A

Swt

YRa

b

c

a

ARA

KEWAS

1

laohe

ite

h0

ARTICLE IN PRESSG ModelNTAGE-4439; No. of Pages 8

International Journal of Antimicrobial Agents xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

International Journal of Antimicrobial Agents

j o ur nal ho me pag e: ht tp : / /www.e lsev ier .com/ locate / i jant imicag

urveillance of antimicrobial resistance among Escherichia coli inastewater in Stockholm during 1 year: does it reflect the resistance

rends in the society?

oung-Keun Kwaka, Patricia Colquea, Sara Byforsb, Christian G. Giskea,c,oland Möllbya, Inger Kühna,∗

Department of Microbiology, Tumor and Cell Biology (MTC), Karolinska Institutet, Box 280, SE-171 77 Stockholm, SwedenDepartment of Microbiology, Public Health Agency of Sweden, Solna, SwedenClinical Microbiology, Karolinska University Hospital Solna, Stockholm, Sweden

r t i c l e i n f o

rticle history:eceived 28 June 2014ccepted 13 September 2014

eywords:scherichia coliastewater

ntibiotic resistanceurveillance

a b s t r a c t

The resistance patterns of Escherichia coli in untreated (raw) urban wastewater (UW) was monitored byrepeated sampling during 1 year. Comparison with data from wastewater samples collected from hospitalwastewater (HW) in the same urban area was made. A total of 1326 E. coli isolates from 17 UW samplesand 451 isolates from six HW samples were analysed by typing using the PhenePlateTM system, and theirsusceptibility towards 10 antibiotics was determined. Resistance to at least one antibiotic was observedin 34% of the UW isolates and 55% of the HW isolates. For UW isolates, phenotypic diversity was loweramong antibiotic-susceptible than among antibiotic-resistant isolates, indicating a higher presence ofclonal groups among susceptible isolates. Total antibiotic resistance measured as the Multiple AntibioticResistance (MAR) index was 0.08 for UW compared with 0.19 for HW, and increased over time for UWisolates, indicating increasing resistance among E. coli in the urban population during the studied timeperiod. Resistance to all included �-lactam antibiotics was detected in 2.4% of UW isolates and 14.0% ofHW isolates, and 73/75 (97%) analysed isolates were confirmed to be extended-spectrum �-lactamase

(including plasmid-mediated AmpC)-producing E. coli. Thus, by cultivating samples from wastewater andanalysing many independent isolates per sample, increasing frequencies of antibiotic resistance in UWwere detected during 1 year that may reflect increasing faecal carriage of resistant bacteria in the society.Surveillance of antibiotic resistance in wastewater could be a valuable tool for screening of resistancetrends on a population level.

© 2014 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved.

. Introduction

Antimicrobial resistance in bacteria is an increasing global prob-em both for human and veterinary medicine. The rapid increase inntibiotic-resistant bacteria has developed as a result of the usef antibiotics in human and animal populations, and resistanceas thereafter spread among bacteria in humans, animals and thenvironment.

To be able to detect changes in resistance patterns over time

Please cite this article in press as: Kwak Y-K, et al. Surveillancewater in Stockholm during 1 year: does it reflect the resistanhttp://dx.doi.org/10.1016/j.ijantimicag.2014.09.016

n bacteria that cause infections in humans, several interna-ional and national surveillance programmes have been conducted,.g. the European Antimicrobial Resistance Surveillance Network

∗ Corresponding author. Tel.: +46 8 524 871 54; fax: +46 8 33 15 47.E-mail address: inger.kuhn@ki.se (I. Kühn).

ttp://dx.doi.org/10.1016/j.ijantimicag.2014.09.016924-8579/© 2014 Elsevier B.V. and the International Society of Chemotherapy. All rights

(EARS-Net), ECO-SENS [1] and CAESAR (Central Asian and East-ern European Surveillance of Antimicrobial Resistance). There arealso surveillance programmes for animals, e.g. in Europe throughthe European Food Safety Authority (EFSA). In these veterinaryprogrammes, antimicrobial resistance is monitored in commensalbacteria such as Escherichia coli or enterococci collected from fae-ces or caecal contents of healthy animals. However, less is knownabout the normal intestinal microbiota in humans, which also actsas a reservoir of resistant bacterial strains and resistance genes.The search for resistant bacteria in the faecal microbiota is a verycumbersome procedure, especially when the prevalence of resis-tance is low, since this approach means that large numbers of faecal

of antimicrobial resistance among Escherichia coli in waste-ce trends in the society? Int J Antimicrob Agents (2014),

samples must be collected and analysed. For example, if a certainresistant strain occurs in 1% of the population, 300 samples have tobe analysed to have a 95% probability of finding at least one samplethat contains the resistant strain.

reserved.

ING ModelA

2 l of An

uwWgwbocwtf

ibaihfcocsSb5efihowatrcagfe

aierswmtdscwiti(a

2

2

ci

ARTICLENTAGE-4439; No. of Pages 8

Y.-K. Kwak et al. / International Journa

A possible alternative method is to analyse incoming water torban wastewater treatment plants (WWTPs) since such waste-ater contains high numbers of bacteria of mainly human origin.astewater may be a niche for resistant bacteria and resistance

enes which have their origin in the population that produces theastewater [2–5], and consequently samples from wastewater can

e used to create an early warning system for the emergence of newr rare bacterial antibiotic resistance in the target population. Thisoncept was proposed in the seventies by Linton et al. [6]. Later itas also found that increasing resistance rates in urban wastewa-

er correspond well with increasing resistance rates among isolatesrom humans [5].

