Measurement of antibodies to human immunodeficiency virus: an international collaborative study to evaluate WHO reference sera.

Bulktin of the World Health Organization, 66 (2): 197-202 (1988) i World Health Organization 1988 Measurement of antibodies to human immuno- deficiency virus: an international collaborative study to evaluate WHO reference sera A. J. GARRETr,1 V. SEAGROArr,1 E. M. SUPRAN,2 K. 0. HABERMEHL,3 H. HAMPL,3 & G. C. SCHILD1 Two preparations ofhuman sera, one reactive against human immunodeficiency virus (HIV) and the other unreactive, were evaluated as potential international reference reagents (IRR) in an international collaborative study. Twenty-one laboratories participated and tested these andfive other human sera which were found to range from highly reactive to unreactive. The proposed 'positive ' IRR wasfound to react strongly in all immunoassays and gave all the expected bands in immunoblot systems using HTLV-III, LA V-I or similar virus strains as antigens. The 'unreactive' serum wasjudged to be negative by ELISA and immunoblots. The end-points determined by ELISAs varied considerably between laboratories, even between those using the same commercial kit. This variation was reduced somewhat when the reactivities ofthe samples were expressed relative to the proposed IRR. Human immunodeficiency virus (HIV), the causa- tive agent of the acquired immunodeficiency syn- drome (AIDS) (1, 9, 12), is transmitted primarily through sexual contact or the injection of contami- nated blood or blood products such as anti-haemo- philic factors (5). Since 1985, the screening of blood donations for anti-HIV has been instituted in many countries in order to minimize the risk of transmis- sion ofAIDS via blood transfusions or treatment with blood products. The detection of antibodies to HIV is also of major importance as a relatively simple and WHO Collaborating Centre on AIDS, National Institute for Biological Standards and Control, South Mimms, Potters Bar, Herts. EN6 3QG, England. Requests for reprints should be sent to Dr A. J. Garrett at this address. 2 WHO Collaborating Centre on AIDS, Division of Microbio- logical Reagents and Quality Control, Central Public Health Service Laboratory, Colindale, London, England. 3 WHO Collaborating Centre on AIDS, Institute for Clinical and Experimental Virology, Free University of Berlin, Berlin (West). rapid determination of the extent and spread of HIV infections (3) and many commercial and 'in house' immunochemical tests are now in use throughout the world. At present, the most commonly used assays are based on enzyme-linked- or radio-immunosor- bance, immunofluorescence, immunoblotting, or im- munoprecipitation, and variations in the specificities and sensitivities of the techniques reflect inherent differences between the principles of the assays as well as batch-to-batch variations in the preparations of reagents and kits (10, 11). Thus, there is an urgent need for well-characterized reference materials that can be used to define the reliability and sensitivity of the tests, for quality control of batches of kits or reagents, and as common references between laboratories. This report presents an assessment oftwo proposed reference preparations of sera, one reactive and the other unreactive to HIV, in a collaborative study involving 21 laboratories in 11 countries. 4872 197- A. J. GARRETT ET AL. MATERIALS AND METHODS Proposed reference materials The proposed preparations of antibody-positive and antibody-negative human sera, freeze-dried in sealed glass ampoules, were supplied by Professor K. 0. Habermehl (Institute for Clinical and Experi- mental Virology, Berlin (West)). Each preparation was derived from a single donor; one an asympto- matic carrier of HIV and the other a donor with no known risk factors. Each serum was unreactive when tested for HBsAg antigen by a standard immunoassay and was heated at 56 °C for 1 h before being freeze- dried. When reconstituted as recommended (in 0.2 ml water) these preparations represented con- centrations one-twentieth of the original sera; this dilution factor is not included in the calculations presented in this report. Coded preparations Seven freeze-dried serum preparations, coded A to G, were supplied to each participant. Preparations A and C were duplicate samples of the reactive pro- posed reference material, preparation E was the non- reactive proposed reference material. Samples D and F were prepared by the Central Public Health Service Laboratory, Colindale, London. Sample D, derived from a single donor, showed weak reactivity in im- munosorbent assays. Sample F, derived from sera