Application of enzyme immunoassays for the confirmation of clinically suspect plague in Namibia, 1982*

Bulletin of the World Health Organization, 64 (5): 745-752 (1986) Application of enzyme immunoassays for the confirmation of clinically suspect plague in Namibia, 1982* JAMEs E. WILLIAMS,' LORRAINE ARNTZEN,2 GuY L. TYNDAL,3 & MARGARETHA ISAACSON4 An outbreak of plague occurred in Ovamboland, northern Namibia, late in 1982. Blood cultures, sera and blood clots were tested to obtain laboratory confirmations for clinically suspect cases of the disease. Isolation of the bacillus (Yersinia pestis) was attempted from blood cultures; sera were tested for antibody by passive haemagglutin- ation (PHA) and enzyme-linked immunosorbent assay (ELISA). Sera and clots also were tested by ELISA for the specific F1 plague antigen. All the ELISA procedures were based on a monoclonal antibody to F1 antigen to ensure specificity. Thirty-eight cases were confirmed as plague: 50% by isolation, 34% by antibody responses, and 16% by the detection of antigenaemia. All isolates of Y. pestis were capable of producing F) antigen, and significant antibody responses were observed in bacteriologically confirmed cases with paired sera. Patients who experienced sero- conversion had a higher IgM titre than IgG titre during the first nine days of hospital- ization, while patients hospitalized for 17 or more days had IgG titres that were higher than the IgM titres. The relationship between IgM and IgG antibody titres is discussed with reference to identifying very recent infections. PHA titres increased and declined with IgM titres but were lower and more transient. ELISA procedures increased laboratory confirmations ofplague by 23% above the numbers achieved using blood cultures and PHA tests alone. The ELISA to detect F) antigen accounted for 86% of this increase by confirming cases where bacteriological isolation was not done. This ELISA did not replace the requirement for bacteriological isolation, since seven bacteraemic patients did not demonstrate antigenaemia. Recent evaluations of enzyme-linked immuno- sorbent assays (ELISA) to detect both antibodies and antigen of the plague bacillus in patients' sera (1, 2), and antibodies in the sera of wild rodents and domestic dogs (3), have indicated that these pro- cedures could have practical applications in the sur- veillance and control of the disease. An ELISA to detect antibody in patients' sera increased the diag- * From the Department of Hazardous Microorganisms, Walter Reed Army Institute of Research, Washington, DC 20307-5100, USA, and the Departments of Tropical Diseases and Epidemiology, School of Pathology, South African Institute for Medical Research and the University of the Witwatersrand, PO Box 1038, Johannesburg 2000, South Africa. l Chief, Department of Hazardous Microorganisms, Walter Reed Army Institute of Research. Requests for reprints should be sent to this author. 2 Senior Research Technologist, Department of Tropical Dis- eases, South African Institute for Medical Research. 3 Medical Laboratory Specialist, Department of Hazardous Microorganisms, Walter Reed Army Institute of Research. 4 Head, Departments of Tropical Diseases and Epidemiology, South African Institute for Medical Research. noses of plague by 742% over what was achieved with the passive haemagglutination (PHA) test (1), while an ELISA to detect antigen in the blood demon- strated antigenaemia in 25% of patients who were confirmed as plague by the isolation of Yersiniapestis from bubo aspirates but who had not developed de- tectable antibody (2). In tests of rodent and dog sera, the ELISA and PHA techniques were of comparable sensitivity for detecting antibody, but the ELISA gave significantly fewer non-specific reactions with dog sera (3). Although the ELISA procedures could be used alone, greater success in confirming clinically suspect illnesses as cases of plague might be obtained if ELISA were applied with traditional bacteriological and serological methods of laboratory diagnosis. During an investigation of a recrudescence of plague that occurred in northern Namibia from October to December 