Oral Microbiota Distinguishes Acute Lymphoblastic

Leukemia Pediatric Hosts from Healthy Populations

Yan Wang1., Jing Xue1., Xuedong Zhou1, Meng You1, Qin Du1, Xue Yang2, Jingzhi He1, Jing Zou3,
Lei Cheng1, Mingyun Li1, Yuqing Li1, Yiping Zhu2, Jiyao Li1, Wenyuan Shi4, Xin Xu1*
1 State Key Laboratory of Oral Diseases, West China Hospital of Stomatology, Sichuan University, Chengdu, China, 2 Department of Pediatric Hematology and Oncology,
West China Second University Hospital, Sichuan University, Chengdu, China, 3 Department of Pediatric Dentistry, West China Hospital of Stomatology, Sichuan University,
Chengdu, China, 4 UCLA School of Dentistry, Los Angeles, California, United States of America

Abstract
In leukemia, oral manifestations indicate aberrations in oral microbiota. Microbiota structure is determined by both host and
environmental factors. In human hosts, how health status shapes the composition of oral microbiota is largely unknown.
Taking advantage of advances in high-throughput sequencing, we compared the composition of supragingival plaque
microbiota of acute lymphoblastic leukemia (ALL) pediatric patients with healthy controls. The oral microbiota of leukemia
patients had lower richness and less diversity compared to healthy controls. Microbial samples clustered into two major
groups, one of ALL patients and another of healthy children, with different structure and composition. Abundance changes
of certain taxa including the Phylum Firmicutes, the Class Bacilli, the Order Lactobacillales, the Family Aerococcaceae and
Carnobacteriaceae, as well as the Genus Abiotrophia and Granulicatella were associated with leukemia status. ALL patients
demonstrated a structural imbalance of the oral microbiota, characterized by reduced diversity and abundance alterations,
possibly involved in systemic infections, indicating the importance of immune status in shaping the structure of oral
microbiota.

Citation: Wang Y, Xue J, Zhou X, You M, Du Q, et al. (2014) Oral Microbiota Distinguishes Acute Lymphoblastic Leukemia Pediatric Hosts from Healthy
Populations. PLoS ONE 9(7): e102116. doi:10.1371/journal.pone.0102116
Editor: Yolanda Sanz, Instutite of Agrochemistry and Food Technology, Spain
Received December 30, 2013; Accepted June 15, 2014; Published July 15, 2014
Copyright: ß 2014 Wang et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by the International S&T Cooperation Program of China (grant number: 2011DFA30940), the National Basic Research Program
of China (‘‘973 Pilot Research Program,’’ grant Number: 2011CB512108), the National Science & Technology Pillar Program during the 12th Five-year Plan Period
(grant number: 2012BAI07B03), National Natural Science Foundation of China (grant number: 81200782) and Doctoral Fund of Ministry of Education of China
(grant number: 20120181120002). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* Email: [email protected]
. These authors contributed equally to this work.

Introduction information can be obtained on the oral microbiota per se under

the diseased condition. To determine the role of oral cavity
Leukemia is a cancer of the early blood-forming cells. Acute microbiota to local and systemic complications in ALL patients, it
lymphoblastic leukemia (ALL), a malignant disorder of lymphoid is first necessary to generate a more complete picture of the oral
progenitor cells, is the most common type of leukemia among cavity microbiota population that is specific to this disease.
children, accounting for 75% of all childhood leukemia and 25% In ALL patients, lymphoid progenitor cells are affected, which
of all malignancy in childhood [1]. Among some individuals can partially impair the immune system of the host [14,15]. It is
leukemia first manifests in oral cavity [2,3]. Oral manifestations known that microbiota structure within a host is determined by
that frequently occur in leukemia patients include gingival both host and environmental factors. If any one of these factors is
bleeding, oral ulceration, gingival enlargement, candidiasis and greatly perturbed, a drastic composition shift is expected in the
periodontitis [4,5,6]. Oral microbes are believed to be involved in oral microbiota, and disease occurs. In the human host, the role of
the occurrence or exacerbation of such complications [7]. Certain
health status in shaping the composition of oral microbiota is still
oral microbiota have been shown to contribute to septicemia,
largely unknown. Characterization of the oral microbiome in ALL
which might delay antineoplastic treatment, compromise treat-
patients provides information about oral microbiota-host interac-
ment efficiency, or even jeopardize the patients’ life [8,9].
tions in immunocompromised individuals, and thus contributes to
Therefore, an adequate treatment of oral lesions could lead to a
better management of oral and systemic complications associated
more favorable resolution of both oral and systemic diseases. At
with immunodeficiency.
present, limited knowledge is available on the oral microbiota of
To elucidate how a compromised host immune status leads to a
leukemia patients. Previous studies dependent on a culture-based
perturbed microbial homeostasis necessitates a thorough compar-
approach have mainly focused on limited cultivable bacterial
species and failed to determine the holistic pattern of oral ison of the oral microbial composition of ALL patients and ALL-
microbiota in leukemia patients [10,11,12,13]. Furthermore, free population. Remarkable advances in high-throughput se-
because most previous studies were primarily focused on the quencing have recently improved practicality in analysis of
effect of antineoplastic treatment on oral microbes, limited microbiota from a variety of biological and environmental samples

