American Journal of Infection Control 46 (2018) 54-9

Contents lists available at ScienceDirect

American Journal of Infection Control American Journal of

Infection Control

j o u r n a l h o m e p a g e : w w w. a j i c j o u r n a l . o r g

Major Article

Stalking a lethal superbug by whole-genome sequencing and

phylogenetics: Influence on unraveling a major hospital outbreak of
carbapenem-resistant Klebsiella pneumoniae
Thorsten Kaiser MD a, Knut Finstermeier PhD a, Madlen Häntzsch MSc a,
Sarah Faucheux MD b, Martin Kaase MD c, Tim Eckmanns MD d, Sven Bercker MD, PhD e,
Udo X. Kaisers MD, PhD e, Norman Lippmann MD f,g, Arne C. Rodloff MD, PhD f,g,
Joachim Thiery MD, PhD a, Christoph Lübbert MD, PhD, DTM&H g,h,*
a Institute for Laboratory Medicine, Clinical Chemistry, and Molecular Diagnostics, Leipzig University Hospital, Leipzig, Germany
b Institute for Hospital Hygiene, Leipzig University Hospital, Leipzig, Germany
c Department of Medical Microbiology, National Reference Center for Multidrug-Resistant Gram-negative Bacteria, Ruhr University Bochum, Bochum, Germany
d
Department for Infectious Disease Epidemiology, Robert Koch Institute, Berlin, Germany
e Department of Anesthesiology and Intensive Care Medicine, Leipzig University Hospital, Leipzig, Germany
f Institute for Medical Microbiology and Epidemiology of Infectious Diseases, Leipzig University Hospital, Leipzig, Germany
g
Interdisciplinary Center for Infectious Diseases, Leipzig University Hospital, Leipzig, Germany
h
Division of Infectious Diseases and Tropical Medicine, Department of Gastroenterology and Rheumatology, Leipzig University Hospital, Leipzig, Germany

Key Words: Background: From July 2010-April 2013, Leipzig University Hospital experienced the largest outbreak
Carbapenemase of a Klebsiella pneumoniae carbapenemase 2 (KPC-2)-producing Klebsiella pneumoniae (KPC-2-Kp) strain
Enterobacteriaceae observed in Germany to date. After termination of the outbreak, we aimed to reconstruct transmission
Colonization
pathways by phylogenetics based on whole-genome sequencing (WGS).
Infection
Methods: One hundred seventeen KPC-2-Kp isolates from 89 outbreak patients, 5 environmental KPC-
Epidemiologic analysis
Microbiologic screening 2-Kp isolates, and 24 K pneumoniae strains not linked to the outbreak underwent WGS. Phylogenetic analysis
Klebsiella pneumoniae was performed blinded to clinical data and based on the genomic reads.
Results: A patient from Greece was confirmed as the source of the outbreak. Transmission pathways for
11 out of 89 patients (12.4%) were plausibly explained by descriptive epidemiology, applying strict defi-
nitions. Five of these and an additional 15 (ie, 20 out of 89 patients [22.5%]) were confirmed by phylogenetics.
The rate of phylogenetically confirmed transmissions increased significantly from 8 out of 66 (12.1% for
the time period before) to 12 out of 23 patients (52.2% for the time period after; P < .001) after imple-
mentation of systematic screening for KPC-2-Kp (33,623 screening investigations within 11 months). Using
descriptive epidemiology, systematic screening showed no significant effect (7 out of 66 [10.6%] vs 4 out
of 23 [17.4%] patients; P = .465). The phylogenetic analysis supported the assumption that a contami-
nated positioning pillow served as a reservoir for the persistence of KPC-2-Kp.
Conclusions: Effective phylogenetic identification of transmissions requires systematic microbiologic screen-
ing. Extensive screening and phylogenetic analysis based on WGS should be started as soon as possible
in a bacterial outbreak situation.
© 2018 Association for Professionals in Infection Control and Epidemiology, Inc. Published by Elsevier
Inc. All rights reserved.

* Address correspondence to Christoph Lübbert, MD, PhD, DTM&H, Division of Infectious Diseases and Tropical Medicine, Department of Gastroenterology and Rheumatology,
Leipzig University Hospital, Liebigstr. 20, D-04103 Leipzig, Germany.
E-mail address: christoph.luebbert@medizin.uni-leipzig.de (C. Lübbert).
