T helper type cytokines in sepsis: time-shared variance and correlation with organ dysfunction and hospital mortality

w w w . e l s e v i e r . c o m / l o c a t e / b j i d

The

Brazilian

Journal

of

INFECTIOUS

DISEASES

Original

article

T

helper

type

cytokines

in

sepsis:

time-shared

variance

and

correlation

with

organ

dysfunction

and

hospital

mortality

Ramon

Teixeira

Costa

_,

_Orlei

_Ribeiro

_de

_Araújo

_,

_Milena

_Karina

_Coló

_Brunialti

_,

Murilo

Santucci

Cesar

Assunc¸ão

_,

_Luciano

_César

_Pontes

_Azevedo

_,

_Flávio

_Freitas

_,

Reinaldo

Salomão

a_{UniversidadeFederaldeSãoPaulo,EscolaPaulistadeMedicina,DepartmentodeMedicina,SãoPaulo,SP,Brazil} b_{ACCamargoCancerCenter,UnidadedeCuidadosIntensivos,SãoPaulo,SP,Brazil}

c_{InstitutodeOncologiaPediátrica(IOP),GrupodeApoioaoAdolescenteeaCrianc¸acomCancer(GRAACC),SãoPaulo,SP,Brazil} d_{HospitalIsraelitaAlbertEinstein,UnidadedeTerapiaIntensiva,SãoPaulo,SP,Brazil}

e_{HospitalSírioLibanês,UnidadedeTerapiaIntensiva,SãoPaulo,SP,Brazil}

f_{UniversidadeFederaldeSãoPaulo,HospitalSãoPaulo,UnidadedeDoreTerapiaIntensiva,SãoPaulo,SP,Brazil}

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received4February2019 Accepted28April2019 Availableonline18May2019

Keywords: Sepsis Cytokines Thresponse Organdysfunction SOFA

a

b

s

t

r

a

c

t

Objective:WeevaluatedthekineticsofcytokinesbelongingtotheThelper1(Th1),Th2,and Th17profilesinsepticpatients,andtheircorrelationswithorgandysfunctionandhospital mortality.

Methods:Thiswasaprospectiveobservationalstudyinacohortofsepticpatients admit-tedtotheintensivecareunits (ICU)ofthreeBraziliangeneralhospitals. Atotalof104 septic patientsand53healthvolunteers(controls)wereincluded.Plasmasampleswere collectedwithinthefirst48hoforgandysfunctionorsepticshock(0D),afterseven(D7) and14 days(D14)offollow-up.Thefollowingcytokinesweremeasuredbyflow cytom-etry:Interleukin-1␤(IL-1␤),IL-2,IL-6,IL-8,IL-10,IL-12/23p40,IL-17,IL-21,tumornecrosis factor-␣(TNF-␣),granulocyte-macrophagecolonystimulatingfactor(GM-CSF),granulocyte colony-stimulatingfactor(G-CSF).

Results:IL-6,IL-8,G-CSFandIL-10concentrationswerehigherinsepticpatientsthanin controls(p<0.001),whileIL-12/23p40presentedhigherlevelsinthecontrols(p=0.003).IL-6, IL-8andIL-17correlatedwithSequential[Sepsis-related]OrganFailureAssessment(SOFA) D0,D1andD3(exceptforIL-6atD0).IL-8wasassociatedwithrenalandcardiovascular dysfunction.Inamixedmodelanalysis,IL-10estimatedmeanswerelowerinsurvivors thanindeceased(p=0.014),whileIL-21hadanestimatedmeanof195.8pg/mLforsurvivors and98.5fordeceased(p=0.03).Cytokinesweregroupedinfourfactorsaccordingtotheir kineticsoverthethreedosages(D0,D7,D14).Group1encompassedIL-6,IL-8,IL-10,IL-1␤, andG-CSFwhileGroup3encompassedIL-17andIL-12/23p40.BothcorrelatedwithSOFA

∗ _{Correspondingauthor.}

E-mailaddress:rsalomao@unifesp.br(R.Salomão).

https://doi.org/10.1016/j.bjid.2019.04.008

1413-8670/©2019SociedadeBrasileiradeInfectologia.PublishedbyElsevierEspa ˜na,S.L.U.ThisisanopenaccessarticleundertheCC BY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

(D0)(p=0.039andp=0.003,respectively).IL-21(Group4)washigherinthosewhosurvived. IL-2,TNF-␣andGM-CSF(Group2)showednocorrelationwithoutcomes.

Conclusion: Inflammatoryandanti-inflammatorycytokines sharedco-varianceinseptic patientsandwererelatedtoorgandysfunctionsandhospitalmortality.

©2019SociedadeBrasileiradeInfectologia.PublishedbyElsevierEspa ˜na,S.L.U.Thisis anopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/ licenses/by-nc-nd/4.0/).

