Leukemia (2005) 19, 1745–1750 & 2005 Nature Publishing Group All rights reserved 0887-6924/05 $30.00 www.nature.com/leu Common genetic variants in the interleukin-6 and chitotriosidase are associated with the risk for serious infection in children undergoing therapy for acute myeloid leukemia

T Lehrnbecher1, T Bernig2, M Hanisch1, U Koehl1, M Behl3, D Reinhardt4, U Creutzig5, T Klingebiel1, SJ Chanock2 and D Schwabe1

1Pediatric Hematology and Oncology, University of Frankfurt, Frankfurt, Germany; 2Pediatric Oncology Branch, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA; 3Competence Center Statistics and OR, University of Applied Sciences, Frankfurt, Germany; 4Pediatric Hematology and Oncology, Hannover Medical School, Hannover, Germany; and 5Pediatric Hematology and Oncology, University of Mu¨nster, Mu¨nster, Germany

Infectious complications represent a substantial cause of therapy, the single most important risk factor for infection in the morbidity and mortality in children undergoing therapy for immunocompromised child. acute myeloid leukemia (AML). Since it has been shown that alterations in innate immune pathways contribute to the risk for In the process of generating a sequence of the human serious infections, we analyzed well-characterized variants in genome, a remarkable set of tools has emerged, the annotation 6 innate immune genes (TNF, IL6, IL8, MPO, CHIT, FCGR2A, of common genetic variants. The catalogue of annotated TLR2, and TLR4) to determine their possible contribution to variants across the genome has already identified host genetic infectious complications during therapy for pediatric AML. The factors that influence clinical outcomes. Even though it is study population consisted of 168 North European Caucasian estimated that there are more than 6 million single nucleotide children enrolled on the clinical trial AML-BFM 93. We found an polymorphisms (SNPs) with a minor allele frequency of greater association between Gram-negative bacterial infection and 6 common, functional variants in two genes, IL6 and CHIT. The than 5–10%, only a small subset probably has functional risk for infection was significantly higher in children with the G implications, namely an alteration in structure or allele in the IL6 promoter at –174 bp (P ¼ 0.026) and in patients expression.7 Until recently, the choice of genetic variants for with the H allele of CHIT (P ¼ 0.033). The promoter variant in IL6 study was limited, but with intensive investigation into key has been shown to increase expression while the H allele genes in well-defined pathways, such as innate immune genes, disrupts both function and circulating levels. Our data suggest it is possible to examine a set of variants in which there is that variant alleles of both IL6 and CHIT could influence 8 susceptibility to infection with Gram-negative bacteria in substantially strong in vitro and prior association data. children undergoing therapy for AML. Follow-up studies, Common, well-defined candidate genetic variants in innate namely replication association studies and in vitro investiga- immune pathways critical for host defense have been identified tion of these common polymorphisms, are warranted to confirm as risk factors for susceptibility and/or outcome to pathogens, these observations. such as malaria, the human immunodeficiency virus, or Leukemia (2005) 19, 1745–1750. doi:10.1038/sj.leu.2403922; 9–11 published online 18 August 2005 Candida infection in leukemia patients. Moreover, common Keywords: polymorphism; infectious complication; child; acute genetic variants in proinflammatory cytokine genes, such as myeloid leukemia tumor necrosis factor-alpha (TNF), interleukins 6 and 8 (IL6 and IL8) contribute to interindividual differences in circulating levels of these cytokines.12,13 The reported findings provide correlative data in support of genetic variation as a risk factor for clinical Introduction outcome. It is likely that small changes in the expression or function of these key components can perturb the delicate Infectious complications remain a substantial cause of morbidity balance between circulating cytokines and thus, modulate the and mortality in patients undergoing therapy for acute myeloid inflammatory response, which in turn, could impact suscept- 1–3 leukemia (AML). It has been estimated that children with ibility to infectious complications.12,14,15 Others have reported AML experience at least three infectious episodes during that common polymorphisms in the genes of the lysosomal 2 intensive therapy. In the past, infection-related mortality has enzyme myeloperoxidase (MPO) and the human chitinolytic 1,2,4 been reported to be as high as 31%. Whereas Gram-positive enzyme chitotriosidase (CHIT) have been associated with bacteria are the most common isolate from the bloodstream, bacterial infection in marrow transplantation and in a region Gram-negative bacteria and fungal pathogens are associated endemic for a filarial parasite.16,17 Similarly, common variants 2 with increased risk for mortality. Studies over the past 30 years in the low-affinity Fc receptor (FCGR2A), and the toll-like have identified a set of clinical risk factors for serious infection receptors 2 and 4 (TLR2 and TLR4) have been associated with during leukemia therapy. These include neutropenia, indwelling immunological complications as well as bacterial and fungal central venous catheters, and therapeutic regimens that induce infections.18–21 2,5 severe mucosal damage (eg, high-dose cytarabine). However, We have chosen to study common genetic variants in genes of it is probable that additional factors contribute to the differences innate immunity, which might influence the risk for infectious in clinical outcomes in children receiving comparable intensive complications in patients undergoing treatment for AML. The rationale is that genetic variants resulting in subtle changes in Correspondence: Dr T Lehrnbecher, Pediatric Hematology and healthy, normal individuals could assume greater significance in Oncology, Johann Wolfgang Goethe-University, Theodor-Stern-Kai immunocompromised patients, such as children undergoing 7, D-60590 Frankfurt, Germany; Fax: þ 49 69 6301 6700; 8 E-mail: [email protected] therapy for AML. We therefore conducted an exploratory study Received 6 June 2005; accepted 19 July 2005; published online 18 with variants from eight genes for which there was excellent August 2005 prior data performed to determine if any of these common Polymorphisms in infections in childhood AML T Lehrnbecher et al 1746 variants influence the risk for serious infection during therapy for the analyis of IL6 levels of the individual groups categorized by pediatric AML. the type of infection established at the time of admission.23