TK, KF, ACR, JT, and CL participated in the study conception and design. TK, SF, SB, and CL were responsible for patient identification and acquisition of clinical data. SB,
UXK, and CL performed the clinical evaluation of study patients. TK, MH, and NL carried out the major laboratory experiments. TK, KF, MK, TE, SB, ACR, JT, and CL analyzed
the data. CL wrote the manuscript. All authors read and approved the final version.
TK and KF contributed equally as first authors.
ACR, JT, and CL contributed equally as senior authors.
Conflicts of interest: None to report.

0196-6553/© 2018 Association for Professionals in Infection Control and Epidemiology, Inc. Published by Elsevier Inc. All rights reserved.
https://doi.org/10.1016/j.ajic.2017.07.022
 T. Kaiser et al. / American Journal of Infection Control 46 (2018) 54-9
Klebsiella pneumoniae (Kp) is a major cause of nosocomial in-
fections, primarily among debilitated patients.1 Kp carbapenemases
(KPCs) have rapidly spread since 19961,2; outbreaks in Europe are
mainly associated with the blaKPC-2 gene and the globally domi-
nant multilocus sequence type 258 (ST258).2
From July 2010-April 2013, Leipzig University Hospital (LUH) ex-
perienced the largest outbreak of a KPC-2-Kp strain (ST258) in
Germany to date.2,3 The outbreak was linked to a patient trans-
ferred from a hospital in Rhodes, Greece, where KPC-producing
pathogens have been endemic since 2007.2,4 A total of 105 pa-
tients were identified as being either colonized (60 out of 105 [57.1%])
or infected (45 out of 105 [42.9%]) with KPC-2-Kp. Significant clin-
ical features of the outbreak have been described in previous
studies.5-10
Genome sequencing technologies and phylogenetic analysis now
allow researchers to discriminate between closely related bacteri-
al clones of the same lineage, helping to analyze outbreak scenarios.11
For instance, applying this method, a small outbreak of KPC-
producing Kp at the National Institutes of Health Clinical Center in
Bethesda, MD, during 2012 could be understood better and trans-
mission pathways were discovered more easily.12
This article describes the phylogenetic analysis of KPC-2-Kp out-
break isolates based on their genomic sequences and analyzes its
additional value to reconstruct transmission routes in the 34-
month outbreak at LUH.
METHODS
Study design
This was a retrospective study. Eighty-nine out of 105 KPC-2-
Kp positive outbreak patients could be included. Data were collected
by medical record review.
Setting
LUH is a tertiary care facility with approximately 1,400 beds, more
than 50,000 inpatients, and approximately 310,000 outpatients per
year. It has an extensive transplant program involving liver, kidney,
pancreas, lung, bone marrow, and peripheral blood stem cell trans-
plantation. There is a tight exchange of patients between the
gastroenterology/hepatology wards, peripheral surgical wards, and
the interdisciplinary surgical intensive care unit (SICU) by the trans-
plant program and the joint care of patients with visceral diseases.
Patients and definitions
A KPC-2-Kp case was defined as any hospitalized patient in whom
KPC-2-Kp was isolated from a clinical or screening specimen between
June 28, 2010, and April 2, 2013. Clinical infection was deter-
mined based on isolating KPC-2-Kp from a clinical specimen and
a medical diagnosis, whereas colonized patients showed no clini-
cal manifestation. We determined KPC-2-Kp carriage by isolating
the pathogen and assessing carbapenem-resistance by applying sus-
ceptibility testing according to International Organization for
Standardization guideline 20776, confirmed by KPC-2–specific poly-
merase chain reaction (PCR) tests.
In accordance with Centers for Disease Control and Prevention
recommendations, we started active surveillance for all patients on
June 1, 2012, as screening for KPC-2-Kp in fecal samples or rectal
swabs,13 employing CHROMagar KPC chromogenic agar plates
(CHROMagar, Paris, France) and KPC-2–specific PCR tests. Cultures
and PCR tests were applied as systematic screening procedures that
were repeated on a weekly basis for each hospitalized patient
55
(leading to 33,623 microbiologic screening investigations within 11
months).