pooled from several donors, was highly reactive. Samples B and G, which were NIBSC reference prep- arations freeze-dried from pooled sera in 1973 and 1967, respectively, were both unreactive. Design of the study The study was designed to identify the coded pre- parations that reacted with HIV antibodies and to ascertain the minimum amount of the reactive sam- ples that could be detected in the methods routinely used by the participants. The 21 participating laboratories (see Annex) were supplied with duplicate sets of the coded preparations and requested to assay them by the procedures usually employed in their laboratories. Assay methods All but three participants assayed the preparations by indirect or competitive ELISA. Nine different commercial kits for ELISA were used: Abbott, Dupont, ENI, Genetic, Organon, Ortho, Pasteur ('ordinary' and 'Rapide'), Travenol and Wellcome. Two laboratories carried out ELISAs using their own 'in house' methods and one included in its series of assays the Abbott 'confirmatory' ELISA, a competi- tive ELISA based on envelope and core antigens derived from recombinant DNA. Immunoblots were carried out by 15 laboratories. All except two, whose techniques involved the use of a mouse monoclonal antibody specific for human IgG and labelled with 125I or protein A labelled with 1251, used peroxidase-linked anti-human IgG for the identification of antigen-antibody complexes and eight used biotin-avidin amplification of the enzyme system. One participant used the Karpas method (8) and one used an assay based on particle agglutination (PA). Method of analysis For each test the reactivities of the coded samples A to G and the end-point titres for samples A, C and F were taken to be those recorded by the participants. End-points were defined as the reciprocals of the highest dilutions of the reconstituted original ma- terials in normal serum (not the final dilutions in the assay wells) that gave positive responses in the assays. Potency ratios of C and F were expressed as ratios of their titres to that of A in the same assays. RESULTS Classification of sera by reactivity in ELISA and the immunoassays Samples A, C and F were found reactive in all tests and samples B, E and G were reported as negative in all but one test. The exception was a test based on particle agglutination (PA) in which sample E wasjudged to be weakly reactive. Sample D was found to be reactive or weakly reactive in 40 of the 48 ELISAs performed, in both the PA and Karpas test and in two of the four fluorescence microscopy (FM) tests (Table 1). In the eight ELISAs in which sample D was identified as unreactive, it had a higher optical density (OD) than the negative control (although not of course as high as the OD of the cut-off limit), and in all but one the ODs were at least twice that of the negative control. Titration ofsamples A, C and F Participants carried out single or duplicate assays for individual manufacturers' ELISAs or by their local methods. Some laboratories used several manu- facturers' ELISAs; in particular, laboratory 1 used seven different kits. For each kit, the geometric means of the titres for A, C and F and of the potency 198 REFERENCE SERA FOR MEASUREMENT OF HIV ANTIBODIES Table 1. Assessment of reactivity of sample D by im- munoassays No. with stated reactivity No.of No. of Assay method laboratories assays + ± ELISA kits: Abbott 9 12 9 2 1 Dupont 2 3 3 - - ENI 1 2 - - 2 Genetic 1 2 - 2 Organon 3 4 1 1 2 Ortho 2 3 3 - - Pasteur' 5 8 6 1 1 Travenol 1 1 - - 1 Wellcome 6 10 8 2 1 ELISA 'in house' 2 3 1 1 1 All ELISAs 18 48 31 9 8 Fluorescence microscopy 4 4 1 1 2 Karpas method 1 1 1 - - Particle agglutination 1 2 2 ' Includes assays using the 'Rapide' version. A 6 T H X E X XF A~ i0 -4 6 -423 -38 -3 o30 -2 6 -2 2 - 8 - 4 - 0 -6 -0.2 LOG END POINT c . -346 -432 -3 8 -3 4 -3 0 - 32.2 -1.8 -0A -1o -0.6 -o 2 LOG END POINT F LG N G e ratios for C and F obtained by individual laboratories were calculated. The frequency distributions of these values are shown in Fig. 1 and 2. One of the partici- pants using the Abbott kit obtained much higher titres for A and F than did the other participants using both this and other kits. Further, its titres for C were ten- fold and 100-fold lower than those for A. The results from this participant were, therefore, considered atypical and excluded from the subsequent analyses. The laboratory mean titres varied considerably over about a 20-fold range. There were, however, no obvious differences between the titres from different kits: for instance, the ranges of titres for Abbott, Pasteur and Wellcome overlapped with each other. The laboratory mean