Journal of Infection (2017) e 366

74, 358

www.elsevierhealth.com/journals/jinf

Blood culture-PCR to optimisetyphoid fever

diagnosis after controlled human infection
identies frequent asymptomatic cases and
evidence of primary bacteraemia
a,b,
Thomas C. D a rton *, Liqing Zhoua , Christoph J. Blo hmke
a
,
a a b,c
Claire Jones , Cla ir e S. Waddington , Stephen Baker ,
Andrew J. Pollard a

a
OxfordVaccineGroup, Department ofPaediatrics andtheNIHR OxfordBiomedical ResearchCentre,
University of Oxford, Oxford, United Kingdom
b
The Hospital for Tropical Diseases, Wellcome Trust Major Overseas Programme, Oxford University
Clinical Research Unit, Ho Chi Minh City, Viet Nam
c
Centre for Tropical Medicine and Global Health, Nufeld Department of Clinical Medicine, Oxford
University, Oxford, United Kingdom

Accepted 14 January 2017

Available online 24 January 2017

KEYWORDS Summary Background:Improved diagnostics for typhoid are needed; a typhoid controlled
Typhoid fever; humaninfectionmodelmayacceleratetheirdevelopmentandtranslation.Here,weevaluated
Controlled human abloodculture-PCRassayfordetectinginfectionaftercontrolledhumaninfectionwithS. Ty-
infection model; phi and compared test performance with optimally performed blood cultures.
Diagnostics; Methodology/Principal ndings: Culture-PCR amplication of blood samples was performed
Polymerase chain alongsidedailybloodculturein41participantsundergoingtyphoidchallenge.Studyendpoints
reaction; for typhoid diagnosis (TD) were fever and/or bacteraemia. Overall, 24/41 (59%) participants
Febrile disease; reachedTD,ofwhom21/24(86%)had 1 positivebloodculture(53/674,7.9%ofallcultures)
SalmonellaTyphi or 18/24 (75%) had 1 positive culture-PCR assay result (57/684, 8.3%). A further ve non-
bacteraemic participants produced culture-PCR amplicons indicating infection; overall sensi-
tivity/specicityoftheassaycomparedtothestudyendpointswere70%/65%.Wefoundnosig-
nicantdifferencebetweenbloodcultureandculture-PCRmethodsinabilitytoidentifycases
(12 mismatching pairs,p Z 0.77, binomial test). Clinical and stool culture metadata demon-
strated that additional culture-PCR amplication positive individuals likely represented true

* Correspondingauthor.OxfordVaccineGroup,DepartmentofPaediatricsandtheNIHROxfordBiomedicalResearchCentre,Universityof
Oxford, Oxford, OX3 7LE, United Kingdom.
E-mail address:thomas.darton@paediatrics.ox.ac.uk (T.C. Darton).

http://dx.doi.org/10.1016/j.jinf.2017.01.006
0163-4453 / 2017The Author(s). Published by Elsevier Ltd on behalf of The British Infection Association. This is an open access article
under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Culture-PCR assay for typhoid 359

cases missed by blood culture, suggesting the overall attack rate may be 30/41 (73%) rather
than 24/41 (59%). Several participants had positive culture-PCR results soon after ingesting
challenge providing new evidence for occurrence of an early primary bacteraemia.
Conclusions/Signicance:Overall the culture-PCR assay performed well, identifying extra
typhoid cases compared with routine blood culture alone. Despite limitations to widespread
eld-use, the benets of increased diagnostic yield, reduced blood volume and faster turn-
around-time,suggestthatthisassaycouldenhancelaboratorytyphoiddiagnosticsinresearch
applications and high-incidence settings.
2017TheAuthor(s).PublishedbyElsevierLtdonbehalfofTheBritishInfectionAssociation.
ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/
by/4.0/).