Using descriptive epidemiology, KPC-2-Kp cases were assigned
to a specific transmission chain if direct room contact with another
KPC-2-Kp-positive case and/or documented care by the same nursing
staff in the same shift could be established. Applying phylogenetics,
cases were considered to belong to the same transmission chain if
they were within the same phylogenetic clade and patients had direct
ward contact with another KPC-2-Kp–positive case and/or had docu-
mented care by the same nursing staff members.
Data sources
Clinical and microbiologic data were retrieved using patients’
charts and LUH’s patient data management system (i.s.h.med; SAP,
Walldorf, Germany).
Bacterial strain typing and sequencing
In total, 152 Kp isolates were available for sequencing. Six samples
were excluded due to insufficient sequencing coverage (<30×). The
remaining 146 samples included 117 specimens from 89 KPC-2–
positive outbreak patients (66 patients with 1 isolate, 19 patients
with 2 isolates, 3 patients with 3 isolates, and 1 patient with 4 iso-
lates; all isolates were obtained within the outbreak period), 5
environmental KPC-2-Kp samples from the outbreak period, 7 speci-
mens of KPC-producing Kp from other countries (Greece, Italy,
Poland, and Brazil), 5 KPC-2–producing Kp isolates from surround-
ing hospitals, 8 Kp isolates without KPC-mediated resistance from
LUH, and 4 Kp isolates from the Faculty of Veterinary Medicine, Uni-
versity of Leipzig. Bacterial DNA was extracted using either the
DNeasy Blood Kit (Qiagen, Hilden, Germany) or the Tissue Gentra
Puregene Blood Kit (Qiagen) for sequencing on the 454 platform
(Roche Sequencing, Pleasanton, CA) and the HiSeq 2000 platform
(Illumina, San Diego, CA), respectively, or phenol chloroform DNA
extraction for sequencing on the PacBio RSII platform (Pacific Bio-
sciences, Menlo Park, CA). DNA samples were sheared using the
Bioruptor (Diagenode, Seraing, Belgium) with 5 cycles (20 seconds
on and 20 seconds off), producing adequate fragment lengths with
an average library size of 650 bp. Shearing was not applied for
samples regarding long read sequencing.
In silico multilocus sequence typing of individual KPC-2-Kp out-
break isolates (based on polymorphisms of rpoB, gapA, mdh, pgi, phoE,
infB, and tonB genes) was performed as described elsewhere.14
Sequencing the reference genome
The first detected KPC-2-Kp isolate from the patient trans-
ferred from Rhodes, Greece, was used to sequence the reference
genome. In a first strategy, DNA samples were sequenced with PacBio
RSII, which supports modes that generate circular consensus se-
quencing short reads and long reads. Circular consensus sequencing
reads were collapsed into consensus read sequences and as-
sembled with long reads into contigs using Pacific Bioscience’s
proprietary software. This returned 131,138 consensus reads (average
read length, 546 bp and maximum read length, 2,567 bp). All reads
were assembled into 125 contigs (average length, 46,232 bp and
maximum length, 482,734 bp). In the second strategy, an aliquot
of the same DNA sample was sequenced on the 454 platform. Quality
filtered reads were assembled into contigs using 454’s proprietary
software Newbler (Roche Sequencing) with default settings. The as-
sembly returned 1,070 contigs (average size, 5,414 bp and maximum
size, 49,594 bp). Contigs from both approaches were further as-
sembled using Minimus2 (https://sourceforge.net/projects/amos/).
All contigs were identified as either of genomic origin or plasmids
56 T. Kaiser et al. / American Journal of Infection Control 46 (2018) 54-9
using the National Center for Biotechnology Information’s Basic Local
Alignment Search Tool (BLAST) database to query the full contig
sequence (https://blast.ncbi.nlm.nih.gov/Blast.cgi). Genes of each
contig were annotated using the bacterial annotation system
(http://wishart.biology.ualberta.ca/basys).15 A total of 109 final contigs
were generated (79 genomic contigs and 30 plasmid contigs). The
average contig lengths were 67,809 bp and 19,018 bp, respective-
ly. Twenty-three contigs with length <500 bp were excluded from
further processing. The final 86 contigs were used as reference se-
quences for read mapping.
Sequencing isolates belonging to the outbreak
Outbreak isolates were sequenced on the HiSeq 2000 platform.
