A CXCL2 tandem repeat promoter polymorphism is associated with susceptibility to severe sepsis in the Spanish population

A CXCL2 tandem repeat promoter polymorphism is associated with susceptibility to severe sepsis in the Spanish population
of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  ORIGINAL ARTICLE A CXCL2 tandem repeat promoter polymorphism is associated with susceptibility to severe sepsis in the Spanish population  C Flores 1 , N Maca-Meyer 1 , L Pe´rez-Me´ndez 1 , R Sangu¨esa 2 , E Espinosa 2 , A Muriel 3 , J Blanco 3 and J Villar 1,4,5 on behalf of the GRECIA and GEN-SEP groups (see Appendix A) 1 Research Institute (Research Center associated to Centro de Investigaciones Biolo´  gicas, Consejo Superior de Investigaciones Cientı´  ficas, Madrid, Spain), Hospital Universitario NS de Candelaria, Tenerife, Spain;  2 Department of Anesthesia, Hospital Universitario NS deCandelaria, Tenerife, Spain;  3  Intensive Care Unit, Hospital Universitario Rı´ o Hortega, Valladolid, Spain;  4 Research Center, St Michael’s Hospital, Toronto, Canada and  5 Canarian Institute for Biomedical Research, Canary Islands, Spain Sepsis describes a complex clinical syndrome resulting from a systemic inflammatory response to bacteria. Functional studies in animal models of sepsis have catalogued CXCL2 as a candidate gene for the development of the disease. We hypothesized that CXCL2 polymorphisms may confer susceptibility to sepsis and performed an association study using 178 severe sepsis patients and 357 population-based controls. We selected two polymorphisms from the promoter of the gene (   437A/G and   665(AC)  n   ), and analyzed whether haplotypes or single loci were associated with disease susceptibility. An overall test of differentiation showed that haplotype distribution was not different between cases and controls (P  ¼ 0.407). Likewise,  437A/G was not associated with disease susceptibility (heterozygote odds ratio (OR) 0.68 (0.47–1.03), and homozygote OR 0.86 (0.56–1.32); P  ¼ 0.706). However, for the   665(AC)  n  , we found that the 24  7 1 repeat alleles were associated with susceptibility (heterozygote OR 2.82 (1.10–7.24), and homozygote OR 3.65 (1.41–9.43); P  ¼ 0.0006). This association remained significant when using a multiple logistic regression analysis (OR 2.23; 95% confidence intervals (95% CI) 1.22–4.03; P  ¼ 0.008) and after a genomic control adjustment (P  ¼ 0.017). Although replicate studies and functional assays are needed, these results suggest that CXCL2 gene variants may contribute to the development of severe sepsis. Genes and Immunity (2006)  7,  141–149. doi:10.1038/sj.gene.6364280; published online 19 January 2006 Keywords:  sepsis; CXCL2; MIP-2; haplotype; lung injury  Introduction Sepsis is the most common cause of organ dysfunction inhospitalized patients and a major cause of death world-wide. Despite numerous advances in medicine, sepsisremains an unconquered challenge for the clinician withan unacceptable high mortality rate of 30–50% for severesepsis and septic shock. 1 Sepsis describes a complexclinical syndrome as a result of a systemic inflammatoryresponse to live bacteria and/or bacterial products. Themost common sites of infection are the lungs and theabdomen. Infectious agents stimulate both proinflamma-tory cytokine production and chemokine release. Thisproinflammatory response contributes to the swiftrecruitment and arming of innate immune cells at aninflamed site. Sepsis develops when the appropriateamplified host response to an infection becomes uncon-trolled causing tissue damage and multiple organfailure. 2 The individual response is determined by manyfactors, including the virulence of the organism, the sizeof the inoculum and the patient’s coexisting conditions.Resistance to bacterial infection is a heritable trait thatseems to be controlled by multiple genes. 1 Hence, geneticvariation that modify or disrupt innate immune sensingof infectious organisms could explain the ability of theimmune system to act in response to infection, thediversity of the clinical presentation of sepsis, theresponse to current medical treatment and the geneticpredisposition to infection in each individual patient.Currently, there is growing evidence that commongenetic variations in humans also play an important rolein the determination of individual susceptibility andoutcome to sepsis. Owing to nature of the trait, thedetermination of such variants have been based oncohorts or case–control association studies using poly-morphisms with a presumed functional effect in candi-date genes relevant to immune response. 