4,21
Introduction escalation study, as previously described. Briey,
healthy consenting adult volunteers aged between 18 and
60 years were challenged by ingesting a fresh preparation
Typhoid fever, a non-specic febrile illness caused by 3 4
of10 or10 CFUof S. Typhisuspendedinsodiumbicarbon-
infectionwithSalmonellaentericaserovarTyphi(S. Typhi),
1 ate solution. After challenge, participants were reviewed
is common in tropical regions. A key limitation to
daily for symptoms and signs of typhoid fever and clinical
improvingthecontroloftyphoidfeveristhelackofreliable
2,3 samples were collected (Table 1). If the study endpoint of
diagnostictests. Inadditiontoconrminginfectioninin-
typhoid diagnosis (TD) was reached, additional samples
dividuals, accurate laboratory diagnostics are needed to
were collected and antimicrobial treatment was initiated.
ascertain true disease burden, to improve understanding
All remaining participants were given antimicrobial treat-
of the natural history of infection in humans, and to eval-
1,2,4 ment on day 14. TD criteria were clinical (temperature
uate vaccine efcacy.
38 Csustainedfor12h)and/ormicrobiological (positive
Diagnostic approaches for typhoid infection are broadly
blood culture).
aimed either at directly detecting bacteria or bacterial
products or measuring the host response in clinical sam-
2,4,5
ples. Bloodcultureremainsthediagnostictechniqueof Diagnostic blood culture (reference standard)
choice,butonlyidenties45e 70%ofconrmedcases,even
with the availability of newer continuous automated cul- Blood for culture was collected daily for 14 days or up to
5e 7
ture systems. Serological tests including the Widal test 96 h after TD, whichever was the later, and processed
are widely available in endemic settings, although in the according to national standard methods as previously
absence of paired clinical samples or background popula-described.21,22 At all time points after challenge 10 mL
tion serosurveillance data these tests perform poorly withblood was collected, except at TD when this was reduced
5,8
low sensitivity and specicity. to5mL(Table1).Stoolculturewasperformedaspreviously
Given the poor accuracy of currently available diag- described.21 Bacterial isolates were identied by pheno-
nostic tests, attempts have been made to develop PCR- typic, biochemical and serological testing according the
9e 13
based assays to detect bacterial DNA. Few, if any, of Kauffmann-White classication, following standard
23,24
these approaches have been instituted in clinical settingsmethods.
mainly due to the difculty of validating tests when using
real-life specimens, in which only few, mostly intracel-
lularbacteria (median,0.5 CFU/mL blood) arepresent.
2,14 Culture-PCR assay
OnemethodtoincreasethesensitivityofS. Typhidetection
from blood is to use ox-bile as a selective culture me- To perform the culture-PCR assay, 5 mL of heparinised
Ox-bile reduces both coagulation and serum com-peripheral venous blood was collected at daily after
15,16
dia.
plement killing activity and causes the selective lysis of challenge19,20,25
(Table 1), and performed as previously
human rather than Salmonella cells.
17,18
Recently, we described. Briey,bloodwasaddedtoculturemedia
developed a culture-PCR assay incorporating a brief pre- (20 mL 3% (w/v) ox bile/tryptone soya broth containing
incubation in ox-bile along with PCR amplication of the 1.5 m L micrococcal nuclease) and incubated for 5 h
Here, we have evaluated (37 C, 220 rpm New Brunswick, Excella e24). Bacteria
19,20
S. Typhi agellin gene,iC.
thisculture-PCRassayasadiagnosticfordetectingS. Typhi were then concentrated by centrifugation at 6000 g for
inthebloodofhealthyadultvolunteersdevelopingtyphoid 20 min and the supernatant removed. DNA was extracted
while participating in a human challenge model. 21 from the bacterial pellet using UltraClean BloodSpin
kits following the manufacturers instructions, except that
elution of the nal DNAwas performed usingm 50
L of pre-
Methods heated Buffer 5 (65 C for 5 min) prior to centrifugation.
PCR amplication was performed with primers targeting
Participants and challenge 26
the S. Typhi agellin gene, iC-d (GenB ank L21912, )as
10
described by Levy et .al
Challengeofhealthyadultswithasingleoraldoseofawild- Amplicationreactionswereperformedin50 mLvolumes
type S. Typhi Quailes strain was performed in a dose- containing10mLDNAtemplateand0.2mMeachofH-forand
360 T.C. Darton et al.

Table 1 Assay schedule and associated blood volumes for laboratory diagnostic tests performed during the study. A) Assays
performed in all participants, and B), assays performed in participants reaching clinical or microbiological TD endpoint. Chal-
3 4
lenge:oralingestionof10 or10 CFU S. TyphiQuailesstrainsuspendedin30mL/0.53gNaHCO 3 (aq).Antimicrobials:rst-line,
ciprooxacin 500 mg twice daily for 14 days.
A) All participants
Day 0 0 0 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 28
Hour 0 6 12 0 12 e e e e e e e e e e e e e e
Procedure Challenge Antimicrobials Antimicrobials
started completed
Blood culture e 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 e
Culture-PCR e 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5

B) Additional time points in typhoid-diagnosed participants

Timepoint hours TD 0TD 6TD 12 TD 24 TD 36 TD 48 TD 72 TD 96
Procedure Antimicrobials started
Blood culture 5 10 10 10 10 10 10 10
Culture-PCR 5 5 5 5 5 5 5 5

Hd-revprimers(TopTaqPCRMasterMixKit,Qiagen).Where discordant pairs analysis using a two-tailed binomial test.

required, distilled H2 O and/or non-study DNA extracted Data were analysed using Prism v6.0e (GraphPad Software
from healthy donor blood was used as negative controls; Inc).
genomicDNAextractedfromS. TyphiQuailesstrainculture
wasusedasapositivecontroltemplate.DNAamplication
Ethics
wasperformed usingstandardthermocycler equipmentat:
95 C for 5 min followed by 35 cycles of 93 C for 30 s,
ThechallengestudywasapprovedbytheNationalResearch
55 C/30 s, and 72 C/40 s and terminating with 1 cycle
Ethics Service (Oxfordshire Research Ethics Committee A,
of72 Cfor5min.ThespecictargetPCRampliconscould
10/H/0604/53) and was performed in accordance with the
be observed at 763 bp when separated by 1% (w/v)
principles of the ICH-Good Clinical Practice guidelines and
ethidium bromide agarose gel electrophoresis and visual-
amendments.All study participants provided written
izedusingUVlighttransillumination.Ampliconswerecate-
informed consent in accordance with the Declaration of
gorised as either being present (positive) or absent
Helsinki. While all blood culture and PCR data were
(negative) by visualisation with the naked eye.
generatedprospectivelyduringthestudy,onlybloodculture
resultswereusedtomakeclinicalmanagementdecisions.
Bloodinoculationexperimentsanddetectionlimits

