Influence of genetic variability at the surfactant proteins A and D in community-acquired pneumonia: a prospective, observational, genetic study

Introduction Genetic variability of the pulmonary surfactant proteins A and D may affect clearance of microorganisms and the extent of the inflammatory response. The genes of these collectins (SFTPA1, SFTPA2 and SFTPD) are located in a cluster at 10q21-24. The objective of this study was to evaluate the existence of linkage disequilibrium (LD) among these genes, and the association of variability at these genes with susceptibility and outcome of community-acquired pneumonia (CAP). We also studied the effect of genetic variability on SP-D serum levels. Methods Seven non-synonymous polymorphisms of SFTPA1, SFTPA2 and SFTPD were analyzed. For susceptibility, 682 CAP patients and 769 controls were studied in a case-control study. Severity and outcome were evaluated in a prospective study. Haplotypes were inferred and LD was characterized. SP-D serum levels were measured in healthy controls. Results The SFTPD aa11-C allele was significantly associated with lower SP-D serum levels, in a dose-dependent manner. We observed the existence of LD among the studied genes. Haplotypes SFTPA1 6A2 (P = 0.0009, odds ration (OR) = 0.78), SFTPA2 1A0 (P = 0.002, OR = 0.79), SFTPA1-SFTPA2 6A2-1A0 (P = 0.0005, OR = 0.77), and SFTPD-SFTPA1-SFTPA2 C-6A2-1A0 (P = 0.00001, OR = 0.62) were underrepresented in patients, whereas haplotypes SFTPA2 1A10 (P = 0.00007, OR = 6.58) and SFTPA1-SFTPA2 6A3-1A (P = 0.0007, OR = 3.92) were overrepresented. Similar results were observed in CAP due to pneumococcus, though no significant differences were now observed after Bonferroni corrections. 1A10 and 6A-1A were associated with higher 28-day and 90-day mortality, and with multi-organ dysfunction syndrome (MODS) and acute respiratory distress syndrome (ARDS) respectively. SFTPD aa11-C allele was associated with development of MODS and ARDS. Conclusions Our study indicates that missense single nucleotide polymorphisms and haplotypes of SFTPA1, SFTPA2 and SFTPD are associated with susceptibility to CAP, and that several haplotypes also influence severity and outcome of CAP.


Introduction
Community-acquired pneumonia (CAP) is the most common infectious disease requiring hospitalization in developed countries. Several microorganisms may be causative agents of CAP, and Streptococcus pneumoniae is the most common cause [1]. Inherited genetic variants of components of the human immune system influence the susceptibility to and the severity of infectious diseases. In humans, primary immunodeficiencies (PID) affecting opsonization of bacteria and NF-Bmediated activation have been shown to predispose to invasive infections by respiratory bacteria, particularly S. pneumoniae [2]. Conventional PID are mendelian disorders, but genetic variants at other genes involved in opsonophagocytosis, with a lower penetrance, may also influence susceptibility and severity of these infectious diseases with a complex pattern of inheritance [3].
In the lung, under normal conditions, microorganisms at first encounter components of the innate immune response, particularly alveolar macrophages, dendritic cells and the lung collectins, the surfactant protein (SP)-A1, -A2 and -D. SP-A1, -A2 and -D belong to the collectin subgroup of the C-type lectin superfamily, and contain both collagen-like and carbohydrate-binding recognition domains (CRDs) [4]. Upon binding to pathogen-associated molecular patterns (PAMPs), SP-A and SP-D enhance the opsonophagocytosis of common respiratory pathogens by macrophages [5,6]. Mice rendered SP-A or SP-D deficient exhibit increased susceptibility to several bacteria and viruses after intratracheal challenge [7][8][9]. SP-A1, -A2 and -D also play a pivotal role in the regulation of inflammatory responses [4,10,11] and clearance of apoptotic cells [4,12,13]. In mice, SP-A and SP-D have been shown to be nonredundant in the immune defense in vivo [9].
The human SP-A locus consists of two similar genes, SFTPA1 and SFTPA2, located on chromosome 10q21-24, within a cluster that includes the SP-D gene (SFTPD) [11]. The nucleotide sequences of human SFTPA1 and SFTPA2 differ little (96.0 to 99.6%) [14]. Single nucleotide polymorphisms (SNP) at the SFTPA1 codons 19, 50, 62, 133 and 219, and at the SFTPA2 codons 9, 91, 140 and 223 have been used to define the SP-A haplotypes, which are conventionally denoted as 6A n for the SFTPA1 gene and 1A n for the SFTPA2 gene (see Table E1 in Additional File 1) [15]. Variability at the SFTPD gene has been also reported. Particularly, the presence of the variant amino acid (aa)-11 (M11T) has been shown to lead to low SP-D levels [16].
In the present study, we assessed the potential association of missense polymorphisms of the SFTPA1, SFTPA2 and SFTPD genes as well as the resulting haplotypes, with the susceptibility to and the severity and outcome of CAP in adults. In addition, we evaluated the existence of linkage disequilibrium (LD) among these genes, and the effect of genetic variability on SP-D serum levels.

