Microbial Ecology

https://doi.org/10.1007/s00248-019-01448-x

HUMAN MICROBIOME

Persistent Gut Microbial Dysbiosis in Children with Acute

Lymphoblastic Leukemia (ALL) During Chemotherapy
Seesandra V. Rajagopala 1,2 & Harinder Singh 1 & Yanbao Yu 1 & Keri B. Zabokrtsky 3,4 & Manolito G. Torralba 5 &
Kelvin J. Moncera 5 & Bryan Frank 1 & Rembert Pieper 1 & Leonard Sender 3,4,6,7 & Karen E. Nelson 1,5

Received: 5 February 2019 / Accepted: 1 October 2019

# Springer Science+Business Media, LLC, part of Springer Nature 2019

Abstract
Prophylactic or therapeutic antibiotic use along with chemotherapy treatment potentially has a long-standing adverse
effect on the resident gut microbiota. We have established a case-control cohort of 32 pediatric and adolescent acute
lymphoblastic leukemia (ALL) patients and 25 healthy siblings (sibling controls) to assess the effect of chemotherapy as
well as antibiotic prophylaxis on the gut microbiota. We observe that the microbiota diversity and richness of the ALL
group is significantly lower than that of the control group at diagnosis and during chemotherapy. The microbiota
diversity is even lower in antibiotics-exposed ALL patients. Although the gut microbial diversity tends to stabilize after
1-year post-chemotherapy, their abundances were altered because of chemotherapy and prophylactic antibiotic treat-
ments. Specifically, the abundances of mucolytic gram-positive anaerobic bacteria, including Ruminococcus gnavus and
Ruminococcus torques, tended to increase during the chemotherapy regimen and continued to be elevated 1 year beyond
the initiation of chemotherapy. This dysbiosis may contribute to the development of gastrointestinal complications in
ALL children following chemotherapy. These findings set the stage to further understand the role of the gut microbiome
dynamics in ALL patients and their potential role in alleviating some of the adverse side effects of chemotherapy and
antibiotics use in immunocompromised children.

Keywords Pediatric leukemia . Gut microbiota . 16S rRNA gene sequencing . Acute lymphoblastic leukemia (ALL)

Introduction
Electronic supplementary material The online version of this article
(https://doi.org/10.1007/s00248-019-01448-x) contains supplementary Every year, more than 15,000 children and adolescents under
material, which is available to authorized users. the age of 19 years are diagnosed with lymphoma, leukemia,
and other cancers in the USA, and an estimated 175,000 chil-
* Harinder Singh dren are diagnosed worldwide. In children, acute lymphoblas-
hsingh@jcvi.org
tic leukemia (ALL) is one of the most common leukemia types
in the USA, accounting for 26% of all cancers in children up
1
J. Craig Venter Institute (JCVI), Rockville, MD, USA to 14 years of age and for 75% of all pediatric leukemia cases
2
Division of Infectious Diseases, Department of Medicine, Vanderbilt [1, 2]. Children and adolescents diagnosed with ALL undergo
University Medical Center, Nashville, TN, USA chemotherapy as part of their treatment plan, which can se-
3
Hyundai Cancer Genomics Center, Children’s Hospital Orange verely compromise their health due to the chemotherapeutic
County (CHOC Children’s), Orange, CA, USA drugs. The cytotoxic effects of these treatments lead to further
4
Division of Hematology-Oncology, Department of Medicine, School immunosuppression, which entails occurrences of febrile neu-
of Medicine, University of California-Irvine, Orange, CA, USA tropenia and potentially life-threatening bloodstream infec-
5
J. Craig Venter Institute (JCVI), La Jolla, CA, USA tions [3]. Furthermore, prophylactic antibiotic usage can dis-
6
Division of Oncology, Hyundai Cancer Institute, CHOC Children’s, rupt the ecological balance of the gastrointestinal (GI) micro-
Orange, CA, USA biota [4]. Short-term health outcomes of chemotherapy in-
7
Department of Pediatrics, School of Medicine, University of clude Clostridium difficile–associated diarrhea (CDAD) and
California-Irvine, Orange, CA, USA disruptions of the GI microbial ecology. Late effects include
 Rajagopala S. V. et al.