Materials and methods Results The study group consisted of children with AML treated on clinical trial AML-BFM 93 in whom samples were available.22 Our analysis included 168 North European Caucasian pediatric Clinical data were collected from chart review at each center. patients treated on clinical trial AML-BFM-93 (Table 1). Seven The study protocol was approved by the local ethics committee. patients underwent allogeneic stem cell transplantation, which Briefly, the treatment protocol was as follows. All patients was performed after consolidation therapy. The clinical received induction therapy with either daunorubicin or idaru- characteristics of the study population did not differ from the characteristics of the previously reported subjects drawn from bicin in combination with cytarabine and etoposide. After 22 induction, patients were stratified according to risk groups; high- clinical trial AML-BFM 93. risk patients received both HAM (high-dose cytarabine and Detailed hospital chart review identified 508 infectious mitoxantrone) and consolidation (6-thioguanine, prednisolone, episodes, which resulted in roughly three episodes per patient. vincristine, cytarabine, and cyclophosphamide), whereas stand- Only three patients escaped significant infectious complication, ard risk patients received consolidation without HAM.22 All whereas seven patients died due to infection-related complica- patients received intensification with HAE (high-dose cytarabine tions. In all, 110 patients had at least one microbiologically and etoposide), cranial irradiation and maintenance therapy for documented infection (68 patients with one, 23 with two, 16 a total of 18 months of therapy.22 with three, two with four, and one patient with six episodes). A detailed analysis of the infectious complications of this Gram-positive bacteria were isolated in the bloodstream of 84 study population has been reported previously.2 Infectious children, whereas Gram-negative bacteria were recovered from events were categorized according to the presence of neutro- the bloodstream of 24 patients. More than one episode of bacteremia occurred in 28 children, specifically, 26 with Gram- penia (absolute neutrophil count p500/ml), microbiological or clinical documentation of infection or fever without an positive bacteremia and two with Gram-negative bacteremia. identifiable source (FUO). The predominant Gram-positive pathogens isolated in the ¼ Genomic DNA was extracted from cryopreserved lymphocyte bloodstream were staphylococci and streptococci (n 54 and ¼ ¼ pellets by a salt precipitation-extraction method (Gentra n 46, respectively), whereas Klebsiella spp. (n 6), Escheri- ¼ ¼ Systems); a minority of extractions was performed with blood chia coli (n 6) and Pseudomonas aeruginosa (n 5) accounted smears using a standard phenol–chloroform extraction method. for most of the Gram-negative isolates. Standard risk and high- Standard genotype analysis for each of the seven genes, TNF G- risk patients were comparable regarding infectious complica- 308A,19 IL6 C-174G,23 IL8 A-251 T,23 MPO G-463A,19 CHIT tions with the exception of a higher incidence of Gram-positive (24-bp duplicate exon 10),17 FCGR2A H131R,19 and TLR2 bacteremia in the high-risk group (18/53 in the standard risk R753Q24 was performed in duplicate. TLR4 D299G genotyping group (34.0%) vs 66/115 in the high-risk group (57.4%); s ¼ was performed by real-time PCR (Taqman ) with unique P 0.005). Â oligonucleotides (50-ctcgatggtattattg-30 and 50-cctcgatgatattatt-30), Genotype frequencies were analyzed using 3 2 contingency labeled with 6-carboxyfluorescein (FAM) and VICs and tables according to the presence or absence of infectious amplification primers (F: 50-ccattgaagaattccgattagcata-30,R: outcomes (Table 2). An analysis of the distribution of genotypes 50-cactcaccagggaaaatgaagaa-30). Amplification conditions for each genetic variant conformed to Hardy–Weinberg included a 10 min denaturation at 951C, followed by 40 cycles equilibrium. The distribution of these variants was comparable of 921C for 30 s and 601C for 60 s. Assay validation included to the distribution of the same variants in blood donors from sequencing of standard samples.25 Each of the polymorphisms Germany (data not shown). was chosen because of prior data on its functional importance Our analysis revealed that genetic variants in two distinct and a minor allele frequency greater than 5%.12,13,16–21,26,27 genes, both with high priors for inclusion, IL6 and CHIT, were In a subset of patients, serum levels of IL6 were assessed using associated with documented bacteremia in children undergoing a commercially available enzyme immunoassay (ELISA) (R&D therapy for AML. We observed an association between Systems, Minneapolis, MN, USA). Serum was obtained at the microbiologically documented infections and the IL6 promoter À time of diagnosis of AML and/or at the time of admission of variant at 174, known to alter expression of IL6: the risk for febrile neutropenia, as reported previously in detail.23 Children infection was significantly higher in patients heterozygous or with fever for longer than 24 h prior to admission of febrile homozygous for G–174 of IL6 promoter: 18 of 29 homozygotes neutropenia were excluded. for the allele G (62.1%) and 67 of 94 heterozygotes (71.3%), but Differences in proportions (using Fisher’s exact test), odds ratios (ORs) and their exact 95% confidence interval (CI) were calculated with the software SAS System (Version 8.2 for Table 1 Characteristics of the study population Windows). In this exploratory analysis, a value for Po0.05 (two-tailed) was considered to be significant.19 The primary Sex (male/female) 94/74 analysis was conducted for each gene separately and included Age (years; median (range)) 6.2 (0.1–17.8) consideration for false-positive findings using the method FAB morphologya (patients M0 5 (1) 28 described by Wacholder et al. This robust tool (false-positive with Down’s syndrome) M1/M2 61 (1) rate probability (FPRP)) was chosen for analysis in genes in M3 5 (0) which there was a high prior probability that the finding is M4/M5 66 (1) noteworthy.29 Cumulative incidence functions were constructed M6/M7 30 (12) Risk group (standard risk/high risk) 53/115 by the method of Kalbfleisch and Prentice and were compared with Gray’s test.30,31 The Wilcoxon sum test was used for aOne patient was not definitely subclassified.