1982, a study was undertaken to quantify the benefits that can be derived from using ELISA 4719 -745 746 J. E. WILLIAMS ET AL. with methods for bacteriological isolation of Y.pestis and serodiagnosis by PHA testing. The epidemiology of plague in the northern region of Namibia has been described elsewhere (4). MATERIALS AND METHODS Clinical specimens Specimens were collected in -Ovambo, northern Namibia, by physicians with experience in the clinical diagnosis and treatment of plague. As a rule, blood was drawn for serology and culture at the time of (or soon after) admission to hospital. Blood cultures were prepared when bacteraemia was considered as likely and during follow-up after treatment of patients who had bacteraemia when acutely ill. A convalescent serum sample was obtained for serology if conditions permitted. Blood clots, sera and cultures were sent to the South African Institute for Medical Research in Johannesburg, South Africa, for evaluation. A total of 36 blood cultures from acutely ill patients were examined. PHA tests of 96 sera and ELISA on 95 sera were done to detect antibodies. ELISA on 93 sera and 88 blood clots were done to detect the specific fraction 1 (Fl) envelope antigen of Y. pestis. Isolation of Y. pestis All isolations of Y. pestis were obtained from blood cultures streaked on blood agar. Isolates were characterized by standard procedures (5), including tests for Fl antigen in an Ouchterlony gel diffusion agar-plate system using anti-Fl rabbit serum. PHA tests The procedure recommended by WHO for the serodiagnosis of plague was used (5, 6). Titres of > 1:16 were considered as positive indicators of antibody to Fl antigen. ELISA for antibody An improved assay based on anti-Fl monoclonal antibody 3G8 (2) was employed to ensure specificity in the measurement of IgM and IgG plague antibody titres. ELISA platesa were sensitized with 3G8 mouse ascitic fluid diluted to 1:10 000 in phosphate buffered saline (PBS), pH 7.0 (100 ul /well). Plates were sealed with tape and stored at 4 °C for at least 48 hours before use. Sealing tape was removed, and the sensitizing solution was washed out of the plates with ' Immulon 2 flat bottom plates, Dynatech Laboratories, Inc., Alexandria, VA, USA. PBST (PBS containing 0.5 ml/litre of Tween 20). Washes consisted of three rinses followed by a wash in which the PBST was left in the wells for five minutes. A solution of Fl antigen (1 yg/ml in PBST) was added (50 ,l/well), after which the plates were incubated at 37 °C for 30 minutes and washed. At this point, the plates were ready for testing the sera. For each serum tested, two identical series of twofold dilutions were prepared from the 1:4 dilution, with final volumes of 50 I1 /well. One set of dilutions was made in PBST; the other, to observe for ELISA inhibition (1, 2, 7), used PBST containing Fl antigen (25 jig/ml.). The plates were then incubated at 37 IC for 30 minutes and washed. Affinity-purified, horse- radish peroxidase-conjugated antibody to human IgM (j)b or IgG(H + L),b diluted 1:5000 in PBS, was added (50 tl/well). The plates were incubated at 37 °C for 15 minutes and then washed with PBST, as before, plus an extra rinse and soak to ensure total removal of excess conjugate. ABTSC indicator solution was added (50 1l/well), the plates were in- cubated at 37 °C for 40 minutes, and results of the tests were read at 414 nm using a photometerd set to a maximum absorbance of 1.4. Titre was taken as the highest dilution of serum demonstrating a difference between ELISA and ELISA inhibition absorbance values of > 20%. The probability of such a difference being due to chance was less than 1% (P<0.01) (2). However, a test was considered positive only if three consecutive dilutions from the endpoint demonstrated the >20% dif- ference. Visually, a positive test appeared to have a , fourfold depression of the endpoint in the ELISA inhibition series of dilutions. Earlier studies (1) em- ployed this visual method to establish endpoints. In effect, both machine and visual endpoint determin- ations were generally identical (an important consid- eration for field applications without a photometer). The