Previous studies performed by us have also shown that bacterian wastewater are good representatives of the indigenous micro-iota of the individuals contributing to the wastewater [7,8]. In

previous multicentre European study on enterococci includ-ng almost 3000 samples and more than 20 000 isolates fromumans, animals, wastewater, manure and the environment, we

ound clear evidence that the population structure (with regard tolonal diversity, species distribution and population similarities)f wastewater well resembled that of a large population of fae-al samples from individual humans and animals [8]. In anothertudy, a widespread ampicillin/fluoroquinolone multiple-resistantwedish clone of Enterococcus faecium of hospital origin [9] coulde directly traced to hospital wastewater where it was found in0% of all studied samples [10]. This clone appeared to be furthernriched in urban wastewater and was also found in many samplesrom receiving waters [10], suggesting that surface water contain-ng bacteria from wastewater may be a source for colonisation ofumans and animals with antibiotic-resistant bacteria of hospitalrigin. In the same study, vancomycin-resistant enterococci (VRE)ere isolated from 60% of urban wastewater samples in Sweden

nd from 35% of hospital wastewater samples [11], although athat time it was said that no VRE existed in Sweden owing to itsestrictive antibiotic policy. The situation changed later, and 487ases of healthcare-related VRE were detected in the Stockholmrea in the period 2007–2009, the majority of them due to a clonalroup of VanB E. faecium [12], the same resistance type that wasound in urban and hospital wastewater in Stockholm some yearsarlier.

Thus, previous studies have shown a clear correlation betweenntibiotic-resistant bacteria in wastewater and in the correspond-ng human population. To be able to use this concept to make anarly warning system for the emergence of new or rare bacte-ial antibiotic resistances, we have developed a simple screeningystem for studying antibiotic resistance indicator bacteria inastewater (the AREB system) [13]. The system consists of 96-wellicroplates containing eight sets of breakpoint concentrations of

en different antibiotics. The incubated microplates are read by aesktop scanner and the plate images are analysed by a specialoftware that presents the resistance data. The AREB method isombined with a rapid typing method, the PhenePlateTM system,hich yields information on the diversity of the bacteria in the stud-

ed samples and on the possible prevalence of resistant clones. Inhe present study, these tools were used to compare resistance ratesn samples collected on different occasions at a WWTP in StockholmSweden) and in wastewater collected from a hospital in the samerea.

. Materials and methods

.1. Collection of samples

Please cite this article in press as: Kwak Y-K, et al. Surveillancewater in Stockholm during 1 year: does it reflect the resistanhttp://dx.doi.org/10.1016/j.ijantimicag.2014.09.016

Samples from raw (untreated) urban wastewater (UW) wereollected 17 times during 1 year at the Henriksdal WWTP, whichs one of the biggest WWTPs in Europe and receives wastewater

PRESStimicrobial Agents xxx (2014) xxx–xxx

from central Stockholm as well as from the southern and easternsuburbs. Approximately 250 000 m3 of wastewater from >750 000people enters Henriksdal WWTP every day. There are no agricul-tural sources that contribute significantly to the wastewater. Theplant is equipped with mechanical, chemical and biological treat-ments, and an additional sand filter removes remaining particulatematter. Samples for the present study were collected during 24 hfrom the influent wastewater by a sampling device at the treatmentplant that allowed continuous sampling. Another three sampleswere collected in the same manner from the treated urban waste-water (TW) at the exit to the Baltic Sea. Six samples from hospitalwastewater (HW) were collected from an 800-bed university hos-pital in Stockholm (Karolinska University Hospital) by lowering a50 mL flask every 10–15 min during 4 h into the wastewater flowand pooling the aliquots.

2.2. Isolation of Escherichia coli

Samples were kept at 4 ◦C and were analysed within 4 h. Serialdilutions of the samples in phosphate-buffered saline were made,and aliquots of 250 �L from appropriate dilutions (10−2 and 10−3

for wastewater) were plated on 14-cm Petri dishes containinga chromogenic medium for E. coli (CHROMagarTM Orientation;CHROMagar Microbiology, Paris, France) using the Plating Beadmethod (http://www.zymoresearch.com/devices-instruments/plating-beads/rattler-plating-beads). The plates were incu-bated overnight (16–18 h) at 44 ◦C to inhibit growth ofnon-thermotolerant bacteria.