Leukemia Table 2 Genotype distributions of genes of molecules of innate immunity in pediatric patients with acute myeloid leukemia

Genes, loci Patients with and without microb. Patients with and without clinically Patients with and without Patients with and without Patients with and without fungal docum. infectionsa documented infectionsa Gram-negative bacteremiaa,b Gram-positive bacteremiaa,b infectiona

PPP PP

Chitotriosidase HH 5 (5)c 3 (5) 5 (11) 3 (2) 1 (4) 7 (5) 3 (4) 5 (6) 0 (0) 8 (5) HT 28 (25) 20 (34) 0.39c 11 (24) 37 (30) 0.08 12 (50) 36 (25) 0.044 20 (24) 28 (33) 0.26 4 (31) 44 (28) 1.00 TT 77 (70) 35 (60) 29 (64) 83 (67) 11 (46) 101 (70) 61 (73) 51 (61) 9 (69) 103 (66)

Myeloperoxidase GG 44 (56) 23 (56) 20 (65) 47 (53) 9 (53) 58 (56) 34 (53) 33 (59) 5 (56) 62 (56) GA 33 (42) 16 (39) 0.82 10 (32) 39 (44) 0.55 7 (41) 42 (41) 0.69 28 (44) 21 (38) 0.84 4 (44) 45 (41) 1.00 AA 2 (3) 2 (5) 1 (3) 3 (3) 1 (6) 3 (3) 2 (3) 2 (4) 0 (0) 4 (4)

