Determination of TNF-alpha levels in dengue virus infected patients(1).pdf

Journal of Virological Methods 90(2000)5157Determination of tumor necrosis factor-alpha levels indengue virus infected patients by sensitive biotin-streptavidinenzyme-linked immunosorbent assayL.Kittigula,*,W.Temproma,D.Sujiraratb,C.KittigulcaDepartment of Microbiology,Faculty of Public Health,Mahidol Uni6ersity,Bangkok,ThailandbDepartment of Epidemiology,Faculty of Public Health,Mahidol Uni6ersity,Bangkok,ThailandcDepartment of Microbiology,Faculty of Science,Kasetsart Uni6ersity,Bangkok,ThailandReceived 31 March 2000;received in revised form 16 June 2000;accepted 16 June 2000AbstractA modified sandwich enzyme-linked immunosorbent assay using biotin-streptavidin system(BS-ELISA)wasdeveloped to determine levels of tumor necrosis factor-alpha(TNF-a)in serum samples of children infected withdengue virus(n=99)and healthy controls(n=41).The minimum detectable concentration of TNF-a by theBS-ELISA was 3.3 pg/ml.The mean TNF-a level was highest in those patients with dengue shock syndrome(DSS)or dengue hemorrhagic fever(DHF)grade III(37.44942.0 pg/ml).Lower levels were found in DHF grade I(28.44942.7 pg/ml),DHF grade II(24.21925.4 pg/ml)and dengue fever(DF)(14.10924.0 pg/ml).TNF-a in thesera of DF and DHF patients could be detected on days 26 after the onset of fever,the high level occuring on day5.TNF-a was detected in 41.4%(24.01935.2 pg/ml)of dengue virus infected patients and 7.3%(4.2915.6 pg/ml)of control subjects.The sera of patients contained significantly higher levels of TNF-a than the sera of controls,P-valueB0.001.DHF patients had significantly higher levels of TNF-a than DF patients(P-value=0.020)but nodifference in the TNF-a levels from sera of DHF grades IIII patients was observed(P-value=0.295).The resultsindicate that the BS-ELISA is a very sensitive method for determining TNF-a in serum samples of DF and DHFpatients.The TNF-a levels might be associated with dengue virus infection and related to disease severity of DHF.2000 Published by Elsevier Science B.V.Keywords:Tumor necrosis factor-alpha;Dengue hemorrhagic fever;Dengue fever;Biotin-streptavidin ELISA virus infection is a common diseasewhich occurrs worldwide with an increasing inci-dence in tropical and subtropical countries.Itproduces a spectrum of clinical illness rangingfrom a mild disease as dengue fever to the more*Corresponding author.Tel.:+662-2-2455525;fax:+662-2-2458351.E-mail address:phlktmahidol.ac.th(L.Kittigul).0166-0934/00/$-see front matter 2000 Published by Elsevier Science B.V.PII:S0166-0934(00)00215-9L.Kittigul et al./Journal of Virological Methods90(2000)515752serious dengue hemorrhagic fever(DHF)anddengue shock syndrome(DSS)(Kautner et al.,1997).Dengue virus antigens were detected inmononuclear cells(Kittigul et al.,1997)and it wassuggested that mononuclear phagocytes are thetargets of dengue virus infection and contribute tohomeostatic defects in DHF and DSS(Halstead,1989).The pathogenesis of DHF is not under-stood fully.However,cytokines have been sug-gestedasresponsibleforhemorrhagicmanifestations and plasma leakage that may leadto shock in dengue virus infected patients.Thepathophysiology of DHF and DSS patients isprobably associated with the high levels of cytoki-nes produced by monocytes and macrophages(Hober et al.,1993,1996,1998;Kurane et al.,1993).Exposure of monocytes and macrophagesto dengue virus in vitro enhanced production ofTNF-a(Lee et al.,1996).TNF-a in culture fluidsfromdenguevirus-infectedperipheralbloodmonocytes activated endothelial cells which mightbe a target in the pathogenesis of DHF(Andersonet al.,1997).However,similar levels of TNF-awere observed in patients with dengue fever andDHF(Laur et al.,1998).A study of