Patients and controls
We studied 682 patients and 769 controls, all of them Caucasoid Spanish adult individuals from five hospitals in Spain. Foreigners and individuals with ancestors other than Spanish were previously excluded in the selection process. The diagnosis of CAP was assumed in the presence of acute onset of signs and symptoms suggesting lower respiratory tract infection and radiographic evidence of a new pulmonary infiltrate that had no other known cause. A detailed description of the exclusion criteria and clinical definitions are shown in Methods in Additional File 1 [17][18][19]. The control group was composed of healthy unrelated blood donors from the same hospitals as patients.
For susceptibility, a case-control study was performed. Severity and outcome were evaluated in a prospective study of CAP patients. Demographic and clinical characteristics of CAP patients included in the study are shown in Table E2 in Additional File 1.

Measurement of SP-D serum levels
In order to analyze the effect of the SFTPD aa11 on SP-D levels in our population, protein levels were measured in serum samples from individuals in the control group by means of a Surfactant Protein D ELISA kit (Antibodyshop ® , Gentofte, Denmark).

Genotyping
Four haplotypes of SP-A1 (6A, 6A 2 , 6A 3 and 6A 4 ) and six of SP-A2 (1A, 1A 0 , 1A 1 , 1A 2 , 1A 3 and 1A 5 ) are found frequently (>1%) in the general population [15]. On the basis of the differences in non-synonymous SNPs (SFTPA1-aa19, -aa50, -aa219, SFTPA2-aa9, -aa91, -aa223) the most frequent conventional haplotypes of these genes, except 1A and 1A 5 , can be unambiguously identified (see Table E1 in Additional File 1). However, this method does not allow for the differentiation of some of these haplotypes from those rare haplotypes (frequency equal or lower than 1%) identified with the SNPs indicated in Table E1 in Additional File 1. For comparative purposes, in our study each haplotype was denoted by the name of the most frequent haplotype for a given combination of non-synonymous SNPs. Genomic DNA was isolated from whole blood according to standard phenol-chloroform procedure or with the Magnapure DNA Isolation Kit (Roche Molecular Diagnostics, Pleasanton, CA, USA). Genotyping of polymorphisms in SFTPA1 (aa19, aa50, aa219), SFTPA2 (aa9, aa91, aa223) and SFTPD (aa11) genes was carried out using minor modifications of previously reported procedures [15,20]. The accuracy of genotyping was confirmed by direct sequencing in an ABI Prism 310 (Applied Biosystems, Foster City, CA, USA) sequencer.
Haplotypes for each individual were inferred using PHASE statistical software (version 2.1) [21]. The haplotype of SFTPA1, SFTPA2 or the haplotype encompassing SFTPA1, SFTPA2 and SFTPD was ambiguous or could not be assigned in 12 individuals, who were excluded from the study. The order used for the haplotypes nomenclature is SFTPD-SFTPA1-SFTPA2. Linkage disequilibrium (LD) was measured by means of Arlequin (version 3.11) [22] and Haploview [23] softwares in the control group. In addition, pairwise LD between haplotypes of SFTPA1 and SFTPA2 as well as with the SFTPD SNP was characterized using Arlequin 3.11. The existence of LD was considered if D' >0.4.
Informed consent was obtained from the patients or their relatives. The protocol was approved by the local ethics committee of the five hospitals. All steps were performed in complete accordance to the Helsinki declaration.

Statistical analysis
Bivariate and multivariate statistical analyses were performed using SPSS (version 15.0) (SPSS, Inc, Chicago, Ill, USA) and R package [24]. A detailed description of the statistical methods is shown in Methods in Additional File 1.