The libraries were constructed using a previously established
protocol.16 Raw paired-end reads were filtered, separated based
on sample-specific barcodes, and mapped to the reference genome
using Burrows-Wheeler Aligner version 0.7.5a-r405 (http://bio
-bwa.sourceforge.net/).17 Then, single nucleotide variants were ex-
tracted using a customized script and applying majority consensus
calls (≥75%). If no majority base could be identified, the nucleo-
tide was designated as “N.” In addition, the following criteria were
applied: present in >1 sample, not in genes with 1 of the 4 anno-
tations (transposon, transposase, intron, or bacteriophage), not in
or directly adjacent to a homopolymer of ≥9 bp. A site was also not
included in the analysis if a minimum sequencing coverage of 30×
or a minimum read mapping score of 30 was not achieved (apply-
ing the manufacturer’s quality scores [https://www.illumina.com/
documents/products/technotes/technote_Q-Scores.pdf]).
Reconstructing the phylogenetic tree
All nucleotide positions of informative value were included in
the phylogenetic analysis. If a position in the whole alignment did
not provide information (ie, isolates showing either the same nucleo-
tide or an “N”), the position was excluded. Alignment gaps were kept
and used if they were outside of homopolymers and provided in-
formation as described before.18 Contigs that were identified as being
of plasmid origin were excluded from phylogenetic reconstruc-
tion. A total of 390,044 informative sites of chromosomal origin were
extracted. A Bayesian phylogenetic tree19 was reconstructed via
Bayesian Evolutionary Analysis Sampling Trees version 1.8.2
(http://beast.bio.ed.ac.uk). The resulting consensus tree was visu-
ally inspected with Figtree (http://tree.bio.ed.ac.uk/software/figtree).
Clades were identified based on posterior branch probabilities ≥ 0.8.
A plasmid contig of 49,399 nucleotides was used as a reference to
analyze the blaKPC-2 gene. For each isolate, reads were mapped against
the full set of reference contigs.
Nucleotide sequence accession numbers
All sequence data were submitted to the European Nucleotide
Archive (www.ebi.ac.uk/ena) under accession No. PRJEB20234.
Environmental sampling and staff screening
Environmental swabs were taken using ESwabs (Copan, Brescia,
Italy); swabs were streaked out directly on the agar plate and were
incubated for 24 hours at 37°C. Staff screening was performed by
collection of 2 separate stool samples per person after obtaining per-
mission. All staff members who had contact to KPC-2-Kp–positive
patients or their direct environment were involved. Hands of staff
members were not systematically investigated for KPC-2-Kp. The
microbiologic analysis was carried out employing CHROMagar KPC
agar plates and KPC-2–specific PCR tests.
Statistical analysis
Statistical analysis was performed using SPSS 20.0 for Windows
(IBM-SPSS Inc, Armonk, NY). Numerical variables were summa-
rized as medians, and categorical variables were given as frequencies
or proportions. Confidence intervals (CIs) for frequencies were cal-
culated based on binomial distribution. Categorical data were
analyzed using Pearson χ2 test or Fisher exact test. The nonpara-
metric Mann-Whitney U test was used to compare 2 independent
groups. A P value (2-sided) < .05 was set as statistically significant.
Ethical approval
This retrospective study was performed in accordance with the
ethical guidelines of the 1964 Declaration of Helsinki and its later
amendments and was approved by the local ethics committee (Uni-
versity of Leipzig register No. 411-12-11032013). The need for
informed consent was waived according to the ethics approval.
Data availability
The sequence data and metadata of this study can be accessed
at http://www.ebi.ac.uk/ena/view/PRJEB20234. All other data are
available from the corresponding author on reasonable request.
RESULTS
Descriptive epidemiologic analysis
The index case appeared to be a 66-year-old patient trans-
ferred from a hospital in Rhodes, Greece, to LUH on June 28, 2010.5,9
He had nosocomial pneumonia and required mechanical ventila-
tion. An isolate of KPC-2-Kp was recovered from bronchoalveolar
lavage fluid from this patient 10 days later. By the time the out-
break ended (April 2013), 104 further KPC-2-positive cases had been
detected. Forty-eight of the 105 cases (45.7%) were identified as being
KPC-positive in the interdisciplinary SICU (sections A-C, with a total
of 58 beds). The remaining cases were hospitalized in 19 different
wards. The proportion of cases with bacteremic KPC-2-Kp infec-
tions was 27.7% (18 out of 65) before systematic screening was
implemented (95% CI, 17.3-40.2) and decreased to 2.5% (1 out of 40)
afterward (95% CI, 0.1-13.2; P = .005).