Introduction

Sepsisis a majorcause of morbidityand mortality world-wide, with increased burdenin low and medium incomes countries: in Brazilianintensive care units (ICU), mortality ratesashighas55%havebeenreported,whichemphasizes theneedforclinicalstudiesinthispopulation.1_Theimmune

responsetoinfectionisacomplexprocess,involvingvarious typesofcellsandproductionofacascadeofcytokines, lead-ing tocelldamageand organ dysfunction.2–4 _Thelper(Th)

cellshavelongbeenrecognizedasvitalcomponentsofthe adaptiveimmunesystem.MajorThcellsubsetsincludeTh1, Th2, and Th17,and each subset produces specificeffector cytokines.Th1lymphocytesproduceIFN-␥,TNFandIL-2;Th2 cellssecreteIL-4,IL-5,IL-10,andIL-13;andTh17cellsproduce effectorcytokinesIL-17,IL-21,IL-22,andIL-23.5_Studies

sug-gestthatThlymphocytesprofilescontributetoautoimmune andinflammatorydiseases,includingsepsis.6_{Septicpatients}

havebeenreportedtopresentincreasedpercentageof IL-17-producingThelperlymphocyteswhencomparedtohealthy volunteers.7_{Contrarily,Guptaetal.showedthatpatientswith}

post-traumaticsepsis(traumahemorrhagicshock),presented lower Th cells response comparedto non-septicpatients.8

RecentstudyevaluatedtheTCD4+cellsinpatientswithacute respiratorydistresssyndromeandfoundthatthosepatients withanimbalancefavoringtheTh17profilepresentedhigher mortality.9 _{There is a paucity of studies evaluating the}

kinetics of Th cytokines in sepsis; most of the studies focusedonintermittentmeasurementsoffewcytokines, pro-viding only a partial picture ofthe complex immunologic cascade.

Herein,weevaluated aprofileofcytokinesinplasmaof septicpatients,encompassingthoserelatedtoTh1,Th2and Th17,ondayszero,sevenand14followingdiagnosisofsepsis inadultsandanalyzedtheirtime-sharedvarianceandtheir associationwithsepsis-inducedorgan dysfunctionsand in-hospitalmortality.

Methods

PatientsadmittedtotheICUofthreetertiaryhospitalsfrom SãoPaulo,Brazil,withthediagnosisofseveresepsisor sep-ticshockwereincludedinaprospectiveobservationalstudy betweenSeptember2007andSeptember2010.10_Severe

sep-sisandsepticshockweredefinedaccordingtoBoneetal.11

androughlycorrespondtothesepsisandsepticshockofthe revisedsepsisdefinition.12

After obtaining informed consent, blood samples were drawnfrom104adultsepticpatientswithinthefirst48hof organdysfunctionorsepticshock,andfrom53healthy vol-unteers, matched forage and sex. Sampleswere collected fromthesepticpatientsatthetimeofinclusion(D0),onthe 7th (D7)and 14thdays(D14)offollow-up.Healthycontrols hadsamplescollectedonlyonce.Thisstudywasapprovedby theEthicsCommitteesofUniversidadeFederaldeSaoPaulo (CEP1477/06),HospitalAlbertEinstein(CEP07/549),and Hospi-talSírioLibanês(CEP2006/27).

Forpatientsandcontrols,10mLofbloodwerecollectedin heparintubes.Thesampleswerecentrifugedforseparationof plasma,whichwasfrozeninaliquotsat−80◦_{Cuntiluse.The}

severityofthediseaseandorgandysfunctionswereassessed using the Sequential [Sepsis-related] Organ Failure Assess-ment(SOFA)scoresatdays0,1and3afterICUadmission.13

CytokinesmeasurementswereperformedbytheCytometric BeadArray(CBA)FlexSetfromBDBiosciences,SanJose,CA, USA.

Data were analyzed with statistical software SPSS ver-sion20.0(SPSSInc.,Chicago,IL,USA).Thesignificancelevel was p<0.05.Differences between distributions and means were evaluated by the Mann–Whitney test or T-test. Cor-relations were evaluated by the Spearman test and linear regression.The99%confidenceintervalsofthepvalueswere obtainedbybootstrapresampling,with1000tablesfromthe sample.Logisticregression(univariateandmultivariate)was performedtoevaluateriskfactorsforin-hospitalmortality. Inordertomaximizetheexplanatorypowerofthecytokine dataset in the threetime-points and toidentify correlated subgroups,afactor analysiswasperformedusing the prin-cipalcomponentsmethodandVarimaxrotation,reducingthe numberofvariablesaccordingtothecorrelationorcovariance patterns.

We used mixed model analysis to evaluate oscillations betweencytokinesmeasurementsoverthethreetime-points andthefactor’seffectsuchassepticshockanddeath.14,15

Theareaunderthereceiver-operatingcharacteristiccurve (AUROC)wasusedtoevaluatethecytokinesaccuracyonD0to predicthospitalmortality.

Results

PatientdemographicsandclinicaldataareshowninTable1. PlasmalevelsofIL-2, IL-6,IL-8,TNF-␣,GM-CSF,IL-10,IL-1␤, IL-17,G-CSF,IL-21,andIL-12/23p40insepticpatients(D0,D7 andD14)andinhealthyvolunteersareshowninTable2.

Table1–Clinicalvariablesanddemographicdatafrom

septicpatients.