impaired growth and development, impaired neurocognitive ALL and two HC subjects). Study participants were enrolled
development, loss of fertility, and negative effects on organ at the Hyundai Cancer Institute at the Children’s Hospital
function [5]. A common sequela in pediatric and adolescent Orange County (CHOC), CA, USA. The study was approved
survivors of leukemia is the development of metabolic syn- by the Internal Review Boards at CHOC and the J. Craig
drome and the subsequent risk for cardiovascular disease [6]. Venter Institute (JCVI). Human subject protocols along with
It is well established that the resident GI microbiota has a consent forms were established. Written informed consent
profound effect on human metabolism and can play an impor- was obtained from a parent and/or legal guardian for all the
tant role in activating, training, and modulating the host im- study participants, and all methods were performed in accor-
mune response [7]. Several studies have suggested a correla- dance with the relevant guidelines and regulations. There was
tion between the microbiota and various diseases including no exclusion of participants based on sex, racial, or ethnic
metabolic disorders, GI complexities, and infectious diseases group. Exclusion criteria included participants with other ma-
[8–11]. Studies have suggested that the composition of the GI lignancies and tumors, those receiving other therapies includ-
microbiota may affect and is modulated by both innate and ing radiation therapy, those with additional underlying dis-
adaptive immune system [12]. Additional recent research sug- eases such as diabetes or inflammatory bowel disease, and
gests that the GI microbiota influences the outcome of cancer those afflicted by infectious or diarrheal diseases. Subject de-
therapy by modulating the host inflammatory response [13, mographics by age and gender are shown in Table 1.
14]. Chemotherapy treatment, coupled with prophylactic or We collected stool samples at the time of diagnosis and
therapeutic antibiotic use, can perturb the GI microbiota, before chemotherapy was administered (“diagnosis”) which
resulting in altered gut microbiota composition and function. is our baseline microbiota data for each patient. HC stool
This, in turn, permits the outgrowth of, and persistent coloni- samples were collected only once at the time point matching
zation by, opportunistic pathogens and can also impact the the patient’s baseline sampling. To assess gut microbiota
long-term healthy development and maintenance of immune changes during the chemotherapy regimen, each patient’s
function. stool was collected every month up to 12 months. This sam-
There is a limited understanding of the gut microbiota and pling period represents several key chemotherapy time points:
their modulation in pediatric and adolescent patients diag- induction chemotherapy (chemotherapy given to induce a re-
nosed with ALL and the impact, if any, of prescribed chemo- mission), consolidation chemotherapy (chemotherapy given
therapeutic treatments. In a recent study, we have shown that once a remission is achieved), and maintenance therapy (che-
the gut microbiota diversity of ALL patients at the time of motherapy given in lower doses to assist in prolonging a re-
diagnosis is significantly lower as compared to healthy sibling mission). During treatment, the patients received antibiotic
controls [15]. Another study that examined the gut microbiota prophylaxis with trimethoprim and sulfamethoxazole and ste-
in children with newly diagnosed ALL showed that the rela- roid prophylaxis at the induction stage. The usage of antibi-
tive abundance of Proteobacteria before chemotherapy initia- otics along with the occurrence of infections in the month
tion predicted the development of febrile neutropenia, while prior to each visit was recorded. The details of the sampling
domination of the gut microbiota by Enterococcaceae or visits for each patient over the period of enrollment are pro-
Streptococcaceae during chemotherapy predicted infection in vided in Supplementary Table 1; some patients did not pro-
subsequent phases of chemotherapy [16]. In the present study, vide samples during maintenance therapy (Supplementary
we assessed the gut microbiota composition and dynamics Table 1). In order to have comparable samples representation
during chemotherapy by comparing stool samples obtained and avoid any biases in the statistical analysis, we created two
for ALL patients and their healthy siblings at varying time datasets: (1) Twenty-one patients with diagnosis and three-
points over 1 year. time points after diagnosis (dataset 1) and (2) fourteen patients

Results Table 1 Characteristics of study subjects

Diagnosis: acute lymphocytic leukemia (total n=29)

Study Population
Age group Ages 1–19 (n = 29, median age 5.00)
We designed a case-control cohort study to evaluate the effect
Gender Male: 21 (72%)
of leukemia and chemotherapy on the gut microbiota of pedi- Female: 8 (28%)
atric and adolescent patients diagnosed with pre-B acute ALL. Healthy control: sibling (total n = 23)
The study cohort consisted of 57 participants, made up of 32
Age group Ages 1–19, median age 6.000
patients and 25 healthy siblings (healthy controls; HC). Both
Gender Male: 6 (26%)
ALL and sibling control samples 19 years of age or older were Female: 17 (74%)
removed from the cohort for the analysis (this included three
Persistent Gut Microbial Dysbiosis in Children with Acute Lymphoblastic Leukemia (ALL) During Chemotherapy

with diagnosis and three-time points after diagnosis and one stabilize by 1-year post-chemotherapy (Fig. 1a;
sample after 1 year (dataset 2). Both datasets were matched Supplementary Figure 1), whereas species richness re-
with an equal number of age-matched sibling HC mains significantly lower throughout the first year of
(Supplementary Table 1, Supplementary Table 2). The analy- treatment with no substantial increase (p value < 0.05
sis primarily describes the analysis of dataset 2, unless dataset [Chao1]). The species richness never recovered during
1 is mentioned. the entire year (PC-1M–PC-12M) of post-chemotherapy
(Fig. 1a; Supplementary Figure 1). We also observed sim-
Comparison of Microbial Diversity in ALL and Controls ilar results with the 21 patient datasets (dataset 1), which
at Diagnosis and After Use of Chemotherapy consist of 21 patients with matched sibling’s samples at
diagnosis and samples collected at three time points after
A total of 249 fecal samples were analyzed in this study. diagnosis. The species evenness is not significant after 2
We observed a total of 699 operational taxonomic units months of first chemotherapy induction (p value > 0.05
(OTUs) across these 249 fecal samples using the [Shannon]), whereas species richness remains significant
UPARSE pipeline [17]. The OTUs identified were used during all time point (p value < 0.05 [Chao1]).
to compute the alpha diversity of the microbial communi-
ties using the Chao1 (species richness) and Shannon (spe-
cies evenness) indices. The mean Shannon and Chao1 Influence of Antibiotics on the Microbial Diversity
indices reveal that the microbiota diversity of the ALL in ALL
group at diagnosis is significantly lower than that of the
HC group (p value = 0.026 [Shannon] and 0.001 [Chao1], To determine the impacts of antibiotic administration on
Supplementary Figure 1), which further confirms the ob- the GI microbiota, the ALL group was further partitioned
servations of our earlier study [15]. Both Chao1 and into those who reported taking antibiotics in the 2-month
Shannon diversity indices were further decreased from period at diagnosis or during chemotherapy and those
diagnosis to the end of the first intensive chemotherapy who did not take antibiotics during the 2-month time pe-
induction (PC-1M) (p value = 0.019 [Shannon] and 0.001 riod. The microbiota diversity in each of these groups was
[Chao1]). However, species evenness tends to increase compared with that of the healthy sibling control group.
after 2 months of first induction and is not significantly The species evenness of ALL is not significantly different
different than (p value > 0.05 [Shannon]) the HC group at for ALL patients whether they took antibiotics or not (p
2 and 3 months after first induction (PC-2M and PC-3M; value > 0.05 [Shannon diversity index]) at the end of first
Fig. 1a). The species evenness of gut microbiota tends to year of treatment (Fig. 1b).