2.3. Identification and analysis of Escherichia coli isolates

Presumptive E. coli isolates were identified on the CHROMagarOrientation plates as red or pink colour colonies releasing a sur-rounding halo, and they were further subject to biochemicalfingerprinting using PhP-RE plates of the PhenePlateTM sys-tem (PhPlate Microplate Techniques AB, Saltsjö-Boo, Sweden;http://www.phplate.se). From the incubated Petri dishes, 80 iso-lated E. coli colonies per sample were picked by sterile tooth sticksand were inoculated into the first columns of ten PhP-RE plates thathad previously been filled with PhP suspending medium. Aliquotsof 10 �L were transferred to all other columns in the PhP-RE plateand to all columns in an AREB plate, containing one column formaking bacterial suspensions, 10 antibiotics in columns 2–11, andone control column [13]. The plates were incubated overnight at37 ◦C and on the next day they were scanned by a desktop scanner(HP G4050) and the resulting plate images were transformed intoabsorbance data by the PhenePlateTM software (PhPlate MicroplateTechniques AB). A majority of the colonies presumed to be E. coliaccording to visual inspection of the CHROMagar plates were con-firmed to be E. coli by the PhP-RE method, and in total 1326 E. coliisolates from UW, 451 from HW and 117 from TW were includedin the present study (Table 1).

For characterisation of extended-spectrum �-lactamase (ESBL)genes, a Check-Points CT101 microarray (Check-Points HealthB.V., Wageningen, The Netherlands) was used in accordance withthe manufacturer’s instructions. The microarray targets blaCTX-M,blaTEM, blaSHV, blaNDM and blaKPC and it groups the blaCTX-M genesinto groups 1, 2, 9 and 8/25. The array also distinguishes betweennon-ESBL blaTEM/blaSHV and ESBL blaTEM/blaSHV. Finally, the arrayidentifies plasmid-mediated AmpC (pAmpC) genes (DHA, CMY-1/MOX, CMY-2, ACT/MIR, FOX and ACC).

of antimicrobial resistance among Escherichia coli in waste-ce trends in the society? Int J Antimicrob Agents (2014),

2.4. Data analysis

The PhenePlateTM software measured the colour of each wellin the scanned images of the PhP-RE plates as well as the amount

ARTICLE IN PRESSG ModelANTAGE-4439; No. of Pages 8

Y.-K. Kwak et al. / International Journal of Antimicrobial Agents xxx (2014) xxx–xxx 3

Table 1Isolates studied and diversity of each population.

Wastewater sample No. of samples No. of isolates % resistance Diversity

Total a Median b Minimum b

Untreated (raw) urban wastewater 17 1326 34 0.97 0.97 0.95Treated urban wastewater 3 117 40 0.94 0.95 0.93Hospital wastewater 6 451 55 0.97 0.94 0.90

oTttaopr(fiTMtpt

ddscfaohldp

2

ceiTczs

3

3

2waif3(ohps

the UW isolates, 2.9% were suspected ESBL-EC, resistant to at leastone of the antibiotics cefotaxime or ceftazidime, and 2.4% of theUW isolates were resistant to all three of the above antibiotics.For HW isolates the corresponding numbers were 14.9% and 14.0%.

0

5

10

15

20

25

30

35

40

amp ctx caz chl cip gen nal pod tet tmp

% re

sista

nt is

olat

es

a Value for all isolates.b Median and minimum value for all samples.

f bacterial growth in each well of the images of the AREB plates.he pattern obtained from utilisation of the 10 different reagents inhe PhP-RE plate by each isolate was used to obtain a biochemicalype (PhP-type) of the isolate [13]. The software also compared themount of bacterial growth in each well of the AREB plate with thatf the control well for the same bacterial isolate (column 12) andrinted out resistance rates as 0 (no growth, susceptible), 1 (weakesponse, result should be controlled by visual inspection) and 2similar amount of growth as in the control well, resistant). For thenal analysis, weak growth was regarded as indicating resistance.otal antibiotic resistance in a population was measured using theultiple Antibiotic Resistance (MAR) index [14,15], calculated as

he mean proportion of resistances for the isolates. The maximumossible MAR value is 1.00, obtained when all isolates are resistanto all antibiotics tested.

The PhenePlateTM software was furthermore used to cluster theata obtained from the PhP-RE plates, as well as to calculate theiversity and homogeneity of the samples. The diversity was mea-ured as Simpson’s diversity index (Di), and the homogeneity wasalculated as mean value of all correlation coefficients obtainedrom N × (N − 1) comparisons among N isolates. These two valuesre useful measures of the population structure: Di is a measuref the possible prevalence of common clones in a population, andomogeneity measures the mean similarity among all studied iso-

ates and thus indicates whether isolates, although belonging toifferent types, are more similar than expected from a randomopulation and thus might have a common clonal origin.

.5. Statistical analysis

The increase in resistance rate with time was calculated as aorrelation between resistance rates at a certain time point and thelapsed time after the first sample. The resistance rates are given forndividual antibiotics (Table 2) or as MAR indices (see Section 2.4).he degree of correlation was calculated as Pearson’s momentumoefficient, and the statistical significance of its slope differing fromero was given as a P-value, all calculated using GraphPad Prism v.5oftware (GraphPad Software Inc., La Jolla, CA).