Numberofpatients 104

Age(years,median,IQR) 62.5(47.8–78)

Female(N,%) 47(45)

Siteofinfection(N,%)

Respiratory 40(38.5) Abdominal 28(26.9) Urinarytract 16(14.5) Skinand/orsofttissue 8(7.7) Bloodstreaminfection 4(3.8) Othersorunknown 8(7.7) Septicshock(N,%) 76(69.9) 7thdaymortality(N,%) 15(14.4) 28thdaymortality(N,%) 27(26) Hospitalmortality(N,%) 39(37.5) Vasopressors(N,%) 81(77.9) Organ/systemdysfunction(N,%) Cardiovascular 90(86.5) Renal 48(46.5) Respiratory 46(44.2) Hematologic 28(26.9) Hepatic 29(27.9) Neurological 36(34.6)

APACHEII(AcutePhysiology,Age, ChronicHealthEvaluationII)score (mean,SD)

18±5.6

SOFA(Sequential[Sepsis-related] OrganFailureAssessment)score day0(mean,SD)

7.7±3.8

SOFA(Sequential[Sepsis-related] OrganFailureAssessment)score day1(mean,SD)

7.3±4.1

SOFA(Sequential[Sepsis-related] OrganFailureAssessment)score day3(mean,SD)

6.4±4.3

IQR,interquartilerange;SD,standarddeviation.

Cytokinesconcentrationinsepticpatientsandhealthy

volunteers

IL-6,IL-8,IL-10,andG-CSFpresentedhigherplasma

concen-trations inseptic patients (D0) than inhealthy individuals

(p<0.001),whereas IL-12/23p40presentedhigher

concentra-tionsinhealthysubjectsthaninsepticpatients(p=0.003).

CytokineslevelsonD0andcorrelationswithorgan

dysfunctions

Individually,IL-6(D0)wascorrelatedwithSOFA(D1)

(Spear-man rho=0.36; p=0.004) and D3 (rho=0.33; p=0.008), but

not with SOFA on D0. IL-8 (D0) correlated with SOFA (D0)

(rho=0.27; p=0.007), SOFA (D1) (rho=0.36; p<0.001), and

SOFA (D3) (rho=0.34; p=0.001); IL-10 correlated with SOFA

(D0) (rho=0.3; p=0.01), SOFA (D1) (rho=0.4; p=0.001), and

SOFA D3 (rho=0.26; p=0.045); G-CSF (D0) correlated with

SOFA (D1) (rho=0.25; p=0.01) and SOFA (D3) (rho=0.24;

p=0.02); IL-17 (D0) correlated with SOFA (D0) (rho=0.23;

p=0.01),buttherewasnocorrelationwithSOFA(D1)orSOFA

(D3).

IL-8(D0)showedhigherconcentrationsinthosepatients

who developed renal dysfunction (patients with renal

dysfunction: median=82.3, IQR: 41.8–138.5, without

dys-function:median=45.6pg/mL,IQR:16.5–168.3pg/mL;p=0.04)

andcardiovasculardysfunction(patientswithcardiovascular

dysfunction:median=68.3pg/mL,IQR:28.7–68.3;without

dys-function: median=23.9pg/mL,IQR:5.6–68.4; p=0.01).G-CSF

(D0)concentrationswerehigherinpatientswith

cardiovas-culardysfunction(patientswithcardiovasculardysfunction:

median=30.4pg/mL, IQR: 11.2–73.7; without dysfunction:

median=11.0,IQR:3.3–41.2;p=0.03).

Factorandmixedmodelanalysis

Cytokineswereincludedinafactoranalysismodel

compris-ingthethreemeasurements.Thecumulativetotalvariance

Table2–Valuesofmediansandinterquartileranges(pg/mL,IQR,25–75percentiles)ofcytokinesinpatients(D0,D7,D14)

andhealthycontrols.

Cytokines D0 D7 D14 Healthy volunteers Median Interquartile

range

Median Interquartilerange Median Interquartile range Median Interquartile range Interleukin-2 0 0–0 0 0–0 0 0–0 0 0–0 Interleukin-6 157 44.9–436.5 43.1 14.9–84.1 71.9 29.4–341.9 2 0–4 Interleukin-8 57.6 24.1–134.5 27.1 12.7–46.5 34.2 21.8–99.4 8.1 1.1–14.4 Tumornecrosisfactor 0 0–0 0 0–1.98 0 0–1.1 0 0–0 Granulocyte-macrophage

colonystimulatingfactor

0 0–0 0 0–0 0 0–0 0 0–0 Interleukin-10 6.5 0–23.7 0 0–3.8 2.3 0–18.4 0 0–0 Interleukin-1B 0 0–0 0 0–0 0 0–0 0 0–0 Interleukin-17 0 0–10.9 0 0–4.1 0 0–0.2 1.2 0–7.7 Granulocyte colony-stimulatingfactor 27.4 10.4–69.2 6.6 2.7–13.6 15.6 3.2–25.3 10.5 3–16.5 Interleukin-21 0 0–244.5 0 0–281.8 0 0–0 0 0–524.1 Interleukin-12/Interleukin-23p40 27.4 0–61.4 25.2 0–62.28 41.2 0–90.9 67.7 14.4–108.7

Table3–Factoranalysisloadingsofthefourfactorsofthesolution,afterVarimaxrotation.Valuesinboldshowthemost

significantloadingforeachfactor.