Fig. 1 Alpha diversity. a

Microbial richness at diagnosis
and during the chemotherapy
regimen was calculated based on
the Chao1 (species richness) and
Shannon (species evenness)
index. b Microbial richness and
evenness for the ALL groups
partitioned into those who
reported taking antibiotics in the
2-month period at diagnosis or
during chemotherapy and those
who did not take antibiotics in this
period. The y-axis represents the
alpha diversity unit scale either
Shannon or Chao1
 Rajagopala S. V. et al.

Beta Diversity Analysis of ALL and Controls considered for analysis. At the time of diagnosis, the pa-
at Diagnosis and After Use of Chemotherapy tients with ALL harbored significantly higher relative
abundances of Bacteroidetes (especially Bacteroides ge-
As shown in our previous study, the microbiota of the nus) than HCs. After the induction chemotherapy, the
ALL patients at diagnosis is different from their healthy Bacteroidetes ratio decreased, and their relative abun-
siblings. Using dataset 2, we performed beta diversity dance moved towards that of the HCs at 2 months post-
analysis to determine differences between ALL patients chemotherapy (Fig. 3). Firmicutes followed the reverse
at diagnosis and during chemotherapy. We performed trend with lower relative abundance during diagnosis as
PERMANOVA and ANOSIM using Bray-Curtis dissimi- compared to HCs (Fig. 3). The Bacteroidetes genus
larity matrices and found no difference between the gut Alistipes decreased substantially during chemotherapy: −
microbiota of ALL at diagnosis and during PC-1M and 3.7 log2fold at PC-1M, − 4 log2fold at PC-2M, and − 2.6
PC-2M, PC-3M, as well as PC-12M time point during the log2fold at PC-3M (Fig. 3; Supplementary Tables 5–9).
first year of treatment (Fig. 2; Supplementary Table 4). Another Bacteroidetes genus, Parabacteroides, increased
When we performed presence/absence standardization after the induction chemotherapy in its relative abundance
analysis using Bray-Curtis dissimilarity matrices, we ob- and began to decline to the levels seen in HCs after 1 year
served a minor difference between the diagnosis and the of chemotherapy. The Firmicutes Faecalibacterium and
PC-3M time point (PERMANOVA’s R2 = 0.074, p = 0.05; Pseudobutyrivibrio were significantly decreased at the
ANOSIM’s R = 0.082, p = 0.54). The difference was even time of diagnosis in ALL patients compared to HC con-
more significant between the ALL patients at diagnosis trols (Fig. 2; Supplementary Tables 5–9). Relative abun-
and the PC-12M time point (PERMANOVA’s R 2 = dance changes in Faecalibacteria have been linked to
0.090, p = 0.001; ANOSIM’s R = 0.11, p = 0.024) using dysbiosis in inflammatory bowel disease [18, 19] and ul-
Bray-Curtis similarity matrices (Fig. 2; Supplementary cerative colitis [20] together with the induced secretion of
Table 4). the cytokine IL-10 [21]. Relative abundance levels of
Faecalibacterium and Ruminococcus negatively correlat-
Differentially Abundance in Gut Microbiota ed with cytokine IL-6 and CRP [22]. Strikingly, several
During Chemotherapy genera of the family Lachnospiraceae substantially
changed in abundance at diagnosis and during chemother-
The differentially abundant OTUs at the genus level were apy, among which Lachnospiraceae_UCG-005 and
identified using a negative binomial model implemented Lachnoclostridium were more abundant at PC-1M, PC-
in DESeq2 package, and only those genera that constitut- 12M as compared to HCs (Supplementary Tables 5–9).
ed more than 1% of total genera relative abundance were Lachnoclostridium was consistently more abundant in

Fig. 2 PCA plots between ALL patients at the time of diagnosis vs their at diagnosis and after 3 months of diagnosis. c PCA plot between the ALL
healthy siblings and during the 1st and 12th month of the treatment course patients at the time of diagnosis and after 12 months of their diagnosis. In
using dataset 2. a PCA plot between the healthy siblings and ALL all three cases, the PERMANOVA p value is less than 0.05
patients at the time of diagnosis. b PCA plot between the ALL patients
Persistent Gut Microbial Dysbiosis in Children with Acute Lymphoblastic Leukemia (ALL) During Chemotherapy

Fig. 3 Box and whisker plot showing differential abundant genera at diagnosis and post-chemotherapy up to 1 year as compared to age-controlled HCs

ALL patients during chemotherapy, and, even after 1-year decrease of a microbe would be observed only in ALL pa-
post-chemotherapy, the genus abundance was 3.1 log2fold tients. After applying the mixed model, we performed
higher than that of the HCs (Table 2; Fig. 4). A N O VA t o o b ta i n th e p v a l u e s . R u m i n o c o c c u s ,
We also performed longitudinal analyses using the mixed Lachnospiraceae_UCG-005, and Lachnoclostridium abun-
random model implemented in the nlme R package [23]. dance increase during the first year of treatment (p value <
Longitudinal analysis was performed using only ALL sam- 0.05). When the analysis accounted for antibiotics usage, only
ples; we removed HC from the analysis so that any increase or Lachnoclostridium genera have a p value of less than 0.05.
 Rajagopala S. V. et al.