potency ratios of C, a coded duplicate of A, were mostly unity. All but one lab- oratory (mentioned above) found the titres ofA and C to be within one dilution step (Fig. 1) although some- times these dilution steps were as large as 5- and 10- fold. The laboratory mean potency ratios for F were more variable than those for C. However, the potency ratios were less variable than the titres (Fig. 1 and 2). One laboratory's results gave a potency ratio, based on a single assay, of256, tenfold higher than the other estimates, and had the highest titre for F (16 384). o -4 * -4 .2 -3 .0 -3 .4* -3 o -2 6 -2P.2 LOG ENO POINT Fig. 1. Frequency distributions of the end-point dilutions obtained for samples A, C and F. Each square denotes an estimate from one test; the letters in the squares refer to the type of assay. A, D, E, G, H, N, P, T and W denote the following commercial kits of ELISA: A, Abbott; D, Dupont; E, ENI; G, Genetic, H, Ortho; N, Organon; P, Pasteur; T, Travenol; and W, Wellcome. L and LG denote 'in house' versions of ELISA. Letters FM, K, and PA denote the following methods other than ELISA: FM, fluorescence microscopy; K, Karpas; and PA, particle agglutination test. This titre and potency ratio were, therefore, con- sidered atypical and excluded from the subsequent analysis. Differences between the laboratories' estimations of titres were found to be significant by analyses of variance, even between those using kits from the same manufacturer. Expressing the reactivities of samples C and F relative to A showed that the dif- ferences between laboratories using the same com- mercial kit were no longer statistically significant. However, there were still significant differences between the potency ratios from different tests. For o.~ ~~ ....... . l E R ;_ D 199 to cG H N A.I. 2 a _ G RA 0 p W L 1. [F-. FA] FAI A A A FM L 0 N w p w p A W F-. F,A A A A. J. GARRETT ET AL. example, the overall mean potency ratio, i.e., the geometric mean of the laboratory mean potencies, for Weilcome was 22, about three times higher than those for Abbott and Pasteur kits (6 and 7, respectively). The overall mean potency ratios for the other kits fell within this range. Immunoblots Fifteen participants tested samples A to G by immunoblot techniques. The results are given in Table 2. The use of control antigens ('mock' anti- C -1.3 -1.0 0.7 0.4 0.1 0.2 0.5 0. LOG POTENCY 1.1 1.4 1.7 2.0 gens) was reported from only two laboratories. Faint reactive bands in the regions of relative molecular mass (Mr) 24x 103 and 64x 103 were detected con- sistently by one participant using 'mock' antigen from H9 cells; more information is required on this im- portant aspect of the assays. The relative molecular masses recorded in Table 2 are those assigned by the individual participants. For convenience of presentation, bands reported within narrow ranges of Mr are not differentiated and are classified in groups (e.g., (32-34) x 103 and (110-160) x 103). Samples A and C were duplicates of the proposed reactive reference serum. As was hoped, they pro- duced identical results for each immunoblot system and are considered together. One participant reported on the detection of antibodies to peptides p24 and gp41 only. All participants, except one who reported a weak reaction in the p65 region for sample E, and another who observed a reaction to p24 antigen in sample G, detected no antibodies to HIV in samples B, E and G. Of the positive samples, F reacted most strongly in all immunoblots but, except for less fre- quent detection of antibodies to the envelope antigens gpl I0-pl60 in A and C than in F, samples A, C and F were qualitatively identical. All participants detected antibodies to p24, gp41 and p53/55 in A, C and F, three reported no anti- bodies to p17/ 18, and only one reported no antibodies to p65. Additional results Essex and colleagues included in their study an investigation of the reactivity of samples A to G in immunoblots in which the recently isolated strain HTLV-IV was used as antigen. No reactions with any F 14 12 10 a 6 U_ 0, 0 z -1.6 .13 K FIM L. A [FMA.A _I; PA T H H N E D W W1 G |N IWI A AI WIW -1.0 40.7 40.4 40.1 0.2 0.5 0A 1.1 IA 1.7 2.0 LOG POTENCY Fig. 2. Frequency distributions of the potency ratios of samples C and F in terms of A obtained from individual tests. (See legend in Fig. 1 for description of the numbers and letters in the squares, etc.). Table 2. Detection of anti-HIV by immunoblot' Peptide or glycopeptide (approximate Reactive bands/total reports relative molecular mass x 103) Sample A (C) Sample D Sample F 17,18 11/14 6/14 13/14 24 15/15 13/15 15/15 32, 34 12/14 11/14 11/14 38, 39 9/13 5/14 10/13 41 15/15 6/15 15/15 53, 55 11/14 12/24 14/14 65 13/14 13/14 13/14 110,120,160 9/13 7/13 12/13 e All participants did not report the presence or absence of each peptide or glycopeptide. 