FcgRIIa HH 23 (34) 10 (29) 7 (25) 26 (35) 4 (31) 29 (33) 19 (34) 14 (30) 2 (29) 31 (33) HT 31 (46) 16 (47) 0.89 17 (61) 30 (41) 0.22 7 (54) 40 (45) 0.93 27 (48) 20 (43) 0.60 3 (43) 44 (46) 0.89 TT 14 (21) 8 (24) 4 (14) 18 (24) 2 (15) 20 (22) 10 (18) 12 (26) 2 (29) 20 (21)

TNF (À308) 11 79 (80) 44 (86) 23 (77) 90 (82) 17 (77) 106 (83) 65 (76) 63 (88) 10 (91) 113 (82) 12 19 (19) 7 (14) 0.67 7 (23) 19 (17) 0.56 5 (23) 21 (16) 0.61 19 (22) 9 (13) 0.14 1 (9) 25 (18) 0.69 22 1 (1) 0 (0) 0 (0) 1 (1) 0 (0) 1 (1) 1 (1) 0 (0) 0 (0) 0 (0) oyopim nifcin ncidodAML Lehrnbecher childhood T in infections in Polymorphisms IL6 (À174) GG 18 (19) 11 (21) 9 (22) 20 (19) 8 (36) 21 (17) 15 (20) 14 (19) 2 (22) 27 (19) GC 67 (70) 27 (52) 0.037 26 (63) 68 (64) 0.87 14 (64) 80 (63) 0.011 50 (67) 44 (60) 0.50 7 (78) 87 (63) 0.53

CC 11 (11) 14 (27) 6 (15) 19 (18) 0 (0) 25 (20) 10 (13) 15 (21) 0 (0) 25 (18) al et

IL8 (À256) TT 10 (19) 8 (26) 7 (29) 11 (19) 2 (18) 16 (22) 9 (21) 9 (23) 2 (50) 16 (20) TA 26 (50) 13 (42) 0.73 10 (42) 29 (49) 0.60 6 (55) 33 (46) 0.92 19 (44) 20 (50) 0.78 2 (50) 37 (47) 0.25 AA 16 (31) 10 (32) 7 (29) 19 (32) 3 (27) 23 (32) 15 (35) 11 (28) 0 (0) 26 (33)

TLR2d wt/wt 59 (89) 32 (94) 0.71 24 (89) 67 (92) 0.70 11 (85) 80 (92) 0.33 50 (89) 41 (93) 0.73 6 (86) 85 (91) 0.49 wt/w1 7 (11) 2 (6) 3 (11) 6 (8) 2 (15) 7 (8) 6 (11) 3 (7) 1 (14) 8 (9)

TLR4 AA 65 (93) 28 (82) 27 (93) 66 (88) 12 (92) 81 (89) 55 (95) 38 (83) 0 (0) 3 (3) AG 4 (6) 4 (12) 0.17 2 (7) 6 (8) 0.86 1 (8) 7 (8) 1.00 3 (5) 5 (11) 0.06 0 (0) 8 (8) 1.00 GG 1 (1) 2 (6) 0 (0) 3 (4) 0 (0) 3 (3) 0 (0) 3 (7) 6 (100) 87 (89) aSubcolumns on the left side show the number (%) of patients who experienced a certain infectious complication, whereas subcolumns on the right side represent the patients without a certain infectious complication. Not all samples amplified for certain genes. bPatients who experienced both Gram-negative and Gram-positive bacteremia during intensive treatment for AML (n ¼ 12) were counted in both groups. cP-values are an explorative measure for evidence of an association based on comparison of patients with AML and a specific infectious complication and control patients with AML without a specific infectious complication, using Fisher’s exact test. dNo individuals homozygous for w1 were observed. Leukemia 1747 Polymorphisms in infections in childhood AML T Lehrnbecher et al 1748 only 11 of 25 homozygotes for the allele C (44.0%) experienced infection, Gram-positive or Gram-negative bacteremia or due to a microbiologically documented infection (2 Â 2 contingency FUO (Table 3). table: GG þ GC (85/123; 69.1%) vs CC (11/25; 44.0%); The second genetic variant to be identified in our analysis is P ¼ 0.022 (OR 2.85 95% CI 1.08–7.58)). When examined by the H allele in CHIT, which is a 24 bp insertion in the coding genotype, specifically evaluating GG or GC, we observed an region. The risk of bacteremia with Gram-negative organisms association with Gram-negative bacteremia: eight of 29 patients was significantly higher in patients with the H allele of CHIT: homozygous for G (27.6%) and 14 of 94 heterozygotes (11.7%) one of eight homozygotes for H (12.5%) and 12 of 48 experienced Gram-negative bacteremia, whereas none of the heterozygotes (25%), but only 11 of 112 homozygotes for the homozygotes for C suffered from this infection (2 Â 2 con- allele T (9.8%) suffered from Gram-negative bacteremia (2 Â 2 tingency table: GG þ GC (22/123; 17.9%) vs CC (0/25; 0%); contingency table: HH þ HT (13/56; 23.2%) vs TT (11/112; P ¼ 0.026 (OR not defined)). The analysis for allele frequencies 9.8%); P ¼ 0.033 (OR 2.78; 95% CI 1.05–7.40)). Similar results was also significant (data not shown). Similar results were were obtained for the analysis of the cumulative incidence obtained when analyzing the standard risk and the high-risk (HH þ HT vs TT; P ¼ 0.015). group separately, but it is not surprising that the analysis was not We did not observe significant associations in the analysis of significant because of the small numbers: (9/37 (24.3%) vs 0/13 common variants in MPO, FCGR2A, TNF, IL8, TLR2, and TLR4 (0%) for standard risk patients P ¼ 0.05 (OR not defined), and and the major infectious end points of study in children with 13/86 (15.1%) vs 0/12 (0%) for high risk; P ¼ 0.15). In contrast, AML. In particular, there was no overrepresentation of any of the an analysis of the cumulative incidence revealed a significant variant genotypes in patients with invasive fungal infection, in difference between patients heterozygous or homozygous for patients with a severe clinical course (eg, hypotension, G–174 of IL6 promoter and patients homozygous for the C allele respiratory failure) or in patients with fatal outcome due to an (P ¼ 0.026). We did not observe an association between the IL6 infectious complication (data not shown). promoter genotype and the serum level of IL6 measured at the Further analysis of risk factors for Gram-negative bacteremia time of diagnosis of AML or at the time of admission for in our retrospective cohort did not reveal any discernible neutropenia and fever due to microbiologically documented difference with respect to gender, age, Down’s syndrome, FAB

