An evaluation of dipstick-dot immunoassay in the detection of antibodies to HIV1 and 2 in Zimbabwe

Tropical Medicine and International Health

volume 2 no. 1 pp 83–88 january 1997

An evaluation of dipstick-dot immunoassay in thedetection of antibodies to HIV-1 and 2 in ZimbabweC. Sunanda Ray1, Peter R. Mason1,2, Heather Smith1, Luanne Rogers1, Ocean Tobaiwa1,2 andDavid A. Katzenstein1

1 Zimbabwe AIDS Prevention Project, University of Zimbabwe, Harare, Zimbabwe2 Biomedical Research and Training Institute, Harare, Zimbabwe

Summary There is a need, in many developing countries, for simple and inexpensive HIV serologytests for use at the district level of health care. The Programme for Appropriate Technologyin Health has developed a simple dipstick ELISA to detect antibodies to HIV-1 and 2, at acost considerably lower than current ELISAs, which requires no specialized washing orreading equipment. In order to evaluate this dipstick under local conditions we used a panelof 546 sera selected from frozen stocks maintained by the Zimbabwe AIDS PreventionProject in Harare, Zimbabwe. Prior to storage, the sera had been tested by Abbottrecombinant peptide HIV-1 and 2 ELISA and Enzygnost synthetic peptide HIV-1 and 2ELISA. The panel included sera that were positive by both (including symptomatics andasymptomatics), negative by both, and sera showing discrepant test results. The panel wasnot representative of a ‘normal’ batch of sera in Zimbabwe, and in particular included anabnormally high number of sera showing discrepant results. Thawed sera were retestedusing the Abbott recombinant peptide HIV-1 and 2 ELISA and concurrently with thesynthetic peptide ICL Dipstick ELISA. Both the sensitivity and specificity of the ICL Dipstickexceeded 99% when using sera that were positive or negative in all 3 plate ELISAs as thegold standard. When using sera that gave discrepant results between the two pre-storageELISAs, most results with the ICL Dipstick concurred with findings from other test systems,including Western blot and p24 antigen detection. Considering the accuracy, low cost andease of operation of the ICL Dipstick ELISA, this test can be recommended for use for therapid detection of antibodies to HIV at district level in developing countries.

keywords HIV, immunoassay, ELISA, antibodies

correspondence Professor P. R. Mason, c/o ZAPP, 114 Baker Ave/5th Street, Harare,Zimbabwe, Fax: 263-4-739406, e-mail: [email protected]

Introduction

The prevalence of infection with the humanimmunodeficiency virus (HIV) in southern Africa hasrisen rapidly over the last decade. In Zimbabwe ithas been estimated, on the basis of antenatal sentinelsurveillance and ad hoc studies, that 15–35% ofthe sexually active adult population is infected

(Mahomed et al. 1991; Latif et al. 1995). TheZimbabwe AIDS Prevention Project has demon-strated a seroprevalence of 19% in a cohort of 2500adult, predominantly male, factory workers inHarare, and the incidence of infection in this groupis 3% per year (Mbizvo et al. 1996).There are several key roles that serological testing

can play in the prevention and control of HIV.

,C 1997 Blackwell Science Ltd83

These include the monitoring of blood for trans-fusion and maintenance of a pool of negativedonors; the detection of HIV infection in patientspresenting with clinical indications of infection; sen-tinel surveillance programmes and research into epi-demiological, clinical, virological or other relatedstudies. More recently, programmes are being devel-oped for voluntary testing of people who think theymay have been exposed to infection but as yet showno signs or symptoms—the ‘worried well’— and foridentification of people infected with HIV for whomprophylactic therapy for tuberculosis or other infec-tious diseases can be recommended. A major con-straint for such programmes is the cost of serologicaltesting for HIV, which must include not just the costof reagents but also the cost of trained personneland capital equipment.The seroprevalences in the groups mentioned

