Research Report

Journal of International Medical Research

2015, Vol. 43(3) 364–377
Randomized controlled ! The Author(s) 2015
Reprints and permissions:
clinical trial evaluating sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0300060514561135
multiplex polymerase chain imr.sagepub.com

reaction for pathogen

identification and therapy
adaptation in critical care
patients with pulmonary or
abdominal sepsis
Sascha Tafelski1,*, Irit Nachtigall1,*,
Thomas Adam2, Stefan Bereswill2, Jana Faust1,
Andrey Tamarkin1, Tanja Trefzer1, Maria Deja1,
Evgeny A Idelevich3, Klaus-Dieter Wernecke4,5,
Karsten Becker3, Claudia Spies1 for the
Molecular Diagnostics of Sepsis Study Group

Abstract
Objective: To determine whether a multiplex polymerase chain reaction (PCR)-based test could
reduce the time required for initial pathogen identification in patients in an intensive care unit
(ICU) setting.
Methods: This double-blind, parallel-group randomized controlled trial** enrolled adults with
suspected pulmonary or abdominal sepsis caused by an unknown pathogen. Both the intervention
and control groups underwent the standard blood culture (BC) testing, but additional pathogen
identification, based on the results of a LightCyclerÕ SeptiFast PCR test, were provided in the
intervention group.

1
Department of Anaesthesiology and Intensive Care
Medicine, Charité–Universitaetsmedizin Berlin, Berlin,
Germany *These authors contributed equally to the manuscript.
2
Institute for Microbiology and Hygiene Berlin, Charité– **Trial registration number: ISRCTN70694559.
Universitaetsmedizin Berlin, Berlin, Germany Corresponding author:
3
Institute of Medical Microbiology, University Hospital Dr Claudia Spies, Department of Anaesthesiology and
Münster, Münster, Germany Intensive Care Medicine, Charité–Universitaetsmedizin
4
Institute of Medical Biometry, Charité– Berlin, Campus Charité Mitte and Campus Virchow-
Universitaetsmedizin Berlin, Berlin, Germany Klinikum, 1 Augustenburger Platz, 13353 Berlin, Germany.
5
SOSTANA GmbH, Berlin, Germany Email: claudia.spies@charite.de

Creative Commons CC-BY-NC: This article is distributed under the terms of the Creative Commons Attribution-NonCommercial
3.0 License (http://www.creativecommons.org/licenses/by-nc/3.0/) which permits non-commercial use, reproduction and
distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access page
(http://www.uk.sagepub.com/aboutus/openaccess.htm).
Tafelski et al. 365

Results: The study enrolled 37 patients in the control group and 41 in the intervention group.
Baseline clinical and demographic characteristics were similar in both groups. The PCR-based test
identified a pathogen in 10 out of 41 (24.4%) patients in the intervention group, with a mean
duration from sampling to providing the information to the ICU of 15.9 h. In the control group, BC
results were available after a significantly longer period (38.1 h).
Conclusion: The LightCyclerÕ SeptiFast PCR test demonstrated a significant reduction in the
time required for initial pathogen identification, compared with standard BC.

Keywords
LightCyclerÕ SeptiFast, multiplex polymerase chain reaction, sepsis, blood culture, pneumonia,
abdominal sepsis