PLOS ONE | www.plosone.org 1 July 2014 | Volume 9 | Issue 7 | e102116

 Oral Microbiota of Leukemia Affected Children

under various conditions. Whereas conventional culture-depen- criteria are presented in Table 1. Detailed clinical data of each
dent approaches underestimate microbial composition, new subject is in Table S1.
culture-independent molecular techniques are capable of investi-
gating entire bacterial communities and characterizing the Sample Collection and Oral Examination
biodiversity of oral microbiota. High-throughput sequencing has For each of the 26 subjects, microbial samples were collected at
been widely and successfully applied to the exploration of oral the same time of day, approximately 2 hours after breakfast, using
microbial diversity in health [16,17,18] and disease-associated the method mentioned in the Manual of Procedures for Human
communities from those associated with dental caries [19,20], Microbiome Project (http://hmpdacc.org/tools_protocols/
periradicular lesions [21], atherosclerosis [22], and head-neck tools_protocols.php) with minor modifications. Briefly, the sam-
tumor undergone radiotherapy [23,24]. However, application of pling sites, the teeth in upper right and lower left quadrants or
these techniques in investigating the oral microbiome under upper left and lower right quadrants, were isolated with cotton
immunocompromised conditions has so far been limited. In this rolls and dried before sampling. A sterile Gracey curette was used
study, we characterized the biodiversity of supragingival plaque to collect a pooled supragingival plaque sample from the mesial
microbiota in pediatric clinic ALL patients using a high- surfaces of each of these teeth in turn. The collected plaque
throughput sequencing technique (454 pyrosequencing). The oral samples were released from the curette by agitation in 700 ml of
microbiota compositional profiles were compared to those of ALL- TE buffer (10 mM Tris-Cl [pH 7.5] and 1 mM EDTA). The
free healthy subjects. microbial samples were immediately transported on ice to the
laboratory and stored at 280uC until further DNA extraction and
Methods pyrosequencing analysis.
Oral conditions of each subject, including dentition status,
Ethics Statement number of teeth, plaque index, caries status, and presence or
This study was approved by the Ethics Committee of State Key absence of gingivitis/periodontitis, were recorded after sampling.
Laboratory of Oral Diseases, Sichuan University, Chengdu, The presence of caries was further confirmed by periapical
China, and conducted in compliance with the Declaration of radiographs. Plaque-Check+pH kit (GC Corporation, Japan) was
Helsinki. Written informed consent was obtained from the parents used for testing plaque fermentation based on a colourimetric
or guardians of all subjects before the study. readout of the pH of supragingival plaque corresponding to each
sampling site [25].
Study Population
Potential study subjects of the leukemia group were selected DNA Extraction
from a group of patients who were newly diagnosed with acute Bacteria were pelleted from dental plaque samples by centri-
lymphoblastic leukemia (ALL) in the Department of Pediatric fugation (Thermo Electron Corporation, Boston, MA, USA) at full
Hematology and Oncology, West China Second University speed (more than 10,0006g) for 10 min. Bacterial DNA was
Hospital, Sichuan University. All subjects received no previous extracted using QIAamp DNA Micro Kit (QIAGEN, Hilden,
antineoplastic treatment. Demographic information was obtained Germany) according to the manufacturer’s instructions with minor
and oral examination was performed. A gender-, age-, and caries modification. Briefly, 30 ml lysozome solution (50 mg/ml) was
status matched healthy counterpart for each leukemia child was added to the mixture at the first step to increase the yield of
recruited from Department of Pediatric Dentistry, West China bacterial DNA from Gram-positive bacteria by hydrolyzing the
Hospital of Stomatology, Sichuan University. The detailed eligible peptidoglycan cell wall. The amount of DNA extracted per sample
was determined using Quant-iT PicoGreen dsDNA Assay Kit

Table 1. Admission criteria.

ALL patients group

Inclusion criteria Exclusion criteria

(1) Newly diagnosed with acute lymphoblastic leukemia based on bone marrow samples (1) Previous antineoplastic treatment
(2) No more than eighteen years old (2) Receiving antibiotics within 3 months before the study
(3) Free of other systemic diseases (including systemic infection) (3) Local antimicrobial treatment within 2 weeks
(4) Written informed consent

Healthy control group

Inclusion criteria Exclusion criteria

(1) Age, socioeconomic and caries status comparable with ALL patients (1) Systemic or local disorders that cause oral mucosal lesions such as lichen planus
(2) Free of systemic diseases (2) Periodontal pockets equal to or greater than 4 mm
(3) Written informed consent (3) Acute oral infection such as dental abscess at enrollment
(4) Evidence of oral candidiasis
(5) Receiving antibiotics within 3 months before the study
(6) Local antimicrobial treatment within 2 weeks

ALL patients: acute lymphoblastic leukemia patients.

doi:10.1371/journal.pone.0102116.t001

PLOS ONE | www.plosone.org 2 July 2014 | Volume 9 | Issue 7 | e102116

 Oral Microbiota of Leukemia Affected Children

(Invitrogen, Carlsbad, USA). Size and integrity of DNA were Sequence Deposition
checked by 1% (w/v) agarose gel electrophoresis in 0.05% (v/v) Sequences were deposited in the NCBI sequence read archive
GoldView. The extracted DNA was then stored at 220uC before (http://www.ncbi.nlm.nih.gov/Traces/sra/) under accession
further analyses. number SRP034724.