30 28 26 24 22 20 18 16 14 12 10 a 6 4 -1 U1m 0 z ... . . . .-. .. .. .. .. .. .. .. .. . . .-. . 200 REFERENCE SERA FOR MEASUREMENT OF HIV ANTIBODIES viral antigens were reported for Western blots and only two reactive regions, gpl60 for sample A and p24 for sample F, were detected in radioimmuno- precipitation using "S-labelled HTLV-IV. DISCUSSION ELISA kits from nine manufacturers were used in this study. The sensitivities of the various kits were assessed by comparing end-point titres for the highly reactive samples A and F and by whether or not a weakly reactive serum was found 'positive' in the tests. This sensitivity varied between laboratories, even between those using the same commercial kit. Expressing the reactivity of sample F relative to A reduced the variation between tests and resulted in agreement between the laboratories using the same kit. Nevertheless there were still consistent differences between the kits in the comparison of the reactivity of A and F. This possibly reflected the differences in specificities of the ELISA systems. Overall, immunoblots revealed reactions of sera A (C) and F to all the expected HIV antigens. The presence of antibodies to p24, gp4l and p55 is con- sidered by most workers an important indication of infection with HIV (6, 11, 13). However, for repro- ducible results the source of antigens, the standardi- zation of electroblotting procedures, and the pro- vision of 'control' antigens require careful attention (6); variations of these factors may well explain the differences shown in Table 2. The results for HTLV-IV confirm earlier findings (2, 4, 7) and emphasize the urgent need for infor- mation on the responses of immunoassays to sera from AIDS patients from different geographical areas and for characterization of the genetic and immuno- logical differences between viral isolates. The pro- posed reference preparation, A, reacted strongly in all ELISAs and related immunoassays and reacted with all the major HIV antigens in immunoblots. Its use will depend on individual requirements but it may be of value as a qualitative check on the specificity of the assays, to calibrate positive controls included in kits and other assays in arbitrary units, to calibrate detection limits (cut-off) in arbitrary units (as done for HBsAg), and to calibrate immunoblots, particu- larly for defining the optimal amounts of antigen and for determining the relative mobilities of the major peptides and glycopeptides from HIV. Because the unreactive preparation (E), reconstituted in 0.2 ml water as recommended, represents diluted serum, its use as a reference material will be limited. The WHO Expert Committee on Biological Stan- dardization reviewed the report of this collaborative study in December 1986 and agreed that preparations A and E would be of value as reference preparations for, respectively, positive and negative anti-HIV sera. The preparations, under code numbers 86/6302 (reactive) and 86/6238 (unreactive) are available from the Director, National Institute for Biological Standards and Control, South Mimms, Potters Bar, Herts. EN6 3QG, England. JME TITRAGE DES ANTICORPS DIRIGtS CONTRE LE VIRUS DE L'IMMUNODEFICIENCE HUMAINE: ETUDE COLL,CTIVE INTERNATIONALE POUR L'tVALUATION DE StRUMS DE REFERENCE OMS Deux pr6parations de s6rum humain, l'une r6agissant avec le virus de l'immunodeficience humaine (VIH) et l'autre inactive, ont fait l'objet d'une 6tude collective a laquelle ont participe 21 laboratoires de 11 pays. Ces pr6parations ont 6t6 fournies, sous forme lyophilis6e en ampoules de verre scell6es, par le Professeur Habermehl (Institut de virologie clinique et exp6rimentale, Berlin- Ouest). Chaque serum provenait d'un donneur unique; l'un d'eux 6tait un porteur de VIH asymptomatique et l'autre ne pr6sentait aucun facteur de risque. L'6tude 6tait conque de facon a identifier, parmi sept pr6- parations codees, celles qui contenaient des anticorps anti- VIH et, dans le cas des 6chantillons positifs, a d6terminer les quantites minimales d6tectables par les m6thodes employees habituellement par les participants. Pour ces essais, la plupart des participants ont choisi une m6thode ELISA indirecte ou par comp6tition, et ils ont utilis6 9 n6cessaires d'analyse commerciaux diff6rents. Quinze laboratoires ont effectu6 des immunotransferts en utilisant principalement l'IgG antihumaine liMe a la p6roxi- dase ou l'amplification du systeme enzymatique par la biotine-avidine