2 and corrected for related individuals using the ‘cluster' command (www.stata.com/). In this study, models of CF pulmonary disease were built using previously reported relationships to establish which variable was designated as ‘exposure' and which variable was designated as ‘outcome' for each pairwise relationship. Potential confounders were identified from pairwise statistical associations in the CF Twin and Sibling Study population in concert with background knowledge of causal networks that link exposure variables, outcome variables, and potential confounder variables.30 To be considered a confounder in the exposure–outcome relationship, a variable had to be associated with exposure in our population, associated with outcome in our population, and not assumed to be in the causal pathway between exposure and outcome.30 Associations between variables were evaluated by simple linear regression for two continuous variables, ANOVA for dichotomous and continuous variables, and logistic regression for two dichotomous variables. Power was computed using UnifyPow.31

Results

Modeling effects on lung disease in a CF population

Individuals in this study consist of 848 Caucasian siblings, dizygous twins, and monozygous twins enrolled in the US CF Twin and Sibling Study.1, 32, 33 For each of the 60 monozygous twin pairs, one twin was randomly removed from analysis. The remaining 788 individuals are 53.0% male, 50.8% ΔF508 homozygotes, mean age 16.0±7.8 with a median age of 14.4 years. In addition, 85.4% had a positive culture for Pa and 70.3% of the patients with a positive Pa culture had a culture positive for mucoid Pa. The prevalence of Pa is comparable to the rates seen in studies of unrelated CF patients.15

To construct a model to examine the genetic and nongenetic effects on CF pulmonary disease, we analyzed the pairwise relationships among four key variables, CFTR genotype, age, Pa infection and lung function, and displayed them in a graphical framework to guide our statistical modeling. However, as these variables are related, we first evaluated and adjusted for potential confounding effects in each pair-wise relationship using a combination of a priori knowledge and statistical analysis.30

As CFTR genotype was the only constant variable in these models, the first set of tests explored relationships between CFTR genotype (classified as ΔF508 homozygote or non-ΔF508 homozygote) and the three key outcome variables: age, Pa infection, and lung function. ΔF508 homozygotes are 2.56 times more likely to be infected with Pa than non-ΔF508 homozygotes (95% confidence interval 1.62, 4.03; P=0.0001), but CFTR genotype was not associated with age or lung function (Figure 1). Age was associated with Pa infection status as CF patients infected with Pa are on average 2.44 years older than uninfected CF patients (P=0.0066). Next, we considered the relationship between Pa infection status and lung function. As age was also associated with lung function, we adjusted for age using the CF-specific percentile for FEV1.22 For every year increase in age, there is a decrease of 1.53 units of FEV1% predicted (P≤0.0001). Finally, CF patients infected with Pa have a decrease in lung function of 7.82 CF-specific percentile points on average when compared with uninfected patients once age has been corrected for (P=0.0025; Figure 1).

Figure 1.

Figure 1

Open in a new tab

Model of the relationships between age, CFTR, P. aeruginosa (Pa) infection, and lung function in 788 cystic fibrosis (CF) patients. Significant associations between variables in a CF population are represented by arrows. The arrows indicate the assumed directionality of the effect, showing the ‘exposure' variable having an effect on the ‘outcome' variable. Values shown below each arrow quantify the effect and are explained fully in the text.

Modeling effects on lung disease in a Pa-positive CF population

The above model is relevant for Pa infection. However, many CF patients infected with Pa convert over time to a mucoid form. Conversion to mucoid Pa in patients infected with Pa is correlated with a significant decrease in lung function.5 To evaluate the role of CFTR and age at mucoid Pa conversion, we developed a second modeling framework specifically for CF patients infected with Pa. In CF patients infected with Pa, CFTR genotype did not associate with patient age, infection with mucoid Pa, or lung function. In pairwise analysis, CF patients infected with mucoid Pa are on average 5.44 years older than uninfected patients (P≤0.0001; Figure 2). Age was associated with lung function in patients infected with Pa. To address confounding attributed to patient age, we used the age-corrected FEV1% predicted21 measurement when evaluating whether presence of mucoid Pa altered lung function. For every year increase in age, there is a decrease of 1.67 units of FEV1% predicted (P≤0.0001). Further, conversion to mucoid Pa is associated with reduced lung function. CF patients infected with mucoid Pa have a decrease in lung function of 6.24 CF-specific percentile points on average when compared with uninfected patients once age has been corrected for (P=0.0071; Figure 2).