Performing an epidemiologic analysis based on a line list tem-
plate by using an electronic spreadsheet (as proposed by the Centers
for Disease Control and Prevention [https://www.cdc.gov/urdo/
downloads/linelisttemplate.pdf]), transmission via the hands of
the health care personnel was suspected as the most likely way
of spread, presumably with the contribution of undetected KPC-2-Kp
cases before systematic screening was established (silent
dissemination20,21). This assumption is supported by the literature.21-24
Descriptive epidemiology clearly assigned 11 patients to a specif-
ic transmission chain.
Strain typing
Multilocus sequence typing confirmed the ST258 lineage in KPC-
2-Kp outbreak isolates.
Phylogenetic analysis
Phylogenetic analysis based on the genomic whole-genome se-
quencing (WGS) data confirmed the blaKPC-2-positive isolate obtained
from the index patient to be the source of the outbreak. Affected
patients were assigned to 20 phylogenetic clades, 1 of which was
epidemiologically related to a positioning pillow (part of a set of
 T. Kaiser et al. / American Journal of Infection Control 46 (2018) 54-9
pillows designed for the prone position of intensive care unit pa-
tients with severe gas exchange disorders)10 (Fig 1).
All KPC-producing isolates from different countries were con-
firmed as outgroups. However, 2 isolates from Greece were
phylogenetically more closely related to the Leipzig outbreak strain
than were the other foreign isolates. All other samples included to
root the phylogenetic tree were genetically different from the out-
break strain.
Phylogenetics versus epidemiologic reconstruction of transmission
chains
Clinically relevant clades and related transmission pathways iden-
tified by WGS and phylogenetics are represented in a normalized
phylogenetic tree (Fig 1). Transmission pathways for 11 patients
(12.4%) could be plausibly explained using descriptive epidemiol-
ogy (95% CI, 6.3-21.0). Transmission pathways comprising 5 of these
11 patients were confirmed using phylogenetic analysis. For 15 ad-
ditional patients, transmission pathways were reconstructed via
phylogenetics only. The exact modes of transmission could not be
unraveled in 63 patients (70.8%; 95% CI, 60.2-79.9); thus, only 26
patients (29.2%) were assigned to a specific transmission chain (95%
CI, 20.1-39.8).
The epidemiologic unraveling ratio did not increase signifi-
cantly after systematic screening for KPC-2-Kp was implemented:
from 7 out of 66 (10.6% during the time period before; 95% CI, 4.4-
20.6) to 4 out of 23 patients (17.4% during the time period after;
95% CI, 5.0-38.8; P = .465). However, detected transmissions by phy-
logenetic analysis increased significantly from 8 out of 66 (12.1%;
95% CI, 5.4-22.5) to 12 out of 23 patients (52.2%; 95% CI, 30.6-
73.2; P < .001).
Evolutionary distance of the outbreak strain
The mean number of single nucleotide variants within the out-
break samples (clade 1) was 34 (median, 32; range, 3-155).
Comparing the sample of the index patient of the outbreak with
the 7 specimens of KPC-producing Kp from other countries (Greece,
Italy, Poland, and Brazil) resulted in a mean number of single nucleo-
tide variants of 9,514 (median, 1,158; range, 46-30,499). Considering
the whole dataset including all outgroups, the mean number of single
nucleotide variants was 16,767.
Phylogenetic analysis of other KPC-2-producing isolates from
Western Saxony
We included 5 KPC-2-Kp isolates from other hospitals of the same
region in Germany into our analysis. Notably, these isolates could
be clearly assigned to the Leipzig scenario. One isolate showed a
close relationship with an outbreak isolate from a hematology patient
(case #27); however, no strong epidemiologic links within the out-
break setting were demonstrated. The other 4 isolates were
genetically closely related to the index patient and to each other,
but they did not have a direct link to individual outbreak patients.
Environmental sampling and staff screening
One thousand thirty environmental swabs were taken from the
patient environment of KPC-2-Kp–positive cases during the out-
break period. Five of 1,030 environmental cultures (0.5%) turned out
to be positive (1 from a patient bed, 1 from a medical nebulizer, 1
from a stethoscope, 1 from a radio button, and 1 from a toilet chair);
all of them were associated with clade 1_1_5_2 (Fig 1). One thou-
sand two hundred fifty-eight stool samples from 629 staff members
57
were screened for KPC-2-Kp, but no evidence for individuals being
colonized by the outbreak strain was found.