2 However, asrule of thumb in association studies, they are plagued bythe impression that they are not consistently reproduci- ble due to methodological limitations.Based on gene expression data derived from multipleanimal models of acute lung injury, Grigoryev  et al. 3 identified a list of candidate genes that are involved in Received 2 June 2005; revised 26 October 2005; accepted 17November 2005; published online 19 January 2006Correspondence: Dr J Villar, Canarian Institute for BiomedicalResearch, Toma´s Morales, 6-1, 35003 Las Palmas de Gran Canaria,Canary Islands, Spain.E-mail: jesus.villar@canarias.org Genes and Immunity (2006) 7,  141–149 & 2006 Nature Publishing Group All rights reserved 1466-4879/06  $ 30.00 www.nature.com/gene  the development of acute lung injury, and that arecommon to other lung stress models such as sepsis. 4 About 85% of candidate genes reported by Grigoryev et al. 3 were found upregulated. One of such gene codesfor the CXCL2 chemokine (also known as the macro-phage inflammatory protein-2 or MIP-2). CXCL2 is a6kDa heparin binding protein that exhibits a pivotal rolein a number of inflammatory and immunoregulatoryactivities, including a potent chemotactic activity forinflammatory and immune effector cells. 5,6 Severalstudies using animal models of lung inflammation andsepsis have implicated CXCL2 as an important mediatorof lung protection. In a septic model, Walley  et al. 7 foundthat increased expression of CXCL2 was associated withthe severity of sepsis, and that antibody blocking of CXCL2 activity decreased mortality. 8,9 Congruent withthese results, different studies have shown that theattenuation of CXCL2 activity, by attenuating its expres-sion by means of glucocorticoids 6 or by siRNA genesilencing, 10  by using inhibitors of its receptor 11 or using amodel of CD8 þ T/NK cell-deficient mice in which theexpression of   CXCL2  is decreased, 12 leads to a reductionin inflammatory response and injury.We tested the hypothesis that variations in  CXCL2 gene (4q13.3) may contribute to the development of severe sepsis. In order to identify risk-associated var-iants, we first determined the common haplotypes alongthe B 2.2kb region containing the gene in local samplesand in a wider region genotyped in Utah CEPHsamples. 13 Then, to assess its contribution as a risk factorin severe sepsis, we genotyped two polymorphismslocated in the promoter region of the gene, an A/G SNPat position   437 and a short tandem repeat (STR) (AC) n at position   665, in 179 patients with established severesepsis and in 364 population-based controls from theSpanish population. To avoid false-positive results due topopulation stratification, we additionally genotyped a battery of 20 unlinked polymorphisms in both groupsand implemented a double-correction approach. Ourresults show that common alleles of the tandem repeatpolymorphism located at position   665 may predisposeto severe sepsis and suggest that they may modulate CXCL2  promoter activity. Results Patients and clinical phenotype It is known that human populations from Spain,including the Canary Islands, are highly homogeneousin their genetic background. 14,15 The principal factorscontributing to differences among Spanish populationsrelate to the levels of external influences, both from theNear East and North Africa. 16 It is well recognized that,with some exceptions such as the Basques, 17 the NearEast influence is homogeneous in the current Spanishpopulation. However, it has been shown that bothcontinental and Island populations have suffered fromNorth African influences at different levels. 17–21 TheCanary Islanders, the populations from Castilla and thesouthern Spanish populations from Andalusia have beenreported to have the highest percentage of North Africaninfluences in Spain. 