. Results

.1. Population structure

The number of E. coli in the samples from UW varied between.8 × 104 and 6.6 × 105 CFU/mL (median 7.4 × 104 CFU/mL),hereas the three samples from TW had only between 40 CFU/mL

nd 120 CFU/mL. Thus, >99.9% of the total number of E. colin UW was removed during the treatment process. Samplesrom HW had between 2.6 × 104 and 5.5 × 104 CFU/mL (median.7 × 104 CFU/mL). The total Di among all E. coli in UW was 0.965Table 1), a value that can be expected from a diverse population

Please cite this article in press as: Kwak Y-K, et al. Surveillancewater in Stockholm during 1 year: does it reflect the resistanhttp://dx.doi.org/10.1016/j.ijantimicag.2014.09.016

f E. coli when typed with the PhP-RE plates [13], and totalomogeneity was 0.549, also a normal value for a diverse E. coliopulation. The mean Di and homogeneity in individual UWamples were approximately the same as in the whole population

(Table 1), indicating that the E. coli in individual samples wererepresentative for normal diverse populations of individuals.Adequate samples from HW were more difficult to obtain, whichwas reflected in lower diversities among the E. coli in some HWsamples (Table 1).

3.2. Comparisons of resistance rates

In total, 34% of the E. coli isolates from UW and 55% of theE. coli isolates from HW were resistant to at least one of the studiedantibiotics. The total antibiotic resistance, calculated as the MARindex, was 0.19 for HW isolates compared with 0.08 for UW iso-lates, and resistance rates were higher to all antibiotics for HWisolates (Fig. 1). Only 117 isolates were available from TW samplesand these isolates appear to carry antibiotic resistance to the samedegree as isolates from samples collected simultaneously enter-ing the treatment plant (data not shown). Depending on samplingtime, the UW samples were divided into five different groups, eachgroup consisting of between 2 and 8 samples and between 154 and628 E. coli isolates. The MAR index for each time period is shown inFig. 2. As a comparison, the MAR index from a previous study, usingthe same methods, calculated on nine antibiotics (cefpodoxime wasnot included in that study) in the year 2011 is also given [13]. Thereappears to be an increase over time in the total prevalence of antibi-otic resistance (Fig. 2) as well as in resistance rates to individualantibiotics among E. coli in UW (Table 2).

3.3. Extended-spectrum ˇ-lactamase-producing Escherichia coli(ESBL-EC)

Three broad-spectrum �-lactam antibiotics were included in theAREB plates (cefotaxime, ceftazidime and cefpodoxime). Among

of antimicrobial resistance among Escherichia coli in waste-ce trends in the society? Int J Antimicrob Agents (2014),

Fig. 1. Resistance rates among 1326 Escherichia coli in untreated (raw) urban waste-water (light grey) and 451 E.coli in hospital wastewater (dark grey). amp, ampicillin;ctx, cefotaxime; caz, ceftazidime; chl, chloramphenicol; cip, ciprofloxacin; gen, gen-tamicin; nal, nalidixic acid; pod, cefpodoxime; tet, tetracycline; tmp, trimethoprim.

Please cite this article in press as: Kwak Y-K, et al. Surveillancewater in Stockholm during 1 year: does it reflect the resistanhttp://dx.doi.org/10.1016/j.ijantimicag.2014.09.016

ARTICLE IN PRESSG ModelANTAGE-4439; No. of Pages 8

4 Y.-K. Kwak et al. / International Journal of Antimicrobial Agents xxx (2014) xxx–xxx

Tab

le

2R

esis

tan

ce

rate

s

amon

g

Esch

eric

hia

coli

in

un

trea

ted

(raw

)

urb

an

was

tew

ater

.

Sam

pli

ng

tim

eN

o.

of

sam

ple

sN

o.

of

E.

coli

anal

ysed

%

E.

coli

isol

ates

resi

stan

t

An

tibi

otic

(con

cen

trat

ion

in

mg/

L)

AM

P

(32)

CTX

(2)

CA

Z

(4)

CH

L

(32)

CIP

(4)

GEN

(16)

NA

L

(32)

POD

(3)

TET

(16)

TMP

(16)

Jan

. 201

33

224

11.2

2.5

2.2

1.1

3.6

1.3

12.5

3.6

9.4

7.1

Feb.

2013

8

628

15.0

2.1

2.5

2.0

1.6

1.1

13.3

4.4

13.9

10.8

July

2013

2

160

17.5

4.4

3.8

2.5

5.0

2.5

22.5

6.6

12.2

14.1

Jan

. 201

42

154

16.2

2.3

2.6

4.9

2.6

2.6

19.8

4.5

16.2

12.3

Feb.

2014

2

160

21.9

3.8

3.1

4.7

5.0

3.1

14.4

6.6

14.4

13.8

Tota

l

17

1326

15.9

2.9

3.0

2.8

2.9

1.7

17.0

5.6

13.4

11.3

Cor

rela

tion

betw

een

%

resi

stan

ce

and

tim

e

of

sam

pli

ng

(Jan

. 201

3–Fe

b.

2014

)0.

82

0.38

0.45

0.97

0.41

0.95

0.45

0.62

0.73

0.76

P-va

lue

for

slop

e

<>0

0.10

0.55

0.51

0.01

*0.

52

0.02

*0.

48

0.30

0.15

0.16

AM

P,

amp

icil

lin

;

CTX

, cef

otax

ime;

CA

Z,

ceft

azid

ime;

CH

L,

chlo

ram

ph

enic

ol;

CIP

, cip

rofl

oxac

in;

GEN

, gen

tam

icin

;

NA

L,

nal

idix

ic

acid

;

POD

, cef

pod

oxim

e;

TET,

tetr

acyc

lin

e;

TMP,

trim

eth

opri

m.

*Si

gnifi

can

t.