TNF-a levelsin relation to immunopathologic mechanisms isrequired to support the severity of DHF but thecommercial enzyme immunoassay for this cy-tokine is expensive in developing countries.The present study describes the development ofa biotin-streptavidin enzyme-linked immunosor-bent assay(BS-ELISA)for sensitive determina-tion of TNF-a levels in sera of Thai patients withdengue virus infection and healthy subjects.Theassociation between TNF-a levels and the severityof dengue diseases was examined.2.Materials and methods2.1.Study populationBlood specimens were collected from 99 pa-tients admitted to hospitals and from 41 healthycontrol children under 15 years of age after in-formed consent was obtained,between June andNovember,1997.The patients were diagnosed ashaving dengue virus infection by clinical findingsand laboratory confirmation with fourfold rise inhemagglutination inhibition(HI)antibody titersin paired sera(Clarke and Casals,1958).Thesequence of dengue infection was categorized andthe disease severity was graded according toWorld Health Organization(WHO)Guidelines(World Health Organization,1997).Of the patients enrolled in the study,15%hadprimary dengue virus infection and 85%had sec-ondary antibody response.There were 37 patientsclassified as DF,34 as DHF grade I,12 as gradeII,and 16 as grade III or DSS.2.2.Serum samplesA blood sample(35 ml)was obtained fromeach patient by venepuncture within a day ofhospital admission.It was allowed to clot at roomtemperature and the serum was separated andthen stored frozen at 70C until use.Details ofthe patientss case histories,physical examinationsand laboratory findings were collected.2.3.Assay for TNF-aTNF-a levels in human sera were measuredusing a BS-ELISA.Assay conditions were opti-mised by checkerboard titration of recombinanthuman TNF-a standard(rh TNF-a).The BS-ELISA method was modified from the proceduredescribed essentially by Kittigul et al.(1998)andPisa et al.(1990).Briefly,a microtiter plate(NuncInter Med,Roskilde,Denmark)was coated at24C overnight with 100 ml of mouse mono-clonal antibody anti-TNF-a(Genzyme Diagnos-tics,Cambridge,MA)at a concentration of 5mg/ml in 0.05 M carbonate bicarbonate buffer,pH9.6.After washing the plate with 0.15 M phos-phate-buffered saline(PBS),nonspecific bindingsites were blocked by adding 2%bovine serumalbumin(BSA)PBS(250 ml)and incubated at37C for 1 h.Then,the plate was washed six timeswith PBS-Tween 20(PBST).Following the lastwash,rh TNF-a or serum sample diluted 1:10 in1%BSA-PBST was added(100 ml/well)and theplate was incubated at 37C for 1 h.After wash-ing,100 ml of chicken IgY anti-TNF-a polyclonalantibody(2mg/ml)(PromegaCorporation,L.Kittigul et al./Journal of Virological Methods90(2000)515753Madison,WI)was added to each well,incubatedat 37C for 1 h and washed again.Biotin conju-gatedanti-chickenIg(PromegaCorporation,Madison,WI)at a concentration of 0.5 mg/ml(100 ml/well)was added to each well.The platewas incubated at 37C for 1 h and washed again.Then,100 ml of streptavidin-horseradish perox-idase(Vector laboratory,Burlingame,CA)diluted1:3000 was added to each well and incubated at37C for 1 h.Finally,the plate was washed andthe color was developed by adding the substratebuffer solution containing tetramethylbenzidine(TMB)and hydrogen peroxide(100 ml/well).Thecolor reaction was stopped after 10 min incuba-tion by the addition of 0.46 M sulfuric acid(50ml/well).Optical densities(ODs)were read at 450nm in an ELISA plate reader(Biotek,Winooski,WT).Minimaldetectableconcentrationwasdefined by the standard deviation of dose mea-surement at zero dose or background(Larsson etal.,1997).The levels of TNF-a in serum sampleswere interpolated from the rh TNF-a standardcalibration curve.2.4.Statistical analysisThe