Cytokines Factors

1 2 3 4

Interleukin-2 −0.06 0.74 −0.14 0.26 Interleukin-6 0.95 −0.04 −0.10 0.03 Interleukin-8 0.91 0.22 0.13 −0.01 Tumornecrosisfactor-␣ 0.25 0.62 0.03 −0.34 Granulocyte-macrophagecolonystimulatingfactor 0.00 0.84 0.11 −0.02 Interleukin-10 0.52 0.07 0.37 −0.18 Interleukin-1␤ 0.55 0.37 0.00 0.50 Interleukin-17 −0.07 −0.08 0.87 0.14 Granulocytecolony-stimulatingfactor 0.94 −0.09 −0.10 0.03 Interleukin-21 −0.02 −0.01 0.15 0.86 Interleukin-12/23p40 0.11 0.55 0.57 0.10 1.0 0.5 0.0 -0.5 -1.0 -1.0 -0.5 0.0 0.5 1.0 _1.0 0.5 0.0 -0.5 -1.0

Fig.1–Componentplotswithfactordiagramsfromfactor analysiswithVarimaxrotation,highlightingthefourlatent variables.

explainedbythemodelwas72.7%,withaKaiser–Meyer–Olkin

test of sampling adequacy of 0.70 and Bartlett’s

spheric-ity test withp<0.001. Themodel reduced the 11 variables

to four factors, or latent variables, presented in Table 3.

Factor 1 corresponds to the factorial loadings of IL-6, IL-8, IL-10, IL-1␤ and G-CSF. Factor 2represents IL-2, TNF-␣ and GM-CSF loadings; Factor 3, the loadings of IL-17 and IL-12/23p40. IL-21 was the cytokine that shared the least variance, being the highest loading in Factor 4.

Fig. 1 shows the plot of components after Varimax rota-tion.

ThevariableobtainedforFactor1showedpositive corre-lationwiththeSOFAscoreonD1(rho=0.25,p=0.04).Factor3 wasalsocorrelatedtoSOFAD1(rho=0.32,p=0.003).Factors2 and4showednocorrelationwiththeSOFAscore.

Table4–Accuracyofcytokinesconcentrationsoverthe

dosagestopredicthospitalmortality.

Cytokinedosage AUROC 95%IC Interleukin-17(D0) 0.71 0.43–0.99 Interleukin-1␤(D7) 0.63 0.34–0.91 Interleukin-6(D7) 0.77 0.47–1 Interleukin-10(D7) 0.64 0.36–0.93 Granulocytecolony-stimulatingfactor(D7) 0.61 0.32–0.9 Interleukin-12/23p40(D7) 0.75 0.49–1 Interleukin-1␤(D14) 0.63 0.34–0.91 Interleukin-6(D14) 0.71 0.44–0.98 Interleukin-8(D14) 0.66 0.37–0.94 Interleukin-10(D14) 0.66 0.37–0.94 Granulocytecolony-stimulatingfactor(D14) 0.62 0.33–0.9

Sepsisandin-hospitalmortalityandcytokines

concentration

IL-6 levels (D0) were higher in non-survivors

(median=785pg/mL, IQR: 114.1–2980.6) than in survivors

(median=133.7pg/mL, IQR: 40.2–293.3) (p=0.02). The same

was true for IL-8, with higher levels in non-survivors

(median=121.3pg/mL, IQR: 48.8–321) than in survivors

(median=50pg/mL,IQR:18.8–114.6)(p=0.01).Amongthe11

cytokines evaluated over the three time-points, IL1-␤, IL-6

and IL-10on D7 and IL-1␤,IL-6, IL-8, IL-10, G-CSF on D14

exhibited AUROC values above 0.6 for predicting hospital

mortality.Thecytokinesthatpresentedadequateaccuracyto

predicthospitalmortality(AUROC>0.7)wereIL-17onD0,IL-6

andIL-12/23p40onD7,andIL-6onD14(Table4).

Inthemixedmodelanalysis,IL-10showeddifferencesin the estimatedmeans forthe outcomein-hospitalmortality (8.5pg/mLforsurvivors,17.4pg/mLforthedeceased,p:0.014). IL-21 presented anestimated mean of195.8pg/mL for the survivors and 98.5pg/mL forthose who died in the hospi-tal(p:0.03).Fig.2showstheactualmeansofIL-10(top)and IL-21(down)observedinthemeasurementsofthethree time-points. There were no significant differences forthe other cytokines

100 100 1000 10 10 1 1 0 1 2 3 4 5 6 Days Survivors Deceased Survivors Deceased Days 7 8 9 10 11 12 13 14 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Fig.2–ObservedmeansforIL-10(top)andIL-21(down) (pg/mL,logscale)insurvivorsanddeceasedpatient (in-hospitalmortality).