Table 2 The abundance of genus Lachnoclostridium in ALL patients in Ruminococcus biomass; this data shows higher level of
across different time points. Only the first three decimals of p value and
Ruminococcus gnavus in ALL patients (Supplementary
FDR are reported
Figure 2; Supplementary Table 12). The qPCR data correlated
Time point Genus Log2fold change p value FDR with the 16s rRNA data showing increased abundance of
Ruminococcus gnavus in the ALL patients as compared to HC.
HC vs diagnosis Lachnoclostridium 2.640 0.000 0.000
HC vs PC-1M Lachnoclostridium 1.890 0.000 0.003
HC vs PC-2M Lachnoclostridium 3.557 0.000 0.000
Discussion
HC vs PC-3M Lachnoclostridium 2.874 0.000 0.000
HC vs PC-12M Lachnoclostridium 3.173 0.000 0.000
The strength of this study is using healthy sibling controls to
compare and contrast the gut microbiota in children with ALL
Other genera with a p value of less than 0.05 were present in undergoing chemotherapy. In previous studies, we observed
very low abundance and thus did not correlate with the ALL. that microbial diversity of the ALL patients was significantly
lower than that of the sibling control group at diagnosis [15].
After the first induction, species richness and evenness are
Validating 16S Data Using Quantitative PCR significantly low at PC-1M; however, only species richness
remained significantly low after 1 year of treatment (p value <
Quantitative real-time PCR (qPCR) was performed on a subset 0.05 [Chao1 index]; Fig. 1). Decreased microbial diversity
of ALL samples from diagnosis through 12 months post- after receiving induction chemotherapy was also reported in
chemotherapy as well as healthy sibling’s samples to assess the other studies that examined adult AML [25], non-Hodgkin’s
relative abundance of Ruminococcus. Species-specific primers lymphoma patients [26], and children with ALL [16]. Limited
for Ruminococcus gnavus–specific genes were described by samples were partitioned into two groups: antibiotics taking
Rekha et. al. [24]. The qPCR results show an elevated level of and not taking. Further some, patients are exposed to antibi-
Ruminococcus gnavus in ALL patients samples as early as 1- otics only during the first few months and other patients were
month post-chemotherapy and their abundance remain high at 12 exposed during the end of first year of treatment. Due to un-
months post-chemotherapy (Fig. 4b, c; Supplementary Table 11), even distribution of antibiotics usage in ALL patients as well
which correlates with 16s rRNA data. Further, based on the 16s as due to small sample size, we were not able to discover the
rRNA data we did not observe any differential abundance of specific effects of antibiotics or chemotherapy on the gut mi-
Clostridium in HC and ALL patients. The qPCR of the small crobial diversity in ALL patients.
subunit rRNA of Clostridium was used as the housekeeping gene The relative abundance of the gut microbiota in ALL pa-
to normalize sample input and to calculate the Δct value. ΔΔct tients during the chemotherapy regimen revealed
values were calculated across samples to determine fold change Lachnospiraceae to be consistently increased an average of 3

Fig. 4 a Abundance of different species in Lachnospiraceae genera in using the RUMGNA primers. c The average ct. values of
HCs and ALL patients at diagnosis and post-chemotherapy. b The Ruminococcus gnavus in HCs and ALL patients using the S-G-Rum
average ct. values of Ruminococcus gnavus in HCs and ALL patients primer
Persistent Gut Microbial Dysbiosis in Children with Acute Lymphoblastic Leukemia (ALL) During Chemotherapy

to 4 log2fold at diagnosis and during chemotherapy. Even limitations in this study. Our study’s sample size was small,
after 12 months, post-chemotherapy Lachnoclostridium was so it is possible that we were underpowered to detect some
3.1 log2fold higher compared to HCs (Table 2). Among the differences between the longitudinal samples. The microbiota
Lachnoclostridium, R. gnavus relative abundance was signif- changes may be associated with age-related changes as well as
icantly increased after chemotherapy. R. gnavus is an anaero- by gender bias in the comparison. Because of the inherent
bic gram-positive coccus commonly present in the GI tract of nature of 16s rRNA sequencing technology, the species level
animals and humans. Previous studies have suggested an in- result presented in the study would be better explored using
creased relative abundance of R. gnavus in the human gut and the metagenomics, which is better suited to explore the mod-
an association with diverticular disease [27]. Importantly, ulation in abundance of various species in a longitudinal
R. gnavus possesses the ability to modulate the glycosylation study. Due to the limited number of samples analyzed in this
profile of the glycoconjugate molecules and mucus in goblet study, it is important to follow up with comprehensive studies
cells [28] and mucolytic functions. We hypothesize that an to corroborate the results as well as explore the species level
increased relative abundance of mucolytic gram-positive an- modulation in the gut microbiome of ALL patients.
aerobic bacteria including R. gnavus and R. torques due to
chemotherapy contributes to the development of clinically
evaluated GI mucositis in immunocompromised children after Materials and Methods
chemotherapy. On the other hand, R. gnavus is a novel
ursodeoxycholic acid producer and plays a pivotal important Sample Collection and Processing
role in the production of this acid in the colon [29].
Ursodeoxycholic acid possesses a hepatoprotective effect Fecal samples were collected using the Human Microbiota
and has been associated with decreased levels of hepatic trans- Project (HMP) Collection protocol section 7.3.3. with no
aminases in ALL treatment [30]. R. gnavus is a lantibiotic and modifications. DNA extraction occurred at CHOC.
produces a bacteriocin Ruminococcin which is effective PowerSoil® DNA Isolation Kit from Qiagen/MO BIO
against pathogens especially Clostridia, which might be the Laboratories, Inc. (catalog no.: 12888) was used to extract
reason for very low incidence of C. difficile infection in ALL the DNA from stool samples as described in Yooseph
patients in this study [31]. The data presented in this study will et al.[32]. Next, amplification of DNA was performed using
aid to design follow-up experiments to examine beneficial primers that targeted the V4 regions of the 16S rRNA gene
versus harmful effects of R. gnavus in ALL patients. [33, 34]. The primers consist of i5 and i7 adaptor sequences
A year after the start of chemotherapy treatment, the gut for Illumina MiSeq pyrosequencing and unique 8 bp indices
microbiota stabilized, but species richness never fully recov- incorporated onto both primers sot that every sample has its
ered. This suggests that the composition of the gut microbiota own unique barcode pair [35]. The primer pairs used in this
is modulated with some species being substantially altered. study are expected to amplify the entire V4 region of bacterial
The microbial diversity started recovering at the PC-3M time and archaeal species. Approximately 100 ng of extracted stool
point. As reported earlier, the observed microbiota changes at DNA and PCR amplicons were generated with Platinum Taq
diagnosis are likely the result of the direct or indirect influence polymerase (ThermoFisher, catalog no.: 11304-011) under the
of leukemia on the immune system, which, in turn, affects the following cycling conditions: 94 °C for 5 min for an initial
microbiota. Data from this study shows that the composition denaturing step followed by 94 °C for 30 s, 57 °C for 30 s, and
of the gut microbiota is further altered in immunocompro- 72 °C for 30 s for a total of 35 cycles followed by a final
mised children during chemotherapy and at one-year post- extension step of 72 °C for 7 min and finally stored at 4 °C.
chemotherapy. This study aimed to determine the overall After the PCR for every sample was completed, the amplicons
changes in the gut microbiota of childhood ALL patients dur- were purified with the QIAquick PCR purification kit
ing chemotherapy by comparing with their healthy siblings. (QIAGEN, catalog no.: 28104) and quantified using Tecan
We have not been able to analyze and identify distinct micro- fluorometric methods (Tecan Group, Männedorf,
biota profiles correlating with clinical outcomes (e.g., Switzerland). Next, we normalize and pooled the amplicons
C. difficile-associated diarrhea, febrile neutropenia) because in preparation for cluster generation, followed by sequencing
of the small sample size and the low incidence of C. difficile with Illumina MiSeq using the dual index 2 × 250 bp format
infections. Our findings emphasize the need to investigate (Roche, Branford, CT) as per the manufacturer’s protocol.
microbiota changes at the species level in survivors of child-
hood leukemia. Further elucidation of the microbial factors 16S rRNA Gene Sequence and Analysis
that drive the progression of adverse outcomes of chemother-
apy regimens in pediatric and adolescent cancer populations is Primers were trimmed, and the paired-end reads were quality
crucial because of the significant implications for acute infec- trimmed using the DynamicTrim program available in the
tions and chronic morbidities. We acknowledge several SolexaQA suite [36]. Operational taxonomic units (OTUs)
 Rajagopala S. V. et al.