Table 3 Children with AML: IL6 C-174G promoter polymorphism and IL6 serum level measured at the time of diagnosis of AML and at the time admitted to the hospital for febrile neutropenia due to microbiologically documented infection, Gram-negative and Gram-positive bacteremia or due to fever of unknown origin (FUO)

Serum levela of IL6 at the time of admission due to

Diagnosis of Microb. docum. Gram-negative Gram-positive FUO (n ¼ 30/17)c AML (n ¼ 10)b infection (n ¼ 21/14)c bacteremia (n ¼ 4/4)c bacteremia (n ¼ 10/8)c

IL6 (À174) genotype GGd 72 (22–271) 377 (85–7251) 612 NA 630 (511–7251) 84 (16–618) GCd 24 (4–115) 136 (2–10 176) 1 911 (375–3 447) 241 (2–585) 96 (2–259) CCd 108 (22–194) 216 (36–12 310) 261 NA 6 285 (261–12 310) 28 (17–58) Children with fever for longer than 24 h prior to admission were excluded. NA ¼ not applicable. aMedian (range) in pg/ml. bNo. of patients. cNo. of episodes/number of patients (some patients experienced more than one infectious complication of the same category). dNo statistically significant difference of IL6 levels was observed among patients with different IL6C-174G promoter genotypes.

Table 4 Comparison of episodes of bacteremia with Gram-negative organisms with infectious complications other than Gram-negative bacteremia in pediatric patients with acute myeloid leukemia

Episodes of bacteremia Infectious complications Pb with Gram-negative other than Gram-negative organisms (n ¼ 26)a bacteremia (n ¼ 482)a

Neutropenia present at onset of infection 22 (84.6)c 350 (72.6) NS Prophylaxis with absorbable antibiotics 11 (73.3) 200 (41.5) NS Prophylaxis with TMPd 23 (88.5) 410 (85.1) NS Central venous line 24 (92.3) 404 (83.8) NS Prior high-dose cytarabine 13 (50) 204 (42.3) NS NS ¼ Not significant. aIn all, 24 patients experienced 26 episodes of Gram-negative bacteremia, whereas 152 children suffered from 482 infectious complications other than Gram-negative bacteremia (11 children experienced both Gram-negative bacteremia and an infectious complication other than Gram-negative bacteremia). bResults are comparable when analyzing standard risk and high-risk patients separately. cData are no. (%) of infectious episodes. dThrimethoprim–sulfamethoxazole.