0.00

0.02

0.04

0.06

0.08

0.10

0.12

2011* Jan 2013 Feb 2013 July 2013 Jan 2014 Feb 2014

R2 = 0.8 3 P = 0.01

* Da ta from 2011 are from 48 isola tes coll ected for a previo us study (Col que-Navarro et al. , 2014)

N = 48 N = 224 N = 628 N = 160 N = 154 N = 160

Fig. 2. Multiple Antibiotic Resistance (MAR) indices for Escherichia coli collected

from untreated (raw) urban wastewater. N denotes the number of isolates studiedwithin each group. A significant increase with time is seen. Note that the x-axis doesnot reflect real time intervals.

Confirmation of ESBL production was performed for 75 of the iso-lates that were resistant to all three broad-spectrum �-lactamantibiotics with the AREB test, and 73 (97%) of these were con-firmed to harbour an ESBL gene (including pAmpC) (Fig. 3). Thus,we estimate that 2.3% of the E. coli in UW and 13.6% in HW areESBL-EC.

The PhP data for the 73 E. coli that were confirmed as ESBL-EC were clustered and the resulting dendrogram is presented inFig. 3. In only a few cases, isolates belonging to the same PhP typeshowed the same resistance patterns, and when this was the casethe isolates were usually of hospital origin and from the same sam-ple. Some isolates from sample HW06 (PhP types 13, 15, 16 and 17)showed the same resistance patterns and belonged to different, butrelated, PhP types (Fig. 3).

3.4. Diversity of Escherichia coli related to resistance

To analyse whether resistant E. coli could partly belong tospecific clonal groups, the phenotypic diversity was calculatedfor isolates carrying each resistance. All of the resistance groupsshowed similar or even higher diversity values than the total E. colipopulation (data not shown) and thus clonal spread does not appearto be the source of resistant strains. However, when diversityand homogeneity for resistant versus susceptible isolates werecompared, susceptible isolates were less diverse and more homoge-neous than the resistant isolates, both among UW and HW isolates(Fig. 4). Thus, clonality among the susceptible isolates appears tobe higher than among resistant ones.

4. Discussion

The quality of antimicrobial susceptibility data obtained fromwastewater depends on whether the analysed bacteria really rep-resent the faecal microbiota of many independent individuals andnot just faeces from one individual or a single household. Sinceuntreated wastewater contains high numbers of E. coli, only 0.25 mLof diluted wastewater could be cultivated for each analysis, andthere always exists a risk that a small piece of solid material (e.g.

of antimicrobial resistance among Escherichia coli in waste-ce trends in the society? Int J Antimicrob Agents (2014),

faeces) is present, which would contain high numbers of E. colibelonging to only one or a few clones. For picking 80 isolates, theideal method would be to cultivate 80 wastewater samples andto analyse one isolate per sample; however, our intention was to

Please cite this article in press as: Kwak Y-K, et al. Surveillance of antimicrobial resistance among Escherichia coli in waste-water in Stockholm during 1 year: does it reflect the resistance trends in the society? Int J Antimicrob Agents (2014),http://dx.doi.org/10.1016/j.ijantimicag.2014.09.016

ARTICLE IN PRESSG ModelANTAGE-4439; No. of Pages 8

Y.-K. Kwak et al. / International Journal of Antimicrobial Agents xxx (2014) xxx–xxx 5

Fig. 3. Dendrogram showing clustered PhP typing data, with resistance patterns and extended-spectrum �-lactamase/plasmid-mediated AmpC (ESBL/pAmpC) type added,for confirmed ESBL-positive Escherichia coli from untreated (raw) urban wastewater (UW), treated urban wastewater (TW) and hospital wastewater (HW). Grey fields indicateresistance (1) and white fields susceptibility (0). amp, ampicillin; ctx, cefotaxime; caz, ceftazidime; chl, chloramphenicol; cip, ciprofloxacin; gen, gentamicin; nal, nalidixicacid; pod, cefpodoxime; tet, tetracycline; tmp, trimethoprim.

ARTICLE IN PRESSG ModelANTAGE-4439; No. of Pages 8

6 Y.-K. Kwak et al. / International Journal of Antimicrobial Agents xxx (2014) xxx–xxx

0.92

0.94

0.96

0.98

1.00

UW HW

a) Diversity index

Any resistance No resistance

0.40

0.45

0.50

0.55

0.60

0.65

0.70

UW HW

b) Homogeneity

Any resistance No resistance

Fig. 4. (a) Diversity indices among resistant and susceptible Escherichia coli popu-lations in untreated (raw) urban wastewater (UW) and hospital wastewater (HW).Note the lower diversities found among isolates that were susceptible to all of the10 antibiotics studied. (b) Homogeneities (=mean similarities among all isolates)ahs

ufmEawdap[ocv

w2HaciWrpaitsa

ipiesiicsfhyrNA

0

10

20

30

40

50

2007 ECDC

2008 ECDC

2009 ECDC

2010 ECDC

2011 ECDC

2013 -HW

2013 -UW

Ampicillin

0

2

4

6

8

10

12

14

2007 ECDC

2008 ECDC

2009 ECDC

2010 ECDC

2011 ECDC

2012 ECDC

2013 -HW

2013 -UW

Ciprofloxacillin

012345678

2007 ECDC

2008 ECDC

2009 ECDC

2010 ECDC

2011 ECDC

2012 ECDC

2013 -HW

2013 -UW

Gentamicin

02468

10121416

2007 ECDC

2008 ECDC

2009 ECDC

2010 ECDC

2011 ECDC

2012 ECDC

2013 -HW

2013 -UW

Cefotaxime

2013/2014

2013/2014

2013/2014

2013/2014

Fig. 5. Resistance rates in Escherichia coli isolates collected from blood samples in

teria in UW and receiving water, respectively, were ESBL-producing

mong resistant and susceptible E. coli populations in UW and HW. Note the higheromogeneities among isolates that were susceptible to all of the 10 antibioticstudied.