pour identifier les complexes antigenes- anticorps. L'6chantillon propose comme r6actif de r6ference "posi- tif' a reagi fortement dans tous les immunotitrages et a donne tous les bandes attendues dans les systemes d'im- munotransfert. La sensibilit des divers essais, mesuree par comparaison des titres de fin de dosage, a varie d'un laboratoire a l'autre, meme lorsque ceux-ci utilisaient le meme n6cessaire d'analyse. Cette variation entre labora- toires a pu etre reduite en exprimant les resultats en fonction de I'activit6 de 1'6chantillon positif propose. Les noms et adresses des laboratoires participants sont indiqu6s dans l'annexe 1 (page 202). Les r6actifs de r6f6rence propos6s, qui portent les nume- ros de code 86/6302 (actif) et 86/6238 (inactif), peuvent etre obtenus sur demande adressee au Directeur du National Institute for Biological Standards and Control, South Mimms, Potters Bar, Herts. EN6 3QG, Angleterre. 201 202 A. J. GARRETT ET AL. ACKNOWLEDGEMENT We thank Jane Bruce for assisting in the statistical analysis of the data from the study. REFERENCES 1. BARRE-SINOUSSI, F. ET AL. Isolation of a T-lympho- tropic retrovirus from a patient at risk for acquired immune deficiency syndrome (AIDS). Science, 220: 868-871 (1983). 2. BIBERFELD, G. ET AL. Findings in four HTLV-IV seropositive women from West Africa. Lancet, 2: 1330-1331 (1986). 3. BIGGAR, R. J. The AIDS problem in Africa. Lancet, 1: 79-83 (1986). 4. CLAVEL, F. ET AL. Isolation of a new human retrovirus from West African patients with AIDS. Science, 233: 343-346 (1986). 5. CURRAN, J. W. ET AL. The epidemiology of AIDS: current status and future prospects. Science, 229: 1352-1357 (1985). 6. ESSEX, M. ET AL. Antigens of human T-lymphotropic virus type HI/lymphadenopathy associated virus. An- nals of internal medicine, 103: 700-703 (1985). 7. KANKI, P. J. ET AL. New human T-lymphotropic reti;o- virus related to simian T-lymphotropic virus type HI (STLV-HI AGM). Science, 232: 238-243 (1986). 8. KARPAS, A. ET AL. Lytic infection by British AIDS virus and development of rapid cell test for antiviral antibodies. Lancet, 2: 695-697 (1985). 9. LEVY, J. A. ET AL. Isolation of lymphocytopathic retroviruses from San Francisco patients with AIDS. Science, 225: 840-842 (1984). 10. MORTIMER, P. P. ET AL. Which anti-HTLV-HI/LAV assays for screening and confirmatory testing? Lancet, 2: 873-877 (1985). 11. PETRICCIANI, J. C. ET AL. An analysis of serum sam- ples positive for HTLV-HI antibodies. New England journal ofmedicine, 313: 47-48 (1985). 12. POPOVIC, M. ET AL. Detection, isolation and continu- ous production of cytopathic retroviruses (HTLV-HI) from patients with AIDS and pre-AIDS. Science, 224: 497-500 (1984). 13. SCHUPBACH, J. ET AL. Antibodies to HTLV-ElI in Swiss patients with AIDS and pre-AIDS and in groups at risk for AIDS. New England journal of medicine, 312: 265-270 (1985). Annex I Participating laboratories National HIV Reference Laboratory, Fairfield Hos- pital, Fairfield, Victoria, Australia Red Cross Blood Transfusion Service, Adelaide, Australia Laboratory Centre for Disease Control, Ottawa, Ontario, Canada Laboratoire National de la Sant6, Departement de Controle des Vaccins a Virus et des Produits Deri- ves du Sang, Paris, France Institut Pasteur, Paris, France Institute for Clinical and Experimental Virology, Free University of Berlin, Berlin (West) Max von Pettenkofer Institute, Munich, Federal Republic of Germany Institute for Virus Research, Kyoto University, Kyoto, Japan Central Laboratory of the Netherlands Red Cross Blood Transfusion Service, Amsterdam, Nether- lands Blood Transfusion Service, Department of Haema- tology, Singapore General Hospital, Singapore Centro Nacional de Microbiologfa, Virologia e Immu- nologia Sanitanras, Majadahonda, Madrid, Spain National Bacteriological Laboratory, Solna (Stock- holm), Sweden North London Blood Transfusion Centre, Edgware, Middlesex, England Scottish National Blood Transfusion Service, Edin- burgh, Scotland Department of Microbiological Reagents and Quality Control, Central Public Health Laboratory, Colin- dale, London, England Department of Haematological Medicine, University of Cambridge, Cambridge, England Regional Blood Transfusion Centre, Royal Infirm- ary, Edinburgh, Scotland National Institute for Biological Standards and Con- trol, Holly Hill, Hampstead, London, England Department of Cancer Biology, Harvard School of Public Health, Boston, MA, USA Division of Virology, Center for Drugs and Bio- logics, Food and Drug Administration, Rockville Pike, Bethesda, MD, USA AIDS Program, Center for Infectious Diseases, Cen- ters for Disease Control, Atlanta, GA, USA