Leukemia Polymorphisms in infections in childhood AML T Lehrnbecher et al 1749 classification, and risk stratification (data not shown). Auxiliary protein production.26 Although the human CHIT gene has been analyses of clinical parameters were also not associated with the shown to be important as an innate immune gene against chitin candidate, functional variants chosen in this study; specifically, containing pathogens, such as yeast or nematodes, its role in the analysis examined the depth or duration of neutropenia, the bacterial host defense mechanisms is under active investigation, presence of a central venous line, prophylactic use of particularly in relation to first-line innate response to pathogens, thrimethoprim–sulfamethoxazole, the use of nonabsorbable such as bacteria. antibiotics, or the frequency of prior chemotherapy with high- We recognize that a limitation of our study is the sample size dose cytarabine (HAM and HAE) (Table 4). and have chosen to analyze the data individually by gene, but using a false-positive discovery tool, which provides an estimation of the degree of noteworthiness.28,29 Accordingly, Discussion we present our data as a first step and plan to seek replication in subsequent studies. We also recognize that an additional causal In our analysis, we have shown that genetic variants in two variant, in linkage disequilibrium could contribute to the genes, IL6 and CHIT, are associated with Gram-negative observed effect. Thus, it will be important to examine haplotype bacteremia in children undergoing therapy for AML. Our structures of these and any other genes in follow-up studies.33,34 findings for the promoter variant of IL6 are consistent with prior In conclusion, we report that functionally important variant studies that demonstrated a contribution of the same promoter alleles in both IL6 and CHIT could influence susceptibility to variant in IL6 to bacterial sepsis in newborns.14 Interestingly, infections with Gram-negative bacteria in children undergoing both newborns and children undergoing intensive chemother- therapy for AML. Our analysis was restricted to Caucasian apy for acute leukemia are considered immunosuppressed and children undergoing AML therapy in Germany. Thus, applica- often suffer infection with opportunistic and serious bacterial tion to other study populations will need to take into account infections.2,14 The observed association between a functional differences in underlying population structure. Still, for these variant of CHIT and bacterial infection is striking and suggests alleles, it is unlikely that allele frequencies will vary significantly that the CHIT gene could have more pleotropic effects in innate in subjects of North European ancestry.35 If our findings are immunity than previously appreciated. confirmed in follow-up studies, these data, and perhaps Our observation that the G allele of IL6 variant at À174 could additional common genetic variants, could be considered in be associated with an increased risk for significant bacterial the decision to institute antibacterial prophylaxis in children at infection, specifically with Gram-negative pathogens, is con- especially high risk for Gram-negative bacteremia. These sistent with prior published data that the GG genotype of IL6 is observations might also lead to new strategies designed to associated with bacterial sepsis in preterm infants.14 Since the C intervene at one or more vulnerable points in host response to allele has been associated with decreased transcription, it is infection, particularly in immunocompromised children.36 The plausible that alteration of the IL6 response could perturb the latter are especially vulnerable because the detrimental effect balance of the downstream cascade, particularly during the cytoreductive chemotherapy is superimposed on a developing acute response to Gram-negative infection.13 We did not immune system. observe an association between the IL6 promoter genotype and the serum level of IL6 measured. This finding corroborates other studies and is not surprising for the following reasons.23,32 Acknowledgements Measured IL6 levels are highly variable (even over minutes), and more so in the presence of inflammatory stimuli. Thus, a single We thank Martin Zimmermann for his excellent advice for the point measurement is insufficient to assess a dynamic change, statistical analysis. The study was supported by the Deutsche perhaps associated with underlying genetic variation. The Krebshilfe. effects of the IL6 promoter alleles have been investigated in vitro,12,13 and do not fully account for the factors influencing levels in children with AML and chemotherapy-induced References neutropenia. The increase of the concentration of IL6 after endotoxin challenge, which is up to several thousand-fold, 1 Madani TA. Clinical infections and bloodstream isolates associated probably overrides any genetically controlled fine-tuning effect. with fever in patients undergoing chemotherapy for acute myeloid The fact that the infection risk was only affected by genetic leukemia. Infection 2000; 28: 367–373. variants of IL6, but not of other molecules such as TNF or IL8 2 Lehrnbecher T, Varwig D, Kaiser J, Reinhardt D, Klingebiel T, might be explained with the unique function of each molecule Creutzig U. Infectious complications in pediatric acute myeloid leukemia: analysis of the prospective multi-institutional clinical of innate immunity in the complex network of inflammatory trial AML-BFM 93. Leukemia 2004; 18: 72–77. regulators, although the molecules exhibit broad overlapping 3 Creutzig U, Zimmermann M, Reinhardt D, Dworzak M, Stary J, effects. An important contribution of IL6 variants in leukemia Lehrnbecher T. Early deaths and treatment-related mortality in seems unlikely, since the distribution of IL6 genotypes was children undergoing therapy for acute myeloid leukemia: analysis similar to that of healthy individuals. However, we recognize of the multicenter clinical trials AML-BFM 93 and AML-BFM 98. that the IL6 promoter polymorphism could be in linkage J Clin Oncol 2004; 22: 4384–4393. 4 Riley LC, Hann IM, Wheatley K, Stevens RF. Treatment-related disequilibrium with another, yet unidentified, variant, which, deaths during induction and first remission of acute myeloid in turn could be ‘causal’. leukaemia in children treated on the Tenth Medical Research The observation that the H allele of CHIT was associated with Council acute myeloid leukaemia trial (MRC AML10). The MCR a higher risk for Gram-negative infection is plausible, especially Childhood Leukaemia Working Party. Br J Haematol 1999; 106: since the H allele is characterized by a 24- duplication 436–444. in exon 10 that activates a cryptic 30 splice site, generating a 5 Tunkel AR, Sepkowitz KA. Infections caused by viridans strepto- 26 cocci in patients with neutropenia. Clin Infect Dis 2002; 34: mRNA with an inframe deletion of 87 nucleotides. The 1524–1529. frequency of the minor CHIT variant, H, is unexpectedly high in 6 Taylor JG, Choi EH, Foster CB, Chanock SJ. Using genetic variation many studied populations; its biological effect is to disrupt to study human disease. Trends Mol Med 2001; 7: 507–512.