Discussion

Sepsishasrecentlybeenredefinedasapotentiallylethalorgan dysfunctioncausedbya dysregulatedimmuneresponse to infection.12_{TheSOFAscore(SOFA≥2)wastakenasa}

param-etertodifferentiatepatientswithinfectionfromthosewith sepsisandhighriskofdeath.Inthepresentstudy,Th cell-relatedcytokineswereevaluatedinsepticpatientsoverthe first14daysofsepsis(D0,D7,D14)andcomparedtohealthy volunteers. We demonstrated that the cytokines could be groupedaccordingtotheirvariance,whichmeanstheir biolog-icalbehavioroverthethreemeasurementsduringtheperiod evaluated bythe study,and thatthey alsocorrelated with organdysfunctionspresentnotonlyatthetimeofadmission (D0)butalsoonthefollowingdays(SOFAD1andSOFAD3).

Oursepticpopulation presentedhigher plasma concen-tration of IL-6, IL-8, G-CSF, IL-10 on ICU admission (D0) than healthy volunteers. Sepsis immunopathogenesis has beencharacterizedbyaninitialinflammatoryresponse, trig-gered by the release of large amounts of inflammatory cytokines,followedbyananti-inflammatoryresponse,with secretionofimmunosuppressivemediators,monocyte deac-tivation, T lymphocyte anergy and apoptosis of immune cells.16,17 _{While inflammation leads to organ dysfunction}

andshock,theanti-inflammationincreasessusceptibilityto secondary infection. This sequential model is challenged, however, to the evidence of the simultaneous release of inflammatory and anti-inflammatory responses, in a con-tinuous process to reach balance and homeostasis.8 _IL-6,

IL-8,but alsoIL-10baselineconcentrationswere associated with sepsis intensity and organ dysfunction progression.2

Inanexperimentalmodel,Osuchowskietal.demonstrated

simultaneouslyincreasedplasmalevelsofpro-inflammatory (IL-6,TNF,IL-1␤,KC(CXCL1),MIP-2,MCP-1,andeotaxin)and anti-inflammatory(TNFsolublereceptors,IL-10,IL-1receptor antagonist)cytokinesinmicedyingfrom cecalligationand puncture(days1–5).18

Thefactoranalysisperformedinourstudydemonstrated the simultaneity of these processes, as it analyzed the time-sharedvariancebetweenthedifferentcytokines.Thus, observingthecorrelationsofthemeasurementinthree time-points, we realize that the most expressive inflammatory cytokines on the first day (IL-6, IL-8) are accompanied in time andvariance byIL-10production(Factor1).IL-10isa keyimmunoregulatorduringinfection,modulatingexcessive Th1 responseand CD8+ T cell responses, characterizedby overproductionofIFN-␥andTNF-␣.19_{Besidesthat,IL-10is}

pro-ducedbyTreglymphocytesregulatingtheTh17inflammatory response.Ontheotherhand,IL-6isanessentialcytokinefor thedifferentiationofTh17cells,andIL-8,althoughnota typi-calThelpercytokine,isalsoassociatedwiththeTh17andTh1 cellsdifferentiationinducingIL-17AandIFN-␥production.20

Theindividualanalysisbetweencytokinesandeachorgan dysfunctiondemonstratedthatIL-8plasmalevelsinD0 cor-related with the development of renal and cardiovascular dysfunctions. IL-8 is associated with increased production ofnitricoxideproductionandcalciumhomeostasis.21,22_Our

observationssupportsinvivowhathasbeenobservedby Hoff-mann et al. in cultures ofmouse cardiomyocytesexposed to blood ultrafiltrates from septic patients: the ultrafiltrate rich inIL-1␤, IL-8and C3a induced areduction inthe fre-quency of contraction of cardiomyocytes when electrically stimulated.23 _{An important mechanism related to sepsis}

hypotensionisthemicrovascularendothelialdamage,which may be causedbybacterial productsor directviral action, andaggravatedbyinflammatorymediatorssuchascytokines, adhesionmolecules, chemokines, andcomplement,among others.23_{IL-8bindstoIL-8RAandIL-8RBreceptorsonthe}

sur-faceofneutrophils,andtheseactivatedreceptorsinducethe expressionofchemotacticmediatorsthattriggerlocal inflam-mation.IL-8 wascorrelated withthedevelopmentofacute renalinjuryinpatientsundergoingcardiacby-pass.One pos-siblemechanismwouldbetheattractionofneutrophilstothe siteofischemicinjury.24_{Inamodelofischemicrenalinjury}

(mice),treatmentwithanantibodytotheIL-8ligandprotected againstischemia–reperfusioninjury.22

We observed that G-CSF accompanies the variances of IL-6, IL-8, IL-1␤ and IL-10.G-CSF is agrowth factor forthe productionofneutrophils,producedatthesiteofinfection. Neutrophilsareakeypartoftheinflammatoryresponse,even whenexcessiveandleadtotissuedamage.Theassociation betweenG-CSFandcardiovasculardysfunctionpresentedin ourstudymaybeexplainedbyneutrophilsactivation.In con-trast,G-CSFisalsoananti-inflammatorymodulator,reducing inflammationbyinhibitingtheproductionandactivityof IL-1␤,TNF-␣andIFN-␥.25,26