Prior to the investigation of clinical specimens, assay Results

performanceandlaboratorysensitivitywasevaluatedusing
S. Typhi -negative whole blood (5 mL) containing known Performance of culture-PCR in a typhoid challenge
concentrationsofS. TyphiDNA,addedaftertheincubation study
step (Supplementary Fig. 1). The calculated sensitivity of
PCR amplication was at least 0.015 r g/mL, equivalent to Usingadose-escalationapproachtodeterminetheoptimal
a DNA starting concentration of 30 bacteria per reaction challengedose,41participantswerechallengedwitheither
(each bacterium containing w 5 fg DNA)27,28 or 6 CFU/ 3 4
10 or 10 CFU of S. Typhi Quailes strain, as previously
mLintheoriginalstartingbloodvolume,assumingnomulti- described.21 Attack rates were 55% and 65% for each
plication had taken place during incubation. dose, respectively; the median day of illness onset was
seven days after challenge.
Reporting and statistical analysis Overall, 684 serial samples were collected for culture-
PCR from 41 challenge participants, 24 of whom were
DatapertainingtothediagnosticaccuracyofthePCRassay diagnosed with typhoid fever (TD). One individual was
in comparison with the predened study endpoint of treated before day 14 based on symptoms alone without
typhoid diagnosis (TD) or positive blood culture are re- fullling the study TD criteria (included here in the nTD
29
portedaccordingtotheSTARDcriteria. Thediagnosticac- group, Supplementary Table 1); the remaining 16 partici-
curacy of PCR or blood culture in comparison to TD are pants were treated with antimicrobials at day 14 but did
reportedusingsensitivity,specicity,positiveandnegative not develop infection, dened as not diagnosed with
likelihood ratios, positive and negative predictive values typhoid (nTD, Table 2). From the 684 samples collected
anddiagnosticoddsratio,eachwitha95%condenceinter- 57 (8.3%) were positive by culture-PCR assay, and were
val. The statistical signicance of the differences in sensi-collected from 23 study participants (Fig. 1). From three
tivities of PCR and blood culture was assessed by days after challenge onwards, nTD and TD participants
Culture-PCR assay for typhoid 361

Table2 Number(%)ofculture-PCRpositivesamplesidentiedduringthestudyaccordingtochallengeoutcomeandday/time
of sample collection. Samplesfrom participants diagnosed with typhoid are further described byday relative to typhoid diag-
nosis (and antibiotic initiation) in 48-h blocks. nTD, non-typhoid diagnosed; TD, typhoid diagnosed.
Challenge outcome,
n/N (%)
nTD (17 participants) TD (24 participants) ALL (41 participants)
Time point after challenge
Day 0 to Day 3 4/62 (6.5) 5/87 (5.7) 9/149 (6.0)
Day 3 onwards 11/203 (5.4) 37/332 (11.1) 48/535 (9.0)
Total 15/265 (5.7) 42/411 (10.2) 57/684 (8.3)
Time relative to TD
> 72h e 2/57 (3.5) e
72to 24h e 18/44 (40.9) e
24to 24h e 15/39 (38.4) e
24to 72h e 0/83 (0) e
>72h e 2/71 (2.8) e

41 parcipants challenged

Culture-PCR
n=41

PCR posive PCR negave

n=23 n=18

BLOOD CULTURE BLOOD CULTURE

BC posive BC negave BC posive BC negave

n=16 n=7 n=5 n=13

5 micro dx 2 clinical dx 5 micro dx 1 clinical dx

11 clinical & micro dx 5 nTD 12 nTD

Figure1 STARDowchartdescribingculture-PCRresultsincomparisonwiththereferencestandard(Bloodculture,BC)for
diagnosisofchallengestudyparticipantswithtyphoidinfectionafterchallenge.nTD,non-typhoiddiagnosed;clinicaldx,clinical
diagnosis; micro dx, microbiological diagnosis.

yielded 11/203 (5.4%) and 37/332 (11.1%) positive culture- howeverthemaximaldayofbloodcultureandPCRpositivity

PCRassayresults,respectively.Aftertheinitiation ofanti- wassixdaysafterchallengefor both assays(Fig.3).
microbials 6/24 (25%) TD participants generated nine
further positive culture-PCR assay results, of which 5/9
occurred within six hours of initiated antimicrobial treat- Early primary DNAaemia in typhoid challenge
ment (e.g. Fig. 2). participants
In comparison with the culture-PCR methods, blood
culture alone yielded 53/674 (7.9%) positive samples from Theearliestpositiveculture-PCRsamplewascollectedonly
the 41 study participants. Positive PCR amplications were six hours after challenge; a further two participants were
evident earlier in the challenge course than with culture, positiveby12h.Intotal,ninepositivesamplesfromseven
362 T.C. Darton et al.