Leukemia Polymorphisms in infections in childhood AML T Lehrnbecher et al 1750 7 Chanock S. Candidate genes and single nucleotide polymorphisms induced activation of murine macrophages. Cell Microbiol 2003; (SNPs) in the study of human disease. Dis Markers 2001; 17: 5: 561–570. 89–98. 22 Creutzig U, Ritter J, Zimmermann M, Reinhardt D, Hermann J, 8 Foster CB, Chanock SJ. Mining variations in genes of innate and Berthold F et al. Improved treatment results in high-risk pediatric phagocytic immunity: current status and future prospects. Curr acute myeloid leukemia patients after intensification with high- Opin Hematol 2000; 7: 9–15. dose cytarabine and mitoxantrone: results of Study Acute Myeloid 9 McGuire W, Hill AV, Allsopp CE, Greenwood BM, Kwiatkowski Leukemia-Berlin–Frankfurt–Munster 93. J Clin Oncol 2001; 19: D. Variation in the TNF-alpha promoter region associated with 2705–2713. susceptibility to cerebral malaria. Nature 1994; 371: 508–510. 23 Lehrnbecher T, Fleischhack G, Hanisch M, Deinlein F, Simon A, 10 Lehrnbecher T, Foster CB, Zhu S, Venzon D, Steinberg SM, Wyvill Bernig T et al. Circulating levels and promoter polymorphisms of K et al. Variant genotypes of FcgammaRIIIA influence the interleukin-6 and 8 in pediatric cancer patients with fever and development of Kaposi’s sarcoma in HIV-infected men. Blood neutropenia. Haematologica 2004; 89: 234–236. 2000; 95: 2386–2390. 24 Schroder NW, Hermann C, Hamann L, Gobel UB, Hartung T, 11 Choi EH, Foster CB, Taylor JG, Erichsen HC, Chen RA, Walsh TJ Schumann RR. High frequency of polymorphism Arg753Gln of the et al. Association between chronic disseminated candidiasis in Toll-like receptor-2 gene detected by a novel allele-specific PCR. adult acute leukemia and common IL4 promoter haplotypes. J Mol Med 2003; 81: 368–372. J Infect Dis 2003; 187: 1153–1156. 25 Packer BR, Yeager M, Staats B, Welch R, Crenshaw A, Kiley M 12 Stuber F, Petersen M, Bokelmann F, Schade U. A genomic et al. SNP500Cancer: a public resource for sequence validation polymorphism within the tumor necrosis factor influences and assay development for genetic variation in candidate genes. plasma tumor necrosis factor-alpha concentrations and outcome of Nucleic Acids Res 2004; 32 (Database issue): D528–D532. patients with severe sepsis [see comments]. Crit Care Med 1996; 26 Boot RG, Renkema GH, Verhoek M, Strijland A, Bliek J, de 24: 381–384. Meulemeester TM et al. The human chitotriosidase gene nature 13 Fishman D, Faulds G, Jeffery R, Mohamed-Ali V, Yudkin JS, of inherited enzyme deficiency. J Biol Chem 1998; 273: Humphries S et al. The effect of novel polymorphisms in the 25680–25685. interleukin-6 (IL-6) gene on IL-6 transcription and plasma IL-6 27 Lorenz E, Mira JP, Cornish KL, Arbour NC, Schwartz DA. A novel levels, and an association with systemic-onset juvenile chronic polymorphism in the toll-like receptor 2 gene and its potential arthritis. J Clin Invest 1998; 102: 1369–1376. association with staphylococcal infection. Infect Immun 2000; 68: 14 Harding D, Dhamrait S, Millar A, Humphries S, Marlow N, 6398–6401. Whitelaw A et al. Is interleukin-6 À174 genotype associated with 28 Wacholder S, Chanock