DISCUSSION
From July 2010-April 2013, LUH experienced the largest out-
break of a KPC-2-Kp strain belonging to the globally dominant ST258
lineage observed and published in Germany to date. Epidemio-
logic and phylogenetic analyses consistently indicated high numbers
of additional undiscovered cases, especially before systematic mi-
crobiologic screening. Consequently, with systematic screening for
KPC-2-Kp, 13 new cases were detected within 14 days. The spread
of the outbreak strain from the SICU, which clearly served as the
main distribution hub, likely led to repeated reintroduction to various
wards.
The start of the outbreak with a single patient from Greece high-
lights the importance of screening patients with risk factors for
carrying multidrug-resistant bacteria, such as individuals who were
previously hospitalized in endemic regions.2,13,21 We show that our
phylogenetic analysis is superior to conventional descriptive epi-
demiologic methods for reconstructing transmission pathways; this
is supported by the results of systematic screening for KPC-2-Kp.
Reducing unrecognized (silent) transmissions via screening efforts
significantly increased the transmission identification rate.
Transmission of KPC-2-Kp due to a contaminated positioning
pillow used in the SICU could be retraced in 4 cases without the
results of the phylogenetic analysis. However, we were able to iden-
tify the index patient of this cluster. An additional 4 patients (Fig 1)
were identified only by means of phylogenetic analysis. Unfortu-
nately, these results were not available during the outbreak, and the
particular contaminated pillow was only identified as a source of
perpetuated transmissions of KPC-2-Kp more than 4 months after
the outbreak ended.10 Therefore, no isolate from the pillow could
be included in the phylogenetic analysis.
Moreover, the phylogenetic analysis revealed 1 female patient
(case #80) who had been a silent transmitter for more than 1 year.
This individual was identified as colonized with KPC-2-Kp by sys-
tematic screening during July 2012 in the rheumatology ward, which
was not affected by the outbreak strain until that time. This patient
was most likely colonized with KPC-2-Kp in the SICU much earlier,
during March 2011. This previous stay was overlooked during the
outbreak. Additionally, we detected common risk factors within dif-
ferent phylogenetic clades, which revealed an accumulation of
patients who had undergone endoscopic procedures or hemodi-
alysis on the same day, although there were no positive results for
KPC-2-Kp from microbiologic surveillance of endoscopes as well as
from environmental screenings, with the exception of 5 positive find-
ings that could be assigned to 1 specific clade (1_1_5_2). Therefore,
we assume that other potential reservoirs of KPC-2-Kp such as mobile
hemodialysis machines were not identified during the outbreak.
The resulting outbreak of KPC-2-Kp with a mortality rate of more
than 40% had dramatic consequences not only for the affected pa-
tients, but also for the structure, hygiene management, and costs
at LUH.5,7,9 Moreover, the phylogenetic analysis suggested ongoing
transmission of the outbreak strain in surrounding hospitals. Why
Western Saxony did not develop into a KPC-endemic area due to
the Leipzig outbreak25 remains unanswered, but we assume that bac-
terial fitness factors and delayed, but consistent implementation of
overall hygiene standards played a role.
Limitations
Most of the data were collected from patient records, the ex-
pressiveness of which depends on the quality of documentation.
Furthermore, by extending barrier precautions and establishing ward
58 T. Kaiser et al. / American Journal of Infection Control 46 (2018) 54-9

Fig 1. Phylogenetic analysis combined with epidemiologic information on clinically relevant clades of the Klebsiella pneumoniae carbapenemase (KPC)-2-Kp outbreak at
Leipzig University Hospital (LUH), Leipzig, Germany. One hundred five outbreak-related cases of KPC-2-Kp were detected from July 8, 2010-April 2, 2013. Forty-four out of
105 affected patients (41.9%) died in the hospital, 7 of whom died due to invasive KPC-2-Kp infections. KPC-2-Kp infections were the multifactorial cause of death in 14
other patients. A phylogenetic tree with normalized branches was used for graphic representation. Corresponding sample identification numbers were added to the ter-
minal branches. Clades were identified based on posterior branch probabilities of ≥ 0.8 (red shadings). Information on clinically relevant clades with conclusive epidemiologic
explanatory patterns is given in concise text boxes. Cases with 2 or more different isolates available for whole-genome sequencing may have been assigned to more than 1
clade. The yellow box (clade 1) includes all isolates that could be traced phylogenetically to the index case. Outgroup samples are highlighted with gray background color.
