MAJOR ARTICLE

Chemotherapy Treatment in Pediatric Patients with Acute Myeloid Leukemia Receiving Antimicrobial Prophylaxis Leads to a Relative Increase of Colonization with Potentially Pathogenic Bacteria in the Gut
Michel J. van Vliet,1 Wim J. E. Tissing,1 Catharina A. J. Dun,1 Nico E. L. Meessen,2 Willem A. Kamps,1 Eveline S. J. M. de Bont,1 and Hermie J. M. Harmsen2
1

Department of Pediatric Oncology and Hematology, Beatrix Childrens Hospital, and 2Department of Medical Microbiology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands
Downloaded from cid.oxfordjournals.org by guest on June 9, 2011

Background. Normally, humans are protected against infections by their anaerobic intestinal microorganisms providing colonization resistance. In immunocompromised patients, the endogenous intestinal gram-positive and gram-negative pathogens often cause infectious complications. Therefore, we analyzed the effect of chemotherapy treatment and antimicrobial prophylaxis on intestinal bacterial populations (microbiota) among pediatric patients with acute myeloid leukemia who are prone to intestinal mucositis and infections. Methods. During 36 chemotherapy cycles, fecal samples were collected from pediatric patients with acute myeloid leukemia. Fecal bacterial populations were analyzed by polymerase chain reaction denaturing gradient gel electrophoresis ngerprinting. Fluorescent in situ hybridization analysis with specic bacterial oligonucleotide probes was used to quantify the fecal bacteria. Results. During chemotherapy treatment, the total number of bacteria in fecal samples was 109 per gram of dry weight feces, which was 100-fold lower than than in healthy control samples. Fluorescent in situ hybridization analysis showed that this decrease was the result of an up to 10,000-fold decrease in anaerobic bacteria, partly compensated for by a 100-fold increase in potentially pathogenic enterococci. Additional experiments showed that both prophylactic and therapeutic use of antibiotics could not sufciently explain the tremendous changes in intestinal microbial composition. In vitro tests showed a direct bacteriostatic effect of chemotherapeutics. Conclusions. Patients with acute myeloid leukemia treated with chemotherapy and prophylactic antibiotics are unable to maintain colonization resistance because of a decrease in anaerobic bacteria and an increase in potentially pathogenic aerobic enterococci. We hypothesize that this disturbance in the balance between anaerobic and aerobic bacteria will further increase the risk of gram-positive aerobic infections among immunocompromised patients with cancer. Infections and inammatory complications remain among the most-encountered serious complications of chemotherapy treatment among patients with cancer, accounting for morbidity and mortality, despite the use of prophylactic antibiotics and early empirical use of broad-spectrum antibiotics for febrile neutropenia [1]. Evidence indicating an endogenous origin of gram-negative and gram-positive bacteremia is increasing [2, 3]. To prevent the invasion of endogenous bacteria from the oral cavity and the gastrointestinal tract, 3 defense mechanisms are considered to be relevant: innate immunity, mechanical mucosal barrier, and colonization resistance. Since 1966, it has been known that the main parameters in innate immunity are the duration and nadir of neutropenia, both of which are associated with an increased risk of infection [4, 5]. Nowadays, attention is drawn toward mucositis, as caused by chemo-

Received 4 November 2008; accepted 15 February 2009; electronically published 10 June 2009. Reprints or correspondence: Dr. Wim J. E. Tissing, Beatrix Childrens Hospital, University Medical Center Groningen, PO Box 30.001, 9700 RB Groningen, The Netherlands (w.j.e.tissing@bkk.umcg.nl). Clinical Infectious Diseases 2009; 49:26270 2009 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2009/4902-0016$15.00 DOI: 10.1086/599346

262 CID 2009:49 (15 July) Vliet et al.

Figure 1. Patient samples collected at the beginning of treatment (early) and later during treatment (late) in 4 consecutive chemotherapy cycles. Values for number of patients sampled reect the number of patients with samples/number of possible patients. ADE I, chemotherapy cycle I; ADE II, chemotherapy cycle II; MACE, chemotherapy cycle III; MidAC, chemotherapy cycle IV.