above may differ considerably, and the requirementsof testing may vary. In the case of blood donors forexample, preliminary screening based on question-naire data can identify donors who are at potentialrisk of infection so that they can be excluded fromthe donor pool. The seroprevalence in the donorpool may then be much lower than in the generalpopulation. The purpose of testing in this group is toprevent transfusion of infected blood, and so anyerror must be on the side of sensitivity (to minimizefalse negatives) rather than specificity. Any resultingfalse positives would be acceptable from the trans-fusion point of view since the main consequence isdeferral from donation. Amongst the ‘worried well’,however, false positive reactions are not acceptablebecause of the anxiety and difficulties that may becreated by the result. For patients who have signsand symptoms consistent with HIV infection, thetrue seroprevalence of HIV will be higher than forthe general population, and the purpose of the test isto confirm a clinical diagnosis. Lower sensitivitiesmay be acceptable, since the positive predictive valuewill still be high in this group.There are a large number of test kits available for

the detection of antibodies to HIV. They includeagglutination, indirect immunofluorescence andenzyme linked immunosorbent assay (ELISA). Manyof the latter now use synthetic or recombinantpeptides that incorporate antigens of both HIV-1and HIV-2 and provide high sensitivity and specifi-

city. They are, however, expensive for widespreaduse in developing countries, costing approximatelyUS$2.00–2.50 per test, and they require the servicesof trained technologists and specialized equipmentfor reading the results. Such facilities are usuallyavailable only at main centres, whereas there is greatdemand at district level for less expensive and moreaccessible testing with a fast turn round time forresults. In Zimbabwe, the standard procedures fordetection of HIV infection require two ELISAs, usingdifferent antigen sets, and confirmation by Westernblotting where there is a consistent discrepancybetween the two. The WHO quotes a cost of US$40per Western blot, so that the true cost to pro-grammes of testing, including costs of test combina-tions, staff, equipment and other laboratory costs,can be prohibitive.Recent recommendations by WHO (1992) suggest

that countries consider testing strategies which useELISA and/or rapid, simple assays in place of theELISA/Western blot combination. Three strategiesare designed to maximize accuracy while minimizingcost. In Strategy I, for transfusion/donor safety orsurveillance in populations where the prevalenceexceeds 10%, sera are tested with a single ELISA orrapid/simple assay. Reactive sera are considered anti-body positive and non-reactive sera are consideredantibody negative. Strategy II, for surveillance wherethe prevalence is less than 10%, or for confirmationof a clinical diagnosis of HIV infection, involvesinitial testing with an ELISA or rapid/simple assay,then retesting the reactive sera with a different anti-gen preparation, using a similar assay, for confirma-tion. Strategy III for asymptomatic persons in apopulation with a prevalence under 10%, involvestesting as in Strategy II with two different antigenpreparations, but testing reactive sera with a thirdantigen preparation for confirmation of positivity. Inaddition, when a seropositive person is identifiedthrough testing, a second blood sample should bedrawn and tested to eliminate clerical or laboratoryerror.The Program for Appropriate Technology in

Health (PATH) has introduced a dipstick dotimmunoassay as a rapid/simple method of detectingantibodies to HIV-1 and HIV-2 at reduced costand requiring no specialized equipment. In orderto evaluate the reliability of this test under local

Tropical Medicine and International Health volume 2 no. 1 pp 83–88 january 1997

C. S. Ray et al. Dipstick-dot immunoassay for HIV antibodies in Zimbabwe

84 ,C 1997 Blackwell Science Ltd

conditions, and bearing in mind the WHO strategies,we tested stored sera from a factory worker cohortof blood donors in Zimbabwe using the dipstick andstandard ELISAs.

Materials and methods

Test kits

The HIV-1 and 2 dipstick kits used for this evalu-ation were manufactured by the ImmunoChemicalLaboratory, Thailand, according to the proceduredeveloped by PATH. The dipstick assay (ICLD) usessynthetic HIV-1 and 2 peptides bonded onto the 8tines of a plastic comb designed to fit the wells of amicrotitre plate. Antibodies in serum specimens thatare reactive with the antigen are detected usingcolloidal gold bound to protein A. Positive sera givea distinct red spot on the tine, and the absence ofthis spot indicates a negative serum. The test israpid (c. 30 minutes), easy to perform, requires noelaborate washing procedures and the results can beread by eye and photocopied for permanent storageand record.