Pyrosequencing and Data Analysis Results

The 16S rRNA hypervariable V1–V3 region was amplified
using polymerase chain reaction (PCR) with the forward primer Richness and Diversity of Oral Microbiota in ALL patients
8F (59-AGAGTTTGATCCTGGCTCAG-39) and reverse primer and Healthy Children
533R (59-TTACCGCGGCTGCTGGCAC-39) [18]. Unique 10- In total, 356000 qualified sequence reads were obtained and
bp barcodes were incorporated into the reverse primers so that used for analysis, with an average of 13692 sequence reads for
sequences of different samples can be differentiated. Each 25 ml each sample. One healthy subject sample (H04) had low sequence
PCR reaction consisted of 1 ml of forward primer (10 mmol/l), 1 ml counts making the depth of pyrosequencing not comparable to
of barcoded reverse primer (10 mmol/l), 0.5 ml of dNTP mix other samples. Thus to avoid potential bias caused by uneven
(10 mmol/l), 2.5 ml of FastStart 106 buffer with 18 mmol/l of sequence depth, this sample was not subjected to further
MgCl2, 0.25 ml of FastStart HiFi Polymerase (5 U/ml), 1 ml of bioinformatics analyses, leaving 13 ALL patients and 12 healthy
genomic DNA, and 18.75 ml of water. FastStart HiFi Polymerase, control subjects samples for downstream data analyses.
FastStart 106 buffer with MgCl2 and dNTP mix were included in The species-level operational taxonomic units (OTUs) and
the FastStart High Fidelity PCR System, dNTP Pack (Roche richness and diversity estimators were generated for each sample
Applied Science). The PCR amplification was performed using the (Table S2). Clustering the unique sequences into OTUs at a 0.03
following program: 3 minutes of initial denaturation at 94uC, dissimilarity level formed 2280 OTUs per microbiome on average.
followed by 25 cycles of denaturation (94uC for 15 seconds), Richness and diversity estimation of 16S rRNA gene libraries at
annealing (56uC for 30 seconds) and extension (72uC for 97% similarity was calculated and compared between two groups
(Table 2). Comparison of species-level OTUs between ALL
30 seconds) with a final extension of 5 minutes at 72uC. To
patients and healthy children revealed that the ALL patients group
obtain sufficient PCR products for pyrosequencing, the amplicons
exhibited lower richness (p = 0.004). Furthermore, community
of three replicates were pooled. After PCR, amplicons were gel
richness estimators (Chao 1 and ACE) also revealed a statistically
purified using MinElute Gel Extraction Kit (QIAGEN), and then
significant lower estimate of richness for ALL patients compared
quantified with Quant-iT PicoGreen dsDNA Assay Kit (Invitro-
with healthy children. ALL patients showed less diversity
gen).The PCR products were combined in equimolar ratios to
compared with healthy subjects as demonstrated by comparing
create a DNA pool used for pyrosequencing. Pyrosequencing was
the Simpson Index between the oral microbiota of the two groups
performed according to standard Roche 454 GS-FLX protocols (p = 0.019).
[26].
To optimize raw sequences, the sequence analyzing programs
Structural Comparisons of Oral Microbiota between ALL
Seqcln (http://sourceforge.net/projects/seqclean/) and MOTHUR
(version 1.30.0; http://www.mothur.org) were applied to the data. Patients and Healthy Subjects
The primer sequences and 10-bp barcode were removed. The No statistically significant difference regarding the demographic
sequences were checked and low quality sequences (quality score ,25) information and oral health conditions was observed between the
were discarded. Sequences that contained ambiguous bases, incorrect ALL group and the healthy control (p.0.05), albeit the ALL
patients have a higher incidence of periodontitis compared with
primer sequence, or identified to be shorter than 200 bp, or
the healthy subjects (Table 3). We compared the oral microbial
homologous sequences longer than six nucleotides were also removed.
communities present within each of the 13 ALL patients and 12
UCHIME (http://drive5.com/uchime) was used to detect potentially
control subjects using three different types of cluster analysis:
chimeric artifacts. Sequences were clustered to operational taxono-
OTU-based, taxonomy-based (at the order level) and phylogeny-
mical units (OTU) using CD-HIT-EST at 97% similarity level. RDP
based cluster analyses. The dendrogram from the OTU-compo-
NaÏve Bayesian Classifier was applied to perform read level taxonomic
sition presented two clusters, with one cluster composed of
assignments with an 80% bootstrap score [27]. Qualified sequences
leukemia oral samples and the other mainly comprised of healthy
were submitted to the SILVA database (SILVA 111; http://www. control samples (Figure 1). Moreover, the supragingival plaque
arbsilva.de) for taxonomic alignment. Community richness (ACE, samples from the ALL patients and healthy controls could be
Chao1) and diversity indices (Simpson diversity index) were divided into two subsets based on microbiota composition and
determined by the MOTHUR program at 97% similarity level. The microbe abundance at the order level. This taxonomy-based
statistical significance of these indices was determined by SPSS 19.0 clustering at the order level was better associated with the health
(SPSS Inc, Chicago, IL, USA) with nonparametric Mann-Whitney U or disease status (ALL-affected vs. ALL-free) of subjects as shown
test for independent samples. The heatmap profile was generated by in Figure S1. Principal coordinate analysis (PCoA) based on the
the R program (http://www.r-project.org/). For phylogeny-based weighted UniFrac metric was also performed. Although 5 subjects
cluster comparisons, principal coordinate analysis (PCoA) plots were from ALL patients group clustered with the healthy controls, a
generated with the distance matrix calculated using the weighted segregation trend for ALL patients and healthy subjects was
UniFrac algorithm. The composition of the microbial communities observed, especially by principal coordinate P1 (Figure 2). To
present in the samples from the ALL patients (n = 13) and control further validate the statistical significance of the aforementioned
(n = 12) groups were analyzed using unweighted and weighted sample clustering, we performed three different types of pair-wise
UniFrac analysis [28,29] and Parsimony p-tests [30]. Metastats was comparisons, including parismony test, Unifrac unweighted and
used to compare the relative abundance of each taxon at different Unifrac weighted analyses. A single NJ phylogenetic tree
taxonomic levels between ALL patients and healthy children [31], containing all 16S rRNA sequences from the samples from the
with the p value threshold set at 0.05 and the q value threshold at 0.5 13 ALL patients and 12 control subjects (identical to that used for
[32]. the PCoA analysis) was used as the input for all three methods. All

PLOS ONE | www.plosone.org 3 July 2014 | Volume 9 | Issue 7 | e102116

 Oral Microbiota of Leukemia Affected Children

Table 2. Comparison of richness and diversity estimates of 16S rRNA gene libraries at 97% similarity between the acute
lymphoblastic leukemia patients and healthy subjects.