therapy and radiotherapy, as an independent risk factor for infectious complications [6, 7]. This article focuses on colonization resistance, the third defense mechanism for the prevention of translocation of enteropathogenic bacteria. Colonization resistance is the benecial effect of the predominantly anaerobic resident microbiota of the gastrointestinal tract, which, by their sheer numbers, provide resistance toward invasion and/or outgrowth of potentially pathogenic bacteria from both the environment and the hosts own intestine [8, 9]. An adequate characterization of intestinal microbiota is necessary to investigate the supposed relationship among colonization resistance, mucositis, intestinal inammation, and infection. Previous studies have investigated the effect of chemotherapeutics and/or antibiotics on intestinal microbiota [1012]. However, most of these studies were hampered by the restrictions of culturing techniques, because a large fraction of intestinal bacteria has never been cultured and thus escape detection. The introduction of DNA-based techniques has made it possible to analyze the hosts intestinal microbiota independent of culturing techniques [13, 14]. Denaturing gradient gel electrophoresis (DGGE) is a sequence-dependent method to separate amplied microbial DNA. The resulting ngerprint can be used to qualitatively analyze and compare samples containing a diversity of microbiota. Fluorescent in situ hybridization (FISH) with specic oligonucleotide probes aimed at bacterial ribosomal RNA can be used to determine specic bacterial groups quantitatively. Changes in the intestinal microbiota during chemotherapy treatment might inuence colonization resistance, thereby increasing the risk of infection. Therefore, we characterized the qualitative and quantitative changes in the intestinal microbiota and their causes, using DNA-based techniques, among a cohort of pediatric patients with acute myeloid leukemia (AML), known for their high risk of infection and gastrointestinal complications. MATERIALS AND METHODS Subjects. All consecutive pediatric patients with AML diagnosed from 1 October 2005 through 1 October 2006 were in-

cluded in this study, after approval by the local medical ethics committee and informed consent according to the institutional guidelines. All patients were treated according to the AML-97treatment protocol of the Dutch Childhood Oncology Group, which is similar to the United Kingdom Medical Research Council (UK MRC-12) protocol (i.e., high-dose cytarabine, daunorubicine, and etoposide [ADE I and ADE II]; amsacrine, high-dose cytarabine, and etoposide [MACE]; and mitoxantrone and high-dose cytarabine [MidAC]) [15]. All patients received antibiotic prophylaxis aimed against gram-negative bacteria and fungi (treatment with oral colistin, neomycin, and amphotericin B or with ciprooxacin and itraconazol), viridans group streptococci (oral phenethicillin), and Pneumocystis jiroveci (oral cotrimoxazol). Febrile neutropenia was empirically treated with both vancomycin and ceftazidime. Feces was collected during 4 consecutive chemotherapy cycles (ADE I, ADE II, MACE, and MidAC; gure 1). Early samples were collected on day 2 (25th75th percentiles, days 13) after the start of a cycle. Late samples were collected on day 11 (25th75th percentiles, days 913). An end-of-treatment sample was collected at least 6 weeks after the end of treatment. To show the high intraindividual similarity between fecal samples, 11 samples from a single healthy volunteer were collected during a 2-month period. Patients FISH results were compared with the FISH results for fecal samples from healthy volunteers, previously collected and analyzed in our laboratory [16]. To determine the effect of phenethicillin treatment on intestinal microbiota, 4 healthy volunteers took phenethicillin at a dosage of 500 mg twice a day for 7 days. Fecal samples were collected before and after treatment. Fecal samples. For storage, aliquots were frozen at -80 C. One aliquot of each sample was dried using a freeze dryer (Christ Alpha; Salm and Kipp) to determine the wet-to-dry weight ratio, which was needed to calculate the number of bacteria per gram of dry weight feces, to compensate for a dilution effect due to diarrhea. DNA extraction from fecal samples. DNA was extracted using a DNA stool mini kit (Qiagen), in accordance with the manufacturers protocol, with 2 minor modications. To imIntestinal Microbiota in Patients with AML CID 2009:49 (15 July) 263

Downloaded from cid.oxfordjournals.org by guest on June 9, 2011

Table 1. Oligonucleotide probes used for uorescent in situ hybridization.

Target Most bacteria (universal probe) Anaerobic bacteria Bacteroides or Prevotella species Eubacterium rectale and Clostridium coccoides group Faecalibacterium prausnitzii Bidobacteria Aerobic bacteria Enterbacteriaceae Enterococcus faecalis Enterococcus faecium Streptococci and lactococci group Probe Eub338 Bac303 Erec482 Fprau645 Bif164 Ec1531 En3 Enfm2 Strc493 Target sequence, 3 r5 TGAGGATGCCCTCCGTCG TTCCAGGGGGTGTAACC GCCATGRACTGATTCTTCG CAAAAAGAACTCATCACGTCTCC GTCGTCGGCGCCATTATG ACTACTGCTCCGTGATGCCAC ACGTGAGTTAACCTTTCTCC GGCCTTTTTCTCCTCACCG GGTCTTTCCCTGCCGATTG Reference [20] [21] [22] [23] [24] [25] [26] [26] [22]