Date received: 18 June 2014; accepted: 29 October 2014

SeptiFast PCR test (Roche Diagnostics

Introduction
Based on data from a large international
prevalence study, half of all patients being
treated on an intensive care unit (ICU)
developed an infection and this was asso-
ciated with a doubling of the mortality rate.1
Sepsis remains one of the most common
complications in critically ill patients and is
associated with fatal outcomes.2 The
ongoing continuum from infection to
sepsis and the resulting organ dysfunction
requires prompt pathogen identification and
therapeutic intervention, because time has
been identified as a critical factor that
influences outcomes.3–5 Based on national
and international guidelines, diagnostics
should be tailored according to patient
status, and diagnostic procedures should
never delay initiation of empirical anti-
microbial therapy in patients with septic
shock.5,6 On the other hand, blood culture
(BC) diagnostics and other microbiological
culture-based methods depend on living
organisms; thus, antibiotic therapy reduces
the microbiological detection rate.7 Modern
diagnostic approaches include bacterial
identification based on genomic polymerase
chain reaction (PCR) analytics; these meth-
ods are not dependent on the viability of the
organisms being sampled.8 One of these
innovative methods is the LightCyclerÕ
GmbH, Mannheim, Germany), which is a
commercially available multiplex PCR
system for the detection of bacterial and
fungal DNA in blood. This assay demon-
strated sufficient sensitivity (62–80%)9,10
and excellent specificity (70–95%) in several
observational studies.9,11
Prompt microbiological identification of
the underlying pathogen in infected patients
and the consequent adaption of the anti-
microbial therapy might lead to improved
patient outcomes in the ICU. At a time when
there is a growing problem of drug resistant
micro-organisms, there is a considerable
need for rapid diagnostic tests.12 The time
advantage for PCR diagnostics might be of
importance especially in patients being trea-
ted with empirical antimicrobial therapy
that does not target the specific causative
species. However, the impact of PCR-based
pathogen identification on infection man-
agement in the ICU setting has not been
evaluated in randomized controlled trials
(RCTs) to date.
Against this background, the aim of the
present study was to determine whether
rapid pathogen identification using a multi-
plex PCR test in ICU patients could reduce
the time required for initial pathogen
identification and subsequent antimicrobial
therapy modification to target the specific
366
pathogen. We hypothesized that the rapid
availability of the results of the PCR-based
pathogen identification would be accompa-
nied by a reduced delay of antimicrobial
therapy modification in critically ill patients
with severe infections.
Patients and methods
Study design
This double-blind, parallel-group RCT was
undertaken at two university hospitals in
Berlin, Germany: Charité University
Hospital Campus Charité Mitte; Charité
University Hospital Campus Charité
Virchow Klinikum. A total of six ICUs
enrolled patients between August 2010 and
March 2012; five anaesthesiological ICUs
and one surgical ICU. This study was
approved by the local Ethics Review
Board (Ethics Commission, Charité–
Universitaetsmedizin Berlin Campus
Charité Mitte, Berlin, Germany) and the
data safety authorities; it was also registered
with the International Standard
Randomized Controlled Trial number
ISRCTN70694559. Written informed con-
sent was obtained from every included
patient or their legal representative.
Patient inclusion and randomization
Patients were eligible for study inclusion
when they presented with signs of sepsis of
suspected abdominal or pulmonary origin,
caused by an unknown pathogen when BC
diagnostics were indicated. Sepsis was
defined as suspected or proven infection
causing systemic inflammation with at least
two of the following: (i) leucocyte count <4
or >12 /nl; (ii) body temperature <36 C or
fever >38 C; (iii) tachypnoea >20/min or
hyperventilation (paCO2 <32 mmHg); (iv)
tachycardia >90 bpm.5 Infections were
defined by the treating physicians using
standardized criteria for infection manage-
ment implemented on the participating
Journal of International Medical Research 43(3)
wards since 2006.13 All patients were
included only once even if they had multiple
episodes of sepsis during their ICU stay.
Additionally, infection onset was required to
be <72 h, to reduce the risk of detecting
persistent circulating DNA. Exclusion cri-
teria were age <18 years, pregnancy, police
custody, missing or withdrawn informed
consent or participation in another pro-
spective clinical study.
Patients fulfilling the inclusion criteria
were asked to provide written informed
consent. A substantial number of critically
ill patients are unable to give consent due to
their clinical status, their need for mechan-
ical ventilation, or altered mental status. In
these cases, patients were included in the
randomization process with a delayed con-
sent procedure. Informed consent was
obtained as soon as possible, from the
patient or their legal representative.
Eligible patients 18 years old were
randomized to either the control or inter-
vention group in a 1 : 1 ratio, according to
consecutive sample processing by the
Institute for Microbiology and Hygiene
Berlin, Charité–Universitaetsmedizin Berlin,
Berlin, Germany. Randomization was
achieved using computer-generated random
numbers and performed in blocks of four.
The randomization scheme was not known
to the treating physicians on the ICUs. To
minimize the risk of bias, allocation was
blinded to all clinical study participants and
staff members on the ICU. Furthermore, to
achieve maximum allocation concealment,
treating physicians on the ICUs were not
aware of group assignment until the results
of the multiplex PCR analysis were available
and communicated to the study ICU wards.
Microbiological procedures, study
intervention, data collection
Microbiological diagnostics were performed
on participating ICU wards, whenever the
patient’s clinical condition indicated
Tafelski et al.
possible systemic infection. Typically, in
patients with suspected pneumonia, two
BC series (each containing one anaerobic
and one aerobic bottle) and samples from
the respiratory tract were taken. In sus-
pected abdominal infection, two BC series
(each with 10 ml of blood) and at least one
culture from the infection site (e.g. intrao-
perative sample) were obtained. In all
patients, an additional blood sample was
taken from a venous or arterial site; a sterile
BC puncture was obtained and an add-
itional volume of 1.5 ml of whole blood was
added to a 1.8 mg/ml ethylenediaminetetra-
acetic acid sample tube. Samples were stored
at 7 C until transfer to the laboratory.
In patients assigned to the intervention
group, biological samples were processed
immediately for the LightCyclerÕ SeptiFast
PCR test (see below for details) and results
were communicated to the physician team on
the study ICU ward as soon as they were
available; in the control group, biological
samples were stored at 80 C for later ana-
lysis. All microbiological samples were pro-
cessed using standard methods as required by
the clinical care process. Positive results from
the BC Gram staining were immediately
communicated by the microbiologist to the
treating physicians on the study ICU wards.
BC bottles (BACTECTM; Becton Dickinson,
Franklin Lakes, NJ, USA) were inoculated
for 6 days until growth was detected. Positive
bottles were streaked onto a set of agar plates
and subjected to direct Gram staining. Any
colonies that grew in culture were identified
using matrix-assisted laser desorption
ionization–time of flight equipment
(MALDI-TOF, microflex LT; Bruker
Daltonic, Bremen, Germany) or biochemical
identification tests (VitekÕ 2; bioMérieux, La
Balme-les-Grottes, France). Commercial
procedures were used for antimicrobial sus-
ceptibility testing (VitekÕ 2 or EtestÕ ;
bioMérieux).
The multiplex PCR evaluated in this trial
was the commercially available
367
LightCyclerÕ SeptiFast test using a sequen-
tial semiautomatic workflow (Roche
Diagnostics GmbH, Mannheim,
Germany). The sample processing included
mechanical cell lysis using SeptiFast
MagNALyser followed by magnetic adsorp-
tion DNA purification in the MagNA Pure
Compact System and detection of pathogen-
specific DNA in a LightCyclerÕ PCR instru-
ment. All results were processed using the
LightCyclerÕ SeptiFast identification soft-
ware provided by the manufacturer. The test
system allows identification of 25 common
BC pathogens without analysis of resistance
patterns. The LightCyclerÕ SeptiFast PCR
test system is described in detail elsewhere.14
Clinical data were taken prospectively
from all patients during their ICU stay
based on all available written information
and the electronic patient chart. This system
contains all relevant data for every ICU day
including laboratory data, physiological
parameters (e.g. heart rate, body tempera-
ture), as well as information regarding ven-
tilation, results from physical examinations
and diagnostic or therapeutic intentions).
Measurements included the Simplified
Acute Physiology Score II to assess severity
of disease on admission.15 In patients receiv-
ing any immunosuppressive agents or
chemotherapy within the past 6 weeks, an
impaired immune system was recorded.
Patients fulfilled criteria for septic shock
when hypotension (systolic blood pressure
<90 mmHg) was noted or when there was
the need for vasopressor usage to maintain a
sufficient blood pressure. Medical history of
patients was also evaluated for documented
alcohol abuse. Microbiological information
was taken from the main hospital patient
database.
Study endpoints
The study design was intended to be com-
parable with the clinical situation where
multiplex PCR testing would be an available
368
diagnostic option, comparable with current
standard BCs.3 Multiplex PCR assays are
undergoing rapid development; bedside tests
as well as automated analyses are already
available. BC bottles are transported to the
laboratory throughout the day and are
inoculated even during the night and at
weekends. The multiplex PCR test analysed
in this study required 170 min of skilled
hands-on time and thus was only available
during the daytime and not at weekends.
For the purpose of this study, the availabil-
ity of the multiplex PCR test processing was
requested to be within 12 h after sampling.
Against this background, patients were
included between 18.00 h and 06.00 h for
comparable transport durations and dur-
ation until sample processing for both treat-
ment arms. Therefore, this study included
patients with progression of infection during
this timeframe. However, when BC diagnos-
tics had already been undertaken during
daytime hours, patients did not fulfil the
inclusion criteria. The choice of antimicro-
bial therapy and the clinical-care process
were completely under the responsibility of
the ICU physicians without any interference
from the study team. Members of the study
team were not part of the clinical care staff
on ICU wards.
The primary aim of the study was to
compare the duration from sampling until
availability of the microbiological results.
The time was calculated from the moment of
sterile puncture of the vein and allocation of
the necessary blood sample for both tests
(LightCyclerÕ SeptiFast PCR and BC bot-
tles) until the information from the
LightCyclerÕ SeptiFast PCR test or a posi-
tive BC result were communicated by the
microbiologist to the ICU physician by
telephone. Furthermore, this study aimed
to assess the influence of the additional
information on therapy modification, and
to describe the corresponding time frame for
the bedside decision. Therefore, all related
changes in antimicrobial therapy were
Journal of International Medical Research 43(3)
monitored. Secondary endpoints included a
comparison of the spectrum of pathogens
that were detected in the two groups and
the sensitivity and specificity of the
LightCyclerÕ SeptiFast PCR test.
Statistical analyses
All statistical analyses were performed using
the SPSSÕ statistical package, version 19.0
(SPSS Inc., Chicago, IL, USA) for
WindowsÕ . Variables were analysed using
numbers and percentages, median and quar-
tiles (25%–75% percentile), or mean  SD
based on scale level and distribution of
normality, as appropriate. All tests for stat-
istical significance were performed with a
two-sided alpha < 0.05. Depending on the
scale level, Fisher’s exact 2-test, Student’s
t-test, or Mann–Whitney U-test were used.
It was not possible to calculate the sample
size requirements a priori: the time demand
for the decision-making process on the basis
of pathogen detection using PCR technol-
ogy was unpredictable, based on previous
data. Consequently, a conservative approach
was used by enrolling 100 patient samples.
Based on the findings of the study in the
context of its pilot nature, a power analysis
was performed to evaluate the number of
patients needed to achieve a power of 80%
with a two-sided alpha of 5% (nonparametric
unpaired Wilcoxon test), to demonstrate a
difference in the time requirement for the
PCR technology in a future confirmatory
RCT (nQuery AdvisorÕ software, version
5.0; Statistical Solutions, Boston, MA, USA).
A P-value < 0.05 was considered statistically
significant.
Results
During the study period, 100 consecutive
patients were eligible for inclusion in the
study. Of these, 22 were unable to provide
written informed consent because the
patients or their legal representatives refused
Tafelski et al. 369