Variables ALL patients (n = 13) Healthy control (n = 12) p value*

OTUs 17786874 28246647 0.004

Chao 1a 4738.2762049.02 7215.1161505.13 0.003
ACEb 8305.5063336.40 12455.8262400.76 0.004
Simpsonc 0.039360.0246 0.016860.0082 0.019

Applicable values are means6SD. ALL patients: acute lymphoblastic leukemia patients.
a,b
Richness estimators (Chao 1 and ACE) were calculated using MOTHUR.
c
A higher number indicates less diversity.
*Independent sample nonparametric Mann-Whitney U test was used.
doi:10.1371/journal.pone.0102116.t002

three statistical approaches showed significant difference (paris- between ALL patients and healthy subjects were found for the
mony test, p = 0.006; Unifrac unweighted, p = 0.005; Unifrac two phyla, i.e. Firmicutes (p = 0.001) and Fusobacteria (p = 0.003).
weighted, p,0.001), indicating that the oral microbial structure of Firmicutes was more abundant while Fusobacteria was less abundant
the ALL-affected population is distinct from that of the healthy in ALL patients compared to healthy children (Figure 3). Of all
controls. genera detected from oral cavity samples, Abiotrophia (p = 0.001),
Comamonas (p = 0.001), Granulicatella (p = 0.002), Leptotrichia
Taxonomy-Based Comparisons of Oral Microbiota (p = 0.001) and Veillonella (p = 0.001) were significantly different
between the Two Host Populations between ALL patients and healthy subjects. Leukemia-enriched
Comparisons of oral microbiota between ALL patients and genera included Abiotrophia, Granulicatella and Veillonella, while
healthy subjects at each of taxonomical levels of phylum, class, Comamonas and Leptotrichia constituted leukemia-depleted genera
order, family and genus were performed based on Metastats (Figure 3).
analysis. From the phylum level down to the Genus level, there Taking taxonomic lineages into account, two lineages were
were no ‘leukemia-specific’ or ‘healthy-specific’ taxa unique to found to be more abundant in the ALL patients compared to the
either leukemia or healthy hosts. However, ‘leukemia-associated’ healthy children from the phylum down to the genus level (Table
taxa (present in healthy and leukemia populations but differentially S3). The two leukemia-enriched taxonomic lineages included
Firmicutes (p = 0.001) at Phylum, Bacilli (p = 0.001) at Class,
distributed) were detected, which were either ‘leukemia-enriched’
Lactobacillales (p = 0.001) at Order, Aerococcaceae (p = 0.001) at Family
or ‘leukemia-depleted’ for certain microbe classes (Figure 3; Table
and Abiotrophia (p = 0.012) at genus, as well as Firmicutes (p = 0.001)
S3). A total of 12 phyla were identified in oral microbiota of ALL
at Phylum, Bacilli (p = 0.001) at Class, Lactobacillales (p = 0.001) at
patients and healthy children in our dataset, which were
Order, Carnobacteriaceae (p = 0.002) at Family and Granullicatella
dominated by six major phyla, including Proteobacteria, Firmicutes,
(p = 0.007) at Genus. Moreover, one leukemia-depleted lineage
Fusobacteria, Actinobacteria, Bacterioidetes and candidate division TM7
was found, consisting of Fusobacteria (p = 0.003) at Phylum,
(Figure S2). Particularly, notable differences in abundance
Fusobacteria (p = 0.001) at Class, Fusobacteriales (p = 0.001) at Order

Table 3. Demographic and oral health information of acute lymphoblastic leukemia and healthy subjects.

Variables Characteristics ALL patients (n = 13) Healthy control (n = 12) p value

Age (years) Mean 6 SD 6.6963.82 7.0063.05 0.806a

Dentition stage Primary 6 5 0.975b
Mixed 5 5
Permanent 2 2
Sex Male 8 8 1.000c
Female 5 4
Caries status dmfs/DMFS 3.4663.26 3.0063.36 0.600a
Periodontal condition Healthy 7 10 0.202c
Gingivitis 6 2
Periodontitis 0 0
Plaque index Silness-Löe Index 1.4160.41 1.4860.70 0.743a
Plaque pH Mean 6 SD 5.7560.28 5.8560.23 0.350a

ALL patients: acute lymphoblastic leukemia patients. dmfs/DMFS: decayed-missing-filled surfaces index in primary and permanent teeth, respectively.
a
Independent sample nonparametric Mann-Whitney U test was used.
b
Pearson Chi-Square test was used.
c
Fisher’s Exact test was used.
doi:10.1371/journal.pone.0102116.t003