prove the lysis of gram-positive bacteria, the temperature was increased to 95 C. Second, the amount of elution buffer was decreased to 30 mL, to increase the nal DNA concentration. Polymerase chain reaction (PCR). For the amplication of bacterial DNA, the universal bacterial primers described by Watanabe et al. [17] were used. To make the PCR product eligible for DGGE, a previously described GC clamp was added to the forward primer [13]. The PCR reaction mixture contained 5 mL reaction buffer (Fermentas Life Sciences), 1.25 U Taq polymerase (Fermentas Life Sciences), 1.5 mmol/L MgCl2, 400 nmol/L of both primers, 200 mmol/L of deoxyribonucleotide triphosphate, 0.5 mL bovine serum albumin (10 mg/mL; New England Biolabs), and 20 ng of template DNA in 50 mL. The temperature prole included 2 min at 94 C, followed by 40 cycles at 94 C for 45 s, 49 C for 30 s, and 72 C for 1 min,

followed by 5 min at 72 C. After gel electrophoresis, the ethidium bromidestained PCR products were analyzed by ultraviolet light. DGGE. DGGE of the PCR products was performed as described by Muyzer et al. [13], using a PhorU system (Ingeny). Depending on the concentration, 115 mL of PCR product was loaded on a 8% (weight/volume) polyacrylamide gel with a 10%60% denaturing gradient (100% denaturant equals 7 mol/ L urea and 40% formamide). Electrophoresis was performed for 16 h at 120 V at 60 C. Gels were stained with silver nitrate as described elsewhere [18]. FISH. Fecal samples were 010 times diluted in phosphate buffered saline (8 g NaCl, 0.2 g KCl, 1.44 g Na2HPO4, and 0.24 g KH2PO4, each per liter), dependent on liquidity. FISH was performed on glass slides as described elsewhere [19], except

Downloaded from cid.oxfordjournals.org by guest on June 9, 2011

Table 2. In vitro sensitivity of bacterial strains to daunorubicine, etoposide, and cytarabine.

Daunorubicine Species Aerobic bacteria Streptococcus mitis Enterococcus faecalis Enterococcus faecium Escherichia coli Anaerobic bacteria Bacteroides distasonis Bacteroides vulgatus Bacteroides fragilis Clostridium difcile Clostridium ramosum Lactobacillus acidophilus Bifodobacterium animalis 0.85 0.90 0.82 0.96 0.76 0.92 0.97 0.75 0.77 0.64 0.90 0.52 0.94 0.69 0.36 0.99 1.04 0.03 0.90 0.00 0.66 0.36 0.41 0.88 0.03 0.06 0.01 0.35 0.92 0.97 0.97 0.95 1.15 0.90 0.83 0.98 0.95 0.99 0.95 1.06 0.92 0.82 0.3 mg/mL 0.07 0.98 0.95 ND 3 mg/mL 0.03 0.95 0.91 1.05 Etoposide 4.5 mg/mL 0.04 0.99 0.98 ND 45 mg/mL 0.03 0.03 1.02 0.93 Cytarabine 4.5 mg/mL 0.87 1.03 0.97 0.94 20 mg/mL 0.86 1.02 1.00 0.95

NOTE. Data are ratio of bacterial growth in samples containing the different chemotherapeutics and in control samples containing the same culturing medium but without chemotherapeutics. ND, not determined.

264 CID 2009:49 (15 July) Vliet et al.

Downloaded from cid.oxfordjournals.org by guest on June 9, 2011

Figure 2. A, Example of polymerase chain reaction (PCR) denaturing gradient gel electrophoresis (DGGE) ngerprints of fecal microbiota of samples collected from 1 patient during chemotherapy cycles IIV, showing decreasing bacterial diversity and low similarity in species composition during treatment. Consecutive samples are compared with the end-of-treatment samples, resulting in the depicted percentage similarities. AB, antibiotics; N, the sample collected 6 weeks after the end of treatment. B, PCR-DGGE ngerprints of fecal microbiota of 11 samples collected serially during a 2-month period from a single healthy control individual. DGGE ngerprints show a high diversity of bacterial species in every lane and a high similarity between the consecutive lanes. The percentages reect the measured similarity between the different microbial ngerprints, compared with that of sample 1.

that xed samples were diluted in phosphate buffered saline instead of Tween 5%. Glass slides were hybridized for 16 h with the probes listed in table 1 or for 23 h in the case of the Bac303 and Ec1531 probes [2026]. Before hybridization with En3,Enfm2, and Strc493, samples were treated with lysozyme to permeabilize the cell wall, as described elsewhere [22]. After incubation, slides were washed, air dried, and mounted with Vectashield (Vector Laboratories). Bacteria were counted visually using an Olympus BH2 epiuorescence microscope. At least 10 microscopic elds were counted or as many elds as needed to obtain a total number of at least 300 positive bacteria (to a maximum of 25 elds). The total number of bacteria was

determined by the bacterial count with the universal Eub338 probe, enumerated with the counts of the Strc493, Enfm2, and En3 probes, because their target species do not hybridize with the Eub338 probe without permeabilization. Obligate anaerobic bacteria are referred to as anaerobes. Facultative aerobic and oxygen tolerant bacteria are referred to as aerobes. Chemotherapeutics and bacterial growth. Bacteria, previously isolated from clinical samples, were cultured in their appropriate medium by use of standard laboratory procedures [27]. Inoculation was done by diluting bacterial cultures 100 times in their corresponding medium. Anaerobic bacteria were incubated and cultured using an
Intestinal Microbiota in Patients with AML CID 2009:49 (15 July) 265

Table 3. Similarity in DNA ngerprints of microbiota at different time points.