to participate in the study or the patient
experienced rapid clinical deterioration and
subsequent death with no legal representa-
tive being nominated. According to the
institutional review board vote, all data
from these 22 excluded patients were
deleted. As a consequence, the study fol-
lowed-up and analysed 78 patients: 37 in the
control group and 41 in the intervention
group (Figure 1).
Baseline demographic and clinical char-
acteristics did not differ significantly
between the two groups (Table 1). The
most common clinical condition was sepsis
of pulmonary origin, which was observed in
78% (29 of 41) of patients in the control
group and 63% (26 of 41) in the intervention
group.
In 10 of 41 (24.4%) patients in the
intervention group, the LightCyclerÕ
SeptiFast PCR test identified one of the
following pathogens in the bloodstream
(Table 2): Staphylococcus aureus,
Pseudomonas aeruginosa, Klebsiella spp.,
Enterococcus faecium, Enterobacter spp.,
Escherichia coli, Candida parapsilosis and
Candida albicans. In the intervention group,
corresponding BCs grew pathogens in five of
41 (12.2%) patients. In one BC sample,
S. epidermidis was detected that was not
found by the LightCyclerÕ SeptiFast PCR
test. In all of the other cases, all of the
BC results were also detected by the
LightCyclerÕ SeptiFast PCR test.
In the control group, BCs were positive in
seven of 37 patients (18.9%) growing
S. epidermidis, E. faecium, E. faecalis and
S. aureus (Table 2). In the post-hoc
LightCyclerÕ SeptiFast PCR test of
collected samples, eight pathogens were
identified (E. coli, E. faecium, E. faecalis,
Streptococcus pneumoniae, S. aureus and
S. epidermidis). In one case with BCs grow-
ing two pathogens (E. faecalis and
S. epidermidis), the LightCyclerÕ SeptiFast
PCR test only identified E. faecalis.
The concordance rate between the BC
results and those of the LightCyclerÕ
SeptiFast PCR test were analysed for
the intervention and control groups.