When analyzing the correlations betweenthe cytokines factors andSOFAscore,weobservedthat Factor1(IL-6, IL-8,IL-10,IL-1␤andG-CSF)correlatedwithorgandysfunction progression,basedonSOFAscorewithin24(SOFA0)and72h (SOFA3)ofsepsisdiagnosis.Thisresultissupportedby previ-ousreportshowingthattheconcentrationsofIL-6,IL-8,IL-10

andG-CSFinplasmawereassociatedwithorgandysfunction andthelevelsofIL-6,IL-8andG-CSFinthefirst24hwere pre-dictorsofworseneddysfunctiononthethirddaybasedinthe delta(D3-D0)SOFA.27

ThecovariancesbetweenIL-2andTNF-␣canbeexplained bythefact thatIL-2induces thegeneexpressionofTNF-␣ inmonocytesandmacrophages.28_{TNF-␣,inturn,isapotent}

inducerofGM-CSF,theothercytokinethatsharedvariancein Factor2ofourfactoranalysismodel.29

IL-17andIL-12/23p40presentedtemporalcorrelation (Fac-tor3).Wemeasuredthesubunitp40thatissharedbetween IL-12andIL-23,soitwasnotpossibletodiscriminatethe indi-vidualcontributionofeachofthesecytokinesinourstudy. IL-12inducesthe differentiationofnaïveCD4+cellsinTh1 thusproducinginterferon(IFN-␥).30_{IL-23inducesIL-17once}

IL-23receptorsareexpressedbyvariouslymphoidcellsofthe innateandadaptivesystems,including␥␦T,iNKT,NKp46, lym-phoidtissueinducercells,andTh17,withpromptsecretion ofIL-17inresponsetotheantigenicstimulus.31_IL-23isnot

relatedtothedifferentiationofTh17fromnaivecells,butis relevantfortheclonalexpansionoftheselymphocytesand productionofIL-17.32

Inthe presentstudy,IL-12/23p40 showedhigher plasma concentrationsinhealthyvolunteersthaninsepticpatients (D0).Thisfindingisinlinewithaprevious reportfrom our groupthat evaluated neutropenic patientswith sepsisand comparedwithanon-septicneutropenicgroupshowingthat theIL-23plasmaconcentrationwassignificantlylowerinthe septicpatients,suggestingarepressionofthisaxisonsepsis.33

Reduced levels ofIL-23 mRNAwere previously reported in septicpatientscomparedtohealthyvolunteers,further sup-portingthedetectionoflowerproteinconcentrationsinour cohort.34

TheFactor3,composedbytheTh17cytokinesIL-17and IL-12/23p40,correlatedwithSOFAonD0andwithin24h(D1). Thesecytokinesareeminentlyinflammatoryandare associ-atedwithorgandysfunctioninsepsis.Guoetal.foundthat septic patients with multiple organ dysfunction syndrome (MODS)presentedhigherlevelsofTh17cytokines,IL-17and IL-23,comparedtono-MODSgroup.35_{Sepsisisassociatedwith}

anincreasedproportionofTh17lymphocytesandthemore exacerbatedistheinflammatoryresponse,themoreadverse theoutcomeandorgandysfunctionsbytheSOFAscore.7

IL-21(Factor4),whichissynthesizedbyseveralCD4+and some CD8+ T cell lines under appropriate conditions did not significantlyshared variance with the other evaluated cytokines.IL-21isessentialfortheThcellsdifferentiationand amplificationoftheTh17response.36_{Thiscytokineisstrongly}

associatedwithinflammationandexertsinhibitoryeffectson TregandBcellsreducingIL-10production.37_Despitethe

asso-ciationwithTh17inflammatoryresponse,IL-21didnotvary accordingtootherth17cytokinesevaluatedandwasgrouped inafactordifferentfromIL-17andIL-12/23p40(Factor3).

Interestingly,theFactor4,composedonlybyIL-21,showed anegativecorrelationwithSOFAscore,whichsuggeststhatit isaprotectivefactorfororgandysfunctioninsepsis. Paidi-pally et al. demonstrated, in an experimentalmodel with

Mycobacterium tuberculosisstimulated CD4+ and NK T cells, thatInterleukin-21activatedNKcells,andalsoenhancesthe immune response by augmenting IL-1␤, IL-18, and MIP-1␤

production,andreducing IL-10productionbymonocytesin responsetoanintracellularpathogen.38_{IL-21isrelatedtothe}

moreeffectivecombatofthepathogen andperhapsthisis associatedto abetteroutcome insepsis, withless organic dysfunctionandlowermortality.

Finally,inthemixedmodelanalysis,weobservedhigher estimated concentrations of IL-10 in patients who died in hospitalandhigherIL-21levelsinthosewhosurvived,thus reinforcing their role in immunomodulation in sepsis. As pointedoutabove,IL-10candirectlyregulatetheinnateand adaptiveresponses,limitingtheactivationofTcellsand dif-ferentiation inthelymphnodes,inadditiontosuppressing thepro-inflammatoryactivityintissues,leadingtodifficulty in controlling pathogens.39,19 _{Higher mean levels in}

non-survivorsseemtoindicatethat,inthisgroup,therewasan imbalance infavorofimmunosuppression,withimpacton outcome. In contrast, survivors presented higher levels of IL-21,atrulypleiotropiccytokinewithimmunestimulatory propertiesonbothinnateandadaptiveimmunitywould sug-gestabalancepreservingimmunefunction.40,41

Ourstudy hasstrengths andlimitations.First, thesmall samplesizelimitstheextenttowhichourfindingsmaybe generalized toothergroups ofpatients. Second,onlythree time-points were used for cytokines measurements. Some cytokines,whosepeakofactionoccurswithinafewhours, such as TNF-␣ and IL-1␤, may have had their evaluation impaired,explainingthesimilarityofthevaluesobservedin healthycontrolsandpatients.