Sera

Initially, 550 sera were selected from frozen("20)C) serum banks administered by ZAPP-UZsince January 1993. Prior to storage the sera had allbeen tested by both a recombinant peptide HIV-1and 2 ELISA (Abbott Labs, USA) and a syntheticpeptide HIV-1 and 2 ELISA (Enzygnost, Behring,Germany) known for high specificity (Maskill et al.1988). We included an abnormally high number ofsera giving discrepant results in order to investigatethe performance of ICLD under these circumstances.For these sera, there were also data from Westernblotting (Ancoscreen, Ancos Laboratories, Denmark)following the standard ZAPP-UZ algorithm. Inaddition, for some of the discrepant sera theresults of p24 antigen detection assays (AbbottLaboratories, USA) were available, as these sera hadbeen included in a programme to examine p24antigenaemia in suspect seroconverters. All serawere marked in code and tested without knowledgeof previous test results.

Procedure

Sera were kept frozen until the day of testing, andthey were tested in batches to avoid differences infreeze–thaw cycles. They were tested concurrentlywith the Abbott recombinant peptide HIV-1 and 2ELISA and ICLD, following the manufacturer’sinstructions carefully in each case. Where considerednecessary, sera were retested.

Results

Four of the original 550 sera gave results on post-storage ELISAs that were repeatedly inconsistentwith both of the pre-storage ELISA results, 3 serabeing positive by both pre-storage ELISAs, butnegative by both ELISAs and ICLD on retesting, andone serum being negative by both pre-storageELISAs, and positive by both ELISAs and ICLD onretesting. Because this may have resulted from poorstorage conditions, or mis-labelling of specimens atthe time of storage, the pre-storage results were con-sidered unreliable and they were excluded from theanalysis.On the basis of the three ELISA results (two pre

and one post-storage), the remaining 546 sera wereclassified into 3 groups: 266 true negative sera(defined as negative by all three ELISAs); 250 truepositive sera (defined as positive by all 3 ELISAs;174 of these sera came from asymptomatic clients,and 76 came from symptomatic patients); 30 prob-lematic sera (sera that gave discrepant resultsbetween the 2 pre-storage ELISAs). Of these 546sera, 529 (96.9%) gave identical results in the twoconcurrent (Abbott and ICLD) tests. After retestingthe ICLD positives, as recommended by the manu-facturer, the number of concordant results increasedto 542 (99.3%).The results of the initial analysis using only the

516 sera defined as true positive or negative asabove are shown in Table 1. Although 12 truenegative sera gave a false positive result on testingwith ICLD, 11 of these were negative on retesting.From these data the ICLD gave a sensitivity of99.6% (CI 98.82–100.38) and a specificity of 99.6%(CI 98.84–100.36).The results of ICLD testing of the 30 ‘problematic’

sera are shown in Table 2. The inclusion in the

Tropical Medicine and International Health volume 2 no. 1 pp 83–88 january 1997

C. S. Ray et al. Dipstick-dot immunoassay for HIV antibody in Zimbabwe

85 ,C 1997 Blackwell Science Ltd

Abbott 3rd generation ELISA of peptides reactivewith IgM antibodies gives this test a high sensitivityin detecting early seroconverters. Specimens ingroups 1 and 2 may represent such converters, andWestern blots and/or p24 antigen detection con-firmed early seroconversion in 3/25 of them. Onlyone serum from this group was consistently positiveby ICLD, and this serum showed bands indicative ofseroconversion in the Western blot. Of 6 sera givinginitially a weak positive reading, 5 were consistentlynegative on repeat testing. The remaining serum,which continued to give a weak reading by ICLD,showed no bands on Western blot but continued togive a result in the high negative range (20–50%below the cut-off OD) in the Abbott ELISA. Serumcollected one month later from this client waspositive by all tests.The Enzygnost OD value was <1.6 times the

cut-off with the 3 sera in group 3, and so may havebeen false Enzygnost positives. They were repeatedlynegative by Abbott ELISA and ICLD and theyshowed no bands on Western blot. Of the 2 sera ingroup 4, one was repeatedly negative by ICLD,despite being repeatedly positive by post-storageAbbott and Enzygnost ELISAs. There were nobands on Western blot, and the p24 antigen assaywas negative. The other serum in this grouprepeatedly gave a faint spot on the ICLD, but againboth Western blot and p24 antigen tests werenegative.On a practical basis the dipstick was simple and

straightforward to perform and appropriate for anon-specialist laboratory where small numbers ofsera are to be tested. The kits require storage at2–8)C, so refrigeration is necessary.