Results
Susceptibility to CAP related to SFTPA1, SFTPA2 and SFTPD gene variants Seven non-synonymous SNPs were genotyped across the region containing the SFTPD, SFTPA1 and SFTPA2 genes ( Table 1). None of the SNPs showed a significant deviation from Hardy-Weinberg equilibrium in controls. Several major alleles were overrepresented in controls compared with patients, but only SFTPA1 aa50-G, SFTPA2 aa9-A and aa91-G remained significant after Bonferroni correction for multiple comparisons. A dominant effect of SFTPA2 aa9-A, and a recessive effect of SFTPA1 aa50-G and aa219-C as well as SFTPA2 aa223-C were associated with a lower risk of CAP (see Table 1). When haplotypes were inferred, seven different haplotypes were found for SFTPA1 and eight for SFTPA2 (see Table 2). All haplotypes except 6A 5 , 6A 15 , 1A 10 and 1A 13 had frequencies higher than 1% in our population. The most frequent haplotype for SFTPA1 and SFTPA2 were respectively TGC and AGC, which correspond mainly with the 6A 2 and 1A 0 haplotypes respectively. The frequencies of both haplotypes were significantly lower in patients compared to controls (P = 0.0009, OR = 0.78; 95% confidence interval (CI) 0.67 to 0.91, for SFTPA1 6A 2 . P = 0.002, OR = 0.79; 95% CI 0.68 to 0.92, for SFTPA2 1A 0 ), even when Bonferroni correction was applied. Several haplotypes were overrepresented in patients compared with controls, but only 1A 10 (P = 0.00007, OR = 6.58; 95% CI 2.24 to 26.22) remained significant after Bonferroni correction. For the observed odd-ratios, the power of the tests with a significance level of 1% were 84.16%, 79.09% and 94.04% for the haplotypes 6A 2 , 1A 0 and 1A 10 respectively. In addition, dominant and recessive models showed a significant dominant effect on CAP susceptibility for haplotypes 6A 3 , 1A 0 , 1A 7 and 1A 10 and a recessive effect for haplotype 6A 2 (see Table 2).
Linkage disequilibrium of SFTPA1, SFTPA2 and SFTPD genes Pairwise LD (D') measured by means of Arlequin confirmed the existence of LD among several SNPs at SFTPA1 and SFTPA2, whereas SFTPD aa11 was only observed in LD with SFTPA1 aa19 (see Figure 1). A similar pattern of LD was observed when D' was measured by means of the Haploview software (data not shown). SFTPA1 and SFTPA2 were previously found to be in LD [25,26]. The value of LD measured as r 2 was very low for every pair of SNPs (data not shown), and none of the studied SNPs could be used as haplotypetagging SNP to infer the observed haplotypes. When pairwise LD was measured among haplotypes instead among SNPs, SFTPA1 was found to be in LD with SFTPD aa11, but only a marginal LD was found between SFTPA2 1A and SFTPD aa11 (see Table E3 in Additional File 1).
Susceptibility to CAP related to haplotypes encompassing SFTPA1, SFTPA2 and SFTPD When haplotypes encompassing both SFTPA genes were studied, we observed 39 of the 64 expected haplotypes, and only 14 haplotypes had frequencies higher than 1% (data not shown). The most common SFTPA1-SFTPA2 haplotype, 6A 2 -1A 0 , was underrepresented in patients (P = 0.0005, OR = 0.77; 95% CI 0.66 to 0.90), whereas 6A 3 -1A was overrepresented (P = 0.0007, OR = 3.92; 95% CI 1.63 to 10.80) (see Table 3). Both differences remained significant after Bonferroni correction. For the observed odd-ratios, the powers of the tests with a significance level of 1% were 87.76% and 84.04% for the haplotypes 6A 2 -1A 0 and 6A 3 -1A respectively. On the other hand, dominant and recessive logistic regression models showed a significant dominant effect on CAP susceptibility for haplotypes 6A 3 -1A and 6A-1A 1 and a recessive effect for haplotype 6A 2 -1A 0 (see Table 3). We also intended to analyze whether phased variants encompassing the three genes were involved in susceptibility to CAP. Only 68 of the 128 expected haplotypes were observed, and 16 of them had a frequency over 1%. Chromosomes containing C-6A 2 -1A 0 were decreased in patients when compared with controls (P = 0.00001, OR = 0.62; 95% CI 0.50 to 0.77), a difference that remained significant after Bonferroni correction. C-6A 2 -1A 0 was also significantly associated with protection against CAP in a dominant model (see Table 3).
A similar pattern of haplotype distribution was observed when individual as well as two-and three-gene based haplotypes were compared between pneumococcal CAP patients and healthy controls (see Table E4 in Additional File 1), though no significant differences were now observed after Bonferroni corrections.
Outcome and severity of CAP patients related to genetic variants at SFTPA1, SFTPA2 and SFTPD genes When fatal outcome was analyzed, patients who died within the first 28 days showed a higher frequency of haplotypes 6A 12 , 1A 10 and 6A-1A, and a lower frequency of the major SFTPA1aa19-T and aa219-C alleles and of haplotypes 6A 3 and 6A 3 -1A 1 (see Table 4). Similar results were observed when 90-day mortality was analyzed (see Table 4). For the observed odd-ratios, the power of the tests with a significance level of 5% was 82.64% when the protective effect of 6A 3 -1A 1 on 28-day mortality was evaluated, and 81.45% and 80.79% concerning the effect of 6A 3 and 6A 3 -1A 1 on 90-day mortality respectively. Kaplan-Meier analysis ( Figure 2) and log-rank test (Table 4) also showed significantly different survival for the above mentioned alleles and haplotypes. Cox Regression for 28-day survival, adjusted for age, gender, hospital of origin and co-morbidities, was significant for haplotypes 6A 12 and 6A-1A, and it remained significant for haplotypes 6A 3 and 6A-1A when 90-day survival analysis was performed (see Table 4). We also analyzed Cox Regression adjusted for hospital of origin, PSI and pathogen causative of the pneumonia, and we found similar results: for 28-day Figure 1 Genomic organization, location of SNPs, and linkage disequilibrium (D') map for SFTPD, SFTPA1 and SFTPA2 genes. SNPs: Single-nucleotide polymorphisms. All the D' values higher than 0.3 were statistically significant (P < 0.05). Linkage disequilibrium was measured in the control group.
survival it remained significant for haplotype 6A-1A (P = 0.029, OR = 2.45; 95% CI 1.10 to 5.46), although for 6A 12 haplotype it was not significant (P = 0.072); for 90-day survival it was significant for both 6A 3 (P = 0.038, OR = 0.52; 95% CI 0.28 to 0.96) and 6A-1A (P = 0.045, OR = 2.12; 95% CI 1.02 to 4.44) haplotypes. No effect of the SFTPD aa11 SNP was observed. Due to the high number of observed haplotypes, and because of the limited sample size in the patient groups when they were stratified on the basis of severity and outcome, the haplotypes including SFTPA1, A2 and D were not studied.
The relevance of these genetic variants in the severity of CAP was also evaluated by analyzing predisposition to acute respiratory distress syndrome (ARDS) and to multiorgan dysfunction syndrome (MODS) (see Tables 5 and  6). The SFTPD aa11-C allele was significantly overrepresented in patients with MODS or ARDS. Haplotypes 6A and 6A-1A, were also associated with the development of ARDS, and SFTPA2 1A and 1A 10 were associated with the development of MODS. For the observed odd-ratios, the power of the association of 1A with predisposition to MODS was 89.29%. However, the number of individuals included in the analysis of outcome was relatively small and the power of the tests with a significance level of 1% was lower than 80%. These associations remained significant in multivariate analysis adjusted for age, gender, hospital of origin and co-morbidities, as well as for hospital of origin, PSI and causative microorganism (see Tables 5 and  6). By contrast, 6A 3 -1A 1 was associated with protection against MODS, although this difference was not significant in the multivariate analysis.