For cases #59, #67, #69, #71, #72, #73, #74, #75, #76, #77, #78, #79, #81, #85, #93, and #96 there were no isolates available for whole-genome sequencing. *Sample
identification numbers of outgroup KPC-2-Kp isolates from Greece (provided by the German National Reference Center for Multidrug-Resistant Gram-Negative Bacteria).
+KPC-2-Kp–positive environmental samples. These samples could also be plausibly linked to patients of clade 1_1_5_2 by descriptive epidemiology. SICU-C, surgical inten-

sive care unit section C.

 T. Kaiser et al. / American Journal of Infection Control 46 (2018) 54-9
sections especially designed for cohorting KPC-2-Kp–positive in-
dividuals and contact patients, medical procedures, and exposures
changed during the outbreak. The results of the phylogenetic anal-
ysis could have been improved by including multiple isolates from
each KPC-2-Kp–positive patient, which was possible only in a mi-
nority of cases.
CONCLUSIONS
The efficiency of WGS and phylogenetic analysis may be signifi-
cantly improved by the combined application of systematic screening
measures. Due to advances in sequencing and in information tech-
nologies, respective results could be available almost in real time
in future outbreak situations. This would improve infection control
and prevention measures by elucidating the start and extent of the
outbreak, confirming possible transmission routes and identifying
most probable sources.
Acknowledgments
The authors thank Professor Svante Pääbo, PhD, and his staff at
the Max Planck Institute for Evolutionary Anthropology (Leipzig,
Germany) for sequencing on the HiSeq 2000 platform (Illumina, San
Diego, CA), and GATC Biotech (Konstanz, Germany) for sequencing
on the PacBio RSII platform (Pacific Biosciences, Menlo Park, CA).
The authors also thank Andreas Knaust, MD (previously head of the
Hospital Hygiene Staff Unit, Leipzig University Hospital, Leipzig,
Germany), and Thilo Busch, PhD (Department of Anesthesiology and
Intensive Care Medicine, Leipzig University Hospital, Leipzig,
Germany), for their contributions to this work.
References
1. Nordmann P, Cuzon G, Naas T. The real threat of Klebsiella pneumoniae
carbapenemase producing bacteria. Lancet Infect Dis 2009;9:228-36.
2. Munoz-Price LS, Poirel L, Bonomo RA, Schwaber MJ, Daikos GL, Cormican M, et al.
Clinical epidemiology of the global expansion of Klebsiella pneumoniae
carbapenemases. Lancet Infect Dis 2013;13:785-96.
3. Steinmann J, Kaase M, Gatermann S, Popp W, Steinmann E, Damman M, et al.
Outbreak due to a Klebsiella pneumoniae strain harbouring KPC-2 and VIM-1 in
a German University Hospital, July 2010 to January 2011. Euro Surveill
2011;16:pii=19944.
4. Souli M, Galani I, Antoniadou A, Papadomichelakis E, Poulakou G, Panagea T,
et al. An outbreak of infection due to beta-lactamase Klebsiella pneumoniae
carbapenemase 2-producing K. pneumoniae in a Greek University Hospital:
molecular characterization, epidemiology, and outcomes. Clin Infect Dis
2010;50:364-73.
5. Ducomble T, Faucheux S, Helbig U, Kaisers UX, König B, Knaust A, et al. A large
hospital outbreak of KPC-2-producing Klebsiella pneumoniae: investigating
mortality and the impact of screening for KPC-2 with polymerase chain reaction.
J Hosp Infect 2015;89:179-85.
59
6. Lübbert C, Faucheux S, Becker-Rux D, Laudi S, Dürrbeck A, Busch T, et al. Rapid
emergence of secondary resistance to gentamicin and colistin following selective
digestive decontamination in patients with KPC-2-producing Klebsiella
pneumoniae: a single-centre experience. Int J Antimicrob Agents 2013;452:565-
70.
7. Lübbert C, Becker-Rux D, Rodloff AC, Laudi S, Busch T, Bartels M, et al. Colonization
of liver transplant recipients with KPC-producing Klebsiella pneumoniae is
associated with high infection rates and excess mortality: a case-control analysis.