Figure 2.

Figure 2

Open in a new tab

Model of cystic fibrosis (CF) pulmonary disease in 630 P. aeruginosa (Pa)-positive patients. Significant associations between variables in patients with at least one positive culture for Pa are represented by arrows. The arrows indicate the assumed directionality of the effect, showing the ‘exposure' variable having an effect on the ‘outcome' variable. Values shown below each arrow quantify the effect and are explained fully in the text.

Incorporating MBL2 into models of CF pulmonary disease

Our primary objective was to establish the relationship between MBL2 diplotype and CF disease in the context of the other variables evaluated so far. We used the models of CF pulmonary disease described above as a framework in which to analyze MBL2 diplotypes for effect on survival, infection, and lung function. To identify potential confounders, we performed analyses in the same order as the CF pulmonary disease models.

Age

We first considered the relationship between MBL2 genotype and age, given that earlier studies suggesting CF patients with MBL2 genotypes corresponding to low levels of MBL have reduced survival.2, 6 MBL-deficient CF patients in our study had a lower average age at their last PFT than MBL sufficient CF patients (2.25 years younger; P=0.0014).

Infection and age at infection

No association was found between MBL2 genotype (sufficient/deficient) and Pa infection status. However, when we analyzed age at first positive Pa culture, MBL-deficient CF patients acquired Pa infection 1.94 years of age earlier than did MBL sufficient patients (P=0.0034). We then considered the association between MBL2 genotype and mucoid Pa only in patients who were infected with Pa. MBL2 was not associated with mucoid Pa status, but was associated with age at first positive mucoid Pa culture (P=0.0003). CF patients who are MBL deficient convert to mucoid Pa an average of 2.72 years of age earlier than MBL-sufficient patients. When analyzing the time between acquisition of Pa and conversion to mucoid Pa, there was no significant difference by MBL2 genotype. We saw no difference in the relationship between age at infection and MBL2 genotype when stratified by the TGFB1 codon10 genotype. These results show that MBL-deficient patients both acquire Pa and convert to mucoid Pa approximately 2 years earlier than do MBL-sufficient patients (Figure 3).

Figure 3.

Figure 3

Open in a new tab

Effect of mannose-binding lectin 2 (MBL2) genotype groups on mean age of acquisition of P. aeruginosa (Pa) and conversion to mucoid Pa in 630 cystic fibrosis (CF) patients. MBL-deficient CF patients acquire Pa at a mean age of 4.67±5.76 years of age compared with 6.61±5.82 years of age for MBL-sufficient CF patients. MBL-deficient CF patients convert to mucoid Pa at a mean age of 8.90±4.89 years of age compared with 11.62±7.00 years of age for MBL-sufficient CF patients. The mean transition time between Pa and mucoid Pa is 4.58±4.00 years for MBL-deficient CF patients and 4.98±4.54 years for MBL-sufficient CF patients. The P-values are corrected for family structure but estimated means are not.

Effect of shared environment

To determine whether cross-infection with a sibling infected with Pa34, 35 confounded the effect of MBL deficiency, we analyzed only the first infected sibling from each family and evaluated for association with age at Pa infection, age at mucoid Pa infection, and the transition time between them. MBL2 genotype was associated with age at Pa infection, with deficient individuals acquiring infection 2.08 years of age earlier than sufficient patients (P=0.0067; Figure 4). When analyzing MBL2 genotypes and age at mucoid Pa infection, deficient patients showed a trend toward conversion to mucoid Pa earlier than did sufficient patients (P=0.0982). The transition time between Pa infection and conversion to mucoid Pa was not significantly different by MBL2 genotype (P=0.8214). Thus, first infected siblings show the same association with age at Pa infection as do the entire cohort of CF siblings.

Figure 4.

Figure 4

Open in a new tab

Effect of mannose-binding lectin 2 (MBL2) genotype groups on mean age of acquisition of P. aeruginosa (Pa) and conversion to mucoid Pa in 339 first infected cystic fibrosis (CF) siblings. MBL-deficient CF patients acquire Pa at a mean age of 3.43±5.02 years of age compared with 5.52±5.17 years of age for MBL-sufficient CF patients. MBL-deficient CF patients convert to mucoid Pa at a mean age of 9.27±5.61 years of age compared with 11.25±6.64 years of age for MBL-sufficient CF patients. The mean transition time between Pa and mucoid Pa is 5.83±4.04 years for MBL-deficient CF patients and 5.64±4.69 years for MBL-sufficient CF patients. None of the individuals in this analysis is related.