Association of genetic variants at SFTPD with serum levels of SP-D
In order to study whether variants at the pulmonary collectins were associated with differences of serum levels of SP-D, this protein was measured in serum from healthy controls with known genotypes. The SFTPD aa11-C SNP associated with lower SP-D serum levels (905.10 ± 68.38 ng/ml for T/T genotype, 711.04 ± 52.02 ng/ml for T/C, and 577.91 ± 96.14 ng/ml for C/C; ANOVA P = 0.017) (see Figure 3).

Discussion
This study is unique in reporting a genetic association between non-synonymous SNPs at SFTPD, SFTPA1 and SFTPA2, as well as of haplotypes encompassing these genes, with the susceptibility, severity and outcome of CAP. The major alleles of SFTPA1 aa50-G, aa219-C as well as SFTPA2 aa9-A and aa91-G or genotypes carrying these alleles were associated with protection against CAP. The frequencies of the different SNPs and haplotypes of SFTPA1, SFTPA2 and SFTPD observed in our study were similar to those previously reported in European populations [25]. SFTPA1 and SFTPA2 were reported to be in strong LD [26,27], and several haplotypes of these loci tend to segregate together, being 6A 2 -1A 0 the major haplotype [27]. A protective role against CAP was associated with 6A 2 , 1A 0 and 6A 2 -1A 0 in our survey but only the rare 1A 10 and 6A 3 -1A haplotypes were significantly associated with susceptibility to CAP. Similar results were observed in susceptibility to pneumococcal CAP. Several SNPs and   Table 4.   [15]. † P-value for the bivariate comparison. ‡ P-value for multivariate analysis, including the variables age, gender, hospital of origin and co-morbidities. For those bivariate comparisons that resulted in nonsignificant differences, multivariate analysis were not calculated. § P-value for multivariate analysis, including the variables hospital of origin, PSI (Pneumonia Severity Index) and pathogen.  haplotypes were also associated with a higher severity and poor outcome; MODS, ARDS, and mortality were selected because they represent the more severe clinical phenotypes. Particularly, 1A 10 and 6A-1A were overrepresented among patients who died at 28 or 90 days, and they also predisposed to MODS and ARDS respectively. Likewise, 6A was associated with ARDS, and 1A was associated with MODS. By contrast, 6A 3 and 6A 3 -1A 1 were underrepresented in patients who died. The SFTPD aa11-C allele was associated with the development of MODS and ARDS, but no significant effects on mortality were observed. In spite that the power of the test for some associations with outcome and severity were higher than 80% for the observed OR with a significance level of 5%, the number of individuals included in the analysis of outcome was relatively small. Consequently, associations with outcome should be interpreted with caution.
Only a few studies have addressed the role of the genetic variability at SFTPA1, and SFTPA2 in infectious diseases [28][29][30][31]. In bacterial infections, homozygosity for the 1A 1 haplotype was reported to be associated with meningococcal disease [30]. Noteworthy, 6A 2 -1A 0 was protective against acute otitis media (AOM) in children [32]. Haplotypes 6A 2 and 1A 0 may also be involved in protection against respiratory syncytial virus (RSV) disease [29,33]. Considering the high difference in the frequencies with the corresponding alternative alleles and haplotypes, it is tempting to speculate that 6A 2 , 1A 0 and 6A 2 -1A 0 could have been maintained at high frequencies partly by their protective effect against respiratory infections. The 6A and 6A-1A haplotypes were found to be associated with an increased risk of wheeze and persistent cough, presumably triggered by respiratory infections or environmental contaminants, among infants at risk for asthma [27]. Regarding SP-D, the SFTPD aa11-T allele was associated with severe RSV bronchiolitis [34], whereas the SFTPD aa11-C variant was associated with tuberculosis [30].
In sharp contrast to the potentially proinflammatory effects after PAMP recognition by collectins, mice deficient in SP-A or SP-D develop enhanced inflammatory pulmonary responses [35][36][37]. SP-A and SP-D play a dual role in the inflammatory response. They interact with pathogens via their CRD, and are recognized by calreticulin/CD91 on phagocytes through the N-terminal collagen domain, promoting phagocytosis and proinflammatory responses [10,13]. By contrast, binding of the CRD to signal inhibitory regulatory protein α (SIRPα) on alveolar macrophages suppresses NF-B activation and inflammation, allowing the lung to remain in a quiescent state during periods of health [10]. A similar dual