19,20 In order to meet power require-ments (see Materials and methods) while preserving thehomogeneity of the population, we included 179unrelated patients, which met severe sepsis criteria, 22 from different Spanish localities with as similar aspossible genetic backgrounds: Tenerife, one of the majorislands of the Canarian Archipelago (a total of 64patients) and different localities from Castilla (a total of 115 patients). Baseline demographic data including age,gender, risk factors to the development of infections/sepsis (previous surgery and previous hospitalization formore than 24h), previous morbidity (diabetes, hyperten-sion, ischemic cardiopathy and smoking), main diag-nosis (including the source of infection and pathogens)and Acute Physiology and Chronic Health Evaluationscore (APACHE II) on admission were recorded (Table 1and Supplementary Tables A–D). Clinical data wererecorded until death in the critical unit or until hospitaldischarge. As we used population-based controls, therewas no limitation in collecting the required number of samples, as estimated by the power calculations, from just one locality. We thus used as controls 364 unrelatedindividuals from the Canary Islands. Linkage disequilibrium in CXCL2 The sequencing of the two regions of the gene in 86chromosomes from healthy individuals showed variationat four sites (Figure 1). Three of them were detected as biallelic markers: the SNP rs3806792 A/G at position  437, Table 1  Demographic and clinical features of samples included inthe association study Variable Cases (178) Controls(357) P -valueGender Male (%) 57.9 48.7Female (%) 42.1 51.3 0.054 a Age (years) (mean 7 s.d.) 64.7 7 14.8 44.5 7 13.3 0.001  b Predisposition to infection Previous surgery (%) 74.1 37.0 0.001 a Hospitalized 4 24h c (%) 77.8 48.7 0.001 a Past medical history of  Diabetes (%) 18.3 12.0 0.179 a Hypertension (%) 43.3 22.4 0.001 a Ischemic cardiopathy(%)8.5 1.2 0.011 a Smoker (%) 31.7 27.2 0.716 a Clinical features APACHE II 21 (16–26) d Source of infection (%) 45.2 Abdominal40.5 Pulmonary14.3 OtherPathogen (%) 28.6 Gram negative23.6 Gram positive5.5 Fungi42.3 UnknownICU mortality (%) 46.4ICU ¼ intensive care unit; APACHE II ¼ Acute Physiology andChronic Health Evaluation score. a w 2 test.  b Student’s  t -test. c Hospitalized more than 24h. d Median (inter-quartile range). Association of  CXCL2  and severe sepsis C Flores  et al 142 Genes and Immunity  an insertion/deletion of (ATTT) motif at position 1511(rs5859414) and the SNP rs9131 A/G at position 1789.The two SNPs (rs3806792 and rs9131) were used by theInternational HapMap project. 13 The fourth  locus  show-ing variation was a stretch of (AC) repeats (rs10682383)starting at position  665, thereafter denoted as D4S3454.The combination of the allelic states at the three biallelicpolymorphisms (rs3806792, rs5859414 and rs9131)yielded only two haplotypes, thus pointing to an absenceof historical recombination in the region. This highassociation between alleles at different  loci , or linkagedisequilibrium (LD), is maintained across a 57kb block(results not shown), including exclusively the  CXCL2 gene, in samples of Utah residents with ancestry fromnorthern and western Europe (CEPH), 13 in which onlythese two haplotypes define B 96% of the diversity of thesample. This analysis also revealed that rs3806792 willcapture any of the alleles of the block with an averagecorrelation ( r 2 ) of 0.94, thus being an informative taggingSNP of the region. Furthermore, its proximity to D4S3454(approx. 230bp apart) facilitated their combination into asingle direct haplotyping strategy for the associationstudy. This method, in fact, will be advantageous in ouranalysis as, although the available statistical haplotypereconstruction methods are cost-effective and haveshown considerable power, they have a poor perfor-mance when dealing with polymorphisms with minorallele frequencies below 5%. 23 A total of 1086 chromo-somes (728 from controls and 358 from cases) were thenanalyzed for the polymorphisms rs3806792 and D4S3454in order to obtain haplotypes. Both the SNP rs3806792and the STR D4S3454 polymorphisms were found inHardy–Weinberg equilibrium in both sample groups. Forthe SNP rs3806792, the G allele was the most common in both groups. The STR D4S3454 was found to show 18 to30 repeats in the whole sample. Testing the nonrandomassociation of STR alleles with the SNP rs3806792supported that markers with high mutation rates (thecase of STRs) show little or no LD with nearby markers,even in the absence of historical recombination. Values of  D 0 D 1 ( P 4 0.05) were