Pulmonary function measures

Our final step was to assess whether MBL2 genotypes were associated with lung function in Pa-infected CF patients. Only infected patients were analyzed to avoid confounding by the relationship between Pa infection and lung function. We saw no association between MBL2 and lung function (measured as FEV1% predicted21 or as CF-specific percentile for FEV122) in patients stratified by mucoid Pa status. Previous studies have suggested that the effect of MBL2 on lung function occurs only in the adult and/or teenage years.2, 7, 11, 12 To address this potential interaction with age, we stratified Pa-infected patients by mucoid Pa status if they were 15, 16, or 18 years or older. None of these separate age analyses showed association between MBL2 genotype and lung function. Finally, we saw no difference in the relationship between lung function and MBL2 genotype when stratified by TGFB1 codon10 genotype.

Discussion

To evaluate the role of MBL2 in CF, we placed clinical variables underlying the progression of CF into modeling frameworks. The validity of the two models presented is supported by observations of similar relationships among the variables used here in other samples of the CF population. For example, it has been shown that increasing age correlates with decreasing lung function and increasing incidence of infection with Pa5 as observed in our first modeling framework. In addition, lung function has been shown to decrease at a faster rate after infection with Pa.36, 37 A third example is the documented detrimental effect of mucoid conversion on lung function,5, 38 which was replicated in our model of Pa-infected CF patients. This study of 630 Pa-infected patients had 86% power to detect association between MBL2 genotypes and age at Pa infection.

Using the modeling frameworks, we observed a robust association between MBL2 genotype and age at Pa infection. Correlation of low-producing MBL2 genotypes with earlier age of infection is consistent with the findings of three earlier studies, including a recent study of 1019 patients reported by Dorfman et al.2, 8, 9 Although two other studies did not see a relationship between MBL2 and age at Pa infection,11, 13 it is possible that the conclusions drawn by the latter studies may have been due to low power to detect an association due to small sample size (162 and 116 Pa-infected patients11, 13). MBL2 status did not modify Pa infection status or mucoid status in this study, consistent with the conclusions of earlier studies.2, 9, 11, 12, 14 This study is the first to show an association between MBL2 variation and age of mucoid Pa conversion. We saw no effect of MBL status on the interval of time to transition between Pa infection and conversion to mucoid Pa. Furthermore, the rate of transition among patients was quite similar. Together, these findings suggest that MBL modifies age at Pa infection, but once infection has initiated, Pa advances to mucoid form at a rate relatively consistent among CF patients and independent of MBL status. Thus, the observed association between MBL2 variants and age at conversion to mucoid Pa appears to be a consequence of the correlation between MBL2 and initial Pa infection.

One important issue to address was the relationship between lung function measurements and MBL2 genotype. Although association has been reported in seven studies,2, 8, 9, 10, 11, 12, 13 most observed association with lung function only after grouping the patients by infection status or age. For example, Garred et al2 and Trevisiol et al9 saw a difference in lung function by MBL2 genotype only in CF patients infected with Pa and Davies et al11 saw the association only in adults. Infection with Pa was associated with lower lung function in the first model (Figure 1). However, the lack of association between MBL2 status and variation in lung function in this study is likely because of the identification and accounting of confounding factors (ie, infection status and age). Alternatively, a 2-year difference in age of acquisition of Pa infection may not result in statistically significant decreases in lung function for a number of years. Thus, our relatively young patient population (median age of our patients was 14.4 years compared with the 36.9 years median age of survival of the US CF population20) may not have had sufficient time to manifest correlation between lung function and MBL2 genotype. Indeed, in Davies et al,11 the association of MBL with lung function was not seen in children with CF. A third possibility is that MBL2 variation is not associated with variation in lung function, as shown in a well-powered study (Drumm et al,15 808 patients) and two other studies.6, 7

In some studies, MBL deficiency has been associated with reduced survival.2, 6 Although the design of our study did not allow assessment of the effect of MBL2 on survival, MBL2 genotypes corresponding to low levels of MBL were associated with a younger average age that would be consistent with a reduction in longevity. The work presented here, together with the bulk of the studies investigating MBL as a modifier of CF, strongly suggests that the primary effect of MBL deficiency is earlier Pa infection and earlier Pa conversion to mucoid status leading to reduced survival.