Similarity between Chemotherapy early and late samples, cycle, time point median (range) ADE I Early 1 Late 1 ADE II Early 2 Late 2 MACE Early 3 Late 3 MidAC Early 4 Late 4 16.8 (2.482.3) 51.5 (8.872.5) 31.9 (22.770.0) 64.7 (20.286.1) Similarity with end-oftreatment sample, median (range) 34.7 (4.142.0) 16.6 (2.431.3) 24.9 (3.557.4) 21.0 (0.032.3) 20.1 (4.759.0) 27.6 (10.347.1) 31.6 (15.744.3) 36.5 (10.669.1)

rate of intestinal mucosal barrier injury was found for chemotherapy cycle I (6 of 9 patients) and cycle III (5 of 6 patients).
Comparison of Intestinal Microbiota PCR-DGGE Fingerprints

NOTE. Time points are the beginning and end of 4 consecutive chemotherapy cycles, and they identify whether samples were taken before (early) or after (late) the 4 consecutive chemotherapy courses. The similarity between the samples collected at these 8 time points are compared with the end-oftreatment sample. ADE I and ADE II, high-dose cytarabine, daunorubicine, and etoposide; MACE, amsacrine, high-dose cytarabine, and etoposide; MidAC, mitoxantrone and high-dose cytarabine.

anaerobic hood (Concept 400; Ruskinn Technology Limited). To determine the effect on bacterial growth, different concentrations of the chemotherapeutics were added to the culturing media, as listed in table 2. The concentrations of added chemotherapeutics were determined on the basis of relevant plasma levels [2830]. Bacterial growth was quantied by the increase in absorbance of the samples at 492 nm, determined after 24 h of incubation at 37 C, with use of a photospectrometer (Vitatron Nederland), relative to a control, which contained the appropriate culturing medium and bacteria but no chemotherapeutic agents. Data analysis. Intrapatient differences in numbers of bacteria at different time points were compared using the Wilcoxon rank test for paired data. Nonrelated samples were tested using the Mann-Whitney U test, with use of SPSS, version 16.0 (SPSS). A P value !.05 was considered to be statistically signicant. For the analysis of fecal DGGE ngerprints, the similarity was calculated on the basis of a Pearson correlation coefcient with use of Gelcompar II (Applied-maths). RESULTS
Subjects

The diversity of intestinal microbiota in serially obtained fecal samples was severely limited in the early and late samples, compared with the end-of-treatment sample (gure 2A). Each lane of a PCR-DGGE gel represents a microbial ngerprint of a fecal sample; each band within a lane corresponds to 1 bacterial species, although different species may sometimes be represented by 1 band. A band at the same height in different lanes is considered to be the same bacterial species. During chemotherapy treatment, a large decrease in bacterial diversity was seen. Before a new chemotherapy course, the bacterial diversity was restored partially. The median similarity between early and late samples in the patient population was low, in the range 16.8%64.7% (table 3), which was in clear contrast to the high similarity found among 11 samples collected in a single healthy volunteer (gure 2B). The measured similarity between the early and late samples, on the one hand, and the end-of-treatment sample, on the other hand, was even lower, in the range 16.6%36.5% (table 3). The ngerprint of the end-of-treatment sample showed a diversity comparable to that for a healthy control sample (data not shown). Because of the high interindividual variety in PCRDGGE proles, patient samples were not compared with samples from healthy controls [32]. The effect of empirical treatment with broad-spectrum antibiotics. Analyzing the 12 chemotherapy cycles in which no therapeutic antibiotics were given, no signicant difference was found in median similarity between early and late samples (median, 52.4%; range, 11.586.1), compared with the median similarity in the cycles with empirical antibiotic treatment (median, 31.9%; range, 8.872.5). The effect of phenethicillin on gut microbiota. In 4 healthy control individuals, we studied whether phenethicillin treatment might be related to a decreased microbial diversity. In comparison of fecal microbial ngerprints for samples collected on day 0 and day 7, a median similarity of 74.2% (range, 61.2%80.5%) was found, which is higher than the median similarity for the patient samples.
Quantitative Analysis of Intestinal Microbiota by FISH

Downloaded from cid.oxfordjournals.org by guest on June 9, 2011

During 36 chemotherapy cycles, a total of 72 fecal samples were collected from 9 pediatric patients with AML who each received 4 consecutive chemotherapy courses (gure 1). Using National Cancer Institute adverse events criteria [31], intestinal mucosal barrier injury grade IIII was found after 55% of chemotherapy cycles (grade IIIIV was found in 4 [20%] of 20). The highest
266 CID 2009:49 (15 July) Vliet et al.