Assessed for randomization

(n = 100)

Allocaon1:1

Allocated to control (n = 50) Allocated to intervention (n = 50)

♦ Declined to participate (n = 13) ♦ Declined to participate (n = 9)

Analysis

All patients followed-up and All patients followed-up and

analysed (n = 37) analysed (n = 41)

Figure 1. Consort flow diagram for patient enrolment into a randomized controlled clinical trial that
evaluated whether rapid pathogen identification (using a multiplex polymerase chain reaction test) reduced
time required for the initial pathogen identification and subsequent therapy modification to target the specific
pathogen, in patients treated in intensive care units.
370 Journal of International Medical Research 43(3)

Table 1. Baseline clinical and demographic For analysis of the test performance of
characteristics of patients treated in intensive care the LightCyclerÕ SeptiFast PCR test, BC
units (ICUs), randomized to undergo either a results were used as the diagnostic gold
multiplex polymerase chain reaction test for initial standard for detecting bacteraemia. The
pathogen identification (intervention group) or resulting sensitivity for a positive
standard blood culture analysis (control group).
LightCyclerÕ SeptiFast PCR test result
Control Intervention was 58.3% (seven of 12 samples) with a
group group corresponding specificity of 74.2% (49 of 66
Parameter n ¼ 37 n ¼ 41 samples).
For the main study endpoint, the dur-
Age, years 59 (47–67) 67 (53–74)
ation between biological sampling and avail-
ICU stay, days 32 (16–57) 34 (13–65)
ability of microbiological results by
Hospital stay, days 37 (20–76) 53 (33–79)
Duration of 608 (246–990) 522 (76–1206) telephone to the ICU physicians was calcu-
ventilation, h lated. In the intervention group, this process
Female sex 13 (35) 15 (37) took 15.9  5.9 h compared with
SAPS II on 47 (34–65) 40 (32–50) 38.1  11.6 h in the control group
admission (P < 0.001). Antimicrobial therapy modifi-
Immune 6 (16) 6 (15) cations resulting from feedback from the
suppression microbiological tests are summarized in
Alcohol abuse 5 (14) 4 (10) Table 3. Microbiological diagnostic infor-
Origin of sepsis mation lead to four therapy adaptations
Abdominal 8 (22) 15 (37)
immediately after the telephone call in the
Pulmonary 29 (78) 26 (63)
intervention group (9.8%) and to five
Septic shock 25 (68) 20 (49)
ICU mortality 8 (22) 7 (17) changes in the control group (13.5%). For
these patients, the mean  SD time from
Data presented as median (25–75 percentile) or n (%). initially drawing blood to adaptation of
SAPS II, Simplified Acute Physiology Score II.15 antimicrobial treatment was 18.8  5.6 h in
No significant between-group differences (P  0.05).
the intervention group compared with
38.3  14.5 h in the control group (Table 4).
Based on these findings, the time required
Seven samples were positive for the same for therapy modification in the intervention
pathogen in both the LightCyclerÕ group was reduced by 19.5 h compared with
SeptiFast PCR test and the corresponding the control group, although the difference
BC; 49 samples were validated as negative between the two groups was not statistically
for both assays. In six control-group sam- significant.
ples analysed after short-term frozen stor- For the reduction in the time to anti-
age, the LightCyclerÕ SeptiFast PCR test microbial therapy modification, this study
result had a technical signal in the test reached a power of 55% with an exact alpha
summary and was not validated: in two of P ¼ 0.063. Based on these findings, a
samples, no pathogens were detected but the second a priori power analysis was per-
internal control was also negative; in four formed to assess the sample size for a
samples, the negative control channel was future confirmatory RCT evaluating this
positive, indicating potential system con- endpoint. With a conservative assumption
tamination. None of these patients had a of 80% power and a relevant effect size of
pathogen detected in BCs. These results 1.691, a nonparametric unpaired Wilcoxon
demonstrated a concordance rate of 56 out test analysis would be able to determine a
of 78 analysed samples (71.8%). significant difference between the two study
Tafelski et al. 371