S, Garcia-Closas M, El Ghormli L, the development of septicemia in preterm infants? Pediatrics 2003; Rothman N. Assessing the probability that a positive report is 112: 800–803. false: an approach for molecular epidemiology studies. J Natl 15 Jiang ZD, Okhuysen PC, Guo DC, He R, King TM, DuPont HL Cancer Inst 2004; 96: 434–442. et al. Genetic susceptibility to enteroaggregative Escherichia coli 29 Benjamini Y, Hochberg Y. Controlling the false discovery rate: a diarrhea: polymorphism in the interleukin-8 promotor region. practical and powerful approach to multiple testing. J R Stat Soc J Infect Dis 2003; 188: 506–511. 1995; 57: 289–300. 16 Rocha V, Franco RF, Porcher R, Bittencourt H, Silva Jr WA, 30 Kalbfleisch JD, Prentice RL. The Statistical Analysis of Failure Time Latouche A et al. Host defense and inflammatory gene poly- Data. John Wiley: New York, 1980, 163–188. morphisms are associated with outcomes after HLA-identical 31 Gray RJ. A class of K-sample tests for comparing the cumulative sibling bone marrow transplantation. Blood 2002; 100: incidence of a competing risk. Ann Stat 1988; 16: 1141–1154. 3908–3918. 32 Schluter B, Raufhake C, Erren M, Schotte H, Kipp F, Rust S et al. 17 Choi EH, Zimmerman PA, Foster CB, Zhu S, Kumaraswami V, Effect of the interleukin-6 promoter polymorphism (À174 G/C) on Nutman TB et al. Genetic polymorphisms in molecules of innate the incidence and outcome of sepsis. Crit Care Med 2002; 30: immunity and susceptibility to infection with Wuchereria bancrofti 32–37. in South India. Genes Immun 2001; 2: 248–253. 33 Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel 18 Norris CF, Surrey S, Bunin GR, Schwartz E, Buchanan GR, B et al. The structure of haplotype blocks in the . McKenzie SE. Relationship between Fc receptor IIA polymorphism Science 2002; 296: 2225–2229. and infection in children with sickle cell disease. J Pediatr 1996; 34 Stram DO, Haiman CA, Hirschhorn JN, Altshuler D, Kolonel LN, 128: 813–819. Henderson BE et al. Choosing haplotype-tagging SNPS based on 19 Foster CB, Lehrnbecher T, Mol F, Steinberg SM, Venzon DJ, Walsh unphased genotype data using a preliminary sample of unrelated TJ et al. Host defense molecule polymorphisms influence the risk subjects with an example from the Multiethnic Cohort Study. Hum for immune-mediated complications in chronic granulomatous Hered 2003; 55: 27–36. disease. J Clin Invest 1998; 102: 2146–2155. 35 Wacholder S, Rothman N, Caporaso N. Population stratification 20 Agnese DM, Calvano JE, Hahm SJ, Coyle SM, Corbett SA, Calvano in epidemiologic studies of common genetic variants and SE et al. Human toll-like receptor 4 mutations but not CD14 cancer: quantification of bias. J Natl Cancer Inst 2000; 92: polymorphisms are associated with an increased risk of Gram- 1151–1158. negative infections. J Infect Dis 2002; 186: 1522–1525. 36 Lehrnbecher T, Foster C, Vazquez N, Mackall CL, Chanock SJ. 21 Meier A, Kirschning CJ, Nikolaus T, Wagner H, Heesemann J, Ebel Therapy-induced alterations in host defense in children receiving F. Toll-like receptor (TLR) 2 and TLR4 are essential for Aspergillus- chemotherapy. J Ped Hematol Oncol 1997; 19: 399–417.

Leukemia