At all time points during treatment (early and late samples), the total number of bacteria was signicantly lower, compared with the healthy control samples (P values ranged from .000 to .001) (gure 3A). No difference in total numbers was found between end-of-treatment samples and healthy control samples, indicating a restored total number of bacteria at 6 weeks after the last chemotherapy cycle. No signicant differences were seen between the consecutive early and late samples.

Downloaded from cid.oxfordjournals.org by guest on June 9, 2011

Figure 3. Number of bacteria in fecal samples collected at the beginning of treatment (early) and later during treatment (late) during 4 consecutive chemotherapy cycles, measured using uorescent in situ hybridization analysis. A, Total bacteria. B, Anaerobic bacteria. C, Enterobacteriaceae. D, Enterococcus faecium and Enterococcus faecalis. E, Streptococci and lactococci. All numbers for early and late time points are signicantly different from those for healthy control samples (P ! .05 ), except for enterococci in the late 1 sample (P p .08 ). Signicant differences between end-of-treatment samples and healthy control samples were found in enterococci and streptococci but not in the total number of bacteria, anaerobes, or Enterobacteriaceae.

Number of anaerobic bacteria. The median number of 4 predominant groups of intestinal anaerobic bacteriaBacteroides species, Clostridium cluster XIVa, Faecalibacterium prausnitzii, and Bidobacterium speciesdecreased 30006000-fold in samples during treatment, compared with the healthy control samples (gure 4). Depending on the different time points, this resulted in a 7020,000-fold decrease in the total number of anaerobic bacteria (P values ranged from .000 to .004) (gure 3B). A recovery was found for both Clostridium XIVa and F. prausnitzii, at 6 weeks after treatment (gure 4). In contrast,

the total numbers of both Bacteroides and Bidobacterium species were still 10300-fold lower in the end-of-treatment samples, compared with the healthy control samples (gure 4). The total number of anaerobic bacteria did not differ signicantly between the end-of-treatment samples and the healthy control samples. Number of aerobic bacteria. In agreement with the prophylactic use of ciprooxacin, no gram-negative Enterobacteriaceae were detected in any of the fecal samples collected during treatment (gure 3C). The number of enterococci was
Intestinal Microbiota in Patients with AML CID 2009:49 (15 July) 267

Downloaded from cid.oxfordjournals.org by guest on June 9, 2011

Figure 4. Number of anaerobic bacteria in fecal samples collected at the beginning of treament (early) and later during treatment (late) during 4 consecutive chemotherapy cycles, measured using uorescent in situ hybridization analysis. A, Clostridium cluster XIVa. B, Bidobacteria. C, Faecalibacterium prausnitzii. D, Bacteroides species. All numbers for early and late time points are signicantly different from those for healthy control samples (P ! .05). Signicant differences between end-of-treatment samples and healthy control samples were found in bidobacteria and Bacteroides species but not in Clostridium XIVa and F. prausnitzii.

signicantly higher in patient samples, compared with healthy control samples, at 7 of 8 time points during treatment (samples early 1, early 2, late 2, early 3, late 3, early 4, and late 4) and the end-of-treatment sample (P values ranged from .001 to .038) (gure 3D). In contrast, a 1001000-fold decrease in the number of streptococci was found in patient samples, compared with healthy control samples (P values ranged from .000 to .016) (gure 3E). The number of streptococci had still not recovered at 6 weeks after treatment. It is noteworthy that microscopic analysis showed that some samples contained only aerobic cocci; the fecal microbiota contained up to 88% enterococci or up to 100% streptococci, which was consistent with DGGE ngerprints showing 1 dominant species.
Chemotherapy and Bacterial Growth

bacteria (table 3). Use of a higher concentration of etoposide, but not daunorubicine, resulted in a further decrease in bacterial growth. In contrast, cytarabine showed no effect on the growth of the tested bacterial strains. DISCUSSION We showed devastating effects of anticancer treatment on intestinal microbiota in patients with AML. The microbial diversity was low during treatment, and consecutive samples showed considerable variations in microbial ngerprints, as demonstrated using PCR-DGGE. This is in sharp contrast with healthy individuals, in whom a very stable intestinal microbiota is seen [32]. The diversity was lowest during chemotherapy cycles I and III, which also showed the highest incidences of intestinal toxicity. This nding is in line with data from the literature showing that these 2 chemotherapy cycles, in particular, cause intestinal toxicity and infection [33]. The dramatic decrease found in the number of anaerobic bacteria might have a large impact on colonization resistance. It might also result in an unfavorable effect on intestinal bu-