difference of TNF-a levels in dengue virusinfected patients and healthy controls was evalu-ated by the MannWhitney U-test.The KruskalWallis test was used when more than two groupswere compared.Differences giving P-valueB0.05were considered significant.3.Results3.1.Optimization of the assay conditionsThe strongest signal of ELISA-OD in determi-nation of rh TNF-a was achieved with the use of5 mg/ml monoclonal mouse anti-human TNF-aand 2 mg/ml anti-human TNF-a polyclonal anti-body.Biotin conjugated anti-chicken IgY at theconcentrationof0.5mg/mlandstreptavidin-horseradish peroxidase diluted 1:3000 gave a bet-tersignal-tobackgroundratiothanotherconcentrations tested.The pooled negative con-trol sera diluted 1:10 gave the lowest backgroundwith the highest difference in OD value of rhTNF-a and was thus chosen to be the standarddiluent in the BS-ELISA.Standard curves of rhTNF-a were plotted with known amounts of rhTNF-a versus their OD values.A linear fit wasobtained at a concentration range up to 64 pg/mlwith r2of 0.9999 as shown in Fig.1.3.2.Sensiti6ity of BS-ELISABackground absorbance was measured usingpooled negative sera diluted 1:10 and gave valuesranging from 0.385 to 0.527 with the average of0.4790.0655(mean9S.D.)in five experiments.The absorbance values of background and threelowest standard concentrations were plotted andare shown in Fig.2.By calculation,the sensitivityof BS-ELISA for detecting TNF-a in humanserum was approximately 3.3 pg/ml.3.3.TNF-a le6els in the sera of dengue fe6er ordengue hemorrhagic fe6er patientsThe serum samples from patients with denguefever or DHF were examined for levels of TNF-a.It was found that TNF-a was detected in 29.7%(11 of 37)of DF(mean9S.D.;14.10924.0 pg/ml)and 48.3%(30 of 62)of DHF patients(29.95939.5 pg/ml).Among 62 DHF patients,the TNF-a level was highest in DHF grade III(37.44942.0 pg/ml)followed by DHF grade I(28.44942.7 pg/ml)and DHF grade II(24.21925.4 pg/ml),as shown in Table 1.Fig.1.Standard curve of recombinant human TNF-a deter-mined by a BS-ELISA with linear fit based on the equation,Y=Bx+A.L.Kittigul et al./Journal of Virological Methods90(2000)515754Fig.2.Determination of the detection limit of the BS-ELISA.Zero dose(background)and absorbance values of the 3 loweststandard concentrations of recombinant human TNF-a werein a linear regression.The detection limit was given by theS.D.of background divided by the slope of the regression line.Fig.3.TNF-a levels in the sera of DF patients on day afterthe onset of fever.The symbol represents the TNF-a valueobtained from each patient on the indicated day.Blood sam-ples were obtained within 24 h of admission,allowed to clotand the sera were separated.The day post onset of fever in thepatients was derived from interviews.observed on the day the patient went into shock(126 pg/ml).The average TNF-a levels werehighest in the sera collected one day before shock(44.75951.8 pg/ml),followed by those collectedon the day of shock(38.5945.4 pg/ml),and 2days before shock(24.30924.0 pg/ml).3.5.Comparison of the TNF-a le6els among thestudied groupsTNF-a levels in the sera of dengue virus in-fected patients(41.4%)were compared with thoseof healthy subjects(7.3%)and are shown in Table3.The presence of TNF-a in patients with denguevirus infection(24.01935.2 pg/ml)were signifi-Based on the time after the onset of fever inindividuals,TNF-a could be detected in patientswith dengue fever whose sera were collected ondays 26,the high value occuring on day 5(Fig.3).Similarly,TNF-a levels were found in DHFsera collected on days 25 with the high valuealso on day 5 after onset of fever(Fig.4).3.4.TNF-a le6els during shock syndromeTable 2 shows the individual levels of TNF-a inserum samples from DSS patients(nine of 16cases).A very high