PLOS ONE | www.plosone.org 4 July 2014 | Volume 9 | Issue 7 | e102116

 Oral Microbiota of Leukemia Affected Children

as a representative microbiota to delineate the influence of the

immune status on oral microbiome.
The holistic pattern of leukemia pediatric patients was presented
and compared with that of healthy controls, revealing key
characteristics of oral microbiota associated with childhood ALL.
ALL patients had lower richness and less diversity of oral
microbiota compared with healthy subjects. Similar results were
reported by Sixou et al, in which the microbial composition of
supragingival plaque from the leukemia patients, analyzed by
traditional culture methods, were found to be less complex than
that of healthy controls [10]. However, Cargill et al. found that the
complexity of oral microbiota was not statistically different
between leukemia patients and healthy controls until initiation of
antineoplastic treatment [33]. This inconsistency might be
attributed to techniques used to profile the oral microbiota.
Previous studies, mainly using culture methods, based their results
on a few selected cultivable bacteria, neglecting characteristics of
other bacteria in oral cavity of leukemia patients. At the time of
sampling, oral bacteria might have already undergone selection
Figure 1. Dendrogram of OTU composition obtained from after the disease manifested, with some bacteria inhibited and
acute lymphoblastic leukemia-affected children and healthy others thriving, resulting in divergent conclusions from different
controls. The dendrogram (left section) indicates the similarity of the
microbial communities of supragingival dental plaque between studies which focused on a few types of bacteria. By using 454
subjects according to their OTU composition, as determined using pyrosequencing, a holistic pattern of oral microbiome was
the Jaccard index (MOTHUR). A summary of patients’ clinical presented and bacteria with low abundance and under the
information is presented in the right section. H, healthy children. L, detection limit of culture methods could be investigated [36]. A
acute lymphoblastic leukemia-affected children. dmfs/DMFS, decayed- decrease in intestinal microbial diversity was found in pediatric
missing-filled surfaces index in primary and permanent teeth,
patients with acute myeloid leukemia [37]. The decrease of
respectively. G, gingivitis. ND, no data obtained.
doi:10.1371/journal.pone.0102116.g001 richness and diversity observed in the current study points to
dysbiosis of oral microbiota in ALL patients. In addition to
richness and diversity, we report altered structure and composition
and Fusobacteriaceae (p = 0.002) at Family, but not Fusobacterium
of oral microbiota in ALL patients. Altered structure and
(p = 0.009; q.0.5) at Genus.
composition of oral microbiota in ALL patients were related to
health status of investigated subjects. This was revealed in the
Discussion separation pattern of samples based on cluster analyses, which
This study compared the oral microbial composition of acute were consistent with the presence/absence of leukemia. Since age,
lymphoblastic leukemia (ALL) and healthy hosts, contributing to a dentition stage, plaque pH, plaque index and caries status (dmfs/
better understanding of the oral microbial profiles in immuno- DMFS) were comparable between ALL patients and healthy
compromised patients. While there have been several reports children, microbial differences observed in supragingival plaque
investigating oral microbiota in leukemia patients are likely directly caused by leukemia itself.
[10,11,12,13,33], most of these studies relied on traditional It is well-known that the host and microbiota have a dynamic
culturing methods and were focused on only a few kinds of interaction, which is closely associated with health [38]. The host
common bacteria, failing to characterize the overall pattern of oral immune system is believed to play an important role in shaping
microbiota in leukemia patients. In addition, most of these cross- human microbiota [39]. A study on the bacterial and fungal
sectional investigations were focused on the effect of antineoplastic microbiota of the stomach fluid indicated that immune status plays
treatment on oral microbiota, and limited information can be a major role in shaping the gastric fluid microbiota in terms of
obtained regarding the oral microbiota under ALL-associated both diversity and composition [40]. Important effectors of innate
immunocompromised conditions. These limitations have severely immunity, a-defensins that are a kind of secreted antibacterial
confounded efforts to pinpoint leukemia-associated oral microbi- protein produced by epithelial cells, can affect the composition of
ota in patients. intestinal microbial communities [41,42]. Immune-driven dysbio-
sis in the intestine has been found in mice deficient for the
Since gingivitis/periodontitis are common oral pathologies
transcription factor T-bet, which is associated with both the innate
experienced by leukemia patients [4,6], it is critical to distinguish
and the adaptive immune system [2]. Here we showed the
whether a detected microbial change is driven by the global
importance of the immune system in shaping the oral microbial
immune status or by the unavoidable existing periodontal
structure. This is not surprising, because like the small and large
pathology. We believe that compared to subgingival and salivary
intestine, the lamina propria of oral mucosa typically has
microbiota, which alter under periodontal pathologies [34,35], the
organized and diffuse lymphoid tissues [43]. In ALL patients,
supragingival microbiota is less affected by gingivitis/periodontitis,
the disorder of lymphoid progenitor cells, the major part of the
but more by the host immune status given that the caries status is
body’s immune system, can lead to impairment of the host
normalized. Therefore, in the present study, we sampled pooled
immunity [14,15]. The malignant proliferation of lymphocytes in
supragingival plaque in carefully controlled ALL and healthy
bone marrow infiltrates into the peripheral lymphoid tissues
populations, and analyzed its microbial composition via 16S-based including the lymphoid tissue of oral mucosa, leading to
454 pyrosequencing. Oral health conditions, except for periodon- impairment of oral immunity [7,44]. Furthermore, the salivary
tal conditions, were well balanced between the ALL patients and defense system which contains various antimicrobial components
healthy groups, further validating the use of supragingival plaque such as secreted immunoglobulin A (SIgA), a-amylase and