Discussion

The sera used in this evaluation were not intended tobe representative of sera seen at hospitals or clinicsin Zimbabwe, where only about 1% of sera givediscrepant results and so require further testing. Thesera were selected to provide a large number ofknown positive, known negative and problematicsera by which to evaluate the performance of theICLD. When compared with a concurrently runELISA, the overall accuracy of the ICLD was >98%,suggesting the ICLD to be an acceptable alternativetest to detect antibodies to HIV-1 and 2. It could beused in combination with another rapid/simple testor conventional ELISA as suggested by the WHOstrategies. The cost of the ICLD is less than US$2.00compared to US$3.00–3.50 for conventional ELISAs,and so the cost of HIV antibody testing may bereduced by 30% when ICLD is used as a confirma-tory test, and by more than 30% when used as theinitial screening test. The sensitivity and specificity ofthe ICLD using sera defined as positive or negativeby repeated ELISAs was >99% in this artificiallycreated set of sera.In general the higher the prevalence of HIV infec-

tion in a population, the greater the predictive valueof a positive serological test. Our data suggest thatin a population of low seroprevalence (3%), thenegative predictive value would be >99%, but thepositive predictive value would be only 88%. Thiswould be acceptable for example for screening ofblood donors, but it does indicate that any positiveswhich might be found would need confirmation byanother test if the results were to be given to thedonor. As the WHO strategies suggest, a secondand/or third test (depending on prevalence) would bevital in testing the ‘worried well’ group, whereavoidable stress and anxiety caused by potential falsepositives would be unacceptable.In the case of symptomatic patients, where the

true seroprevalence may be much higher (40%),both positive and negative predictive values are>99%. Our experience was that each of the ELISA-seropositive symptomatic patients gave a positiveresult with the ICLD, so this test may be of value inconfirming a clinical diagnosis of HIV infection forpurposes of management or chemoprophylaxiswithout recourse to confirmatory testing. This has

Table 1 Comparison of ICL Dipstick HIV-1 and 2 andrepeated ELISAs

ICLD

Number of sera givingrepeated ELISAs

+ve "ve Total

+ve1 249 1 250"ve 1 265 266Total 250 266 516

1ICLD+ve, Positive on initial and repeat testing.

Tropical Medicine and International Health volume 2 no. 1 pp 83–88 january 1997

C. S. Ray et al. Dipstick-dot immunoassay for HIV antibodies in Zimbabwe

86 ,C 1997 Blackwell Science Ltd

become an issue in situations where the overall costof a test may inhibit testing patients with suspectedHIV-related illnesses. In this context patients areoften not informed of the diagnosis so the oppor-tunity to plan their lives or to introduce measures toprevent further transmission of HIV may be lost.There are now a large number of different test kits

available, all giving high sensitivity and specificity inat least some population groups, and high positivepredictive values in populations with high HIV sero-prevalence (Maskill et al. 1988; Van Kerckhovenet al. 1991). Problems arise, however, with sera thatgive equivocal or discrepant results. Recommendedguidelines suggest that such sera should be examinedby Western blotting, but this is costly, time consum-ing, requires sophisticated equipment and may bedifficult to interpret. Our experience has been thatWestern blots often give no additional informationwith sera that give inconsistent ELISA results, withmost being labelled ‘indeterminate’. The problemof indeterminate results with Western blots hasbeen discussed in several studies (Mortimer 1991;Ramirez et al. 1992; Healy & Bolton 1993; Satoet al. 1994).Using the ICLD in combination with another

rapid/simple test or conventional ELISA would lessenthe cost to research the effectiveness of counsellingand testing those showing high risk behaviour asprevention strategies. It may also reduce the costs ofprevention of HIV related diseases, such as identify-ing HIV-infected persons who may benefit from pro-phylactic isoniazid. The simplicity of the test meansthat self-testing would be possible, although sucha procedure would be undesirable for a numberof ethical reasons. Restrictions on the sale and

availability of kits would need to be established andenforced.In summary, when tested with sera that were well