Figure 2 Example of a challenge study participant who had several early positive culture-PCR results ( ), in
yellowsquares
additiontoearlypositivestoolcultureresult( ). Thisparticipantwassubsequentlydiagnosedwithtyphoidinfec-
orangesquares
tion based on both microbiological and clinical criteria.
Red square , positive blood culture;grey squares
, no sample collected;
black line, oral temperature;dashed grey line
, C-reactive protein level;
shaded area, antibiotic treatment initiated.

participants were obtained48 h after the ingestion ofS. (mean, 2.00 vs. 2.73, respectively); this difference was
Typhi with blood samples from a further two participants non-signicant (95%CI,0.75to 2.22,p Z 0.31, T test).
testing positive within 72 h (Table 2 and Fig. 3). There Assuming that the predened TD criteria were the best
wasnosignicantassociationbetweenearlystoolshedding referencestandard,culture-PCRdemonstratedasensitivity
and culture-PCR assay positivityn( Z 3/16, p Z 0.63, and specicity of 70 and 65%, respectively (Table 3).
Fishers exact test; Fig. 2). Similarly, an early positive Despite forming part ofthe TD denition, the use of blood
culture-PCRresultwasnotpredictiveofsubsequentdevel- culture alone as a reference standard did not detect all
opment of typhoid infection (diagnostic odds ratio (DOR) cases of typhoid: routine blood culture was more sensitive
0.57,95%CI0.12 e 2.71,p Z 0.69)ordevelopmentofS. Ty- andspecicthanculture-PCR;87.5and100%,respectively.
phi bacteraemia (DOR 0.50, 95%CI e0.10 2.44,p Z 0.45). Interestingly, addition of the culture-PCR results to the
study endpoint denitions suggested an overall attack
Conrmation of typhoid diagnosis rate after challenge of 73% rather than 59%. Discordant
pairs analysis identied 12 mismatching culture-PCR and
Of the 24 participants diagnosed with typhoid infection, bloodcultureresults,andconrmedthattherewasnosig-
21/24 (87.5%) had a bacteraemia with or without reaching nicantdifferenceinyieldbetweenculture-PCRandblood
the clinical endpoint (Fig. 1, Supplementary Table 1). culture sensitivity if either was used alone to diagnose
Z
While culture-PCR and blood culture results concurred in infection(p 0.77,binomialtest;SupplementaryTable2).
participants with bacteraemia and fever, discrepancies
arose in participants diagnosed by positive blood culture Asymptomatic DNAaemia after typhoid challenge
or temperature criteria alone. In three participants diag-
nosed by the clinical endpoint, 2/3 generated a positive In participants who remained nTD throughout the 14-day
culture-PCR result supporting the clinical diagnosis while observation period, 6/17 generated 1 positive culture-
all blood cultures remained negative throughout the chal-PCR result (Fig. 1). Further investigation of participants
lengeperiod(Fig.1andSupplementaryFig.3).Inaddition, clinical features demonstrated that several had features
the participant treated early based on symptoms alone indicative of the development of typhoid infection in this
(i.e. no bacteraemia or fever) also had several positive controlled challenge scenario (Supplementary Table 3).
culture-PCR assay results after the start of antimicrobial These features included a single elevated temperature
treatment. reading within 12 h of the positive culture-PCR result
Incontrast,veparticipantshadpositivebloodcultures (Supplementary Fig. 4), maximal symptom reporting on
on at least one occasion but remained PCR amplication the day of the positive culture-PCR result (Supplementary
negative throughout. In general, these ve individuals had Table 3) and the additional participant who was treated
fewer positive blood cultures than those who generated based onsymptoms without meeting the TDendpoint de-
positive blood cultures and were culture-PCR positive nition. The remaining 2/6 participants had only mild
Culture-PCR assay for typhoid 363

Figure3 Numberofpositiveculture-PCRandbloodculturesamplescollectedafterchallengebytyphoidoutcome. The6and

12hpositiveresultshavebeenpooledintothe0.5daygroup.ThemaximumnumberofTDsamples/dayexceedsthenumberofTD
participants as more than one sample was collected per day following initiation of antibiotic treatment. TD, typhoid diagnosed;
nTD, non-typhoid diagnosed; PCR, culture-PCR assay; BC, blood culture.

symptoms,reportingconstipationandacoughonthedayof The application of PCR-based laboratory methods to