To determine whether the decrease in the numbers of bacteria was a result of chemotherapy, the sensitivity of different bacterial species to chemotherapeutics used in 11 chemotherapy cycledaunorubicine, etoposide, and cytarabinewas determined. In vitro, both daunorubicine and etoposide showed a negative effect on the growth of both anaerobic and aerobic
268 CID 2009:49 (15 July) Vliet et al.

tyrate production, an essential trophic factor for the formation of enterocytes and an important inhibitor of local inammation, which could further increase the risk of bacterial translocation and infection in neutropenic patients with mucosal barrier injury [34, 35]. The increase in enterococci during treatment for AML might imply an increased risk of infection with vancomycin-resistant enterococci, because all patients with AML are given empirical treatment with vancomycin during periods of febrile neutropenia [3638]. For this small group of patients, we could not test this hypothesis, making it a focus for further research. Several factors can be responsible for the decrease in both diversity and number of intestinal microbiota in patients with cancer: (1) nasal tube feeding, (2) (prophylactic) antibiotic treatment, and (3) chemotherapy itself [10, 30, 39]. First, prior research suggests that nutrition is, at best, partly responsible for the devastating effect on microbiota in pediatric patients with cancer [39, 40]. Second, in our study, the devastating effect on the intestinal microbiota cannot be contributed to solely by the use of prophylactic antibiotics, because the diversity of bacteria differed between the early and late samples in all 4 chemotherapy cycles, although prophylaxis was not discontinued between them. Moreover, in another study, only a small effect of oral ciprooxacin treatment on intestinal microbiota was found [41]. In addition, we found only a small effect of oral phenethicillin treatment on the composition of fecal microbiota in healthy volunteers. Thus, we conclude that the devastating effect on microbial colonization cannot be exclusively caused by the use of prophylactic antibiotics. The effects on intestinal microbiota also cannot be ascribed to the use of broad-spectrum antibiotics in cases of febrile neutropenia, because a low similarity in microbial ngerprints between early and late samples was found, irrespective of the use of these antibiotics. The third possible cause, chemotherapy itself, was tested in vitro. Both daunorubicine and etoposide showed a direct negative effect on bacterial growth. A direct effect of cytarabine was not seen, which might be explained by the fact that only its metabolites exert cytotoxic activity [42]. Unfortunately, with the currently available laboratory techniques, it is not possible to test chemotherapeutic levels in feces, which complicates the interpretation of the results found in this study. However, even the lowest dosages of chemotherapeutics tested inhibited bacterial growth, suggesting an effect on intestinal microbiota in vivo. The impact of chemotherapeutic drugs on colonization resistance will need further research. In this study, multiple statistical tests were performed for a small population of patients. This could introduce false-positive results. However, even after stringent Bonferroni correction for multiple testing, the decrease in anaerobic bacteria stayed statistically signicant, although changes in the numbers of both

streptococci and enterococci were no longer signicantly different between patient samples and healthy control samples. However, this did not change the overall conclusion of the studythe number of anaerobic bacteria decreases during treatment for AML, thereby increasing the proportion of aerobic and potentially enteropathogenic bacteria, such as enterococci. To our knowledge, we are the rst to demonstrate, using molecular detection techniques (DGGE and FISH) instead of classic culturing techniques (with their drawbacks), that the combination of chemotherapeutics, antibiotics, and articial feeding has a large effect on the intestinal microbiota. Further research will have to be done to show whether changes in bacterial colonization play a role in the development and maintenance of mucosal barrier injury, infection, and inammatory syndromes and whether there might be a role for the use of prebiotics, probiotics, and bacterial products, such as butyrate, to prevent mucosal barrier injury and its complications. In conclusion, a tremendous decrease in intestinal microbial diversity was found in patients with AML during treatment, with a disturbed balance between aerobic and anaerobic bacteria in favor of potentially pathogenic gram-positive aerobic cocci. These changes in microbiota lower the resistance to pathogen colonization. We expect that the diminished colonization resistance increases the risk of gram-positive aerobic infection in immunocompromised patients.
Acknowledgments
We thank D. Dimitropoulou and P. C. M. van den Huijssen for their technical assistance. Financial support. Foundation of Pediatric Oncology Groningen (research grant to M.J.v.V.). Potential conicts of interest. All authors: no conicts.