Table 2. Results of standard blood culture (BC) microbiological analyses, LightCyclerÕ SeptiFast poly-
merase chain reaction (PCR) tests and corresponding clinical pictures for patients treated in intensive care
units and randomized to undergo either a multiplex polymerase chain reaction test for initial pathogen
identification (intervention group) or standard BC analysis (control group).

Pathogen detection method

Initially suspected focus of infection/interpretation
LightCyclerÕ SeptiFast of microbiological findings
PCR BC

Intervention group
Pseudomonas aeruginosa P. aeruginosa Pneumonia, PCR and BC pathogen also detected
in TBS cultures
Escherichia coli E. coli Suspected pneumonia, PCR and BC pathogen
detected in TBS cultures
Candida parapsilosis C. parapsilosis Suspected pulmonary infection; candidaemia
identified in multiple BCs; C. parapsilosis
detected in TBS cultures
C. albicans C. albicans Suspected abdominal sepsis; candidaemia
identified in multiple BCs
Staphylococcus aureus Pneumonia, multiple BCs remained negative,
S. aureus detected in TBS cultures
P. aeruginosa Suspected abdominal infection, TBS and multiple
BC samples remained negative
Klebsiella pneumoniae/oxytoca Pneumonia, BCs and TBS, and pleural puncture
secretion remained negative
E. coli Pneumonia, multiple BCs remained negative,
TBS showed E. coli growth
Enterococcus faecium Abdominal infection, BCs series remained
negative, intraoperative samples identified
Enterococcus spp.
Enterobacter cloacae/aerogenes Pneumonia, BCs remained negative, TBS culture
negative
S. epidermidis Pneumonia, no bacterial growth in TBS culture,
three independent BCs grew S. epidermidis,
catheter-related infection suspected (devices
were changed)
Control group
E. faecalis S. epidermidis, Pneumonia, E. faecalis PCR validated by BC.
E. faecalis S. epidermidis detected in one BC, not detected
in PCR, clinically judged as contamination
S. epidermidis S. epidermidis Pneumonia, BC and PCR results validated by
additional BCs, catheter-related infection
suspected (devices were changed)
E. faecium E. faecium Abdominal infection, BC and PCR results
validated by pathogen identification E. faecium
in ascites aspirate
S. aureus Pneumonia, BC remained negative, TBS and BAL
culture identified S. aureus
S. aureus Pneumonia, multiple BCs, TBS, BAL cultures
remained negative
(continued)
372
Table 2. Continued.
Pathogen detection method
LightCyclerÕ SeptiFast
PCR BC
Streptococcus pneumonia
E. coli
S. epidermidis
S. epidermidis
S. epidermidis
S. epidermidis
S. aureus
TBS, tracheobronchial secretion; BAL, bronchoalveolar lavage.
groups if they included nine patients with
therapy modifications in each study arm
(i.e. a total of 18 patients).
Discussion
This double-blind, parallel-group RCT
demonstrated a significant reduction in the
time required for initial pathogen identifica-
tion using a multiplex PCR detection plat-
form, compared with standard BC
diagnostics, in ICU patients with clinical
sepsis of pulmonary and abdominal origin.
Most importantly, this very early identifica-
tion of the causative pathogen within 24 h of
admission was integrated into the clinical
decision-making process and resulted in
earlier modifications of empirical antimicro-
bial therapy. Using the LightCyclerÕ
SeptiFast PCR test shortened the duration
from sampling to availability of microbio-
logical results by 22.2 h. Two cases of
candidaemia were detected in the
Journal of International Medical Research 43(3)
Initially suspected focus of infection/interpretation
of microbiological findings
Pneumonia, repetitive BCs remained negative, S.
pneumoniae antigen identified in urine samples
Pneumonia, BC series remained negative, repeti-
tive BAL and TBS cultures revealed no bacterial
growth
Pneumonia, BC remained negative; contamination
was suspected; Klebsiella spp. identified in TBS.
Pneumonia, additional BCs showed S. epidermidis,
bloodstream infection, TBS grew Gram-
positive bacteria but no species identification
possible
Pneumonia, TBS remained negative, S. epidermidis
identified in two separate BCs, bloodstream
infection
Pneumonia, multiple BC and TBS samples
remained negative
Pneumonia, TBS and BC identified S. aureus
intervention group, both of which were
validated with subsequent standard BC
diagnostics. PCR-based methods might be
most beneficial for patients with a high
incidence of fungal infections or ongoing
antibiotic therapy.16
Results from the standard BCs and the
LightCyclerÕ SeptiFast PCR test were con-
cordant in seven of 12 BC positive cases (i.e.
the gold standard). Therefore, the
LightCyclerÕ SeptiFast PCR test had a
sensitivity of 58.3%. The specificity of the
LightCyclerÕ SeptiFast PCR test was 74.2%
(49 of 66 samples). Other clinical research
showed a sensitivity of 50–100%, even with
low bacterial counts of 3 colony forming
units/ml, depending on the species.14
Although authors reported a specificity of
100%, contamination during sampling and
sample workflow might be an issue.14 This
difficulty is also a well-described phenom-
enon for standard BC diagnostics and a
contamination rate of 5% is reported for
Tafelski et al. 373

Table 3. Modifications of empirical antimicrobial therapy (standard regimens), based on results of

microbiological analyses, in patients treated in intensive care units and randomized to undergo either a
multiplex polymerase chain reaction test for initial pathogen identification (intervention group) or standard
blood culture analysis (control group).