the positive result.
Discussion
Direct detection of pathogen nucleic acid is a widely-used
method to increase sensitivity, specicity and time-to-
30e 32
resultinmoderndiagnosticlaboratories. Wehavepre-
viously described the rst culture-PCR assay for detecting
20
S. Typhi in the modern era ; here, we demonstrate the
performanceofthismethod,specicallydesignedtosensi-
tivelydetectthe S. Typhi iC-d geneinthebloodofpartic-
ipants undergoing typhoid challenge. While the direct
comparison with blood culture suggested that the sensi-
tivity of the assay was lower for participants reaching the
typhoid diagnosis study endpoints, overall, the culture-
PCR provided useful additional information compared with
blood culture alone. This additional approach enabled the
detection and conrmation of typhoid infection cases that
wouldhavebeenmissedbybloodculture.Inusingahuman
typhoid challenge model to perform the evaluation, we
have provided unique insights into the natural history of
typhoid infection. These include evidence for the fre-
quency of asymptomatic infection after exposure, and the
conrmation of bacterial DNA circulation in blood soon af-
ter exposure.
conrm clinical diagnoses of typhoid fever are not
commonly reported, despite the appeal of detecting non-
cultivable bacteria. This is of special relevance in most
endemic settings, where antimicrobial pre-treatment or
immuneinterferencearecommonandlikelyreduceculture
sensitivitystillfurther.Inthisstudywedemonstrate aPCR
sensitivity of 70% using selective pre-incubation in ox-bile
of a 5 mL blood sample, compared with the study dened
endpoints.Asreportedpreviously,themedianquantitative
bacterial loads at diagnosis are 0.47 and 1.10 CFU/mL in
3
participants ingesting the 10 or 104 CFU dose, respec-
21
tively. With our estimation of a lower detection limit of
6 CFU/mL,itisnotsurprisingthattheassaymayhavefailed
todetectsomecases.Theextremechallengeposedbyde-
tecting such a small number of bacteria has been recently
demonstrated in a eld study evaluating a sensitive qPCR,
capable of detectingw 1 CFU/mL, which, despite favour-
able performance in developmental stages demonstrated
33
eld performance of 40% sensitivity and 91% specicity.
While the detection of bacterial DNA may be increased
by pre-incubation of clinical material prior to performing
27,34,35
DNA extraction, disadvantagesto this method
include loss of the ability to accurately quantify bacterial
numbers in the blood directly, prolongation of assay time
and the requirement for microbiological culture facilities.
364 T.C. Darton et al.

Table3 Contingencytablesdisplayingestimates[95%CIs]ofthesensitivityandspecicityforculture-PCRandroutineblood
culturefordiagnosingparticipantswithtyphoidinfectionduringachallengestudy.*Notethatbacteraemiawasoneofthediag-
nostic criteria. TD, typhoid diagnosed; nTD, non-typhoid diagnosed; LR, likelihood ratio; PPV, positive predictive value; NPV,
negative predictive value; DOR, diagnostic odds ratio.
Challenge outcome Challenge outcome

TD nTD Total TD nTD Total

sult lture result

Posive 17 6 23 Posive 21 N/A* 21
ture-PCR re
Negave 7 11 18 Negave 3 17 20
u
Blood c
Cul Total 24 17 41 Total 24 17 41

Culture-PCR Blood culture

Sensivity 0.70 [0.49 - 0.87] 0.88 [0.67 0.97]
Specicity 0.65 [0.38 0.86] 1.00* [0.80 1.00]
LR+ 2.01 [1.00 4.01] N/A
LR- 0.45 [0.22 0.92] 0.12 [0.04 0.36]
PPV 0.74 [0.52 0.90] 1.00 [0.84 1.00]
NPV 0.61 [0.36 0.83] 0.85 [0.62 0.97]
DOR 4.5 [1.2 17.5] 215 [10.4 4448]

Nevertheless, selective culture to release viable bacteria Cambodia, for example, identied a subpopulation among
from the blood intracellular compartment has been shown children <16 years old presenting with fever who were
36 38
to produce an almost 2-fold rise in bacterial numbers. culture-negativebutpositiveusingareal-timePCRassay.
While broths containing 10% and more of ox-bile have Of note, this subpopulation was younger, had a shorter
beenshowntobeatbestbacteriostaticforS. Typhigrowth, duration of illness prior to presentation and presented
our recent work identied that, at an optimal concentra- with fewer features characteristic of typhoid infection.
tion of w 2.4% and with 5-h incubation, a signicant in- Incompleteimmunityandearlierpresentationwithillness
crease in bacterial numbers may be achieved arecommontobothpatient/participant groupswhichmay
18,19
(Supplementary Table 4). The addition of micrococcal reect higher bacterial loads.
nucleaseduringtheextractionprocesstoremoveremaining The occurrence of a primary, subclinical, bacteraemia
20
human DNA further improved assay sensitivity. that results in the dissemination of bacteria to the
7,16
Our identication of participants with asymptomatic lymphoreticularsystemhaslongbeenspeculated. While
infection, i.e. participants who had evidence of DNA in early studies demonstrated bacteraemia in patients and
the blood (positive culture-PCR result) or bacterial shed- challengestudyparticipantsasearlyas4or3days,respec-
16,37
ding in stool (positive stool culture) in the absence of tively, these cases likely represented early true infec-
fever or clinical signs of infection, may represent false tion. Individuals were febrile and went on to develop
positiveassayresults.WhilevariousS. Typhitargets have overt clinical infection, probably as a result of high expo-
been used previously, Song and colleagues rst described suredose,whichhasbeenshowntocorrelatetoshorterin-
12 21
the use of PCR to detect agellin sequences in 1993. cubationperiods. Similarresultswereobservedinstudies
39
Flagellin expression is monophasic in several Salmonella performed in chimpanzees.Our culture-PCR results sug-
sp. includingS. Typhi,andthephase-1antigendisfound gestthatS. TyphiDNAisdetectableinbloodwithin48haf-
in many species. While the end regions ofiC-d (previ- ter ingestion; data that are supported by corresponding
ouslyH1-d) are identical between species, there are two increasesinplasmacytokinesandhostgeneexpressionac-
hypervariable regions (IV and VI) unique S. to Typhi and tivity.40 Factors affecting the outcome of these early inva-
26
similar to S. Muenchen. Selecting primers targeting sion events, which probably occur more frequently than
this region should thereforeresultingood specicity was recognised here as demonstrated by the ubiquitous
with little overlap with environmental or other Salmo- cytokinesignaturefoundinchallengedparticipants,areun-
nella species. certain but probably include the host inammatory re-
Identication of probable asymptomatic, subclinical sponses and early immune responses.
typhoid patients has been previously noted in other sus- Thereareknownlimitationstoreportingofdetecting S.
ceptible patient groups including the historical typhoid TyphiDNAfromclinicalsamples,whichincludetheinappro-
37
challenge studies performed in Maryland. A study in priate validation ofthe methodology,usingnested primers
Culture-PCR assay for typhoid 365

and use of archived samples rather than a real time com-NHS Trust, Oxford; Clinical Research Fellowships to CSW
13
parison between culture and PCR technique. While some and TCD); CJB is a Marie Curie Research Fellow supported
of these may be overcome by using real-time PCR, these by the European Union (FP7); AJP is a Jenner Investigator
techniques have reiterated the difculty of amplifying and James Martin Senior Fellow. SB is a Sir Henry Dale
very low DNA copy numbers which suggests its continuedFellow,jointlyfundedbytheWellcomeTrustandtheRoyal
13
inferiority to standard bacterial culture.Overreliance on Society (100087/Z/12/Z).
culture methods as the true reference standard is known
33
tobeproblematic, however,andthereforecorroborating Contributions
41
clinical and stool culture data is also valuable.
WhilePCRisarelativelycommonlyperformedtechnique
Conceived and designed the experiments: LZ, AJP, TCD.
inmostdiagnosticlaboratories,includingthoseinlesswell-
Performed the experiments: TCD, CJ, CJB, CSW. Analyzed
resourced settings, the pre-culture step, which is vital to
the data: TCD, CJB. Contributed reagents/materials/anal-
increase assay sensitivity, does not yet render the proced-
ysistools:TCD,LZ,CJ,CSW,SB.Wrotethepaper:TCD,LZ,
ure beyond cross-infection (or cross-contamination) risk.
CJ, CJB, SB, AJP.
Thehighlyselectivenatureofox-bile,whilesuitablefor S.
Typhi and other bile resistant organisms, means that this
assayisofquestionableapplicabilitytomostsettingswhere Conict of interest
clinical material is scant andS. Typhi is not the predomi-
6
nant pathogen. Alternative lysing agents have been pro- All authors report no conict of interest.
posed, including saponin (used for bacterial blood
quantication in this study) and digitonin, which may pro-Acknowledgements
42,43
duce cell lysis without loss of bacterial viability. In
thespeciccontextofvaccineeldorprobestudies,how- We would like to express our thanks to the study partici-
ever, such a mono-directional assay, possibly with the pants for taking an active part in our research. We
44
additional of targetedS. Paratyphi A primers, may offer acknowledgethekindsupportofthemicrobiology,haema-
additional support to validate efcacy at least in small tology and clinical chemistry laboratories at the Oxford
scale studies. Universities NHS Foundation Trust, and public health sup-
An important limitation of our assay was the volume ofportfromtheThamesValleyHealthProtectionUnit.Weare
blood required to perform the test. While the lower gratefultoourcolleagueProfessorMyronMLevineandthe
volumes may be used, the stochastic nature of sampling UniversityofMarylandSchoolofMedicineforprovidingthe
such low-density bacteraemic blood will necessarily result Quailes challenge strain, and to the Health Protection
inareductioninsensitivity.Itisalsoimportanttonotethe Agency (now Public Health England) Porton Down for
difference between blood volumes used in our evaluation: challenge strain manufacture to GMP standard.
blood culture was performed with 10 mL whereas the
culture-PCR assay was performed with 5 mL. While detec-
tionofbacteraemiawaskeytotheendpointfortheclinical
Appendix A. Supplementary data
study, ethical and physiological limitations to sampling
meant that a matched volume could not be collected at Supplementary data related to this articlecan befound at
every time point. This likely underestimates the perfor- http://dx.doi.org/10.1016/j.jinf.2017.01.006.
mance of our culture-PCR assay.
In conclusion, a selective culture-PCR assay targeting References
the agellin gene, while less sensitive than optimally
performed blood culture for the detection of participants
developing typhoid infection after challenge, provides 1. BuckleGC,WalkerCL,BlackRE.Typhoidfeverandparatyphoid
useful additive diagnostic information regarding the fever:systematicreviewtoestimateglobalmorbidityandmor-
tality for 2010.J Glob Health2012;2(1):10401.
outcome of typhoid challenge. Performing evaluation of
2. Baker S, Favorov M, Dougan G. Searching for the elusive
newly available diagnostics in the context of a human
typhoid diagnostic. BMC Infect Dis2010;10:45.
challengestudyhighlightsimportantfeaturesofthenatural 3. ParryCM,HienTT , DouganG,WhiteNJ,FarrarJJ.Typhoidfe-
historyoftyphoidinfection.Theseincludethefrequencyof ver .N Engl J Med2002;347(22):1770 e 82.
asymptomatic cases and new evidence for primary bacter-4. Darton TC, Blohmke CJ, Pollard AJ. Typhoid epidemiology , di-
aemia occurring soon after ingestion. agnosticsandthehumanchallengemodel. CurrOpinGastroen-
terol 2014;30(1):7e 17.
5. Parry CM, Wijedoru L, Arjyal A, Baker S. The utility of diag-
Funding nostic tests for enteric fever in endemic locations.Expert
Rev Anti Infect Ther2011;9(6):711 e 25.
Funding for the challenge study was provided by a Well- 6. WainJ, Hosoglu S.The laboratorydiagnosiso f entericfever J .
Infect Dev Ctries2008;2(6):421e 5.
come Trust Strategic Translational Award (grant number
7. World Health Organisation . Background document: the diag-
092661to AJP; www.wellcome.ac.uk). Additional specic nosis, treatment and prevention of typhoid fever Depart-
. In:
fundingforthisworkwasprovidedbyJesusCollege,Oxford ment of immunization, vaccines and biologicals . Geneva:
(Graduate Scholarship to TCD). WHO; 2003.
TCD, CJ, CJB, CSW, AJP are supported by the NIHR 8. Keddy KH, Sooka A, Letsoalo ME, Hoyland G, Chaignat CL,
Biomedical Research Centre (Oxford University Hospitals Morrissey AB, etal. Sensitivity and specicityo f typhoidfever
366
rapid antibody tests for laboratory diagnosis at two sub-
Saharan African sites.Bull World Health Organ 2011;89(9):
640e 7.
9. Ali A, Haque A, Haque A, Sarwar Y , Mohsin M, Bashir S, et al.
Multiplex PCR for differential diagnosis of emerging typhoidal
pathogens directly from blood samples. Epidemiol Infect
2009;137(1):102e 7.
10. Levy H, Diallo S, Tennant SM, Livio S, Sow SO, Tapia M, et al.
PCR method to identifySalmonella entericaserovars Typhi,
Paratyphi A, and Paratyph i B among Salmonellaisolates from
the blood of patients with clinical enteric fever J Clin
. Micro-
biol 2008;46(5):1861 e 6.
11. Massi MN, Shirakawa T , Gotoh A, Bishnu A, Hatta M,
KawabataM.RapiddiagnosisoftyphoidfeverbyPCRassayus-
ingonepairofprimersfromagellingeneof Salmonellatyphi.
J Infect Chemother2003;9(3):233 e 7.
12. SongJH,ChoH,ParkMY , NaDS,MoonHB , Pa i CH . Detectionof Sal-
monellaTyph i inthebloodofpatientswithtyphoidfeverbypo - 32.
lymerasechainreaction. JC l inMicrobio 199
l 3;31(6):1439 e 43.
13. NgaT , KarkeyA,DongolS,ThuyH,DunstanS,HoltK,etal.The
sensitivity of real-time PCR amplication targeting invasive
Salmonellaserovars in biological specimens. BMC Infect Dis
2010;10(1):125.
14. Wain J, Pham VB, Ha V , Nguyen NM, To SD, Walsh AL, et al. 34.
Quantitation of bacteria in blood of typhoid fever patients
and relationship between counts and clinical features, trans-
missibility , and antibiotic resistance J Clin
. Microbiol 1998;
36(6):1683 e 7.
15. Coleman W . Short-duration typhoid fever Am J .Med Sci1909;
137(6):781 e 9.
16. ColemanW , BuxtonBH.Thebacteriologyofthebloodinconva- 36.
lescencefromtyphoidfever . Withatheoryofthepathogenesis
of the disease.J Med Res1909;21(1):83e 93.
17. WatsonKC.Laboratoryandclinicalinvestigationofrecoveryof
Salmonellatyphi from blood. J Clin Microbiol 1978;7(2):
122e 6.
18. Kaye D, Palmieri M, Rocha H. Effect of bile on the action of 38.
blood againstSalmonella . J Bacteriol 1966Mar;91(3):945 e 52.
19. Zhou L, Pollard AJ. A fast and highly sensitive blood culture
PCRmethodforclinicaldetectionof Salmonellaentericasero-
var Typhi.Ann Clin Microbiol Antimicrob 2010;9:14.
20. ZhouL,PollardAJ.Anovelmethodofselectiveremovalofhu-
manDNAimproves PCR sensitivity for detectionSalmonella of
Typhi in blood samples. BMC Infect Dis2012;12:164.
21. Waddington CS, Darton TC, Jones C, Haworth K, Peters A,
John T , et al. An outpatient, ambulant-design, controlled hu-40.
man infection model using escalating doses Salmonella
of Ty -
phi challenge delivered in sodium bicarbonate solution. Clin
Infect Dis2014;58(9):1230 e 40.
22. Health Protection AgencyInvestigation
. of faecal specimens 41.
forentericpathogens.Vol.B30:UKStandardsforMicrobiology
Investigations. 2013.
23. HealthProtectionAgencySalmonellaidentication:serotypes
. reference.PLoS One2015;10(11):e0142364.
and antigenic formulae: Kauffmann-White Scheme 2007. Lon-
don. 2007.
24. Health Protection AgencyIdentication
. of Salmonella spe-
cies.Vol. ID 24:UK standardsfor microbiology investigations . 43.
2011.
25. ZhouL,DartonT , WaddingtonCS,PollardAJ.Moleculardiagnosis
ofentericfever:progressandperspectives.In:AnnousB,editor .
Salmonellae distribution, adaptation, control measures and
moleculartechnologies . Online:InTech;2012.
26. Frankel G, Newton SM, Schoolnik GK, Stocker BA. Unique se-
quencesinregionVIoftheagellingeneof SalmonellaTyphi.
Mol Microbiol1989;3(10):1379 e 83.
T.C. Darton et al.
27. ZhuQ,LimCK,ChanYN.Detectionof SalmonellaTyphibypo-
lymerase chain reaction. J Appl Bacteriol1996;80(3):244e 51.
28. NairS,PohCL,LimYS,TayL , GohKT . Genomengerprintingof
SalmonellaTyphi by pulsed-eld gel electrophoresis for sub-
typing common phage types. Epidemiol Infect1994;113(3):
391e 402.
29. Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP ,
Irwig LM, et al. The STARD statement for reporting studies of
diagnostic accuracy: explanation and elaboration. Ann Intern
Med2003;138(1):W1e 12.
30. Darton T , Guiver M, Naylor S, Jack DL, Kaczmarski EB,
Borrow R, et al. Severity of meningococcal disease associated
with genomic bacterial load. Clin Infect Dis 2009;48(5):
587e 94.
31. Gray J, Coupland LJ. The increasing application of multiplex
nucleic acid detection tests to the diagnosis of syndromic in-
fections.Epidemiol Infect2014;142(1):1e 11.
Maurin M. Real-time PCR as a diagnostic tool for bacterial dis-
eases.Expert Rev Mol Diagn 2012;12(7):731e 54.
33. TennantSM,ToemaD,QamarF , IqbalN,BoydMA,MarshallJM,
et al. Detection of typhoidal and paratyphoidal Salmonellain
blood by real-time polymerase chain reaction. Clin Infect Dis
2015;61(Suppl 4):S241 e 50.
Chaudhry R, Laxmi BV , Nisar N, Ray K, Kumar D. Standardisa-
tion of polymerase chain reaction for the detection Salmo-
of
nella typhi in typhoid feverJ Clin
. Pathol 1997;50(5):437 e 9.
35. Teh CS, Chua KH, Puthucheary SD, Thong KL. Further evalua-
tionofamultiplexPCRfordifferentiationof Salmonellapara-
typhi A from other salmonellae. Jpn J Infect Dis 2008;61(4):
313e 4.
Wain J, Pham VB, Ha V , Nguyen NM, To SD, Walsh AL, et al.
Quantitation of bacteria in bone marrow from patients with
typhoid fever: relationship between counts and clinical fea-
tures. J Clin Microbiol 2001;39(4):1571 e 6.
37. Woodard WE.Summary of 1886 subjects in Volunteer chal-
lenges. University of Maryland , School of Medicine; 1980.
Moore CE, Pan-Ngum W , Wijedoru LP , Sona S, Nga TV , Duy PT ,
et al. Evaluation of the diagnostic accuracy of a typhoid IgM
owassayforthediagnosisoftyphoidfeverinCambodianchil-
drenusingaBayesianlatentclassmodelassuminganimperfect
gold standard.Am J Trop Med Hyg2014;90(1):114 e 20.
39. EdsallG,GainesS, LandyM, TigerttWD,SprinzH, TrapaniRJ,
et al. Studies on infection and immunity i n experimental
typhoid fever . I. Typhoid fever in chimpanzees orally infected
with Salmonellatyphosa.J Exp Med1960;112:143e 66.
Blohmke CJ, Darton TC, Jones C, Suarez NM, Waddington CS,
Angus B, et al. Interferon-driven alterations of the hosts
amino acid metabolism in the pathogenesis of typhoid fever .
J Exp Med2016;213(6):1061 e 77.
Storey HL, Huang Y , Crudder C, Golden A, de los Santos T ,
Hawkins K. A meta-analysis of typhoid diagnostic accuracy
studies:arecommendationtoadoptastandardizedcomposite
42. WainJ,DiepTS,HoVA , WalshAM,NguyenTT , ParryCM,etal.
Specimens and culture media for the laboratory diagnosis of
typhoid feverJ.Infect Dev Ctries2008;2(6):469 e 74.
MurrayPR,SpizzoAW , Niles AC.Clinicalcomparison ofthere-
coveries of bloodstream pathogens in Septi-Chek brain heart
infusion broth with saponin, Septi-Chek tryptic soy broth,
and the isolator lysis-centrifugation system. J Clin Microbiol
1991;29(5):901 e 5.
44. ZhouL,JonesC,GibaniMM,DobinsonH,Thomaides-BrearsH,
Shrestha S, et al. Development and evaluation of a blood cul-
ture PCR assay for rapid detection Salmonellaparatyphi
of A
in clinical samples.PLoS One2016;11(3):e0150576.