Although accounting for several important factors involved in CF infection and lung disease, our modeling approach is not exhaustive. Infections with other organisms, such as B. cepacia, and other environmental modifiers, such as secondhand smoke or socioeconomic status, have not been tested in this study. Extensive analysis by CFTR mutation variation has also not been performed. These variables will require a more complex version of the basic modeling framework shown in the paper and are worthy areas for future study. However, by establishing a basic modeling framework, we have created a foundation to test, quantify, and incorporate interactions among additional aspects of CF lung disease. Incorporating greater detail in the modeling frameworks should better reflect the interactions among these variables, thereby facilitating a priori hypotheses regarding modifier effects. Modeling frameworks such as those presented here may help reconcile inconsistent results generated by genetic association studies that evaluate different but correlated components of a complex phenotype.

Acknowledgments

We thank the North American Cystic Fibrosis Foundation for use of the Cystic Fibrosis Foundation Data Registry; Ase Sewall, Monica Brooks (both from Cystic Fibrosis Foundation), and the staff at the Data Registry; Nulang Wang for CFTR genotyping; Erica Senat for her assistance with MBL2 genotyping; Michael Kulich (Charles University, Prague, Czech Republic), for providing conversion programs for cystic fibrosis-specific percentiles; Rita McWilliams, Julie Hoover-Fong, and Ada Hamosh, all from Johns Hopkins University, for designing questionnaires; and most importantly the patients with cystic fibrosis and their families, research coordinators, nurses, and physicians participating in the Cystic Fibrosis Twin and Sibling Study. This study was supported by the National Heart, Lung, and Blood Institute Grant HL68927. Genotyping reagents were provided by Roche Molecular Systems Inc (Pleasanton, CA, USA).

GR Cutting is a consultant for Roche Molecular Systems, and received less than US$1000 per year as consulting fees. M Grow and S Cheng are employed by Roche Molecular Systems.