Detected pathogens Empirical therapy Modified therapy

Intervention group
Candida albicans Piperacillin/tazobactam Fluconazole
Fluconazole
Candida parapsilosis Fluconazole Voriconazole
Tigecycline
Ceftazidime
Staphylococcus aureus Piperacillin/tazobactam Gentamicin
Ciprofloxacin Ceftriaxone
Vancomycin
Pseudomonas aeruginosa Piperacillin/tazobactam Ciprofloxacin
Ciprofloxacin Meropenem
Sultamicillin
Control group
S. epidermidis Piperacillin/tazobactam Piperacillin/tazobactam
Tobramycin Tobramycin
þ Enterococcus faecalis Vancomycin
S. epidermidis Meropenem Meropenem
Ciprofloxacin Vancomycin
S. epidermidis Piperacillin/tazobactam Piperacillin/tazobactam
Clarithromycin Clarithromycin
Vancomycin
S. epidermidis Cefuroxime Vancomycin
S. epidermidis Piperacillin/tazobactam Piperacillin/tazobactam
Clarithromycin Clarithromycin
Linezolid

Table 4. Results of time analyses for patients being treated in intensive care units, randomized to undergo
either a multiplex polymerase chain reaction (PCR) test for initial pathogen identification (intervention group)
or standard blood culture analysis (control group).

Control group Intervention group

Parameter n ¼ 37 n ¼ 41

Duration from sampling to Gram staining in positive 38.1  11.6 41.8  12.9
blood cultures, ha
Duration from sampling to positive PCR result, ha – 15.9  5.9
Duration from sampling to therapy modification, hb 38.3  14.5 18.8  5.6

Data presented as mean  SD.