were generated de novo from raw Illumina sequence reads Quantitative PCR
using the UPARSE pipeline [17]. The minimum fastq length
was set to 50, and the mean expected error cutoffs of 0.11 with To determine the relative quantification of the Ruminococcus
41 median Q score were used for clustering performed at 97% biomass in each sample, quantitative real-time polymerase
identity threshold. Sequences were subjected to a chain reaction (qPCR) was performed using 2 μL of each
dereplication step, and abundances were determined. sample (20 μL total reaction volume) with LightCycler®
Chimera filtering of the sequences occurred during the clus- 480 SYBR Green I Master (Roche Diagnostics, Rotkreuz,
tering step. We used the Wang classifier (method = wang) and Switzerland). Reactions were performed in duplicate using
bootstrapped using 100 iterations (iters = 100). We set mothur the LightCycler® 480 (Roche Diagnostics). The following
to report full taxonomies only for sequences where 80 or more amplification protocol was used: 45 cycles each of 95 °C for
of the 100 iterations are the same (cutoff = 80). Taxonomies 10 s, 58 °C for 10 s, and 72 °C for 45 s with single acquisition
were assigned to OTUs with mothur [37] using release 123 of using primers at a final concentration of 200 nM [24]
the SILVA 16S ribosomal RNA database [38] as the reference. (Supplementary Table 10). The crossing point (Cp) values
Tables with OTUs and the corresponding taxonomy assign- were determined by the software using the 2nd derivative
ments were generated and used in subsequent analyses. The max analysis method while primer specificity was determined
next step was to remove non-informative OTUs with an inde- using the melting temperature (Tm) analysis module in the
pendent filtering process. Rare OTUs or taxa are strongly Roche LightCycler 480 Software. Amplification of the small
affected by MiSeq sequencing errors, and any statistical con- subunit rRNA of Clostridium was used as the housekeeping
clusions relying on them are typically unstable. Even in the gene to normalize sample input and to calculate the Δct value.
univariate differential abundance analysis, the presence of ΔΔct values were calculated across samples to determine fold
such taxa increases the penalty from the multiple testing cor- change in Ruminococcus biomass. Species-specific primers
rections applied to the more abundant taxa. We used unbiased, targeting various proteins from Ruminococcus gnavus were
metadata-independent filtering at each level of the taxonomy selected after performing BLAST of Ruminococcus gnavus
by eliminating all features that did not pass these criteria. This genes against the RefSeq non-redundant genes. Only those
included samples with less than 2000 reads and OTUs present Ruminococcus gnavus genes which are unique to
in less than 10 samples. The phyloseq package version 1.16.2 Ruminococcus gnavus were selected for design primers using
in R package version 3.2.3 was used for the microbiota data Primer 3 [42].
analysis [39, 40]. We calculate the Shannon and Chao1 diver-
sity index from the OTUs to calculate the microbial commu- Authors’ Contributions Managed study participant’s consent and sample
collection: LS, KBZ. Performed experiments: SVR, YY, MT, KJM, BF,
nity alpha diversity.
MT. Performed computational analysis: SVR, HS. Analyzed the data and
provided input to the manuscript: SRV, HS, YY, KBZ, MT, BF, RP, LS,
KEN. Wrote the paper: SVR, HS, KEN. Conceived, designed, and super-
Statistical Analysis vised: SVR, KEN, LS. All authors read and approved the final
manuscript.