PLOS ONE | www.plosone.org 5 July 2014 | Volume 9 | Issue 7 | e102116

 Oral Microbiota of Leukemia Affected Children

Figure 2. Weighted Unifrac PCoA analysis. The first two principal coordinates (PCo1 and PCo2) from the principal coordinate analysis of
weighted UniFrac are plotted for each sample. The variance explained by the PCos is indicated in parentheses on the axes. H, healthy children. L,
acute lymphoblastic leukemia-affected children.
doi:10.1371/journal.pone.0102116.g002

lysozyme, has been reported to be significantly impaired in ALL patients predisposes the host to developing oral infection by
leukemia [45]. However, the exact mechanisms involved in the favoring pathogenic bacteria thriving in the oral niche. Prevention
interaction between the oral immune system and microbiota need of oral microbiota dysbiosis might be a promising measure for
further research. decreasing mortality in patients with ALL. It should be noted that
Our investigations of oral microbiota in ALL patients also the 16S rRNA gene assay has its own limitation, i.e. the detection
provided the opportunity for identifying potential microbiota of 16S rRNA sequences does not imply that live bacteria are
associated with systemic infections in leukemia patients. The present. Therefore, either combination of 16S rRNA gene assay
present data suggest two taxonomical lineages (Firmicutes/Bacilli/ with traditional culture-based approach, or further cohort studies
Lactobacillales/Carnobacteriaceae/Granullicatella, and Firmicutes/Bacilli/ are needed to elucidate the exact relationship between oral
Lactobacillales/Aerococcaceae/Abiotrophia) that are much more abun- microbial disequilibrium and risk of systemic infection in ALL-
dant in the supragingival plaque of ALL patients than healthy affected patient.
controls from the phylum down to the genus level; this indicates
that favorable conditions existed for their development in the oral Conclusion
cavity of ALL patients. This could be responsible for an increased
risk of bacteremia in leukemia patients. Granulicatella and In summary, by comparing the oral microbial composition of
Abiotrophia, formerly known as nutritionally variant streptococcus ALL patients and healthy subjects, we have identified a structural
(NVS), have been commonly implicated in endocarditis and imbalance of the oral microbiota, characterized by reduced
bacteraemia, and also in several other infections such as central diversity of microbiota and abundance alterations of certain
nervous system infections, otitis media, cholangitis and arthritis bacteria, possibly involved in systemic infections, indicating the
[46,47]. A high mortality rate for endocarditis by NVS has been important role immune status plays in shaping the structure of oral
reported [48,49]. Previous studies found that oral bacteria are microbiota. Although more works still need to be done, the
responsible for 25% to 50% of systemic infections in neutropenic dysbiosis of oral microbiota identified in this study provides insight
patients [50]. An increase in abundance of opportunistic into the host-microbe interactions related to the infectious
pathogens might be an important factor to the increased risk of complications of this susceptible population.
systemic infections in leukemia patients. Altered health status in

PLOS ONE | www.plosone.org 6 July 2014 | Volume 9 | Issue 7 | e102116

 Oral Microbiota of Leukemia Affected Children

Figure 3. Differential relative abundance of bacterial taxonomy profiles of acute lymphoblastic leukemia and healthy subjects
based on Metastats analysis. Comparisons were performed at each of the taxonomical levels of Phylum (A), Class (B), Order (C), Family (D) and
Genus (E). Means of the relative abundance for each taxon at each taxonomical level between the healthy and acute lymphoblastic leukemia host-
populations are compared, with a p value threshold set at 0.05 (*p,0.05; **p,0.01) and a q value threshold at 0.5. The discriminating taxa (p,0.05
and q,0.5) with relative abundance .1% for at least one group at each level are shown. H, healthy children. L, acute lymphoblastic leukemia-
affected children.
doi:10.1371/journal.pone.0102116.g003

Supporting Information Table S1 Overview of subject clinical information.

(DOC)
Figure S1 Heatmap analysis of the orders detected
among all subjects based on microbiota composition Table S2 The number of OTUs and species richness
and abundance. The colour of each column represents relative and diversity estimates in each supragingival plaque
abundance of the corresponding order according to the scale at the microbiome.
bottom of the plot. Subject metadata are shown using checker- (DOC)
board plots based on the following color codes: ALL subjects Table S3 Differential relative abundance of bacterial
(black) or healthy children (white); ,6 years old (white), 6–12 years taxonomy profiles of acute lymphoblastic leukemia
(gray) or older than 12 years (black); male (black) or female (white); (ALL) patients and healthy (H) subjects based on
primary (white), mixed (gray) or permanent (black) dentition;
Metastats analysis.
plaque index: ,1.00 (white), 1.00–2.00 (gray) or .2.00 (black);
(DOC)
plaque pH: ,5.4 (white), 5.4–5.7 (gray) or .5.7 (black); dmfs/
DMFS: 0–3 (white), 4–7 (gray) or 8–11 (black); gingivitis presence
(gray) or absence (white). See figure 1 and table S1 for additional Acknowledgments
details. H, healthy children. L, acute lymphoblastic leukemia The authors wish to thank Shanghai Majorbio Bio-Pharm Technology
affected children. Co.,Ltd for the assistance in data analysis.
(TIF)
Figure S2 Relative abundance of oral microbiota com- Author Contributions
positions at the phylum level. Comparison of all samples Conceived and designed the experiments: XX XDZ JZ LC MYL YQL
from ALL patients and healthy subjects showed major phyla YPZ JYL WYS. Performed the experiments: YW JX QD XY JZH.
comprised of Proteobacteria, Firmicutes, Fusobacteria, Actinobacteria, Analyzed the data: YW JX MY XX. Contributed reagents/materials/
Bacterioidetes and candidate division TM7. H, healthy children, L, analysis tools: YW JX MY. Wrote the paper: YW XX JX XDZ MY. Agree
acute lymphoblastic leukemia-affected children. with manuscript results and conclusions: YW JX XDZ MY QD XY JZH
(TIF) JZ LC MYL YQL YPZ JYL WYS XX.