characterized by repeated ELISA, using both recom-binant and synthetic peptides, the ICLD was foundto be sensitive and specific. When tested with ‘prob-lematic’ sera, the ICLD was positive in only one casewhere other tests were negative, and there was evi-dence that this may have come from a seroconverter.Where the ICLD was negative, but ELISAs werepositive, there was evidence for either false positiveresults in the ELISA, or an indication of early sero-conversion that the ICLD failed to detect. The ICLDwas simple to perform, required no special equip-ment and a set of tests could be completed within 30minutes. It was noted that when testing large num-bers of sera at the same time, there was a possibilityof error in time-keeping, and the test was consideredless useful in a busy routine laboratory where ahundred or more tests may be performed daily. Theaccuracy, low cost and ease of operation suggeststhat ICLD has an important role in detecting anti-bodies to HIV at district level in developingcountries. This can be underlined by the manufactur-er’s claim that tests can be carried out, without lossof sensitivity or specificity, on whole blood samplesor plasma.

Acknowledgements

We are grateful to PATH, Seattle, for providing theICL Dipsticks through McDonald Scientific, Harare.The ZAPP-UZ is supported by US Public HealthService Award no. A1-33868-02, Preparation forAIDS Vaccine Evaluation. We are indebted to the

ProblemNumber of sera givingICL Dipstick result

Ab1 Behr Ab2 +veweak+ve "ve Total

Group 1 +ve "ve +ve 1 2 5 8Group 2 +ve "ve "ve 0 4 13 17Group 3 "ve +ve "ve 0 0 3 3Group 4 "ve +ve +ve 0 1 1 2Total 1 7 22 30

Table 2 Results of tests withproblematic sera

Ab 1, Pre-storage Abbott ELISA,Behr, Behring ELISA; Ab2,post-storage Abbott ELISA.ICLD weak+ve, faint, butdiscernible spot on testing.ICLD+ve, positive on initial andrepeat testing.

Tropical Medicine and International Health volume 2 no. 1 pp 83–88 january 1997

C. S. Ray et al. Dipstick-dot immunoassay for HIV antibody in Zimbabwe

87 ,C 1997 Blackwell Science Ltd

ZAPP-UZ staff for their support, and particularlyto P. Foroma and R. Chifamba for secretarialassistance.

References

Healy DS & Bolton WV (1993) Apparent HIV-1glycoprotein reactivity in Western blot in uninfectedblood donors. AIDS 7, 655–658.

Latif AS, Marowa E, Mason PR et al. (1995) The aetiologyand pattern of STD among men and women presentingto health centres in Harare, Zimbabwe and risk factorsfor cervicitis among symptomatic and asymptomaticpregnant women. Ministry of Health and Child Welfare,Government of Zimbabwe, Harare, Zimbabwe.

Mahomed K, Kasule J , Makuyana D et al. (1991)Seroprevalence of HIV infection amongst antenatalwomen in greater Harare, Zimbabwe. Central AfricanJournal of Medicine 37, 322–325.

Maskill WJ, Crofts N, Waldeman E et al. (1988) Anevaluation of competitive and second generation ELISAscreening tests for antibody to HIV. Journal ofVirological Methods 22, 61–73.

Mbizvo MT, Machekano R, McFarland W et al. (1996)HIV seroincidence and correlates of seroconversion in acohort of male factory workers in Harare, Zimbabwe.AIDS (in press).

Mortimer PP (1991) The fallibility of the HIV Westernblot. Lancet 337, 286.

Ramirez E, Uribe P, Escanilla D et al. (1992) Reactivitypatterns and infection status of serum samples withindeterminate Western immunoblot tests for antibody tohuman immunodeficiency virus type 1. Journal ofClinical Microbiology 30, 801–805.

Sato PA, Maskill WJ, Tamashiro H & Heymann DL(1994) Strategies for laboratory HIV testing: anexamination of alternative approaches not requiringWestern blot. Bulletin of the World Health Organisation72, 129–134.

Van Kerckhoven I, Vercauteren G, Piot P & van der GroenG (1991) Comparative evaluation of 36 commercialassays for detecting antibodies to HIV. Bulletin of theWorld Health Organisation 69, 753–760.

World Health Organisation, Global Programme on AIDS(1992) Recommendations on the selection and use ofHIV antibody tests. Weekly Epidemiological Record 20,145–149.

Tropical Medicine and International Health volume 2 no. 1 pp 83–88 january 1997

C. S. Ray et al. Dipstick-dot immunoassay for HIV antibodies in Zimbabwe

88 ,C 1997 Blackwell Science Ltd