a
Number of positive blood cultures: control group n ¼ 7; intervention group n ¼ 5.
b
Number of patients with therapy modification based on a positive diagnostic test: control group n ¼ 5; intervention group
n ¼ 4.
374
the ICU setting.17 A Japanese group
reported a higher positive rate using a
PCR-based method than with BC, but
demonstrated a slightly higher sensitivity
and specificity than was observed in this
current study.18 The sensitivity and specifi-
city of the LightCyclerÕ SeptiFast PCR test
in this current study were similar to those
reported in recent studies conducted in
Germany and Spain.19,20 The Spanish
group evaluated patients with complicated
bloodstream infections and reported a sen-
sitivity of 56.0% and a specificity of
79.5%.20 It should be noted that in all
these studies including the current
study,14,18–20 BC results were used as the
reference gold standard, even though the
sensitivity of this standard diagnostic meth-
ods is also limited.7 As shown in Table 2,
there were cases with negative BCs but
positive PCR-based pathogen identification.
These pathogens were also identified in
other microbiological samples. Results
from other biological samples, such as
urine and tracheobronchial or abdominal
drainage secretions, may be included in the
clinical assessment of a specific patient’s
condition. In some cases, BCs may have
shown false-negative results, but analysis of
diagnostic performance remains limited to
the reference diagnostics.
Timely identification of the causative
pathogen provides important information
to ICU physicians that can be used to guide
antimicrobial therapy usage while also redu-
cing nonessential use of antimicrobial
agents. In this current study, two cases of
unsuspected candidaemia were identified
using the LightCyclerÕ SeptiFast PCR test.
In this context, a quasi-experimental, before
and after, observational cohort study
demonstrated that a conservative strategy
with prompt diagnostics and reluctant anti-
biotic prescribing was associated with
reduced overall mortality in ICU patients.12
These authors also reported that their
microbiological results within 24 h.12
Journal of International Medical Research 43(3)
The application of fast pathogen detection
methods might enable physicians to tailor
antibiotic prescribing to individual patients
in the future.
Although the LightCyclerÕ SeptiFast
PCR test was expected to reduce the time
to initial pathogen diagnosis based on the
methodological characteristics of the test,
the effects on the subsequent clinical deci-
sion making process was not predictable a
priori. A reduction in the time to pathogen
identification through the use of PCR-based
technology has been previously reported,
but the study used an open design with the
physicians knowing the results for all
patients and them not reporting the time
until they received the results of the BCs.19
PCR-based methods might be particularly
useful in patients pretreated with antibiotics,
as reported by a Japanese group.18 The best
results in terms of rapid pathogen identifi-
cation would probably require the combin-
ation of both standard BC methods and
PCR-based technololgy.21,22 There remains
an element of uncertainty about PCR-based
pathogen identification because circulating
DNA does not necessarily mean that the
patient is infected with the pathogen. In this
current study, the risk of detecting persistent
circulating DNA was reduced by only
including patients in the first 72 h after the
onset of sepsis.
This study had a number of strengths and
limitations. The strengths included external
patient randomization, which included allo-
cation concealment and separation of clin-
ical care teams from the study team, to
maximize the reduction in bias.
Furthermore, the study groups were similar
with regard to their baseline demographic
and clinical characteristics. The limitations
included the fact that the a priori scenario
assumed that the availability of the multi-
plex PCR-based test was equal to that of
standard BC diagnostics, which it was not.
As a consequence of the limited availability
of trained staff to undertake multiplex
Tafelski et al.
PCR-based test during evenings and week-
ends, and to ensure an adequate comparison
of the two diagnostic techniques, patients
were only enrolled into the study between
18.00 h and 06.00 h because the multiplex
PCR-based test was only available for 12 h
during the daytime on weekdays. This
assumption was made because of the recent
technical developments in the field.
Multiplex PCR technology is used in plat-
forms with improved automation and has
also been developed for point-of-care set-
tings. In contrast, the workflow for standard
BCs allowed for immediate growth of micro-
organisms and automated incubation was
also available during the night and at week-
ends. This limited enrolment period led to a
substantial number of patients not fulfilling
the inclusion criteria because of having
already had BCs taken during the regular
morning shifts on the ICU ward. The limited
enrolment period also meant that patient
recruitment lasted longer than initially
expected. As a result of this particular
limitation, generalizability of these findings
is limited to the precondition that the
multiplex PCR-based test was only available
within a distinct time frame. Compared with
previous trials assessing specific clinical
settings,23,24 this current study focused on
a broader target population, which was
recruited from six different ICUs at two
study sites and represented a broad spec-
trum of patients admitted to postoperative
ICUs. It was not in the scope of this current
study to evaluate different PCR-based diag-
nostic techniques that are currently avail-
able.25 The current study only aimed to
assess the use of DNA-based pathogen
detection using one specific assay, the
LightCyclerÕ SeptiFast PCR test.
Furthermore, this trial did not evaluate the
economic impact of this innovative technol-
ogy. Similarly, the impact on patient
outcomes is one of the most important
aspects, but this was not investigated in
this current study. Morbidity and mortality
375
outcomes should be addressed in future
trials, to demonstrate if advantages of
rapid multiplex PCR-based pathogen diag-
nosis lead to clinical benefits for patients.
This trial was designed to provide an
estimate of the number of patients that
would be needed to be included in a con-
firmatory RCT with a comparable design in
future. Based on these findings, with a power
of 80%, 18 patients with therapy modifica-
tions based on rapid PCR-based diagnostic
information would be needed to demon-
strate a statistically significant difference on
a two-sided alpha level of 5%. If there was a
low pathogen detection rate of 10% in the
population, 180 patients would be required.
In conclusion, multiplex PCR-based
identification of microbial pathogens might
represent a fundamental change in the way