To determine statistical changes in alpha diversity among

Funding Information This project was funded by Hyundai Motor
different groups, we performed a Wilcoxon rank sum test America and Hyundai Hope on Wheels and J. Craig Venter Institute
followed by Benjamini and Hochberg [41] adjustment for p (JCVI).
value correction. The differential abundance test at the ge-
nus level was performed using the DESeq2 package ver- Data Availability The 16S rRNA gene sequencing data utilized in the
sion 1.12.3 in R, followed by multiple testing adjustments study is available via the NIH BioProject Accession PRJNA421352.
The metadata is available in the supplementary files as well as available
using Benjamini and Hochberg correction [41]. We used from the NIH Biosamples.
DESeq2 for the differential abundance test because it uses
a negative binomial model to control for different sequenc- Compliance with Ethical Standards
ing depths that lowers the false positive detection rate. The
permutational multivariate analysis of variance Human subject protocols along with consent forms were established.
(PERMANOVA) calculations and analysis of similarities Written informed consent was obtained from a parent and/or legal guard-
ian for all the study participants, and all methods were performed in
(ANOSIM) were performed using the R VEGAN package accordance with the relevant guidelines and regulations.
version 2.4-2 [19]. The function “lme” in R package Competing Interests
“nlme” is used for performing mixed effect model that The authors declare that they have no competing interests.
allows random effects integration in the model [23]. The Consent for Publication
Informed consent (and assent, where appropriate) and authorization to
“lme” function allows to study any microbe abundance use, create, and disclose health information for research was obtained
modulation in a longitudinal study while also considering from and documented for each research participant enrolled to study at
any given factor. the Hyundai Cancer Institute at CHOC Children’s.
Persistent Gut Microbial Dysbiosis in Children with Acute Lymphoblastic Leukemia (ALL) During Chemotherapy