References

  1. Vanscoy LL, Blackman SM, Collaco JM, et al. Heritability of lung disease severity in cystic fibrosis. Am J Respir Crit Care Med. 2007;10:1036–1043. doi: 10.1164/rccm.200608-1164OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Garred P, Pressler T, Madsen HO, et al. Association of mannose-binding lectin gene heterogeneity with severity of lung disease and survival in cystic fibrosis. J Clin Invest. 1999;104:431–437. doi: 10.1172/JCI6861. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Ohlenschlaeger T, Garred P, Madsen HO, Jacobsen S. Mannose-binding lectin variant alleles and the risk of arterial thrombosis in systemic lupus erythematosus. N Engl J Med. 2004;3:260–267. doi: 10.1056/NEJMoa033122. [DOI] [PubMed] [Google Scholar]
  4. Horiuchi T, Gondo H, Miyagawa H, et al. Association of MBL gene polymorphisms with major bacterial infection in patients treated with high-dose chemotherapy and autologous PBSCT. Genes Immun. 2005;2:162–166. doi: 10.1038/sj.gene.6364165. [DOI] [PubMed] [Google Scholar]
  5. Li Z, Kosorok MR, Farrell PM, et al. Longitudinal development of mucoid Pseudomonas aeruginosa infection and lung disease progression in children with cystic fibrosis. JAMA. 2005;5:581–588. doi: 10.1001/jama.293.5.581. [DOI] [PubMed] [Google Scholar]
  6. Buranawuti K, Boyle MP, Cheng S, et al. Variants in mannose-binding lectin and tumor necrosis factor {alpha} affect survival in cystic fibrosis. J Med Genet. 2006;3:209–214. doi: 10.1136/jmg.2006.046318. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Carlsson M, Sjoholm AG, Eriksson L, et al. Deficiency of the mannan-binding lectin pathway of complement and poor outcome in cystic fibrosis: bacterial colonization may be decisive for a relationship. Clin Exp Immunol. 2005;2:306–313. doi: 10.1111/j.1365-2249.2004.02690.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Dorfman R, Sandford A, Taylor C, et al. Complex two-gene modulation of lung disease severity in children with cystic fibrosis. J Clin Invest. 2008;3:1040–1049. doi: 10.1172/JCI33754. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Trevisiol C, Boniotto M, Giglio L, Poli F, Morgutti M, Crovella S. MBL2 polymorphisms screening in a regional Italian CF center. J Cyst Fibros. 2005;3:189–191. doi: 10.1016/j.jcf.2005.04.001. [DOI] [PubMed] [Google Scholar]
  10. Gabolde M, Guilloud-Bataille M, Feingold J, Besmond C. Association of variant alleles of mannose binding lectin with severity of pulmonary disease in cystic fibrosis: cohort study. BMJ. 1999;7218:1166–1167. doi: 10.1136/bmj.319.7218.1166. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Davies JC, Turner MW, Klein N. Impaired pulmonary status in cystic fibrosis adults with two mutated MBL-2 alleles. Eur Respir J. 2004;5:798–804. doi: 10.1183/09031936.04.00055404. [DOI] [PubMed] [Google Scholar]
  12. Muhlebach MS, MacDonald SL, Button B, et al. Association between mannan-binding lectin and impaired lung function in cystic fibrosis may be age-dependent. Clin Exp Immunol. 2006;2:302–307. doi: 10.1111/j.1365-2249.2006.03151.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Yarden J, Radojkovic D, De Boeck K, et al. Polymorphisms in the mannose binding lectin gene affect the cystic fibrosis pulmonary phenotype. J Med Genet. 2004;8:629–633. doi: 10.1136/jmg.2003.017947. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Olesen HV, Jensenius JC, Steffensen R, Thiel S, Schiotz PO. The mannan-binding lectin pathway and lung disease in cystic fibrosis--dysfunction of mannan-binding lectin-associated serine protease 2 (MASP-2) may be a major modifier. Clin Immunol. 2006;3:324–331. doi: 10.1016/j.clim.2006.08.014. [DOI] [PubMed] [Google Scholar]
  15. Drumm ML, Konstan MW, Schluchter MD, et al. Genetic modifiers of lung disease in cystic fibrosis. N Engl J Med. 2005;14:1443–1453. doi: 10.1056/NEJMoa051469. [DOI] [PubMed] [Google Scholar]
  16. Cutting GR. Modifier genetics: cystic fibrosis. Annu Rev Genomics Hum Genet. 2005;6:237–260. doi: 10.1146/annurev.genom.6.080604.162254. [DOI] [PubMed] [Google Scholar]
  17. Colhoun HM, McKeigue PM, Davey SG. Problems of reporting genetic associations with complex outcomes. Lancet. 2003;9360:865–872. doi: 10.1016/s0140-6736(03)12715-8. [DOI] [PubMed] [Google Scholar]
  18. Greenland S, Morgenstern H. Confounding in health research. Annu Rev Public Health. 2001;22:189–212. doi: 10.1146/annurev.publhealth.22.1.189. [DOI] [PubMed] [Google Scholar]
  19. Vanscoy LL, Lai T, Bowers A, Cutting GR. Modifier genes may contribute substantially to variation in cystic fibrosis (CF) lung disease estimated by CF-specific % for FEV1. Proc Am Thorac Soc. 2006;3:A408. [Google Scholar]