Infection 2014;42:309-16.
8. Lübbert C, Rodloff AC, Laudi S, Simon P, Busch T, Mössner J, et al. Lessons learned
from excess mortality associated with Klebsiella pneumoniae carbapenemase-2-
producing K. pneumoniae in liver transplant recipients. Liver Transpl
2014;20:736-8.
9. Lübbert C, Lippmann N, Busch T, Kaisers UX, Ducomble T, Eckmanns T, et al.
Long-term carriage of Klebsiella pneumoniae carbapenemase-2-producing K.
pneumoniae after a large single-center outbreak in Germany. Am J Infect Control
2014;42:376-80.
10. Lippmann N, Lübbert C, Kaiser T, Kaisers UX, Rodloff AC. Clinical epidemiology
of Klebsiella pneumoniae carbapenemases. Lancet Infect Dis 2014;14:271-2.
11. Punina NV, Makridakis NM, Remnev MA, Topunov AF. Whole-genome sequencing
targets drug-resistant bacterial infections. Hum Genomics 2015;9:19.
12. Snitkin ES, Zelazny AM, Thomas PJ, Stock F, Henderson DK. NISC Comparative
Sequencing Program Group. Tracking a hospital outbreak of carbapenem-resistant
Klebsiella pneumoniae with whole-genome sequencing. Sci Transl Med
2012;4:148ra116.
13. Centers for Disease Control and Prevention (CDC). Guidance for control
of infections with carbapenem-resistant or carbapenem-producing
Enterobacteriaceae in acute care facilities. MMWR Morb Mortal Wkly Rep
2009;58:256-60.
14. Diancourt L, Passet V, Verhoef J, Grimont PA, Brisse S. Multilocus sequence typing
of Klebsiella pneumoniae nosocomial isolates. J Clin Microbiol 2005;43:4178-82.
15. Van Domselaar GH, Stothard P, Shrivastava S, Cruz JA, Guo A, Dong X. BASys:
a web server for automated bacterial genome annotation. Nucleic Acids Res
2005;33(Web Server issue):W455-9.
16. Meyer M, Kircher M. Illumina sequencing library preparation for highly
multiplexed target capture and sequencing. Cold Spring Harb Protoc
2010;2010:pdb.prot5448.
17. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler
Transform. Bioinformatics 2009;25:1754-60.
18. McTavish EJ, Steel M, Holder MT. Twisted trees and inconsistency of tree
estimation when gaps are treated as missing data—The impact of model
mis-specification in distance corrections. Mol Phylogenet Evol 2015;93:289-
95.
19. Didelot X, Gardy J, Colijn C. Bayesian inference of infectious disease transmission
from whole-genome sequence data. Mol Biol Evol 2014;31:1869-79.
20. Viau RA, Hujer AM, Marshall SH, Perez F, Hujer KM, Briceño DF, et al. “Silent”
dissemination of Klebsiella pneumoniae isolates bearing K. pneumoniae
carbapenemase in a long-term care facility for children and young adults in
Northeast Ohio. Clin Infect Dis 2012;54:1314-21.
21. Munoz-Price LS, Quinn JP. Deconstructing the infection control bundles for the
containment of carbapenem-resistant Enterobacteriaceae. Curr Opin Infect Dis
2013;26:378-87.
22. Haverkorn MJ, Michel MF. Nosocomial Klebsiellas. II. Transfer in a hospital ward.
J Hyg 1979;82:195-205.
23. Pittet D, Hugonnet S, Harbarth S, Mourouga P, Sauvan V, Touveneau S, et al.
Effectiveness of a hospital-wide programme to improve compliance with hand
hygiene. Infection Control Programme. Lancet 2000;356:1307-12.
24. Derde LP, Cooper BS, Goossens H, Malhotra-Kumar S, Willems RJ, Gniadkowski
M, et al. Interventions to reduce colonisation and transmission of antimicrobial-
resistant bacteria in intensive care units: an interrupted time series study and
cluster randomised trial. Lancet Infect Dis 2014;14:31-9.
25. Ehrhard I, Karaalp AK, Hackel T, Höll G, Rodewald N, Reif U, et al. Prevalence of
carbapenemase-producing bacteria in hospitals in Saxony, Germany.
Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2014;57:406-
13.