Ethics Approval and Consent to Participate 12. Fulde M, Hornef MW (2014) Maturation of the enteric mucosal
This study was reviewed and approved by the In-House Institutional innate immune system during the postnatal period. Immunol Rev
Review Board at CHOC Children’s (protocol number 120555). 260:21–34. https://doi.org/10.1111/imr.12190
13. Sivan A, Corrales L, Hubert N, Williams JB, Aquino-Michaels K,
Earley ZM, Benyamin FW, Lei YM, Jabri B, Alegre ML, Chang
EB, Gajewski TF (2015) Commensal Bifidobacterium promotes
References antitumor immunity and facilitates anti-PD-L1 efficacy. Science
350:1084–1089. https://doi.org/10.1126/science.aac4255
1. Smith MA RL, Gurney JG, et al. (1999) Leukemia. In: Ries LA, 14. Vetizou M, Pitt JM, Daillere R, Lepage P, Waldschmitt N, Flament
Smith MA, Gurney JG, et al., eds Cancer incidence and survival C, Rusakiewicz S, Routy B, Roberti MP, Duong CP, Poirier-
among children and adolescents: United States SEER Program Colame V, Roux A, Becharef S, Formenti S, Golden E, Cording
1975-1995. Bethesda, Md: National Cancer Institute, SEER S, Eberl G, Schlitzer A, Ginhoux F, Mani S, Yamazaki T, Jacquelot
Program NIH PubNo 99-4649, pp 17-34. N, Enot DP, Berard M, Nigou J, Opolon P, Eggermont A, Woerther
2. Ward E, DeSantis C, Robbins A, Kohler B, Jemal A (2014) PL, Chachaty E, Chaput N, Robert C, Mateus C, Kroemer G,
Childhood and adolescent cancer statistics, 2014. CA Cancer J Raoult D, Boneca IG, Carbonnel F, Chamaillard M, Zitvogel L
Clin 64:83–103. https://doi.org/10.3322/caac.21219 (2015) Anticancer immunotherapy by CTLA-4 blockade relies on
3. Oeffinger KC, Mertens AC, Sklar CA, Kawashima T, Hudson MM, the gut microbiota. Science 350:1079–1084. https://doi.org/10.
Meadows AT, Friedman DL, Marina N, Hobbie W, Kadan-Lottick 1126/science.aad1329
NS, Schwartz CL, Leisenring W, Robison LL, Childhood Cancer 15. Rajagopala SV, Yooseph S, Harkins DM, Moncera KJ, Zabokrtsky
Survivor S (2006) Chronic health conditions in adult survivors of KB, Torralba MG, Tovchigrechko A, Highlander SK, Pieper R,
childhood cancer. N Engl J Med 355:1572–1582. https://doi.org/10. Sender L, Nelson KE (2016) Gastrointestinal microbial populations
1056/NEJMsa060185 can distinguish pediatric and adolescent acute lymphoblastic leuke-
4. van Vliet MJ, Harmsen HJ, de Bont ES, Tissing WJ (2010) The role mia (ALL) at the time of disease diagnosis. BMC Genomics 17:
of intestinal microbiota in the development and severity of 635. https://doi.org/10.1186/s12864-016-2965-y
chemotherapy-induced mucositis. PLoS Pathog 6:e1000879. 16. Hakim H, Dallas R, Wolf J, Tang L, Schultz-Cherry S, Darling V,
https://doi.org/10.1371/journal.ppat.1000879 Johnson C, Karlsson EA, Chang TC, Jeha S, Pui CH, Sun Y,
5. Hudson MM, Ness KK, Gurney JG, Mulrooney DA, Chemaitilly Pounds S, Hayden RT, Tuomanen E, Rosch JW (2018) Gut
W, Krull KR, Green DM, Armstrong GT, Nottage KA, Jones KE, microbiome composition predicts infection risk during chemother-
Sklar CA, Srivastava DK, Robison LL (2013) Clinical ascertain- apy in children with acute lymphoblastic leukemia. Clin Infect Dis.
ment of health outcomes among adults treated for childhood cancer. https://doi.org/10.1093/cid/ciy153
JAMA 309:2371–2381. https://doi.org/10.1001/jama.2013.6296 17. Edgar RC (2013) UPARSE: highly accurate OTU sequences from
6. Gibson TM, Ehrhardt MJ, Ness KK (2016) Obesity and metabolic microbial amplicon reads. Nat Methods 10:996–998. https://doi.
syndrome among adult survivors of childhood leukemia. Curr Treat org/10.1038/nmeth.2604
Options in Oncol 17:17. https://doi.org/10.1007/s11864-016-0393- 18. Miquel S, Martin R, Rossi O, Bermudez-Humaran LG, Chatel JM,
5 S o k o l H , T h o m a s M , We l l s J M , L a n g e l l a P ( 2 0 1 3 )
7. Rajagopala SV, Vashee S, Oldfield LM, Suzuki Y, Venter JC, Faecalibacterium prausnitzii and human intestinal health. Curr
Telenti A, Nelson KE (2017) The human microbiome and cancer. Opin Microbiol 16:255–261. https://doi.org/10.1016/j.mib.2013.
Cancer Prev Res (Phila) 10:226–234. https://doi.org/10.1158/1940- 06.003
6207.CAPR-16-0249 19. Frank DN, St Amand AL, Feldman RA, Boedeker EC, Harpaz N,
8. Karlsson F, Tremaroli V, Nielsen J, Backhed F (2013) Assessing the Pace NR (2007) Molecular-phylogenetic characterization of micro-
human gut microbiota in metabolic diseases. Diabetes 62:3341– bial community imbalances in human inflammatory bowel dis-
3349. https://doi.org/10.2337/db13-0844 eases. Proc Natl Acad Sci U S A 104:13780–13785. https://doi.
9. Le Chatelier E, Nielsen T, Qin J, Prifti E, Hildebrand F, Falony G, org/10.1073/pnas.0706625104
Almeida M, Arumugam M, Batto JM, Kennedy S, Leonard P, Li J, 20. Machiels K, Joossens M, Sabino J, De Preter V, Arijs I, Eeckhaut V,
Burgdorf K, Grarup N, Jorgensen T, Brandslund I, Nielsen HB, Ballet V, Claes K, Van Immerseel F, Verbeke K, Ferrante M,
Juncker AS, Bertalan M, Levenez F, Pons N, Rasmussen S, Verhaegen J, Rutgeerts P, Vermeire S (2014) A decrease of the
Sunagawa S, Tap J, Tims S, Zoetendal EG, Brunak S, Clement K, butyrate-producing species Roseburia hominis and
Dore J, Kleerebezem M, Kristiansen K, Renault P, Sicheritz-Ponten Faecalibacterium prausnitzii defines dysbiosis in patients with ul-
T, de Vos WM, Zucker JD, Raes J, Hansen T, Meta HITc, Bork P, cerative colitis. Gut 63:1275–1283. https://doi.org/10.1136/gutjnl-
Wang J, Ehrlich SD, Pedersen O (2013) Richness of human gut 2013-304833
microbiome correlates with metabolic markers. Nature 500:541– 21. Sokol H, Pigneur B, Watterlot L, Lakhdari O, Bermudez-Humaran
546. https://doi.org/10.1038/nature12506 LG, Gratadoux JJ, Blugeon S, Bridonneau C, Furet JP, Corthier G,
10. Pedersen HK, Gudmundsdottir V, Nielsen HB, Hyotylainen T, Grangette C, Vasquez N, Pochart P, Trugnan G, Thomas G,
Nielsen T, Jensen BA, Forslund K, Hildebrand F, Prifti E, Falony Blottiere HM, Dore J, Marteau P, Seksik P, Langella P (2008)
G, Le Chatelier E, Levenez F, Dore J, Mattila I, Plichta DR, Poho P, Faecalibacterium prausnitzii is an anti-inflammatory commensal
Hellgren LI, Arumugam M, Sunagawa S, Vieira-Silva S, Jorgensen bacterium identified by gut microbiota analysis of Crohn disease
T, Holm JB, Trost K, Meta HITC, Kristiansen K, Brix S, Raes J, patients. Proc Natl Acad Sci U S A 105:16731–16736. https://doi.
Wang J, Hansen T, Bork P, Brunak S, Oresic M, Ehrlich SD, org/10.1073/pnas.0804812105
Pedersen O (2016) Human gut microbes impact host serum metab- 22. Chua LL, Rajasuriar R, Azanan MS, Abdullah NK, Tang MS, Lee
olome and insulin sensitivity. Nature 535:376–381. https://doi.org/ SC, Woo YL, Lim YA, Ariffin H, Loke P (2017) Reduced microbial
10.1038/nature18646 diversity in adult survivors of childhood acute lymphoblastic leu-
11. Turnbaugh PJ, Hamady M, Yatsunenko T, Cantarel BL, Duncan A, kemia and microbial associations with increased immune activa-
Ley RE, Sogin ML, Jones WJ, Roe BA, Affourtit JP, Egholm M, tion. Microbiome 5:35. https://doi.org/10.1186/s40168-017-0250-1
Henrissat B, Heath AC, Knight R, Gordon JI (2009) A core gut 23. Pinheiro J, Bates D, DebRoy S, Sarkar D, Team RC (2018) nlme:
microbiome in obese and lean twins. Nature 457:480–484. https:// linear and nonlinear mixed effects models. R package version 31-
doi.org/10.1038/nature07540 137
 Rajagopala S. V. et al.