se a simple method that can be applied on a routine basis. There-ore, the resistance screening was combined with a simple typing

ethod that could determine whether the diversity among the. coli analysed in each sample was sufficient to allow for a reli-ble analysis. Using combined typing and resistance determinationith PhP and AREB, several hundred isolates can be analysed peray, and automated data analysis allows final results to be availablelready 2 days after sampling. PhP typing has been used in manyrevious studies involving typing of several thousand E. coli isolates7,16–18] and it has been shown that populations of intestinal E. colif diverse origins show high diversities with Di values >0.95. Con-omitantly, for the present study, UW samples showing lower Dialues than 0.95 were excluded from further analysis.

In contrast to UW samples that were obtained from incomingastewater collected in the treatment plant continuously during

4 h, it was more difficult to obtain representative samples fromW, where the samples had to be collected manually by lowering

flask into the wastewater flow. Even though the final samplesonsisted of pooled aliquots from 12–20 subsamples collected dur-ng 4 h, three of the six samples still showed too low diversities.

hether this is due to inadequate samples containing solid mate-ial or whether it is due to low diversity of the population thatroduces the hospital wastewater is so far not known. Data fromll HW samples taken together yielded a diversity as high as 0.97,ndicating that the low diversity in some samples was not dueo persistent clonal groups in the wastewater. For future studies,amples need to be collected using a continuous sampling device,nd preferably over a longer time period.

An important finding in this study was that we could measurencreasing resistance rates among E. coli in UW over the short timeeriod of only 1 year, possibly mirroring increasing resistance rates

n the urban population. Similar results were found by Reinthalert al., who compared resistance rates among E. coli in wastewaterludge from the year 2000 with the year 2009 and found increas-ng resistance rates to most antibiotics [5]. They also found similarncreases in resistance rates among human isolates collected fromlinical samples during the same time period. Resistance data toome of the antibiotics used in the current study were availablerom a European database on human blood isolates (EARS-Net;ttp://ecdc.europa.eu) [19]. Fig. 5 shows resistance data for theears 2007–2012 compared with our data from 2013–2014. High

Please cite this article in press as: Kwak Y-K, et al. Surveillancewater in Stockholm during 1 year: does it reflect the resistanhttp://dx.doi.org/10.1016/j.ijantimicag.2014.09.016

esistance to cefotaxime and gentamicin in isolates from EARS-et is also seen for the isolates from HW in the present study.mpicillin rates appear to be similar, whereas for ciprofloxacin

Sweden during 2007–2012 compared with data from isolates collected from hospitalwastewater (HW) and untreated (raw) urban wastewater (UW) during 2013–2014.ECDC, data from the European Centre for Disease Prevention and Control network(http://ecdc.europa.eu).

lower rates were observed in HW isolates than in the EARS-Netstudy.

Several other recent studies have reported high rates ofantibiotic-resistant bacteria, including ESBL-producers, in waste-water from the urban society and from hospitals, as well as inwastewater effluents and in the receiving waters. A recent studyin Poland reported that as many as 19% and 15% of the enterobac-

of antimicrobial resistance among Escherichia coli in waste-ce trends in the society? Int J Antimicrob Agents (2014),

[20]. In another study from hospitals in Poland, 30% of the studiedEnterobacteriaceae isolates harboured ESBL genes [21]. A studyfrom France found 1.7% ESBL-producing isolates in UW [22], similar

ING ModelA

l of An

tbtepbclwipth

bnaroicartr

nhisbatofbio[et2dn(wharte

aittasbti

A

B

[

[

[

[

[

[

[

[

[

[

[

ARTICLENTAGE-4439; No. of Pages 8

Y.-K. Kwak et al. / International Journa

o the corresponding result for the present study (2.3%). It has alsoeen shown that the higher the bacterial density in the wastewater,he higher the percentage of ESBL-producing isolates. Possibly thisffect could be due to increased horizontal gene transfer from ESBL-roducers to non-producers when concentrations of ESBL genesecome higher [20]. Gene transfer within the wastewater systemould also explain some of the HW data, where ESBL-producing iso-ates of different PhP types showed identical resistance profiles as

ell as the same ESBL genes (Fig. 3). Alternatively, isolates belong-ng to different but related types and showing the same resistancerofile (such as types 13, 15, 16 and 17 in Fig. 3) stem from a clonehat is endemic in the wastewater system or some other part of theospital and has undergone diversification with time.