obtained for low-frequency alleles(less than 5% in the whole sample) containing 18, 19, 20,26, 27, 28 and 30 repeats. Alleles ranging from 21 to 24showed  D 0 B 0.35–0.53 ( P o 0.0001), whereas the allelewith 25 repeats showed the lowest LD with rs3806792( D 0 ¼ 0.018,  P ¼ 0.54).Additionally, we noticed the unusual distribution of STR alleles, as variation was almost constrained to allelescontaining 24 7 1 (AC) repeats, and they covered  B 70–75% of variation of both groups. We thus evaluated theimportance of selection, demographic and/or randomprocesses that may be involved in deviations from theneutral equilibrium at this single  locus . FollowingPayseur  et al. , 24 we assumed that an excess of rare allelescan be used as an evidence of deviations from neutralexpectations. Applying a coalescent method by the use of Bottleneck program, 25 which calculates heterozygotedeficiency as a measure of allele excess, we found thatD4S3454 showed a marginal significant excess of rarealleles in our control sample ( P ¼ 0.042). Recently, Wang et al . 26 postulated that diseases that are mediated byimmune responses might be caused by alleles that have been under positive selection to provide resistance toinfectious diseases and have therefore reached higherpopulation frequencies. Taken together, it seems reason-able to evaluate if the most frequent allele(s) of the STRmay be associated with the disease. For such a purpose,we included all the STR alleles found but selected asalleles ‘at-risk’, the modal allele in both groups (allelewith 24 repetitions) plus the alleles that limit the inter-quartile range in both groups (alleles with 23 and 25repetitions), thereafter called 24 7 1 allele.  Assessing population stratification Before testing the association, we first assessed empiri-cally if systematic differences in genetic backgroundexisted between cases and controls by typing a battery of 20 unlinked neutral polymorphic Alu insertions (PAI).The maximum amount of missing genotypes per  locus was 2.2% (0.89% over all  loci ). No  locus  showeddeviations from Hardy–Weinberg expectations afterBonferroni correction.An overall exact test of genotypic differentiation between the two groups showed no statistically sig-nificant evidence of stratification ( w 2 ¼ 44.661, df  ¼ 40, P ¼ 0.282). In an attempt to homogenize groups, toenhance the power for detecting association, we removedeight individuals (seven from the control group and oneCastillian patient) with abnormal genetic background(see Materials and methods). Thus, association analyseswere performed with 357 control subjects and 178 cases.CXCL2  variants and association with severe sepsis Twenty-one different SNP rs3806792/STR D4S3454 hap-lotypes (Figure 2) were found in the control sample, Figure 1  Schematic diagram of the  CXCL2  gene showing analyzedpolymorphisms and type of variation. Black boxes represent exons.Open circles denote phylogenetically conserved putative transcrip-tion factor binding sites: TFBS1, putative site for NF- k B; TFBS2,putative site for FREAC-2. Figure 2  Frequency distributions of SNP rs3806792/STR D4S3454haplotypes in  CXCL2  gene in controls (black) and severe sepsispatients (white). Numbers on  X  -axis denote the number of (AC) n repeats at D4S3454. Association of  CXCL2  and severe sepsis C Flores  et al 143 Genes and Immunity  giving a heterozygosity estimate of 0.893 7 0.005.Although with nonsignificant differences ( P ¼ 0.083,Bayesian non-credible value after a simulation of 10000realizations of haplotype frequencies), cases showed alower diversity (0.878 7 0.008) and less number of haplo-types. 17 Before performing the analysis at individual  locus ,we first explored whether systematic differences existedin the distribution of haplotype frequencies between bothgroups. Our results showed no evidence of statisticallysignificant differences in the distribution of   CXCL2 haplotypes between cases and controls ( P ¼ 0.407). Usingonly the SNP rs3806792, association tests showed thatgenotypes of this polymorphism were not correlated withthe development of severe sepsis, either using a co-dominant (Table 2) or a recessive model (not shown).For the STR D4S3454, again heterozygosity estimateswere slightly higher in controls (0.799 7 0.007) than incases (0.778 7 0.010), although with nonsignificant differ-ences ( P ¼ 0.074, Bayesian non-credible value after asimulation of 10000 realizations of haplotype frequen-cies). An overall difference in the distribution of STRallele counts in the patients and controls showednonsignificant results ( P ¼ 0.584). However, when geno-types including the 24 7 1 allele were considered for theassociation as risk genotypes and all the remainder of alleles as composing non-risk genotypes, testing fordifferences in genotypes carrying the 24 7 1 alleleshowed a significant crude association with the devel-opment of severe sepsis under an inheritance model withco-dominant effects (Table 2). A crude association Table 2  Summary of association tests of   CXCL2  genotypes and susceptibility to severe sepsis under a co-dominant model. The percentageof genotypes is shown (absolute number in parenthesis) Polymorphism Position Genotype Cases( n ¼ 178)Controls ( n ¼ 357) Compound genotype OR (95% CI) P for trend rs3806792   437 (A/G) A/A 21.9 (39) 17.4 (62) 1A/G 44.4 (79) 51.5 (184) 0.68 (0.47–1.03)G/G 33.7 (60) 31.1 (111) 0.86 (0.56–1.32) 0.706D4S3454   665 (AC) n  18/22 1.12 (2) 0.56 (2) 24 7 1 Hetero and non-carriers 120/22 0 0.28 (1)21/21 0 0.28 (1)21/22 0 0.84 (3)21/27 0 0.56 (2)22/22 1.12 (2) 1.40 (5)22/26 1.12 (2) 2.80 (10)22/27 1.12 (2) 1.40 (5)22/28 0 0.56 (2)26/27 0 0.56 (2)27/28 0 0.28 (1)18/23 0 0.56 (2) 24 7 1 heterozygote carriers 2.82 (1.10–7.24) a 18/24 0.56 (1) 0.84 (3)18/25 0 0.84 (3)19/24 0 0.56 (2)19/25 0 0.28 (1)20/23 0.56 (1) 0.28 (1)20/24 0 0.56 (2)20/25 0.56 (1) 0.84 (3)21/23 0.56 (1) 0.28 (1)21/24 2.25 (4) 1.12 (4)21/25 1.68 (3) 0.84 (3)22/23 5.05 (9) 4.48 (16)22/24 9.55 (17) 9.24 (33)22/25 7.30 (13) 8.12 (29)23/26 1.12 (2) 2.24 (8)23/27 0.56 (1) 0.56 (2)23/28 1.12 (2) 024/26 3.93 (7) 4.48 (16)24/27 1.12 (2) 4.20 (15)24/28 0 0.56 (2)24/30 0 0.28 (1)25/26 1.12 (2) 0.84 (3)25/27 2.25 (4) 1.12 (4)25/28 0.56 (1) 0.56 (2)23/23 2.81 (5) 1.96 (7) 24 7 1 homozygote carriers 3.65 (1.41–9.43) a 0.017  b 23/24 11.80 (21) 10.64 (38)23/25 8.99 (16) 8.40 (30)24/24 11.80 (21) 8.96 (32)24/25 14.04 (25) 14.00 (50)25/25 6.18 (11) 2.80 (10) a Stratified M–H-adjusted OR.  b Significance after a double correction for population stratification. Association of  CXCL2  and severe sepsis C Flores  et al 144 Genes and Immunity  yielded odds ratio (OR) ¼ 2.85 (95% confidence intervals(95% CI): 1.10–7.37) for heterozygote carriers, andOR ¼ 3.71 (95% CI: 1.45–9.48) for homozygote carriers( P  for trend ¼ 0.0003). The observed association wasreplicated stratifying the sample by both gender and theSNP rs3806792 genotype using the stratified Mantel–Haenszel (M–H) approach, obtaining OR M–H ¼ 2.82 (95%CI: 1.10–7.24) for heterozygote carriers, and OR M–H ¼ 3.65(95% CI: 1.41–9.43) for homozygote carriers ( P  fortrend ¼ 0.0006), which implies that the association wasindependent of the SNP background. The associationremained again significant after adjusting for populationstratification ( P ¼ 0.017). Significant risks were alsoobtained for genotypes carrying the 24 7 1 allele undera recessive model: OR ¼ 1.71 (95% CI: 1.19–2.47, P ¼ 0.005). The observed association was replicatedstratifying the sample by both gender and the SNPrs3806792 genotype using the stratified M–H approach(OR M–H ¼ 1.69, 95% CI: 1.17–2.43,  P  for trend ¼ 0.005),and remained significant after adjusting for populationstratification ( P ¼ 0.04).We also tested the association of 24 7 1 genotypes,considering the inclusion of other recorded variables(Table 1 and Supplementart Tables A–D) by means of amultiple logistic regression analysis, and again consider-ing co-dominant and recessive models of inheritance(Table 3). For the two models, the analyses showed thatgender, previous hospitalization for more than 24h,diabetes, hypertension, antecedents of previous ischemiccardiopathy, smoking and SNP background were notsignificantly correlated with the disease, and thus theywere removed from the model. We maintained in theregressions the 24 7 1 STR genotypes, age and previoussurgery, as they remained significant. These analysesshowed that, even with multiple adjustments for otherconfounding variables, the 24 7 1 STR genotypes remainedassociated with the disease (Table 3), confirming the risksobserved using the previous approaches (Table 2).As the rarer variants may be informative for the study,we considered