Downloaded from cid.oxfordjournals.org by guest on June 9, 2011

References
1. Hann I, Viscoli C, Paesmans M, Gaya H, Glauser M. A comparison of outcome from febrile neutropenic episodes in children compared with adults: results from four EORTC studies. International Antimicrobial Therapy Cooperative Group (IATCG) of the European Organization for Research and Treatment of Cancer (EORTC). Br J Haematol 1997; 99:5808. 2. Costa SF, Miceli MH, Anaissie EJ. Mucosa or skin as source of coagulase-negative staphylococcal bacteraemia? Lancet Infect Dis 2004; 4:27886. 3. Elting LS, Bodey GP, Keefe BH. Septicemia and shock syndrome due to viridans streptococci: a case-control study of predisposing factors. Clin Infect Dis 1992; 14:12017. 4. Bodey GP, Buckley M, Sathe YS, Freireich EJ. Quantitative relationships between circulating leukocytes and infection in patients with acute leukemia. Ann Intern Med 1966; 64:32840. 5. Bodey GP, Rodriguez V, Chang HY, Narboni. Fever and infection in leukemic patients: a study of 494 consecutive patients. Cancer 1978; 41:161022. 6. Blijlevens NM, Donnelly JP, De Pauw BE. Mucosal barrier injury: biology, pathology, clinical counterparts and consequences of intensive treatment for haematological malignancy: an overview. Bone Marrow Transplant 2000; 25:126978.

Intestinal Microbiota in Patients with AML CID 2009:49 (15 July) 269

7. Sonis ST, Oster G, Fuchs H, et al. Oral mucositis and the clinical and economic outcomes of hematopoietic stem-cell transplantation. J Clin Oncol 2001; 19:22015. 8. Stecher B, Hardt WD. The role of microbiota in infectious disease. Trends Microbiol 2008; 16:10714. 9. van der Waaij D, Berghuis-de Vries JM, Lekkerkerk L. Colonization resistance of the digestive tract in conventional and antibiotic-treated mice. J Hyg (Lond) 1971; 69:40511. 10. van der Waaij D. The digestive tract in immunocompromised patients: importance of maintaining its resistance to colonization, especially in hospital in-patients and those taking antibiotics. Antonie Van Leeuwenhoek 1984; 50:74561. 11. Stringer AM, Gibson RJ, Logan RM, Bowen JM, Yeoh AS, Keefe DM. Faecal microora and b-glucuronidase expression are altered in an irinotecan-induced diarrhoea model in rats. Cancer Biol Ther 2008; 7:191925. 12. Von Bultzingslowen I, Adlerberth I, Wold AE, Dahlen G, Jontell M. Oral and intestinal microora in 5-uorouracil treated rats, translocation to cervical and mesenteric lymph nodes and effects of probiotic bacteria. Oral Microbiol Immunol 2003; 18:27884. 13. Muyzer G, Dewaal EC, Uitterlinden AG. Proling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-amplied genes coding for 16S ribosomal RNA. Appl Environ Microbiol 1993; 59:695700. 14. Muyzer G. DGGE/TGGE a method for identifying genes from natural ecosystems. Curr Opin Microbiol 1999; 2:31722. 15. Gibson BES, Wheatley K, Hann IM, et al. Treatment strategy and longterm results in paediatric patients treated in consecutive UK AML trials. Leukemia 2005; 19:21308. 16. Harmsen HJ, Raangs GC, He T, Degener JE, Welling GW. Extensive set of 16S rRNA-based probes for detection of bacteria in human feces. Appl Environ Microbiol 2002; 68:298290. 17. Watanabe K, Kodama Y, Harayama S. Design and evaluation of PCR primers to amplify bacterial 16S ribosomal DNA fragments used for community ngerprinting. J Microbiol Methods 2001; 44:25362. 18. Sanguinetti CJ, Dias NE, Simpson AJ. Rapid silver staining and recovery of PCR products separated on polyacrylamide gels. Biotechniques 1994; 17:91421. 19. Jansen GJ, Wildeboer-Veloo AC, Tonk RH, Franks AH, Welling GW. Development and validation of an automated, microscopy-based method for enumeration of groups of intestinal bacteria. J Microbiol Methods 1999; 37:21521. 20. Amann RI, Binder BJ, Olson RJ, Chisholm SW, Devereux R, Stahl DA. Combination of 16S rRNA-targeted oligonucleotide probes with ow cytometry for analyzing mixed microbial populations. Appl Environ Microbiol 1990; 56:191925. 21. Manz W, Amann R, Ludwig W, Vancanneyt M, Schleifer KH. Application of a suite of 16S rRNA-specic oligonucleotide probes designed to investigate bacteria of the phylum cytophaga-avobacter-bacteroides in the natural environment. Microbiology 1996; 142:1097106. 22. Franks AH, Harmsen HJ, Raangs GC, Jansen GJ, Schut F, Welling GW. Variations of bacterial populations in human feces measured by uorescent in situ hybridization with group-specic 16S rRNA-targeted oligonucleotide probes. Appl Environ Microbiol 1998; 64:333645. 23. Suau A, Rochet V, Sghir A, et al. Fusobacterium prausnitzii and related species represent a dominant group within the human fecal ora. Syst Appl Microbiol 2001; 24:13945. 24. Langendijk PS, Schut F, Jansen GJ, et al. Quantitative uorescence in situ hybridization of Bidobacterium spp. with genus-specic 16S rRNA-targeted probes and its application in fecal samples. Appl Environ Microbiol 1995; 61:306975.