TNF-a concentration wasTable 1Levels of TNF-a in sera from dengue virus infected patientsTNF-a(pg/ml)No.of patientsClinical syndromeRangeMean9S.D.Positive(%)Total3711(29.7)14.10924.0DF069DHF0155DHF I3414(41.2)28.44942.707724.21925.47(58.3)DHF II120126DHF III169(56.3)37.44942.030(48.3)620155Total DHF29.95939.524.01935.241(41.4%)99Total(DF+DHF)0155L.Kittigul et al./Journal of Virological Methods90(2000)515755Fig.4.TNF-a levels in the sera of DHF patients on day afterthe onset of fever.The symbol represents the TNF-a valueobtained from individual patients.Blood samples were ob-tained within 24 h of admission,allowed to clot and the serawere separated.The day post onset of fever in the patients wasderived from interviews.Table 3Comparison of TNF-a levels between dengue virus infectedpatients with disease severity and healthy subjectsGroupTNF-a levels(pg/ml)aP-valueB0.001bDengue patients vs24.01935.2 vs.healthy controls4.2915.6DHF vs DF29.95939.5 vs.0.020bpatients14.10924.0DHF grade I vs II0.295c28.44942.7 vs.vs III24.21925.4 vs37.44942.0aValues expressed as mean9S.D.bAnalysed by MannWhitney U-test.cAnalysed by KruskalWallis test.P-value of 0.242,sex:P-value of 0.731.The TNF-a levels in dengue virus infected patients weresignificantly different from healthy controls(P-value,0.001).If the time after the onset of feverwas controlled,the TNF-a levels in the serumsamples of DHF and DF patients were signifi-cantly different(P-value,0.030).4.DiscussionIn this study,TNF-a levels were determined inthe sera of patients with dengue fever and denguehemorrhagic fever by a highly sensitive BS-ELISA.Mouse monoclonal anti-human TNF-aantibody,specifically reactive with human TNF-aand lacking cross-reactivity was used as the cap-ture reagent in the assay.Anti-human TNF-acantly higher than those in healthy subjects(4.2915.6 pg/ml),P-valueB0.001.The sera of DHF patients had higher TNF-alevels than that of dengue fever patients(29.95939.5 versus 14.10924.0 pg/ml).This was statisti-cally significant,P-value=0.020.Although thehighest value of TNF-a was found in the sera ofDHF grade III patients,there was no significantdifference in DHF grade IIII(P-value=0.295).Multivariable technique was used to determinehow well several independent variables,both sep-arately and together,explained the variation inTNF-a levels.By multiple classification analysis(MCA),there was no effect caused by the age andsex of the studied subjects on TNF-a levels;age:Table 2Levels of TNF-a in serum samples from patients with shock symdromeDay of collection(before shock)aTNF-a level(pg/ml)Patient numberTime-post onset of fever1253222248134841449550445064757051260425803769aThe sera were collected on the day of the patients admission.L.Kittigul et al./Journal of Virological Methods90(2000)515756polyclonal antibody was produced in immunizedchicken and purified from egg yolk by resin chro-matography.This antibody specifically reacts withhuman TNF-a,has no cross reactivity with mam-malian IgG(Ambrosius and Hadge,1987),anddoes not bind bacterial or mammalian Fc receptor(Larsson and Sjoguist,1990).To increase thesensitivity,biotin-conjugate anti-chicken IgY andstreptavidin-horseradish peroxidase were includedin the assay.Excellent linearity of the standardcurve was obtained.The BS-ELISA method had ahigh sensitivity and specificity.It was simple toperform and was quantitative,and the cost of themethod was less than that of the commercial kit.Thus,the BS-ELISA might be useful for facilitat-ing the expanded application of clinical TNF-adata.TNF-a could affect both endothelial cells(An-derson et al.,1997)and participate in some mani-festationsofDHFandDSSsuchasthrombocytopenia,hemorrhage and extravasationof fluid from the vascular compartment progress-ing to hypovol