PLOS ONE | www.plosone.org 7 July 2014 | Volume 9 | Issue 7 | e102116

 Oral Microbiota of Leukemia Affected Children

References
1. Pui CH, Robison LL, Look AT (2008) Acute lymphoblastic leukaemia. Lancet 27. Wang Q, Garrity GM, Tiedje JM, Cole JR (2007) Naive Bayesian classifier for
371: 1030–1043. rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl
2. Garrett WS, Lord GM, Punit S, Lugo-Villarino G, Mazmanian SK, et al. (2007) Environ Microbiol 73: 5261–5267.
Communicable ulcerative colitis induced by T-bet deficiency in the innate 28. Lozupone C, Knight R (2005) UniFrac: a new phylogenetic method for
immune system. Cell 131: 33–45. comparing microbial communities. Appl Environ Microbiol 71: 8228–8235.
3. Barrett AP (1986) Leukemic cell infiltration of the gingivae. J Periodontol 57: 29. Lozupone C, Hamady M, Knight R (2006) UniFrac–an online tool for
579–581. comparing microbial community diversity in a phylogenetic context. BMC
4. Hou GL, Huang JS, Tsai CC (1997) Analysis of oral manifestations of leukemia: bioinformatics 7: 371.
a retrospective study. Oral Dis 3: 31–38. 30. Schloss PD, Handelsman J (2006) Introducing TreeClimber, a test to compare
5. Meyer U, Kleinheinz J, Handschel J, Kruse-Losler B, Weingart D, et al. (2000) microbial community structures. Appl Environ Microbiol 72: 2379–2384.
Oral findings in three different groups of immunocompromised patients. J Oral 31. White JR, Nagarajan N, Pop M (2009) Statistical methods for detecting
Pathol Med 29: 153–158. differentially abundant features in clinical metagenomic samples. PLoS Comput
6. Javed F, Utreja A, Bello Correa FO, Al-Askar M, Hudieb M, et al. (2012) Oral Biol 5: e1000352.
health status in children with acute lymphoblastic leukemia. Crit Rev Oncol 32. Krych L, Hansen CH, Hansen AK, van den Berg FW, Nielsen DS (2013)
Hematol 83: 303–309. Quantitatively different, yet qualitatively alike: a meta-analysis of the mouse core
7. Paunica SC, Dumitriu A, Mogos M, Georgescu O, Mogos I (2009) The gut microbiome with a view towards the human gut microbiome. PLoS One 8:
evaluation of the periodontium in patients with leukemia using thermographic e62578.
imaging. Hematology 14: 341–346. 33. Lucas VS, Beighton D, Roberts GJ, Challacombe SJ (1997) Changes in the oral
8. Khan SA, Wingard JR (2001) Infection and mucosal injury in cancer treatment. streptococcal flora of children undergoing allogeneic bone marrow transplan-
JNCI Monographs 2001: 31–36. tation. J Infect 35: 135–141.
9. Greenberg MS, Cohen SG, McKitrick JC, Cassileth PA (1982) The oral flora as 34. Belstrøm D, Fiehn NE, Nielsen CH, Kirkby N, Twetman S, et al. (2014)
a source of septicemia in patients with acute leukemia. Oral Surg Oral Med Oral Differences in bacterial saliva profile between periodontitis patients and a control
Pathol 53: 32–36. cohort. J Clin Periodontol 41: 104–112.
10. Sixou JL, De Medeiros-Batista O, Gandemer V, Bonnaure-Mallet M (1998) The 35. Colombo APV, Boches SK, Cotton SL, Goodson JM, Kent R, et al. (2009)
effect of chemotherapy on the supragingival plaque of pediatric cancer patients. Comparisons of subgingival microbial profiles of refractory periodontitis, severe
Oral Oncol 34: 476–483. periodontitis, and periodontal health using the human oral microbe identifica-
11. Wahlin YB, Granstrom S, Persson S, Sjostrom M (1991) Multivariate study of tion microarray. J Periodontol 80: 1421–1432.
enterobacteria and Pseudomonas in saliva of patients with acute leukemia. Oral 36. Mardis ER (2008) Next-generation DNA sequencing methods. Annu Rev
Surg Oral Med Oral Pathol 72: 300–308. Genomics Hum Genet 9: 387–402.
12. O’Sullivan EA, Duggal MS, Bailey CC, Curzon ME, Hart P (1993) Changes in
37. van Vliet MJ, Tissing WJ, Dun CA, Meessen NE, Kamps WA, et al. (2009)
the oral microflora during cytotoxic chemotherapy in children being treated for
Chemotherapy treatment in pediatric patients with acute myeloid leukemia
acute leukemia. Oral Surg Oral Med Oral Pathol 76: 161–168.
receiving antimicrobial prophylaxis leads to a relative increase of colonization
13. Galili D, Donitza A, Garfunkel A, Sela MN (1992) Gram-negative enteric
with potentially pathogenic bacteria in the gut. Clin Infect Dis 49: 262–270.
bacteria in the oral cavity of leukemia patients. Oral Surg Oral Med Oral Pathol
38. Dethlefsen L, McFall-Ngai M, Relman DA (2007) An ecological and
74: 459–462.
evolutionary perspective on human–microbe mutualism and disease. Nature