  20. Cystic Fibrosis Foundation . Bethesda, MD: Cystic Fibrosis Foundation; 2006. Cystic Fibrosis Foundation Patient Registry Annual Data Report 2006. [Google Scholar]
  21. Knudson RJ, Lebowitz MD, Holberg CJ, Burrows B. Changes in the normal maximal expiratory flow–volume curve with growth and aging. Am Rev Respir Dis. 1983;6:725–734. doi: 10.1164/arrd.1983.127.6.725. [DOI] [PubMed] [Google Scholar]
  22. Kulich M, Rosenfeld M, Campbell J, et al. Disease-specific reference equations for lung function in patients with cystic fibrosis. Am J Respir Crit Care Med. 2005;7:885–891. doi: 10.1164/rccm.200410-1335OC. [DOI] [PubMed] [Google Scholar]
  23. Sumiya M, Super M, Tabona P, et al. Molecular basis of opsonic defect in immunodeficient children. Lancet. 1991;8757:1569–1570. doi: 10.1016/0140-6736(91)93263-9. [DOI] [PubMed] [Google Scholar]
  24. Lipscombe RJ, Sumiya M, Hill AV, et al. High frequencies in African and non-African populations of independent mutations in the mannose binding protein gene. Hum Mol Genet. 1992;9:709–715. doi: 10.1093/hmg/1.9.709. [DOI] [PubMed] [Google Scholar]
  25. Madsen HO, Garred P, Kurtzhals JA, et al. A new frequent allele is the missing link in the structural polymorphism of the human mannan-binding protein. Immunogenetics. 1994;1:37–44. doi: 10.1007/BF00163962. [DOI] [PubMed] [Google Scholar]
  26. Madsen HO, Garred P, Thiel S, et al. Interplay between promoter and structural gene variants control basal serum level of mannan-binding protein. J Immunol. 1995;6:3013–3020. [PubMed] [Google Scholar]
  27. Baxter N, Sumiya M, Cheng S, et al. Recurrent miscarriage and variant alleles of mannose binding lectin, tumour necrosis factor and lymphotoxin alpha genes. Clin Exp Immunol. 2001;3:529–534. doi: 10.1046/j.1365-2249.2001.01663.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Wang X, Myers A, Saiki RK, Cutting GR. Development and evaluation of a PCR-based, line probe assay for the detection of 58 alleles in the cystic fibrosis transmembrane conductance regulator (CFTR) gene. Clin Chem. 2002;7:1121–1123. [PubMed] [Google Scholar]
  29. Bremer LA, Blackman SM, Vanscoy LL, et al. Interaction between a novel TGFB1 haplotype and CFTR genotype is associated with improved lung function in cystic fibrosis. Hum Mol Genet. 2008. [DOI] [PMC free article] [PubMed]
  30. Hernan MA, Hernandez-Diaz S, Werler MM, Mitchell AA. Causal knowledge as a prerequisite for confounding evaluation: an application to birth defects epidemiology. Am J Epidemiol. 2002;2:176–184. doi: 10.1093/aje/155.2.176. [DOI] [PubMed] [Google Scholar]
  31. O'Brien RG. A tour of UnifyPow: a SAS module/macro for sample-size analysis. Proceedings of the 23rd Annual SAS Users Group International Conference, Cary, NC. 1998. pp. 1346–1355.
  32. Collaco JM, Vanscoy L, Bremer L, et al. Interactions between secondhand smoke and genes that affect cystic fibrosis lung disease. JAMA. 2008;4:417–424. doi: 10.1001/jama.299.4.417. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Blackman SM, Deering-Brose R, McWilliams R, et al. Relative contribution of genetic and nongenetic modifiers to intestinal obstruction in cystic fibrosis. Gastroenterology. 2006;4:1030–1039. doi: 10.1053/j.gastro.2006.07.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Renders NH, Sijmons MA, van Belkum A, Overbeek SE, Mouton JW, Verbrugh HA. Exchange of Pseudomonas aeruginosa strains among cystic fibrosis siblings. Res Microbiol. 1997;5:447–454. doi: 10.1016/s0923-2508(97)83875-2. [DOI] [PubMed] [Google Scholar]
  35. Thomassen MJ, Demko CA, Doershuk CF, Root JM. Pseudomonas aeruginosa isolates: comparisons of isolates from campers and from sibling pairs with cystic fibrosis. Pediatr Pulmonol. 1985;1:40–45. doi: 10.1002/ppul.1950010110. [DOI] [PubMed] [Google Scholar]
  36. Nixon GM, Armstrong DS, Carzino R, et al. Clinical outcome after early Pseudomonas aeruginosa infection in cystic fibrosis. J Pediatr. 2001;5:699–704. doi: 10.1067/mpd.2001.112897. [DOI] [PubMed] [Google Scholar]
  37. Kosorok MR, Zeng L, West SE, et al. Acceleration of lung disease in children with cystic fibrosis after Pseudomonas aeruginosa acquisition. Pediatr Pulmonol. 2001;4:277–287. doi: 10.1002/ppul.2009.abs. [DOI] [PubMed] [Google Scholar]
  38. Ballmann M, Rabsch P, von der HH. Long-term follow up of changes in FEV1 and treatment intensity during Pseudomonas aeruginosa colonisation in patients with cystic fibrosis. Thorax. 1998;9:732–737. doi: 10.1136/thx.53.9.732. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from European Journal of Human Genetics are provided here courtesy of Nature Publishing Group

RESOURCES