Another interesting observation in this study was that isolateseing susceptible to all studied antibiotics were more homoge-eous and less diverse than those being resistant to at least onentibiotic (Fig. 4). This finding is in contrast to the expectation thatesistant isolates would show a lower diversity owing to presencef resistant clonal groups. Thus, it appears that the susceptiblesolates in wastewater to a higher extent consist of clonal groupsompared with those that are antibiotic-resistant. Most studies onntibiotic resistance in bacteria are focused on characterisation ofesistant isolates, but perhaps it could also be interesting to studyhe susceptible clones and elucidate why they do not becomeesistant.

Although wastewater treatment processes normally reduce theumber of heterotrophic bacteria in the effluents by up to 3 log,igh numbers of bacteria may still be discharged into the receiv-

ng waters [2,4,5,23]. Of special concern is the fact that isolatesurviving the wastewater treatment process in some studies haveeen reported to be more resistant to antibiotics than bacteria thatre eliminated during the treatment process [15,24–26] and thathe release of antibiotic resistance genes is even higher than thatf antibiotic-resistant bacteria [2,4]. Bréchet et al. reported that asew as 0.1% of E. coli in UW and 7.6% in HW were ESBL-producing,ut still as many as 0.6 × 1012 ESBL-EC per day were discharged

nto the receiving water from their WWTP, which purifies less thanne-tenth of the wastewater amount of the WWTP studied here26]. In the present study, the degree of reduction of E. coli in theffluent was >3 log (99.93%), and the data did not indicate a selec-ion for antibiotic-resistant bacteria in the WWTP. Still, since ca..5 × 1011 mL of wastewater arrives to the studied treatment planturing 24 h, containing in total ca. 4.6 × 1016 E. coli bacteria, theumber of E. coli that will be discharged into the receiving waterin this case the Baltic Sea) every day is as high as 4.6 × 1013, ofhich 34% are resistant to at least one of the 10 antibiotics studiedere, and 2.3% (1.1 × 1012) are presumed ESBL-EC. Little is knownbout the further fate of these bacteria in the environment, but it iseasonable to expect that they accumulate over time and will con-ribute to the load of antibiotic resistance among bacteria that ismerging worldwide.

In conclusion, by cultivating samples from wastewater andnalysing many independent isolates per sample, we could detectncreasing frequencies of antibiotic resistance in UW during 1 yearhat may reflect increasing faecal carriage of resistant bacteria inhe society. We believe that surveillance of antibiotic resistancemong faecal bacteria in wastewater could be a valuable tool forcreening of resistance trends on a population level, and possiblye used as an early warning system to detect new, emerging resis-ances, not least to understand the burden of faecal carriage, whichs normally not measured when focusing on clinical isolates.

Please cite this article in press as: Kwak Y-K, et al. Surveillancewater in Stockholm during 1 year: does it reflect the resistanhttp://dx.doi.org/10.1016/j.ijantimicag.2014.09.016

cknowledgments

The authors thank Maj Anttila Ringman, Sofia Ny, Filippa Lindénergman and Claudia Perez, who helped with the laboratory

[

[

PRESStimicrobial Agents xxx (2014) xxx–xxx 7

analyses; Peter Lindström and all the other helpful staff at WWTPHenriksdal and Karolinska University Hospital Solna (Stockholm,Sweden) for help with the sampling from wastewater; and CajsaWahlberg at Stockholm Vatten AB (Stockholm, Sweden) for con-structive criticism of the manuscript.

Funding: This study was supported by the Chile–Sweden Coop-eration Fund [grant no. 210-0205], the Swedish Research Council[grant no. C0626201] and the Swedish Civil Contingency Agency.

Competing interests: None declared.Ethical approval: Not required.

References

[1] Kahlmeter G, Poulsen HO. Antimicrobial susceptibility of Escherichia coli fromcommunity-acquired urinary tract infections in Europe: the ECO.SENS studyrevisited. Int J Antimicrob Agents 2012;39:45–51.

[2] Gao P, Munir M, Xagoraraki I. Correlation of tetracycline and sulfonamideantibiotics with corresponding resistance genes and resistant bacteria ina conventional municipal wastewater treatment plant. Sci Total Environ2012;421:173–83.

[3] Kummerer K, Henninger A. Promoting resistance by the emission of antibi-otics from hospitals and households into effluent. Clin Microbiol Infect2003;9:1203–14.

[4] Munir M, Wong K, Xagoraraki I. Release of antibiotic resistant bacteria andgenes in the effluent and biosolids of five wastewater utilities in Michigan.Water Res 2011;45:681–93.

[5] Reinthaler FF, Galler H, Feierl G, Haas D, Leitner E, Mascher F, et al. Resis-tance patterns of Escherichia coli isolated from sewage sludge in comparisonwith those isolated from human patients in 2000 and 2009. J Water Health2013;11:13–20.

[6] Linton KB, Richmond MH, Bevan R, Gillespie WA. Antibiotic resistance and Rfactors in coliform bacilli isolated from hospital and domestic sewage. J MedMicrobiol 1974;7:91–103.

[7] Blanch AR, Belanche-Munoz L, Bonjoch X, Ebdon J, Gantzer C, Lucena F, et al.Integrated analysis of established and novel microbial and chemical methodsfor microbial source tracking. Appl Environ Microbiol 2006;72:5915–26.

[8] Kühn I, Iversen A, Burman LG, Olsson-Liljequist B, Franklin A, Finn M,et al. Comparison of enterococcal populations in animals, humans, and theenvironment—a European study. Int J Food Microbiol 2003;88:133–45.