Conclusions

PlasmalevelsofcytokinesencompassingTh1,Th2andTh17 profiles showed that inflammatory and anti-inflammatory cytokines shared co-variance in septic patients and were relatedwithorgandysfunctionsandin-hospitalmortality.

Funding

ThisresearchwassupportedbyFAPESP(Fundac¸ãodeAmparo àPesquisadoEstadodeSãoPaulo–Grant2011/20401-4).

Conflicts

of

interest

Theauthorsdeclarenoconflictsofinterest.

r

e

f

e

r

e

n

c

e

s

1.MachadoFR,CavalcantiAB,BozzaFA,etal.Theepidemiology ofsepsisinBrazilianintensivecareunits(theSepsis

PREvalenceAssessmentDatabase,SPREAD):anobservational study.LancetInfectDis.2017:1180–9.

2.ChoustermanBG,SwirskiFK,WeberGF.Cytokinestormand sepsisdiseasepathogenesis.SeminImmunopathol. 2017;39:517–28.

3.MachadoFR,SanchesLC,LucianoCésarPontesAzevedoLC, etal.Associationbetweenorgandysfunctionandcytokine concentrationsduringtheearlyphasesofsepticshock.Rev BrasTerIntensiva.2011;23:426–33.

4. SalomaoR,BrunialtiMK,RapozoMM,Baggio-ZappiaGL, GalanosC,FreudenbergM.Bacterialsensing,cellsignaling, andmodulationoftheimmuneresponseduringsepsis. Shock.2012;38:227–42.

5. TisoncikJR,KorthMJ,SimmonsCP,FarrarJ,MartinTR,Katze MG.Intotheeyeofthecytokinestorm.MicrobiolMolBiolRev. 2012;76:16–32.

6. ChaudhryH,ZhouJ,ZhongY,AliMM,McGuireF,Nagarkatti PS.Roleofcytokinesasadouble-edgedswordinsepsis.In Vivo.2013;27:669–84.

7. BrunialtiMK,SantosMC,RigatoO,MachadoFR,SilvaE, SalomaoR.IncreasedpercentagesofThelpercellsproducing IL-17andmonocytesexpressingmarkersofalternative activationinpatientswithsepsis.PLoSONE.2012;7:e37393.

8. GuptaDL,BhoiS,MohanT,GalwnkarS,RaoDN.Coexistence ofTh1/Th2andTh17/Tregimbalancesinpatientswithpost traumaticsepsis.Cytokine.2016;88:214–21.

9. YuZX,JiMS,YanJ,etal.TheratioofTh17/Tregcellsasarisk indicatorinearlyacuterespiratorydistresssyndrome.Crit Care.2015;19:82.

10.MachadoFR,SalomãoR,RigatoO,etal.Laterecognitionand illnessseverityaredeterminantsofearlydeathinsevere septicpatients.Clinics.2013;68:586–91.

11.BoneRC,BalkRA,CerraFB,etal.Definitionsforsepsisand organfailureandguidelinesfortheuseofinnovative therapiesinsepsisTheACCP/SCCMConsensusConference Committee.AmericanCollegeofChestPhysicians/Societyof CriticalCareMedicine.Chest.1992;101:1644–55.

12.SingerM,DeutschmanCS,SeymourCW,etal.Thethird internationalconsensusdefinitionsforsepsisandseptic shock(Sepsis-3).JAMA.2016;315:801–10.

13.VincentJL,MorenoR,TakalaJ,etal.TheSOFA(Sepsis-related OrganFailureAssessment)scoretodescribeorgan

dysfunction/failure.OnbehalfoftheWorkingGroupon Sepsis-RelatedProblemsoftheEuropeanSocietyofIntensive CareMedicine.IntensiveCareMed.1996;22:707–10.

14.FrielCM.Notesonfactoranalysis.Availableat:http://www. bama.ua.edu/∼jcsenkbeil/gy523/Factor%20Analysis.pdf. 15.GarsonD.Fundamentalsofhierarchicallinearandmultilevel

modeling.Availableat:http://www.sagepub.com/sites/ default/files/upm-binaries/47528ch1.pdf[accessed28.08.15]. 16.RittirschD,FlierlMA,WardPA.Harmfulmolecular

mechanismsinsepsis.NatRevImmunol.2008;8:776–87.

17.NovotnyAR,ReimD,AssfalgV,AltmayrF,FriessHM, EmmanuelK.Mixedantagonistresponseandsepsis severity-dependentdysbalanceofpro-and

anti-inflammatoryresponsesattheonsetofpostoperative sepsis.Immunobiology.2012;217:616–21.