effect is observed in the promotion or inhibition of apoptosis [12]. SP-A and SP-D can also inhibit inflammation by blocking, through the CRD, Toll-like receptors 2 and 4 [38,39]. In agreement with previous results [16], we have observed that the SFTPD aa11-C allele associates with significantly lower SP-D serum levels than the aa11-T allele, and this effect was dose-dependent. The aa11-C/T SNP, located in the Nterminal domain, influences oligomerization of SP-D and explains a significant part of the heritability of serum SP-D levels [16,40]. Serum from aa11-C homozygotes lack the highest molecular weight (m.w.) forms of the protein, which binds preferentially to complex microorganisms whereas the low m.w. SP-D preferentially binds LPS [16].
The significance and the nature of functional differences between variants at SP-A1 and SP-A2 are poorly understood [14,49,50]. Variants aa50 (SP-A1) and aa91 (SP-A2) are located in the collagen region. These changes may affect the oligomerization pattern and binding to receptors such as calreticulin/CD91 or the functional activity of the protein. Likewise, the variants aa219 (SP-A1) and aa223 (SP-A2) are located in the CRD, and might directly influence the binding properties to microorganisms or to surface receptors such as SIRPα or TLR4. Residue 9, and frequently residue 19, is located in the signal peptide, and it is not know whether these variants may affect the function of the protein   [14,44]. Alternatively all the missense variants could be in LD with SNPs in regulatory regions that might affect translation and RNA stability [51,52].
Native SP-A is thought to consist of hetero-oligomers of SP-A1 and SP-A2, and properties of co-expressed SP-A1/SP-A2 are between those of SP-A1 and SP-A2 [41,46]. However, the extent of oligomerization of SP-A, as well as the SP-A1/SP-A2 ratio, may be altered in various diseases and can vary among individuals [53,54]. The combination of both gene products may be important for reaching a fully native conformation [41]. In fact, it was recently shown that both SP-A1 and SP-A2 are necessary for the formation of pulmonar tubular myelin [55]. Therefore, the effect of a given haplotype may be largely influenced by haplotypes at the other gene. Our results suggest that the 6A 2 to1A 0 haplotype is more protective against CAP than both 6A 2 and 1A 0 .
It was previously reported that the SFTPD aa11 SNP is in LD with SFTPA1 and SFTPA2 [25]. A protective effect of the 6A 2 to 1A 0 haplotype was even higher when this haplotype co-segregates with the SFTPD aa11-C allele. Likewise, one haplotype containing 6A 2 -1A 0 and the G allele of the SFTPD aa160 SNP could be protective against severe RSV disease [29]. Haplotypes at SFTPA1 are in LD with SFTPD aa11 in our population, but only a marginal LD between SFTPA2 and SFTPD aa11 was observed. In addition, no LD between 6A 2 to A 0 and SFTPD aa11 was found in controls (D' = 0.09) or CAP patients (D' = 0.024) in our study. These findings suggest that the protective effect of the co-segregation of SFTPD aa11-C with 6A 2 to 1A 0 on CAP susceptibility may rather reflect genetic interactions. Alternatively, the SFTPD aa11 SNP may be a marker of other SNPs in LD with SFTPA1 and SFTPA2. The gene of another collecting, the mannose-binding lectin (MBL), is located at 10q11.2-q21. We have previously observed that MBL deficiency predisposes to higher severity and poor outcome in CAP [56], and LD of the SP genes with MBL2 cannot be ruled out.
Despite modern antibiotics, CAP remains a common cause of death, and the search for new therapeutic approaches has been redirected into non-antibiotic therapies [57]. SP-A levels are reduced in several pulmonary diseases [58][59][60]. SP-D may also be reduced in some patients with ARDS [59]. In Sftpa -/and Sftpd -/mice, intratracheally administered SP-A or SP-D can restore microbial clearance and inflammation [8,35]. Exogenous surfactant preparation containing the hydrophobic SP-B and -C are nowadays widely used for replacement therapies in infantile RDS. In addition, intratracheal instillation of recombinant SP-C reduced mortality in patients with severe ARDS due to pneumonia or aspiration [61]. Some of the genetic variants analyzed in our survey, such as 1A 10 , although rare, may have a high impact on susceptibility, severity and outcome of CAP. Validation of our results in other populations, and a better knowledge of the functional and clinical significance of the genetic variability at SFTPA1, SFTPA2 and SFTPD could be relevant for future investigations in the use of these collectins in the treatment of respiratory infectious diseases.