25. Poulsen LK, Lan F, Kristensen CS, Hobolth P, Molin S, Krogfelt KA. Spatial distribution of Escherichia coli in the mouse large intestine inferred from rRNA in situ hybridization. Infect Immun 1994; 62: 51914. 26. Waar K, Degener JE, van Luyn MJ, Harmsen HJ. Fluorescent in situ hybridization with specic DNA probes offers adequate detection of Enterococcus faecalis and Enterococcus faecium in clinical samples. J Med Microbiol 2005; 54:93744. 27. Murray P, Baron E, Jorgensen JH, Landry M, Pfaller M. Manual of medical microbiology. 9th ed. Washington, DC: ASM Press, 2007. 28. Bodet CA III, Jorgensen JH, Drutz DJ. Antibacterial activities of antineoplastic agents. Antimicrob Agents Chemother 1985; 28:4379. 29. Calame W, van der Waals R, Douwes-Idema N, Mattie H, van Furth R. Antibacterial effect of etoposide in vitro. Antimicrob Agents Chemother 1988; 32:14567. 30. Vetere A, Giuliano M, Pantosti A, Panichi G. In vitro activity of several cytostatic drugs against aerobic and anaerobic intestinal bacteria [in Italian]. Boll Ist Sieroter Milan 1984; 63:5059. 31. National Cancer Institute. Cancer therapy evaluation program. Bethesda, MD: US National Institutes of Health, 2006. Available at: http: //ctep.cancer.gov/protocolDevelopment/electronic_applications/docs/ ctcaev3.pdf. Accessed 8 August 2008. 32. Zoetendal EG, Akkermans AD, de Vos WM. Temperature gradient gel electrophoresis analysis of 16S rRNA from human fecal samples reveals stable and host-specic communities of active bacteria. Appl Environ Microbiol 1998; 64:38549. 33. Riley LC, Hann IM, Wheatley K, Stevens RF. Treatment-related deaths during induction and rst remission of acute myeloid leukaemia in children treated on the Tenth Medical Research Council acute myeloid leukaemia trial (MRC AML10). The MCR Childhood Leukaemia Working Party. Br J Haematol 1999; 106:43644. 34. Hamer HM, Jonkers D, Venema K, Vanhoutvin S, Troost FJ, Brummer RJ. Review article: the role of butyrate on colonic function. Aliment Pharmacol Ther 2008; 27:10419. 35. Pryde SE, Duncan SH, Hold GL, Stewart CS, Flint HJ. The microbiology of butyrate formation in the human colon. FEMS Microbiol Lett 2002; 217:1339. 36. Matar MJ, Tarrand J, Raad I, Rolston KV. Colonization and infection with vancomycin-resistant Enterococcus among patients with cancer. Am J Infect Control 2006; 34:5346. 37. Sakka V, Tsiodras S, Galani L, et al. Risk-factors and predictors of mortality in patients colonised with vancomycin-resistant enterococci. Clin Microbiol Infect 2008; 14:1421. 38. Worth LJ, Thursky KA, Seymour JF, Slavin MA. Vancomycin-resistant Enterococcus faecium infection in patients with hematologic malignancy: patients with acute myeloid leukemia are at high-risk. Eur J Haematol 2007; 79:22633. 39. Schneider SM, Le Gall P, Girard-Pipau F, et al. Total articial nutrition is associated with major changes in the fecal ora. Eur J Nutr 2000; 39:24855. 40. Whelan K, Judd PA, Preedy VR, Simmering R, Jann A, Taylor MA. Fructooligosaccharides and ber partially prevent the alterations in fecal microbiota and short-chain fatty acid concentrations caused by standard enteral formula in healthy humans. J Nutr 2005; 135: 1896902. 41. Donskey CJ, Hujer AM, Das SM, Pultz NJ, Bonomo RA, Rice LB. Use of denaturing gradient gel electrophoresis for analysis of the stool microbiota of hospitalized patients. J Microbiol Methods 2003; 54:24956. 42. Wiley JS, Taupin J, Jamieson GP, Snook M, Sawyer WH, Finch LR. Cytosine arabinoside transport and metabolism in acute leukemias and T cell lymphoblastic lymphoma. J Clin Invest 1985; 75:63242.

Downloaded from cid.oxfordjournals.org by guest on June 9, 2011

270 CID 2009:49 (15 July) Vliet et al.