14. Kitchingman GR, Rovigatti U, Mauer AM, Melvin S, Murphy SB, et al. (1985)
449: 811–818.
Rearrangement of immunoglobulin heavy chain genes in T cell acute
39. Hooper LV, Littman DR, Macpherson AJ (2012) Interactions between the
lymphoblastic leukemia. Blood 65: 725–729.
microbiota and the immune system. Science 336: 1268–1273.
15. Zhang XL, Komada Y, Chipeta J, Li QS, Inaba H, et al. (2000) Intracellular
cytokine profile of T cells from children with acute lymphoblastic leukemia. 40. von Rosenvinge EC, Song Y, White JR, Maddox C, Blanchard T, et al. (2013)
Cancer Immunol Immunother 49: 165–172. Immune status, antibiotic medication and pH are associated with changes in the
16. Li K, Bihan M, Methe BA (2013) Analyses of the stability and core taxonomic stomach fluid microbiota. ISME J 7: 1354–1366.
memberships of the human microbiome. PLoS One 8: e63139. 41. Salzman NH, Ghosh D, Huttner KM, Paterson Y, Bevins CL (2003) Protection
17. Zaura E, Keijser BJ, Huse SM, Crielaard W (2009) Defining the healthy ‘‘core against enteric salmonellosis in transgenic mice expressing a human intestinal
microbiome’’ of oral microbial communities. BMC Microbiol 9: 259. defensin. Nature 422: 522–526.
18. Lazarevic V, Whiteson K, Hernandez D, Francois P, Schrenzel J (2010) Study of 42. Salzman NH, Hung K, Haribhai D, Chu H, Karlsson-Sjoberg J, et al. (2010)
inter- and intra-individual variations in the salivary microbiota. BMC Genomics Enteric defensins are essential regulators of intestinal microbial ecology. Nat
11: 523. Immunol 11: 76–83.
19. Ling Z, Kong J, Jia P, Wei C, Wang Y, et al. (2010) Analysis of oral microbiota 43. Tlaskalová-Hogenová H, Štěpánková R, Hudcovic T, Tučková L, Cukrowska B,
in children with dental caries by PCR-DGGE and barcoded pyrosequencing. et al. (2004) Commensal bacteria (normal microflora), mucosal immunity and
Microb Ecol 60: 677–690. chronic inflammatory and autoimmune diseases. Immunol Lett 93: 97–108.
20. Yang F, Zeng X, Ning K, Liu KL, Lo CC, et al. (2012) Saliva microbiomes 44. Tjwa E, Mattijssen V (2008) Images in clinical medicine. Gingival hypertrophy
distinguish caries-active from healthy human populations. ISME J 6: 1–10. and leukemia. N Engl J Med 359: e21.
21. Saber MH, Schwarzberg K, Alonaizan FA, Kelley ST, Sedghizadeh PP, et al. 45. Hegde AM, Joshi S, Rai K, Shetty S (2011) Evaluation of oral hygiene status,
(2012) Bacterial flora of dental periradicular lesions analyzed by the 454- salivary characteristics and dental caries experience in acute lymphoblastic
pyrosequencing technology. J Endod 38: 1484–1488. leukemic (ALL) children. J Clin Pediatr Dent 35: 319–323.
22. Koren O, Spor A, Felin J, Fak F, Stombaugh J, et al. (2011) Human oral, gut, 46. Phulpin-Weibel A, Gaspar N, Emirian A, Chachaty E, Valteau-Couanet D,
and plaque microbiota in patients with atherosclerosis. Proc Natl Acad Sci U S A et al. (2013) Intravascular catheter-related bloodstream infection caused by
108 Suppl 1: 4592–4598. Abiotrophia defectiva in a neutropenic child. J Med Microbiol 62: 789–791.
23. Hu YJ, Shao ZY, Wang Q, Jiang YT, Ma R, et al. (2013) Exploring the dynamic 47. Senn L, Entenza JM, Greub G, Jaton K, Wenger A, et al. (2006) Bloodstream
core microbiome of plaque microbiota during head-and-neck radiotherapy using and endovascular infections due to Abiotrophia defectiva and Granulicatella
pyrosequencing. PLoS One 8: e56343. species. BMC Infect Dis 6: 9.
24. Hu YJ, Wang Q, Jiang YT, Ma R, Xia WW, et al. (2013) Characterization of 48. Cargill JS, Scott KS, Gascoyne-Binzi D, Sandoe JA (2012) Granulicatella
oral bacterial diversity of irradiated patients by high-throughput sequencing. infection: diagnosis and management. J Med Microbiol 61: 755–761.
Int J Oral Sci 5: 21–25. 49. Liao CH, Teng LJ, Hsueh PR, Chen YC, Huang LM, et al. (2004) Nutritionally
25. Hague A, Baechle M (2008) Advanced caries in a patient with a history of variant streptococcal infections at a University Hospital in Taiwan: disease
bariatric surgery. J Dent Hyg 82: 22. emergence and high prevalence of beta-lactam and macrolide resistance. Clin
26. Margulies M, Egholm M, Altman WE, Attiya S, Bader JS, et al. (2005) Genome Infect Dis 38: 452–455.
sequencing in microfabricated high-density picolitre reactors. Nature 437: 376– 50. Khan SA, Wingard JR (2001) Infection and mucosal injury in cancer treatment.
380. J Natl Cancer Inst Monogr: 31–36.

PLOS ONE | www.plosone.org 8 July 2014 | Volume 9 | Issue 7 | e102116