the addition of the STR allele with 22repetitions (the only allele that was found over 5%frequency in the whole sample apart from those contain-ing 23–25 repetitions) to the pool of ‘at-risk’ alleles. Thisanalysis resulted in a marginal significant crude associa-tion with the disease under a recessive model (OR ¼ 1.58;95% CI: 1.03–2.41;  P ¼ 0.03). It maintained significationafter stratifying the sample by gender and the SNPrs3806792 genotype (OR M–H ¼ 1.56; 95% CI: 1.09–2.40; P ¼ 0.04) or by using a multiple logistic regression( P ¼ 0.030), but turned to nonsignificant when adjustingfor population stratification ( P ¼ 0.160). An additivemodel could not be evaluated due to the absence of cases with non-risk genotypes. Discussion To our knowledge, this is the first case–control associa-tion study using  CXCL2  as a candidate gene for sepsissusceptibility, although its key role in promoting inflam-matory stress has been extensively studied in differentmodels of the disease, 7–12 and has been listed as acandidate gene for susceptibility to lung injury occurringwithin the setting of critical illnesses. 3 The commondisease–common variant hypothesis 27 proposes thatmost of the genetic risk for complex diseases is owingto common variants of the  loci  involved. Well-supportedexamples favoring this hypothesis include the tandemrepeat at the  INS locus 28 (risk allele reaching 75%frequency) or the variants in  PPAR g 29 (risk allele at 85%frequency) and the association with different types of diabetes, and the satellite polymorphism in the promoterof   SMYD3  (risk allele reaching 79% frequency) and theassociation with some types of human cancers. 30 In somesituations, as is the case for  APOE  and Alzheimerdisease, the high frequency reached by the risk allelemay be explained by virtue of its ancestral state. 31 However, selection may have been involved in increas-ing the frequency of the risk alleles associated with otherdiseases. 32 Here, we report the association of   CXCL2 common variants (over 70% frequency) and the devel-opment of severe sepsis, showing a significant trend( P ¼ 0.017, after a double correction of populationstratification) with the addition of risk alleles to thegenotype. We cannot exclude the possibility that theremay be other non-tested polymorphisms that are thecausal variants – one must take into account that  CXCL2 gene is surrounded by other genes with key roles in theimmune response, such as  IL-8  and other members of the CXC  family, as well as several epidermal growth factors(  Areg  and  Ereg ) – as even within regions of high LD, theremay be variants that have low LD with surroundingpolymorphisms. However, in our view, both the ob-served low LD between D4S3454 common alleles andSNP rs3806792, and the fact that the D4S3454 risk allelesare the most common in the population, point tocausality of the  locus . Note that, although we detectsome evidence of departure from neutral equilibrium inthe STR D4S3454 (which may involve selection as amechanism for increasing the frequency of some of itsalleles in the population), common alleles are generallyconsidered older, and thus there has been moreopportunity for random processes of recombinationand mutation to break down the ancestral haplotypes. 33 We achieved a high power in this study, eitherassuming recessive (83%) or additive (95%) models witha 0.001 incidence of disease and similar risks to thatobserved. However, more than half (64%) of patients Table 3  Results for the association tests of recorded variablesusing a multiple logistic regression analysis  Included variable a B  b OR adjusted  95% CI–OR adjusted  P -valueCo-dominant model 24 7 1 genotypes 0.80 2.23 1.22–4.03 0.008Age 1.74 5.68 3.27–9.86  o 0.001Previous surgery 1.28 3.60 1.74–7.42 0.001 Recessive model 24 7 1 genotypes 0.94 2.26 1.28–5.12 0.008Age 1.73 5.64 3.25–9.77  o 0.001Previous surgery 1.28 3.58 1.74–7.33 0.001 a Gender, hospitalization for more than 24h, hypertension, ante-cedents of diabetes, antecedents of hypertension, antecedents of ischemic cardiopathy, smoking and SNP background were removedfrom the final model since they were not significantly associated( P 4 0.05).  b Regression coefficient. Association of  CXCL2  and severe sepsis C Flores  et al 145 Genes and Immunity
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks