Excellen Diagnos ic Cha ac e is ics o
Ul a as Gene P ofiling o DEFA1-IL1B-
LTF in De ec ion o P os he ic Join
In ec ions
Regina Fille o a,
a
Ji i Gallo,
b
Ma in Rad ansky,
c
Ve onika K aiczo a,
a
Milos Kudelka,
c
E a K iego a
a
Depa men o Immunology, Facul y o Medicine and Den is y, Palacky Uni e si y & Uni e si y Hospi al,
Olomouc, Czech Republic
a
; Depa men o O hopaedics, Facul y o Medicine and Den is y, Palacky
Uni e si y, Olomouc, Czech Republic
b
; Depa men o Compu e Science, Facul y o Elec ical Enginee ing and
Compu e Science, VSB-Technical Uni e si y o Os a a, Os a a, Czech Republic
c
ABSTRACT The imely and exac diagnosis o p os he ic join in ec ion (PJI) is c u-
cial o su gical decision-making. In aope a i ely, deli e y o he esul wi hin an
hou is equi ed. Alpha-de ensin la e al immunoassay o join fluid (JF) is p ecise o
he in aope a i e exclusion o PJI; howe e , o pa ien s wi h a limi ed amoun o JF
and/o in cases whe e he JF is bloody, his es is unhelp ul. Impo an in o ma ion
is hidden in pe ip os he ic issues ha may much be e eflec he cu en s a us o
implan pa hology. We he e o e in es iga ed he u ili y o he gene exp ession pa -
e ns o 12 candida e genes (TLR1,-2,-4,-6, and 10,DEFA1,LTF,IL1B,BPI,CRP,IFNG,
and DEFB4A) p e iously associa ed wi h in ec ion o de ec ion o PJI in pe ip os-
he ic issues o pa ien s wi h o al join a h oplas y (TJA) (n⫽76) eope a ed o
PJI (n⫽38) o asep ic ailu e (n⫽38), using he ul a as quan i a i e e e se
ansc ip ion-PCR (RT-PCR) Xxp ess sys em (BJS Bio echnologies L d.). Ad anced da a-
mining algo i hms we e applied o da a analysis. Fo PJI, we de ec ed ele a ed
mRNA exp ession le els o DEFA1 (P⬍0.0001), IL1B (P⬍0.0001), LTF (P⬍0.0001),
TLR1 (P⫽0.02), and BPI (P⫽0.01) in compa ison o hose in issues om asep ic
cases. A ea u e selec ion algo i hm e ealed ha he DEFA1-IL1B-LTF pa e n was
he mos app op ia e o de ec ion/exclusion o PJI, achie ing 94.5% sensi i i y and
95.7% specifici y, wi h likelihood a ios (LRs) o posi i e and nega i e esul s o 16.3
and 0.06, espec i ely. Taken oge he , he esul s show ha DEFA1-IL1B-LTF gene ex-
p ession de ec ion by use o ul a as qRT-PCR linked o an elec onic calcula o al-
lows de ec ion o pa ien s wi h a high p obabili y o PJI wi hin 45 min a e sam-
pling. Fu he es ing on a la ge coho o pa ien s is needed.
KEYWORDS p os he ic join in ec ion, gene exp ession, pseudosyno ial issues,
diagnos ics, in aope a i e es
P os he ic join in ec ion (PJI) is one o he mos de as a ing and cos ly complica-
ions ollowing o al join a h oplas y (TJA), occu ing in 1 o 2% o cases a e
p ima y TJA and up o 7% o cases ollowing e ision TJA (1).
The majo i y o PJIs a e easily diagnosed using he adi ional diagnos ic a mamen-
a ium (2). Howe e , he e a e cases whe e he diagnosis o PJI lacks he equi ed
accu acy and/o he esul s o blood and join fluid es s a e una ailable o conflic ing.
Fo some pa ien s, join fluid o p eope a i e examina ions is no e en a ailable. On
he o he hand, pa ien s wi h sys emic inflamma o y diseases may ha e ele a ed
inflamma o y ma ke s ega dless o he p esence o absence o in ec ion (3). Finally, in
some cases, he p obabili y o PJI can inc ease jus a e pseudocapsule incision o
Recei ed 5 Ap il 2017 Re u ned o
modifica ion 24 Ap il 2017 Accep ed 13 June
2017
Accep ed manusc ip pos ed online 21
June 2017
Ci a ion Fille o a R, Gallo J, Rad ansky M,
K aiczo a V, Kudelka M, K iego a E. 2017.
Excellen diagnos ic cha ac e is ics o ul a as
gene p ofiling o DEFA1-IL1B-LTF in de ec ion
o p os he ic join in ec ions. J Clin Mic obiol
55:2686–2697. h ps://doi.o g/10.1128/JCM
.00558-17.
Edi o Robin Pa el, Mayo Clinic
Copy igh © 2017 Fille o a e al. This is an
open-access a icle dis ibu ed unde he e ms
o he C ea i e Commons A ibu ion 4.0
In e na ional license.
Add ess co espondence o E a K iego a,
[email p o ec ed].
BACTERIOLOGY
c ossm
Sep embe 2017 Volume 55 Issue 9 jcm.asm.o g 2686Jou nal o Clinical Mic obiology
on Sep embe 11, 2017 by gues h p://jcm.asm.o g/Downloaded om
emo al o he implan on he basis o he appea ance o pe ip os he ic issues and/o
join fluid, ega dless o he esul s o p eope a i e es s. Taken oge he , hese in-
s ances indica e ha he e is a s ong equi emen o a diagnos ic ool ha is exac and
a ailable in a imely manne o he in aope a i e exclusion o PJI.
A p esen , he e is e idence o he use ulness o he leukocy e-es e ase es and,
especially, he alpha-de ensin la e al immunoassay o he in aope a i e exclusion o
PJI, bo h o which deli e esul s wi hin 10 min (4). Howe e , hese es s ha e well-
known limi a ions in e ms o he equi ed wo king condi ions (e.g., he minimum
amoun o join fluid needed and a p oblem wi h blood in e e ence) and he gene -
a ion o alse-posi i e and/o alse-nega i e esul s (5–8). Pe ip os he ic issues should
also be conside ed an impo an sou ce o diagnos ic in o ma ion. The cells wi hin he
issue may di ec ly eflec he cu en s a us o he in ec ion, e.g., by he p oduc ion o
hos de ense an imic obial pep ides and p o eins and ac i a ion o an inna e immune
esponse. Impo an ly, pseudosyno ial and pe ip os he ic issues a e easily a ailable a
he beginning o su ge y o u he analysis; hey a e now used mainly o ou ine
cul u e examina ion o bac e ia (9). Some s udies ha e examined he clinical u ili y o
ozen sec ions, wi h clea limi a ions (10). Recen ly, i was epo ed ha gene p ofiling
o pe ip os he ic issues may allow he p oposal o no el PJI bioma ke s, as shown o
TLR1 (11).
We he e o e in es iga ed he exp ession pa e ns o 12 candida e genes o de ec-
ion o PJI in pe ip os he ic issues o pa ien s wi h TJA. The candida e genes we e
selec ed om p omising bioma ke s o PJI epo ed p e iously o (i) syno ial fluid,
namely, DEFA1 (coding o alpha-de ensin), LTF (lac o ans e in), CRP (C- eac i e p o-
ein), BPI (bac e icidal/pe meabili y-inc easing p o ein), IL1B (in e leukin 1 be a), DEFB4A
(be a-de ensin 4), and IFNG (gamma in e e on) (4, 12), and (ii) pe ip os he ic issue and
se um, namely, TLR1 (Toll-like ecep o 1), TLR2,TLR4,TLR6, and TLR10 (11, 13, 14).
Mo eo e , we s udied whe he ul a as quan i a i e one-s ep e e se ansc ip ion-PCR
(qRT-PCR) may deli e esul s wi hin a easonable ime (up o 45 min) a e sampling,
hus helping su geons in decision-making du ing he ope a ion.
RESULTS
Gene exp ession p ofiling. In o de o de e mine PJI-associa ed gene exp ession
pa e ns in pe ip os he ic issues, we compa ed gene exp ession le els o 12 candida e
molecules (TLR1,-2,-4,-6, and -10,DEFA1,LTF,IL1B,BPI,CRP,IFNG, and DEFB4A)
p e iously epo ed as po en ial bioma ke s o de ec ion o PJI in issues ob ained om
pa ien s wi h TJA du ing e ision su ge y.
Among he s udied genes, enhanced gene exp ession o DEFA1 (P⬍0.0001), IL1B
(P⬍0.0001), LTF (P⬍0.0001), TLR1 (P⫽0.02), and BPI (P⫽0.01) was de ec ed o
pa ien s wi h clinically p o en PJI compa ed o ha o pa ien s wi h asep ic loosening
(AL) by use o he Ro o Gene Q sys em; in addi ion, high in e indi idual a iabili y was
de ec ed in he pa ien subg oups (Fig. 1; Table 1). Gene exp ession le els o IFNG,CRP,
TLR2,TLR4,TLR6,TLR10,and DEFB4A did no di e be ween PJI and non-PJI cases (P⬎
0.05) (Fig. 1).
U ili y o ul a as qRT-PCR. In o de o assess he u ili y o ul a as qRT-PCR o
de ec ion o in ec ed issues wi hin a easonable ime, we in oduced gene p ofiling on
he no el qPCR Xxp ess ins umen . Mo eo e , gene exp ession pa e ns ob ained by
use o he Ro o Gene Q and Xxp ess ins umen s we e compa ed.
The da a om he qRT-PCR Xxp ess sys em we e compa able o he da a om he
Ro o Gene Q sys em ega ding up egula ion/down egula ion and significance, bu hey
di e ed in hei ob ained h eshold cycle (C
T
) alues (Table 1). A di e ence was
achie ed only o TLR2 gene exp ession, as he di e ence be ween PJI and non-PJI
pa ien s eached significance on he Xxp ess ins umen bu no on he Ro o Gene Q
ins umen . Using one-s ep RT-PCR, he qRT-PCR da a we e a ailable wi hin 20 min.
Using he Xxp ess ins umen , he p ocedu e om sampling o a ailabili y o qRT-PCR
da a akes less han 45 min (Fig. 2A).
Gene Pa e n o PJI Diagnos ics Jou nal o Clinical Mic obiology
Sep embe 2017 Volume 55 Issue 9 jcm.asm.o g 2687
on Sep embe 11, 2017 by gues h p://jcm.asm.o g/Downloaded om
Sea ch o genes and combina ions o genes dis inguishing PJI om non-PJI
issues. In o de o find genes o combina ions o genes ha cha ac e ize he g oup o
pa ien s wi h clinically p o en PJI, we applied a neu al ne wo k-based ea u e selec ion
algo i hm. Mo eo e , we calcula ed ecei e ope a ing cha ac e is ic (ROC) cu es o
DEFA1,IL1B,LTF, and TLR1 and neu al ne wo k classifica ion esul s based on he
selec ed combina ion o DEFA1-IL1B-LTF and he op fi e neu al ne wo ks.
The neu al ne wo k-based ea u e selec ion algo i hm iden ified DEFA1-IL1B-LTF as
he bes combina ion o cha ac e iza ion o clinically p o en PJI. Using his gene
combina ion, we we e able o achie e ⬎81% o e all ag eemen o all en olled
pa ien s, on a e age, based on 100 10- old c oss- alida ed neu al ne wo ks (Fig. 3).
FIG 1 Rela i e gene exp ession le els o s udied genes in pe ip os he ic issues om PJI and non-PJI pa ien s, as de e mined wi h he Ro o Gene Q sys em.
G oup means a e indica ed by ho izon al ba s, and e o ba s indica e 95% confidence in e als. Only P alues o significan di e ences be ween g oups o
pa ien s a e s a ed (*,P⬍0.05; **,P⬍0.01; ***,P⬍0.001).
TABLE 1 Rela i e mRNA exp ession le els o candida e genes in pe ip os he ic issues ob ained om pa ien s wi h and wi hou PJI
a
Gene
Ro o Gene Q Xxp ess
Mean (95% CI)
P alue
Mean (95% CI)
P aluePJI (nⴝ38) Non-PJI (nⴝ38) PJI (nⴝ23) Non-PJI (nⴝ25)
DEFA1 1.021 (0.001–2.042) 0.035 (0.017–0.053) ⬍0.0001 19.650 (0.278–57.030) 0.057 (0.013–0.102) ⬍0.0001
IL1B 4.361 (0.717–8.006) 0.201 (0.093–0.309) ⬍0.0001 0.637 (0.042–1.585) 0.013 (⫺0.0001–0.025) 0.0004
LTF 5.599 (0.228–13.710) 0.162 (0.095–0.228) ⬍0.0001 7.606 (0.638–18.380) 0.397 (0.183–0.612) 0.0003
TLR1 1.856 (1.452–2.261) 1.563 (0.970–2.156) 0.019 2.728 (1.313–4.143) 0.927 (0.320–1.535) 0.001
BPI 0.445 (0.039–0.964) 0.041 (0.025–0.057) 0.010 1.524 (0.177–3.559) 0.181 (0.020–0.341) 0.002
IFNG 9.146 (2.438–15.850) 3.690 (2.230–5.150) 0.644 0.954 (0.405–1.503) 1.681 (0.405–2.958) 0.451
TLR2 3.876 (2.276–5.476) 2.005 (1.399–2.610) 0.137 1.680 (0.988–2.373) 0.625 (0.206–1.043) 0.002
TLR4 2.515 (1.817–3.213) 3.060 (2.380–3.740) 0.169 7.549 (4.679–10.420) 5.651 (3.316–7.986) 0.358
CRP 3.912 (1.151–6.673) 5.079 (2.781–7.377) 0.175 NA NA NA
TLR6 4.121 (2.748–5.495) 3.045 (2.387–3.703) 0.983 NA NA NA
TLR10 3.154 (1.504–4.804) 3.164 (2.087–4.242) 0.448 NA NA NA
DEFB4A 12.350 (3.624–21.080) 4.601 (2.397–6.806) 0.537 NA NA NA
a
Resul s a e exp essed ela i e o le els o HPRT1 mRNA. 95% CI, 95% confidence in e al; NA, no a ailable.
Fille o a e al. Jou nal o Clinical Mic obiology
Sep embe 2017 Volume 55 Issue 9 jcm.asm.o g 2688
on Sep embe 11, 2017 by gues h p://jcm.asm.o g/Downloaded om
Impo an ly, all pa ien s wi h posi i e mic obiology findings we e co ec ly classified
in o he PJI g oup (100% posi i e ag eemen ). Fo he combina ion o DEFA1-IL1B-LTF,
he obse ed sensi i i y and specifici y o he op fi e neu al ne wo ks eached 94.5%
and 95.7%, espec i ely. The likelihood a ios (LRs) o posi i e and nega i e esul s o
he op fi e neu al ne wo ks, on a e age, we e 16.3 (95% confidence in e al, 15.9 o
16.7) and 0.06 (95% confidence in e al, 0.06 o 0.06), espec i ely (Table 2; Fig. 4).
E en he sole exp ession o DEFA1 was able o cha ac e ize bo h he PJI and non-PJI
g oups o pa ien s, wi h a sensi i i y o 95.7% and a specifici y o 79.3% ( h eshold ⫽
0.046) (Table 2; Fig. 3). A neu al ne wo k wi h 10- old c oss- alida ion showed he high
u ili y o DEFA1 o sepa a ion o PJI cases, achie ing 80% o e all ag eemen , on
a e age, on bo h ins umen s (Fig. 3). Fo IL1B ( h eshold ⫽0.005), he sensi i i y was
78.1% and he specifici y 75.8%; o LTF ( h eshold ⫽0.483), hese alues we e 82.8%
and 79.9%, espec i ely (Table 2; Fig. 3).
Classifica ion based on neu al ne wo k analysis also showed ha exp ession o TLR1
(Fig. 3)o heTLR1-TLR6-TLR10 combina ion (Fig. 3) is no usable o classifica ion o PJI
and AL. The o e all pe cen ag eemen eached 49% o TLR1 and 57% o he
TLR1-TLR6-TLR10 combina ion; o he TLR genes (TLR2,TLR4,TLR6, and TLR10) and e en
hei combina ions did no each o e all ag eemen le els o ⬎40%. De ec ion o gene
exp ession o TLR1 ( h eshold ⫽0.616) showed a sensi i i y o 90.0% and a specifici y
o 66.6% (Table 2; Fig. 3).
Calcula o o PJI p obabili y. To iden i y pa ien s wi h PJI, we c ea ed an elec onic
decision ool based on he op fi e neu al ne wo ks (Fig. 5). To es he abili y o he
ained neu al ne wo k models o iden i y PJI, we analyzed addi ional blinded samples
om clinical se ings ha we e no p e iously used o he aining p ocedu e (Fig. 6).
Fo he es g oup o 10 pa ien s, we classified 3 PJI and 7 non-PJI pa ien s co ec ly
(Table 3). Using his ool, we we e able o deli e he da a o clinicians o in aope a i e
decision-making immedia ely a e ob aining qRT-PCR esul s (Fig. 2B).
DISCUSSION
In his s udy, we in oduced ul a as qRT-PCR o de ec /exclude PJI in pa ien s wi h
TJA. The exp ession o he DEFA1-IL1B-LTF gene combina ion in pe ip os he ic issues
changes he p e es p obabili y o PJI ma kedly wi hin 45 min a e sampling, making
his examina ion app op ia e o in aope a i e decision-making. In addi ion, we de-
eloped a decision suppo ool o he in e p e a ion o da a ob ained om exp ession
p ofiling.
The imely diagnosis o PJI is c ucial o in aope a i e decision-making, as ea men
o PJI necessi a es unique su gical s a egies compa ed o hose o asep ic eope a ion,
aiming p ima ily a e adica ion o he in ec ing o ganisms (15, 16). The majo i y o
FIG 2 Schema o sample p ocessing om issue sampling o decision-making, using issue gene p ofiling (A) and a diagnos ic wo k-up s a egy o exclusion
o PJI (B). SWCC, syno ial whi e cell coun .
Gene Pa e n o PJI Diagnos ics Jou nal o Clinical Mic obiology
Sep embe 2017 Volume 55 Issue 9 jcm.asm.o g 2689
on Sep embe 11, 2017 by gues h p://jcm.asm.o g/Downloaded om
in ec ions a e easily diagnosed p eope a i ely by use o mo e o less adi ional
diagnos ic ools, and his is ollowed by app op ia e he apy (4). Howe e , o pa ien s
wi h low-g ade PJI o hose ha appea o ha e AL, he p obabili y o PJI can
unexpec edly inc ease in aope a i ely, so we need a sensi i e and as diagnos ic ool
FIG 3 Ne wo ks isualizing he bes combina ions o genes dis inguishing PJI om non-PJI cases, using a neu al
ne wo k-based ea u e selec ion algo i hm. The esul s a e shown o gene pa e ns o DEFA1-IL1B-LTF (A and B),
DEFA1 (C and D), TLR1 (E), and TLR1-TLR6-TLR10 (F) ob ained wi h Xxp ess (A and C) and Ro o Gene Q (B and D o
F) ins umen s. Ne wo k e ices ep esen indi idual gene pa e ns (indi idual pa ien eco ds); edges (links)
be ween he e ices ep esen simila i ies o he co esponding pa ien eco ds, calcula ed by he Gaussian
unc ion. Colo s dis inguish he classes as pa ien s wi h clinically p o en PJI ( ed) and non-PJI pa ien s (g een).
TABLE 2 Value o gene exp ession p ofiles o pe ip os he ic issues ob ained wi h he Xxp ess sys em o diagnosis o PJI, conside ing
he esul s o each sample independen ly
a
Gene(s) es ed
Mean % (95% CI)
Sensi i i y/posi i e
pe cen ag eemen
Specifici y/nega i e
pe cen ag eemen
Posi i e
p edic i e alue
Nega i e
p edic i e alue Accu acy LRⴙLRⴚ
DEFA1 95.7 (95.4–96.0) 79.8 (79.3–80.3) 81.2 (80.8–81.7) 95.3 (95.0–95.6) 87.4 (87.1–87.7) 5.9 (5.6–6.1) 0.05 (0.05–0.06)
IL1B 78.1 (77.6–78.7) 75.8 (75.3–76.4) 74.7 (74.1–75.2) 79.2 (78.7–79.7) 76.9 (76.6–77.3) 3.8 (3.7–3.9) 0.29 (0.28–0.30)
LTF 82.8 (82.3–83.3) 79.9 (79.4–80.4) 79.3 (78.8–79.8) 83.4 (82.9–83.9) 81.3 (81.0–81.6) 5.0 (4.8–5.2) 0.22 (0.21–0.22)
TLR1 90 (89.6–90.4) 66.6 (66.0–67.2) 70.3 (69.7–70.8) 88.3 (87.9–88.8) 77.5 (77.1–77.9) 3.0 (2.9–3.1) 0.15 (0.15–0.16)
DEFA1-IL1B-LTF a g
o NN1 o NN5
94.5 (94.4–94.7) 95.7 (95.5–95.9) 95.0 (94.8–95.2) 95.6 (95.5–95.8) 95.2 (95.1–95.3) 16.3 (15.9–16.7) 0.06 (0.06–0.06)
a
LR⫹, likelihood a io o a posi i e esul ; LR⫺, likelihood a io o a nega i e esul ; NN, neu al ne wo k.
Fille o a e al. Jou nal o Clinical Mic obiology
Sep embe 2017 Volume 55 Issue 9 jcm.asm.o g 2690
on Sep embe 11, 2017 by gues h p://jcm.asm.o g/Downloaded om

o he apid and eliable di e en ia ion o ue PJI om alse-posi i e esul s. Fo his
s udy, we wonde ed whe he exp ession p ofiling o genes p e iously associa ed wi h
in ec ion may help iden i y PJI in pe ip os he ic issue, especially when syno ial fluid is
no a ailable bu issues a e a ailable. Fo ea u e selec ion and classifica ion o gene
exp ession da a in ou s udy, we used a neu al ne wo k app oach shown o ha e
excellen p ope ies o he analysis o gene exp ession pa e ns (17–20). These mi-
c oa ay s udies p o ed ha a neu al ne wo k akes in o accoun he in insic cha ac-
e is ics o gene exp ession da a, confi ms he mos in o ma i e gene subse s, and
imp o es he classifica ion accu acy wi h he bes pa ame e s based on da a se s. Ou
analysis o he exp ession p ofiles o candida e genes p e iously associa ed wi h PJI
e ealed a high u ili y o he exp ession o DEFA1, a gene coding o alpha-de ensin, o
dis inguishing pa ien s wi h and wi hou PJI. This obse a ion is in line wi h he
FIG 4 ROC cu es wi h a ea unde he concen a ion- ime cu e (AUC) alues o DEFA1,IL1B,LTF, and TLR1 gene p ofiles (A) and o he op fi e neu al ne wo ks
o he DEFA1-IL1B-LTF combina ion (B).
FIG 5 Rep esen a i e examples o end-use in e ace esul s o he elec onic calcula o o he DEFA1-
IL1B-LTF gene combina ion exp ession da a o pa ien s om he es coho , showing a high p obabili y
o PJI (le ) and a low p obabili y o PJI ( igh ).
Gene Pa e n o PJI Diagnos ics Jou nal o Clinical Mic obiology
Sep embe 2017 Volume 55 Issue 9 jcm.asm.o g 2691
on Sep embe 11, 2017 by gues h p://jcm.asm.o g/Downloaded om
e idence o an alpha-de ensin es o syno ial fluid as a highly sensi i e and specific PJI
bioma ke . The alpha-de ensin es is in excellen acco dance wi h he Musculoskele al
In ec ion Socie y (MSIS) c i e ia compa ed o o he es s ou inely used o he p e-/
in aope a i e diagnosis o PJI (cul u e, e y h ocy e sedimen a ion a e [ESR], CRP le el,
syno ial whi e cell coun , and leukocy e-es e ase es ) (21, 22). This an imic obial
pep ide is eleased mainly by local neu ophils in esponse o a wide spec um o
mic oo ganisms, ega dless o he o ganism ype, G am ype, species, o i ulence o
he o ganism (23). Impo an ly, some s udies ha e p o en ha alpha-de ensin es ing
main ains i s pe o mance o PJI e en in he se ing o an ibio ic adminis a ion (22).
Mo eo e , he e is e idence ha alpha-de ensin es ing pe o ms conside ably
FIG 6 Algo i hm flow cha (labo a o y es s, neu al ne wo ks/ ea u e selec ion algo i hm, and clinical es ing) used
in his s udy.
TABLE 3 Cha ac e is ics o a “ es g oup” o pa ien s unde going e ision su ge y a e TJA
a
Pa ien
ID
Pa ien
cha ac e is ic P eope a i e es esul In aope a i e es esul
MSIS
classifica ionGende
P ima y
diagnosis
CRP concn
in se um
(mg/li e )
IL-6 concn
in se um
(pg/ml)
SWCC
(10
3
cells/
␮
l)
Neu ophils
in fluid (%)
Lymphocy es
in fluid (%)
CRP concn
in fluid
(mg/li e )
DEFA1-IL1B-LTF
p ofiling
classifica ion
Confidence
(%)
P1 Male RA 51.8 10.4 75.5 97.5 2.5 20.3 In ec ion 90 PJI
P2 Male POA 31.7 16.8 34.5 92.0 8.0 NA In ec ion 60 PJI
P3 Male POA 29.2 11.2 NA NA NA NA In ec ion 100 PJI
P4 Female PD 7.1 3.3 0.5 31.1 68.9 2.0 No in ec ion 100 Non-PJI
P5 Female POA 0.6 2.2 NA NA NA NA No in ec ion 95 Non-PJI
P6 Female PD 0.3 NA 0.4 54.5 45.5 0.2 No in ec ion 100 Non-PJI
P7 Male NA 16.1 3.2 0.5 53.6 46.4 1.5 No in ec ion 95 Non-PJI
P8 Female POA 2.3 7.4 0.3 53.0 47.0 0.7 No in ec ion 100 Non-PJI
P9 Female POA 3.7 5.6 1.6 27.1 72.9 1.0 No in ec ion 85 Non-PJI
P10 Female POA 6.3 NA 0.5 50.0 42.0 NA No in ec ion 63 Non-PJI
a
POA, p ima y os eoa h i is; PD, pos dysplas ic hip a h i is; RA, inflamma o y join disease, heuma oid a h i is; SWCC, syno ial whi e cell coun ; NA, no a ailable.
Fille o a e al. Jou nal o Clinical Mic obiology
Sep embe 2017 Volume 55 Issue 9 jcm.asm.o g 2692
on Sep embe 11, 2017 by gues h p://jcm.asm.o g/Downloaded om
be e when i is combined wi h he syno ial CRP le el (21). Addi ionally, ou esul s
show ha he combina ion o DEFA1 wi h o he genes, such as IL1B, dis inguishes PJI
be e han DEFA1 alone does. IL1B is a c i ical media o o he hos ’s esponse o
mic obial in ec ions and bac e ial clea ance (24). I s p o ein le els in syno ial fluid we e
also epo ed o be highly in o ma i e o PJI in pa ien s wi h sys emic inflamma o y
diseases and p eope a i e an ibio ic ea men s (12). Fu he mo e, LTF possesses an i-
mic obial ac i i y and is an impo an pa o he inna e de ense agains bac e ia, ungi,
and i uses (25). The e is al eady e idence ha LTF, oge he wi h o he bioma ke s
(alpha-de ensin, ELA2, BPI, and NGAL), co ec ly p edic s he occu ence o PJI in
syno ial fluid (26). In ou s udy, he combina ion o DEFA1 wi h CRP did no enhance he
di e en ia ion be ween pa ien s wi h PJI and non-PJI pa ien s, p obably as a esul o
he known nonspecific exp ession o CRP in o he cells (e.g., infil a ing immune cells)
and/o i s in ol emen mainly in he ea ly de ense agains in ec ions (27).
Impo an ly, he gene exp ession le els o DEFA1,IL1B, and LTF we e ele a ed in all
he samples wi h posi i e cul u es; howe e , wo pa ien s wi h nega i e cul u es bu
classified as ha ing PJI we e pu in o he non-PJI g oup. We sugges ha al hough he
issues om hese pa ien s ulfilled he PJI c i e ia, hey may ha e al eady been in he
la e s age o he an i-in ec i e esponse. Ou obse a ion is suppo ed by he ac ha
he main p oduce s o DEFA1 and o he an imic obial molecules a e sho -li ed neu-
ophils, he fi s immune cells o a el o he si es o bac e ial, ungal, and i al
in ec ions (28). One may he e o e sugges ha he issue gene pa e n migh al eady
eflec a (di e en ) la e s age o esponse o PJI, while he syno ial fluid may s ill con ain
ele a ed le els o soluble PJI bioma ke s. On he o he hand, fi e (14%) o he samples
wi hou PJI we e classified in o he PJI g oup. The enhanced exp ession o DEFA1-IL1B-
LTF may deli e a di e en ype o in o ma ion, eflec ing an ea lie phase o PJI o ,
pa icula ly, a low g ade o PJI, which may no be de ec ed ye by me hods analyzing
syno ial fluid o se um. The clinical ele ance o hese obse a ions and hei explana-
ion should be elucida ed u he .
Nex , we analyzed he exp ession pa e ns o membe s o he TLR amily and o he
genes known o play a c ucial ole in he inna e immune esponse o in ading
pa hogens (29). Despi e TLR1 being nomina ed as a po en ial ma ke o PJI in pe ip os-
he ic issues (11), we did no p o e i s u ili y o dis inguishing PJI om non-PJI cases.
Addi ionally, TLR2 and o he TLR membe s nomina ed by p e ious s udies (11, 13) ha
we e examined he e did no show any p edic i e alue o he de ec ion o PJI in ou
pa ien s. The eason may be he in ol emen o TLRs no only in pa hogen ecogni ion
bu also in wea pa icle-induced inflamma ion and os eolysis (30), which may well limi
hei usage as bioma ke s o PJI. Despi e hei an imic obial and/o an i i al ac i i ies,
BPI,DEFB4A, and IFNG did no show u ili y o he iden ifica ion o PJI in issues ei he .
We belie e ha gene p ofiling o pe ip os he ic issue migh help de e mine he
ue s age o issue damage, including ha induced by pa hogen in asion. Al hough
ou explo a o y s udy needs u he es ing on a la ge coho o pa ien s, we sugges
ha issue exp ession p ofiling o a pa icula se o genes coding o an i-in ec i e
p o eins may also be use ul o he de ec ion o in ec ion in o he issues. Mo eo e , ou
da a also show u ili y o dis inguishing be ween issues wi h acu e in ec ion and hose
in la e esol ing and eco e y phases. Whe he gene p ofiling o a issue is sui able o
dis inguishing be ween p esen and pas in ec ions in issues and, pa icula ly, i s
clinical ele ance me i u he in es iga ion. This migh open up a new a enue o
in aope a i e diagnos ics, including cases be o e he eimplan a ion o an implan in
wo-s age su ge y.
This s udy has se e al limi a ions. We a e awa e ha he in ec ion in ou pa ien
coho was based on a defini ion o PJI ha did no in ol e he pa ame e s o syno ial
fluid (31). We used a ela i ely simple his ological defini ion o PJI (32), as he new
his ological defini ion (33) was applied o ou ine specimen analysis la e . Addi ionally,
o some pa ien s we did no analyze syno ial join fluid because he join was d y o
syno ial join fluid leaked because o a su gical mis ake. Mo eo e , he cu en s udy
did no eflec he clinical and causa i e he e ogenei y o PJI ( ype o bac e ia, acu e-
Gene Pa e n o PJI Diagnos ics Jou nal o Clinical Mic obiology
Sep embe 2017 Volume 55 Issue 9 jcm.asm.o g 2693
on Sep embe 11, 2017 by gues h p://jcm.asm.o g/Downloaded om
ness e sus ch onici y o p esen a ion, and du a ion o symp oms). Thus, his explo -
a o y s udy showing excellen diagnos ic cha ac e is ics o he de ec ion o PJI needs
o be alida ed wi h la ge s udy coho s diagnosed acco ding o he MSIS c i e ia and
wi h subg oups eflec ing he he e ogenei y o PJI.
Taken oge he , ou da a show ha gene p ofiling o pe ip os he ic issues p o ides
a p ecise and as enough me hod compa ed o he cu en de ec ion o PJI bioma ke s
in syno ial fluid and se um. This may be ex emely use ul in cases whe e join fluid is
no a ailable o is comp omised by blood, e c., and in esol ing si ua ions wi h
conflic ing da a. Ou ul a as app oach shows conside able po en ial as an al e na i e
in aope a i e es o su geons ha is in ended o he p ecise and imely diagnosis
o PJI.
MATERIALS AND METHODS
S udy coho . The s udy coho consis ed o Czech Caucasian pa ien s unde going e ision su ge y
a e TJA (n⫽76) be ween 2010 and 2016 a a single ins i u ion (Depa men o O hopaedics, Uni e si y
Hospi al, Olomouc, Czech Republic); all pa ien s we e ope a ed on by a single su geon. Among he
pa ien s in he coho , 38 unde wen eope a ion due o AL and 38 due o PJI. The PJI cases we e
classified acco ding o he ollowing defini ion o PJI (31): (i) p esence o a sinus ac communica ing
wi h a join and/o in a-a icula pus; (ii) coinciden ally posi i e esul s o his ological examina ion (fi e
o mo e neu ophils pe high-powe field [HPF]) and cul u e o in aope a i e samples (ⱖ2 cul u es
posi i e o he same mic oo ganism); and (iii) i only in aope a i e cul u e o his ological esul s we e
posi i e, hen a leas wo o he ollowing signs had o be p esen : high clinical suspicion o in ec ion
(acu e onse , e e , e y hema, edema, pe sis en local pain, ea ly p os he ic ailu e, wound healing
dis u bances, e c.), e y h ocy e sedimen a ion a e o ⬎30 mm/h, C- eac i e p o ein le el ele a ed mo e
han 1.5 imes abo e he labo a o y e e ence alue, and posi i e echne ium-99m leukocy e scin ig a-
phy. The in ec ions in pa ien s ec ui ed o he es g oup (n⫽10) we e based on clinical ea u es
ma ching he Musculoskele al In ec ion Socie y (MSIS) c i e ia (34). The en olled AL pa ien s did no ulfill
he PJI c i e ia.
Pe ip os he ic issue specimens we e ha es ed acco ding o he in-house clinical p o ocol o
his opa hological, cul u e, and immunological examina ions. A leas 3 samples ( ange, 3 o 7) o each
me hod unde s udy we e aken om each pa ien , using a su gical kni e and o ceps. Sampling si es
we e as ollows: (i) inne memb ane o a TJA pseudocapsule; (ii) memb ane om he bone-implan
in e ace su ounding he p oximal componen , i a ailable; and (iii) memb ane om he bone-implan
in e ace su ounding he dis al componen , i a ailable. Addi ionally, one o mo e samples o issues
wi h a high suspicion o in ec ion/inflamma ion we e aken om some pa ien s, depending on he
su geon’s decision. Tissues om he PJI pa ien s we e ou inely assessed by cul u e-based analysis o
mul iple specimens as desc ibed p e iously (31). Fo de ailed pa ien cha ac e is ics and cul u e esul s,
see Table 4.
W i en in o med consen abou he usage o pe ip os he ic issues o he pu pose o his s udy was
ob ained om each subjec , and he local e hics commi ee app o ed he s udy.
One-s ep RT-PCR. The pseudosyno ial issue samples (0.1 o 1 mg) ob ained du ing e ision su ge y
we e placed immedia ely in T i eagen (Sigma-Ald ich, Ge many) and used immedia ely o isola ion o
RNA by use o a Di ec -zol RNA minip ep ki (Zymo Resea ch, USA) acco ding o he manu ac u e ’s
ecommenda ions. One-s ep qRT-PCR was pe o med in a o al olume o 20
␮
l con aining Kapa SYBR
Fas One-S ep qRT-PCR mix (Kapa Biosys ems, USA), 200 nM (each) gene-specific p ime s, and 5
␮
lo RNA
(10 ng/
␮
l; absolu e amoun , 50 ng). See Table 5 o p ime sequences (In eg a ed DNA Technologies,
USA). qRT-PCR was pe o med on wo sys ems, i.e., he Ro o Gene Q (Qiagen, USA) and Xxp ess (BJS
Bio echnologies, Uni ed Kingdom) sys ems. Each un included a no- empla e con ol, in which RNA
was eplaced by wa e . In de ail, o e e se ansc ip ion, he sample was held a 42°C o 5 min,
ollowed by enzyme inac i a ion a 95°C o 3 min. PCR was hen pe o med o 35 cycles o 95°C o
3 s and 60°C o 20 s. qRT-PCR akes 60 min on he Ro o Gene Q sys em and 20 min on he Xxp ess
PCR sys em. The da a we e no malized o he housekeeping gene HPRT1 (35,36). A human uni e sal
e e ence RNA (S a agene, USA) was used in iplica e as a calib a o , a 12.5 ng/ eac ion mix,
calcula ed based on inpu RNA.
S a is ical analysis and ea u e selec ion algo i hm. S a is ical analysis (nonpa ame ic Mann-
Whi ney U es ) o ela i e gene exp ession alues was pe o med using G aphPad P ism 5.01 (G aphPad
So wa e, La Jolla, CA, USA). ROC cu es, sensi i i y, specifici y, posi i e p edic i e alue, nega i e
p edic i e alue, accu acy, and posi i e/nega i e likelihood a ios (LRs) o gene exp ession p ofiles we e
calcula ed using R so wa e. The h eshold a e aging me hod on boo s apped alues was used o
calcula ion o confidence in e als (37).
Fo iden ifica ion o he bes combina ion o genes dis inguishing PJI om non-PJI cases, a neu al
ne wo k-based ea u e selec ion algo i hm con aining he co ec class assignmen (PJI/non-PJI) o each
eco d o exp ession alues was used. Neu al ne wo ks a e based on ea u es analogous o hose o
human lea ning. Depending on he lea ning ( aining) da a, neu al ne wo ks ha e he abili y o
gene alize. This is mani es ed by he ac ha he ained ne wo k is able o deduce phenomena ha can
be de i ed in some way, e en hough hey we e no (o only o a limi ed ex en ) pa o lea ning (38).
Fille o a e al. Jou nal o Clinical Mic obiology
Sep embe 2017 Volume 55 Issue 9 jcm.asm.o g 2694
on Sep embe 11, 2017 by gues h p://jcm.asm.o g/Downloaded om