Conclusions
The surfactant proteins A1, A2 and D are key components of innate immune response and the antiinflammatory status in the lung. Genetic variability at the genes of these collectins influences susceptibility and outcome of community-acquired pneumonia. These results could be relevant for future investigations in the use of these collectins in the treatment of respiratory infectious diseases.
• 1A 10 and 6A 3 to 1A haplotypes are associated with increased susceptibility to CAP.
• Haplotypes 6A and 6A to 1A are associated with development of ARDS, while 1A and 1A 10 are associated with MODS in patients with CAP.
• The variant SFTPD aa11-C leads to decreased SP-D serum levels, and predisposes to development of MODS and ARDS in patients with CAP.

Additional material
Additional file 1: Further description of methods, definitions and statistical analysis, and Tables E1-E4. The file contains additional information on exclusion criteria and definitions of PSI, ARDS and MODS. The statistical tests used are described. The additional file also includes four tables. Table E1 defines the resulting haplotypes from SNPs combination in SFTPA1 and SFTPA2 genes. Table E2 presents demographic and clinical characteristics of CAP patients. Table E3 shows the pairwise linkage disequilibrium measure for surfactant proteins A1, A2 and D alleles. Table E4 compares haplotypes of SFTPA1, SFTPA2 and SFTPD between patients with pneumococcal CAP and controls.