that physicians manage bacterial and fun-
gal infections. PCR-based technology is
developing quickly in this field, and sample
processing and workflow have already
improved compared with the LightCyclerÕ
SeptiFast PCR test platform used in this
study.25 With the innovations being made
and the growing clinical availability of these
PCR-based techniques, a time requirement
of <6 h from sampling to pathogen identi-
fication could mean that infection man-
agement shifts from modifying empirical
therapy to providing pathogen-specific tar-
geted regimens right from the initiation of
therapy. However, more clinical data are
required to prove the reliability of PCR-
based technology, as well as its usefulness,
in different clinical settings.
Declaration of conflicting interest
All authors declare that they have no conflicts of
interest concerning the specific subject of this study.
I.N. and S.T. received lecture fees from Roche
Deutschland GmbH and Pfizer Deutschland
GmbH. C.S. received lecture fees or grants for
other projects from Abbott, Aspect, Baxter,
Deltex, Care Fusion, Fresenius, Hutchinson,
376
Köhler, MSD, MCN, Novartis, Sysmex, Pajunk,
Köhler Chemie, Essex Pharma, Pfizer and GSK.
Funding
This study was supported by a restricted grant,
from Roche Deutschland GmbH, which covered
laboratory transportation and microbiological
diagnostic materials.
Acknowledgements
We would like to thank the patients and their
relatives for their willingness to participate in
this study in a difficult clinical situation. Also, we
would like to gratefully mention Sara Casu MD,
Susanne Heim MD, Robert Powollik MD, and
Andreas Rothbart MD for their support in this
study.
References
1. Vincent JL, Rello J, Marshall J, et al.
International study of the prevalence and
outcomes of infection in intensive care units.
JAMA 2009; 302: 2323–2329.
2. Reinhart K, Daniels R, Kissoon N, et al. The
burden of sepsis – a call to action in support of
World Sepsis Day 2013. J Crit Care 2013; 28:
526–528.
3. Liesenfeld O, Lehman L, Hunfeld KP, et al.
Molecular diagnosis of sepsis: new aspects
and recent developments. Eur J Microbiol
Immunol (Bp) 2014; 4: 1–25.
4. Kumar A, Roberts D, Wood KE, et al.
Duration of hypotension before initiation of
effective antimicrobial therapy is the critical
determinant of survival in human septic
shock. Crit Care Med 2006; 34: 1589–1596.
5. Dellinger RP, Levy MM, Rhodes A, et al.
Surviving Sepsis Campaign: international
guidelines for management of severe sepsis
and septic shock, 2012. Intensive Care Med
2013; 39: 165–228.
6. Reinhart K, Brunkhorst FM, Bone HG, et al.
Prevention, diagnosis, treatment, and follow-
up care of sepsis. First revision of the S2k
Guidelines of the German Sepsis Society
(DSG) and the German Interdisciplinary
Association for Intensive and Emergency
Care Medicine (DIVI). Anaesthesist 2010; 59:
347–370. (in German, English Abstract).
Journal of International Medical Research 43(3)
7. Shafazand S and Weinacker AB. Blood
cultures in the critical care unit: improving
utilization and yield. Chest 2002; 122:
1727–1736.
8. Breitkopf C, Hammel D, Scheld HH, et al.
Impact of a molecular approach to improve
the microbiological diagnosis of infective
heart valve endocarditis. Circulation 2005;
111: 1415–1421.
9. Bloos F, Hinder F, Becker K, et al.
A multicenter trial to compare blood culture
with polymerase chain reaction in severe
human sepsis. Intensive Care Med 2010; 36:
241–247.
10. Bravo D, Blanquer J, Tormo M, et al.
Diagnostic accuracy and potential clinical
value of the LightCycler SeptiFast assay in
the management of bloodstream infections
occurring in neutropenic and critically ill
patients. Int J Infect Dis 2011; 15: e326–e331.
11. Tsalik EL, Jones D, Nicholson B, et al.
Multiplex PCR to diagnose bloodstream
infections in patients admitted from the
emergency department with sepsis. J Clin
Microbiol 2010; 48: 26–33.
12. Hranjec T, Rosenberger LH, Swenson B,
et al. Aggressive versus conservative initi-
ation of antimicrobial treatment in critically
ill surgical patients with suspected intensive-
care-unit-acquired infection: a quasi-
experimental, before and after observational
cohort study. Lancet Infect Dis 2012; 12:
774–780.
13. Tafelski S, Nachtigall I, Deja M, et al.
Computer-assisted decision support for
changing practice in severe sepsis and septic
shock. J Int Med Res 2010; 38: 1605–1616.
14. Lehmann LE, Hunfeld KP, Emrich T, et al.
A multiplex real-time PCR assay for rapid
detection and differentiation of 25 bacterial
and fungal pathogens from whole blood
samples. Med Microbiol Immunol 2008; 197:
313–324.
15. Le Gall JR, Lemeshow S and Saulnier F.
A new Simplified Acute Physiology Score
(SAPS II) based on a European/North
American multicenter study. JAMA 1993;
270: 2957–2963.
16. Rath PM, Saner F, Paul A, et al. Multiplex
PCR for rapid and improved diagnosis of
bloodstream infections in liver transplant
Tafelski et al.
recipients. J Clin Microbiol 2012; 50:
2069–2071.
17. Stohl S, Benenson S, Sviri S, et al. Blood
cultures at central line insertion in the
intensive care unit: comparison with periph-
eral venipuncture. J Clin Microbiol 2011; 49:
2398–2403.
18. Obara H, Aikawa N, Hasegawa N, et al. The
role of a real-time PCR technology for rapid
detection and identification of bacterial and
fungal pathogens in whole-blood samples.
J Infect Chemother 2011; 17: 327–333.
19. Lodes U, Bohmeier B, Lippert H, et al.
PCR-based rapid sepsis diagnosis effectively
guides clinical treatment in patients with new
onset of SIRS. Langenbecks Arch Surg 2012;
397: 447–455.
20. Fernández-Cruz A, Marı́n M, Kestler M,
et al. The value of combining blood culture
and SeptiFast data for predicting compli-
cated bloodstream infections caused by
Gram-positive bacteria or Candida species.
J Clin Microbiol 2013; 51: 1130–1136.
21. Pasqualini L, Mencacci A, Leli C, et al.
Diagnostic performance of a multiple
real-time PCR assay in patients with
377
suspected sepsis hospitalized in an internal
medicine ward. J Clin Microbiol 2012; 50:
1285–1288.
22. Yanagihara K, Kitagawa Y, Tomonaga M,
et al. Evaluation of pathogen detection from
clinical samples by real-time polymerase
chain reaction using a sepsis pathogen DNA
detection kit. Crit Care 2010; 14: R159.
23. Chaidaroglou A, Manoli E, Marathias E,
et al. Use of a multiplex polymerase chain
reaction system for enhanced bloodstream
pathogen detection in thoracic transplant-
ation. J Heart Lung Transplant 2013; 32:
707–713.
24. Torres-Martos E, Pérez-Ruiz M, Pedrosa-
Corral I, et al. Evaluation of the
LightCyclerÕ SeptiFast test in newborns and
infants with clinical suspicion of sepsis.
Enferm Infecc Microbiol Clin 2013; 31:
375–379. (in Spanish, English Abstract).
25. Schreiber J, Nierhaus A, Braune SA, et al.
Comparison of three different commercial
PCR assays for the detection of pathogens in
critically ill sepsis patients. Med Klin
Intensivmed Notfmed 2013; 108: 311–318.