[9] Torell E, Kühn I, Olsson-Liljequist B, Haeggman S, Hoffman BM, Lindahl C,et al. Clonality among ampicillin-resistant Enterococcus faecium isolates inSweden and relationship with ciprofloxacin resistance. Clin Microbiol Infect2003;9:1011–9.

10] Iversen A, Kühn I, Rahman M, Franklin A, Burman LG, Olsson-Liljequist B, et al.Evidence for transmission between humans and the environment of a nosoco-mial strain of Enterococcus faecium. Environ Microbiol 2004;6:55–9.

11] Iversen A, Kühn I, Franklin A, Möllby R. High prevalence of vancomycin-resistant enterococci in Swedish sewage. Appl Environ Microbiol2002;68:2838–42.

12] Soderblom T, Aspevall O, Erntell M, Hedin G, Heimer D, Hokeberg I, et al. Alarm-ing spread of vancomycin resistant enterococci in Sweden since 2007. EuroSurveill 2010;15:13–8.

13] Colque Navarro P, Fernandez H, Möllby R, Otth L, Tiodolf M, Wilson M, et al.Antibiotic resistance in environmental Escherichia coli—a simple screeningmethod for simultaneous typing and resistance determination. J Water Health2014 [Epub ahead of print].

14] Krumperman PH. Multiple antibiotic-resistance indexing of Escherichia coli toidentify high-risk sources of fecal contamination of foods. Appl Environ Micro-biol 1983;46:165–70.

15] Zhang R-Q, Ying G-G, Su H-C, Zhou L-J, Liu Y-S. Antibiotic resistance and geneticdiversity of Escherichia coli isolates from traditional and integrated aquaculturein South China. J Environ Sci Health 2013;48:999–1013.

16] Anastasi EM, Matthews B, Stratton HM, Katouli M. Pathogenic Escherichia colifound in sewage treatment plants and environmental waters. Appl EnvironMicrobiol 2012;78:5536–41.

17] Landgren M, Oden H, Kuhn I, Osterlund A, Kahlmeter G. Diversity among2481 Escherichia coli from women with community-acquired lower uri-nary tract infections in 17 countries. J Antimicrob Chemother 2005;55:928–37.

18] Reyes D, Vilchez S, Paniagua M, Colque P, Weintraub A, Möllby R, et al. Diver-sity of intestinal Escherichia coli populations in Nicaraguan children with andwithout diarrhoea. J Med Microbiol 2009;58:1593–600.

19] Brolund A, Edquist PJ, Mäkitalo B, Olsson-Liljequist B, Söderblom T, Wisell KT,et al. Epidemiology of extended-spectrum �-lactamase-producing Escherichiacoli in Sweden 2007–2011. Clin Microbiol Infect 2014;20:O344–52.

20] Korzeniewska E, Harnisz M. Extended-spectrum �-lactamase (ESBL)-positiveEnterobacteriaceae in municipal sewage and their emission to the environ-

of antimicrobial resistance among Escherichia coli in waste-ce trends in the society? Int J Antimicrob Agents (2014),

ment. J Environ Manage 2013;128:904–11.21] Korzeniewska E, Harnisz M. �-Lactamase-producing Enterobacteriaceae in

hospital effluents. J Environ Manage 2013;123:1–7.22] Diallo AA, Brugère H, Kerourédan M, Dupouy V, Toutain PL, Bousquet-Mélou

A, et al. Persistence and prevalence of pathogenic and extended-spectrum

ING ModelA

8 l of An

[

[

[

ARTICLENTAGE-4439; No. of Pages 8

Y.-K. Kwak et al. / International Journa

�-lactamase-producing Escherichia coli in municipal wastewater treatmentplant receiving slaughterhouse wastewater. Water Res 2013;47:4719–29.

Please cite this article in press as: Kwak Y-K, et al. Surveillancewater in Stockholm during 1 year: does it reflect the resistanhttp://dx.doi.org/10.1016/j.ijantimicag.2014.09.016

23] Reinthaler FF, Feierl G, Galler H, Haas D, Leitner E, Mascher F, et al. ESBL-producing E. coli in Austrian sewage sludge. Water Res 2010;44:1981–5.

24] Łuczkiewicz A, Jankowska K, Fudala-Ksiazek S, Olanczuk-Neyman K. Antimi-crobial resistance of fecal indicators in municipal wastewater treatment plant.Water Res 2010;44:5089–97.

[

PRESStimicrobial Agents xxx (2014) xxx–xxx

25] Ferreira da Silva M, Vaz-Moreira I, Gonzalez-Pajuelo M, Nunes OC, Manaia CM.Antimicrobial resistance patterns in Enterobacteriaceae isolated from an urban

of antimicrobial resistance among Escherichia coli in waste-ce trends in the society? Int J Antimicrob Agents (2014),

wastewater treatment plant. FEMS Microbiol Ecol 2007;60:166–76.26] Bréchet C, Plantin J, Sauget M, Thouverez M, Talon D, Cholley P, et al.

Wastewater treatment plants release large amounts of extended-spectrum�-lactamase-producing Escherichia coli into the environment. Clin Infect Dis2014;58:1658–65.