24. Rekha R, Rizvi MA, Jaishree P (2006) Designing and validation of 16S rRNA diversity at a depth of millions of sequences per sample.
genus-specific primers for human gut flora study Proc Natl Acad Sci U S A 108:4516–4522. https://doi.org/10.1073/
25. Galloway-Pena JR, Smith DP, Sahasrabhojane P, Ajami NJ, pnas.1000080107
Wadsworth WD, Daver NG, Chemaly RF, Marsh L, Ghantoji SS, 34. Turner S, Pryer KM, Miao VP, Palmer JD (1999) Investigating deep
Pemmaraju N, Garcia-Manero G, Rezvani K, Alousi AM, Wargo phylogenetic relationships among cyanobacteria and plastids by
JA, Shpall EJ, Futreal PA, Guindani M, Petrosino JF, Kontoyiannis small subunit rRNA sequence analysis. J Eukaryot Microbiol 46:
DP, Shelburne SA (2016) The role of the gastrointestinal 327–338
microbiome in infectious complications during induction chemo- 35. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD
therapy for acute myeloid leukemia. Cancer 122:2186–2196. (2013) Development of a dual-index sequencing strategy and
https://doi.org/10.1002/cncr.30039 curation pipeline for analyzing amplicon sequence data on the
26. Montassier E, Gastinne T, Vangay P, Al-Ghalith GA, Bruley des MiSeq Illumina sequencing platform. Appl Environ Microbiol 79:
Varannes S, Massart S, Moreau P, Potel G, de La Cochetiere MF, 5112–5120. https://doi.org/10.1128/AEM.01043-13
Batard E, Knights D (2015) Chemotherapy-driven dysbiosis in the 36. Cox MP, Peterson DA, Biggs PJ (2010) SolexaQA: At-a-glance
intestinal microbiome. Aliment Pharmacol Ther 42:515–528. quality assessment of Illumina second-generation sequencing data.
https://doi.org/10.1111/apt.13302 BMC Bioinforma 11:485. https://doi.org/10.1186/1471-2105-11-
27. Hansen SG, Skov MN, Justesen US (2013) Two cases of 485
Ruminococcus gnavus bacteremia associated with diverticulitis. J 37. Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister
Clin Microbiol 51:1334–1336. https://doi.org/10.1128/JCM. EB, Lesniewski RA, Oakley BB, Parks DH, Robinson CJ, Sahl JW,
03382-12 Stres B, Thallinger GG, Van Horn DJ, Weber CF (2009)
28. Graziani F, Pujol A, Nicoletti C, Dou S, Maresca M, Giardina T, Introducing mothur: open-source, platform-independent,
Fons M, Perrier J (2016) Ruminococcus gnavus E1 modulates mu- community-supported software for describing and comparing mi-
cin expression and intestinal glycosylation. J Appl Microbiol 120: crobial communities. Appl Environ Microbiol 75:7537–7541.
1403–1417. https://doi.org/10.1111/jam.13095 https://doi.org/10.1128/AEM.01541-09
29. Lee JY, Arai H, Nakamura Y, Fukiya S, Wada M, Yokota A (2013)
38. Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, Peplies
Contribution of the 7beta-hydroxysteroid dehydrogenase from
J, Glockner FO (2013) The SILVA ribosomal RNA gene database
Ruminococcus gnavus N53 to ursodeoxycholic acid formation in
project: improved data processing and web-based tools. Nucleic
the human colon. J Lipid Res 54:3062–3069. https://doi.org/10.
Acids Res 41:D590–D596. https://doi.org/10.1093/nar/gks1219
1194/jlr.M039834
30. Mohammed Saif M, Farid SF, Khaleel SA, Sabry NA, El-Sayed 39. McMurdie PJ, Holmes S (2013) phyloseq: an R package for repro-
MH (2012) Hepatoprotective efficacy of ursodeoxycholic acid in ducible interactive analysis and graphics of microbiome census da-
pediatrics acute lymphoblastic leukemia. Pediatr Hematol Oncol ta. PLoS One 8:e61217. https://doi.org/10.1371/journal.pone.
29:627–632. https://doi.org/10.3109/08880018.2012.713083 0061217
31. Dabard J, Bridonneau C, Phillipe C, Anglade P, Molle D, Nardi M, 40. Dehbi HM, Jones S, Sohaib SM, Finegold JA, Siggers JH,
Ladire M, Girardin H, Marcille F, Gomez A, Fons M (2001) Stegemann B, Whinnett ZI, Francis DP (2015) A novel curve
Ruminococcin A, a new lantibiotic produced by a Ruminococcus fitting method for AV optimisation of biventricular pacemakers.
gnavus strain isolated from human feces. Appl Environ Microbiol Physiol Meas 36:1889–1900. https://doi.org/10.1088/0967-3334/
67:4111–4118 36/9/1889
32. Yooseph S, Kirkness EF, Tran TM, Harkins DM, Jones MB, 41. Love MI, Huber W, Anders S (2014) Moderated estimation of fold
Torralba MG, O’Connell E, Nutman TB, Doumbo S, Doumbo change and dispersion for RNA-seq data with DESeq2. Genome
OK, Traore B, Crompton PD, Nelson KE (2015) Stool microbiota Biol 15:550. https://doi.org/10.1186/s13059-014-0550-8
composition is associated with the prospective risk of Plasmodium 42. Untergasser A, Cutcutache I, Koressaar T, Ye J, Faircloth BC,
falciparum infection. BMC Genomics 16:631. https://doi.org/10. Remm M, Rozen SG (2012) Primer3–new capabilities and inter-
1186/s12864-015-1819-3 faces. Nucleic Acids Res 40:e115. https://doi.org/10.1093/nar/
33. Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Lozupone gks596
CA, Turnbaugh PJ, Fierer N, Knight R (2011) Global patterns of