18.OsuchowskiMF,WelchK,SiddiquiJ,RemickDG.Circulating cytokine/inhibitorprofilesreshapetheunderstandingofthe SIRS/CARScontinuuminsepsisandpredictmortality.J Immunol.2006;177:1967–74.

19.CouperKN,BlountDG,RileyEM.IL-10:themasterregulatorof immunitytoinfection.JImmunol.2008;180:5771–7.

20.GaschM,GorollT,BauerM,etal.GenerationofIL-8andIL-9 producingCD4+TcellsisaffectedbyTh17polarizing conditionsandAHRligands.MediatInflamm. 2014;2014:182549.

21.CannonJG,TompkinsRG,GelfandJA,etal.Circulating interleukin-1andtumornecrosisfactorinsepticshockand experimentalendotoxinfever.JInfectDis.1990;161:79–84.

22.XingD,LiP,GongK,etal.Endothelialcellsoverexpressing interleukin-8(IL8)receptorsreduceinflammatoryand neointimalresponsestoarterialinjury.Circulation. 2012;125:1533–41.

23.HoffmannJN,WerdanK,HartlWH,JochumM,FaistE,Inthorn D.Hemofiltratefrompatientswithseveresepsisand depressedleftventricularcontractilitycontainscardiotoxic compounds.Shock.1999;12:174–80.

24.LiangosO,KolyadaA,TighiouartH,PerianayagamMC,Wald R,JaberBL.Interleukin-8andacutekidneyinjuryfollowing cardiopulmonarybypass:aprospectivecohortstudy. NephronClinPract.2009;113:c148–54.

25.HartungT.Anti-inflammatoryeffectsofgranulocyte colony-stimulatingfactor.CurrOpinHematol.1998;5:221–5.

26.TanakaH,IshikawaK,NishinoM,ShimazuT,YoshiokaT. Changesingranulocytecolony-stimulatingfactor

concentrationinpatientswithtraumaandsepsis.JTrauma. 1996;40:718–25.

27.BrownKA,BrainSD,PearsonJD,EdgeworthJD.Neutrophilsin developmentofmultipleorganfailureinsepsis.Lancet. 2006;368.

28.StrieterRM,RemickDG,LynchJP,SpenglerRN,KunkelSL. Interleukin-2-inducedtumornecrosisfactor-alpha

(TNF-alpha)geneexpressioninhumanalveolarmacrophages andbloodmonocytes.AmRevRespirDis.1989;139:

335–42.

29.ShiY,LiuCH,RobertsAI,etal.Granulocyte-macrophage colony-stimulatingfactor(GM-CSF)andT-cellresponses: whatwedoanddon’tknow.CellRes.2006;16:

126–33.

30.LupardusPJ,GarciaKC.Thestructureofinterleukin-23 revealsthemolecularbasisofp40subunitsharingwith interleukin-12.JMolBiol.2008;382:931–41.

31.Zú ˜nigaLA,JainR,HainesC,etal.Th17celldevelopment: fromthecradletothegrave.ImmunolRev.2013;252: 78–88.

32.LangrishCL,ChenY,BlumenscheinWM,etal.IL-23drivesa pathogenicTcellpopulationthatinducesautoimmune inflammation.JExpMed.2005;201:233–40.

33.deAraujoOR,SalomãoR,BrunialtiMKC,etal.Cytokine kineticsinfebrileneutropenicchildren:insightsonthe usefulnessassepsisBiomarkers,influenceoffilgrastim,and behavioroftheIL-23/IL-17pathway.MediatInflamm. 2017;2017:8291316.

34.O’DwyerMJ,MankanAK,WhiteM,etal.Thehumanresponse toinfectionisassociatedwithdistinctpatternsofinterleukin 23andinterleukin27expression.IntensiveCareMed. 2008;34:683–91.

35.GuoJ,TaoW,TangD,ZhangJ.Th17/regulatoryTcell

imbalanceinsepsispatientswithmultipleorgandysfunction syndrome:attenuatedbyhigh-volumehemofiltration.IntJ ArtifOrgans.2017;40:607–14.

36.KornT,BettelliE,GaoW,etal.IL-21initiatesanalternative pathwaytoinduceproinflammatoryTH17cells.Nature. 2007;448:484–7.

37.SpolskiR,KimHP,ZhuW,LevyDE,LeonardWJ.2009.IL-21 mediatessuppressiveeffectsviaitsinductionofIL-10.J Immunol.2009;182:2859–67.

38.PaidipallyP,TripathiD,VanA,etal.Interleukin-21regulates naturalkillercellresponsesduringMycobacteriumtuberculosis

infection.JInfectDis.2018;217:1323–33.

39.TrinchieriG.Interleukin-10productionbyeffectorTcells:Th1 cellsshowself-control.JExpMed.2007;204:239–43.

40.CroceM,RigoV,FerriniS.IL-21:apleiotropiccytokinewith potentialapplicationsinoncology.JImmunolRes. 2015;2015:6965–78.

41.SpolskiR,LeonardWJ.Interleukin-21:adouble-edgedsword withtherapeuticpotential.NatRevDrugDiscov.