data reported in this paper are from endpoints derived from photometer readings. To confirm the test's sensitivity, every ELISA included a test of a control serum with established titres for IgM and IgG antibodies. Accepted test variation was plus or minus one test dilution. Titres of > 1:16 were considered to be evidence of antibody to Fl antigen. ELISA for antigen ELISA plates were sensitized with monoclonal antibody 3G8, stored at 4 °C and washed, as de- scribed above. Duplicate twofold dilution sequences, b Kirkegaard & Perry Laboratories, Inc., Gaithersburg, MD, USA. C 2.2'-azino-bis(3-ethylbenzthiazoline-6-sulfonate); Kirkegaard & Perry Laboratories, Inc., Gaithersburg, MD, USA. d Titertek Multiskan photometer, Flow Laboratories, Inc., McLean, VA, USA. ENZYME IMMUNOASSAYS FOR DIAGNOSIS OF PLAGUE 747 Table 1. Cases confirmed as plague in Ovambo, 1982 Case Date when Bubo Y. pestis Antibody Antigen No. first seen Age Sex site' isolatedb response' in blood 1 7 Oct 9 F + - 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 9 Oct 12 Oct 12 Oct 14 Oct 17 Oct 18 Oct 18 Oct 18 Oct 20 Oct 20 Oct 22 Oct 23 Oct 25 Oct 1 Nov 2 Nov 8 Nov 9 Nov 11 Nov 14 Nov 14 Nov 15 Nov 17 Nov 19 Nov 23 Nov 29 Nov 30 Nov 1 Dec 4 Dec 5 Dec 8 Dec 9 Dec 10 Dec 10 Dec 13 Dec 13 Dec 14 Dec 20 Dec 7 Adult 12 10 6 3 13 Adult Adult 6 Adult 26 3 31 7 12 4 5 12 10 10 30 13 12 10 Adult 5 56 26 5 7 24 12 Adult M F M F M F F F F F F F M F F F M M F M M F F M M F M M F F F M F M F M F A F + + (B) A F C F A C A C F+C F F A A A A + + + + + + + + + + (B) + (H), - (E) + (B) + (H) + (B) + (B) + (B) + (B) + (B) + (B) + (B) + (B) + + + + + + + (B) +(B) + (B) 4- + + F + (E), - (H) a Axillary (A), cervical (C), femoral (F), inguinal (I). b Isolation from blood sample taken on admission to hospital. ' Plus sign indicates a change in antibody titre of >4-fold by PHA (H), ELISA (E) or both (B); no test of paired sera if blank. from 1: 4, with a final volume of 50 Id /well, were pre- pared to test a specimen. One set of dilutions was made in PBST; the other, to observe for ELISA inhibition, used PBST containing 3G8 ascitic fluid at a 1:1000 dilution. After incubation at 37 °C for 30 minutes, the plates were washed and an anti-plague J. E. WILLIAMS ET AL. rabbit serum globulin solution, equivalent to a 1: 5000 dilution in PBST of the globulin fraction from I ml of serum, was added (50 il/well). The plates were incubated at 37 IC for 15 minutes, washed, and a horseradish peroxidase-conjugated antibody to rabbit IgG(H + L),b diluted 1:5000 in PBS, was added (50 itl/well). The plates were again incubated at 37 IC for 15 minutes and washed. Then, ABTS indicator was added and the plates incubated again, results were recorded, and endpoints were determined as in the ELISA for antibody. A solution of Fl anti- gen (40 ng/ml in PBST) was tested with specimens to ensure that test sensitivity varied only in the narrow range of 0.6-1.3 ng Fl antigen per ml, for an average test sensitivity of 1 ng/ml. In effect, the concen- tration of Fl antigen was the reciprocal of the speci- men dilution at the test endpoint. Concentrations of Fl antigen of > 16 ng/ml were considered significant Fl antigenaemias. Data analysis Data were examined with reference to the date of hospitalization rather than a date when the onset of symptoms might have occurred. Unlike the latter, the date of hospital admission was consistently and accurately determined. RESULTS Y. pestis was isolated from 13 blood cultures obtained during October and from six blood cultures in December 1982; only two blood cultures from acute cases were submitted for examination between 3 and 28 November, and they were negative. However, clinical observations, and serological data obtained retrospectively, indicated that plague infections also occurred in November 1982. Overall, 38 cases were confirmed: 19 (50%) by the isolation of Y. pestis, 13 (34%) by detecting a significant (i.e., >4-fold) change in antibody titre, and 6 (16%) by detecting Fl antigen in the blood (Table 1). Among the 19 cases confirmed by isolation were one case with antigenaemia and four cases which subsequently demonstrated significant antibody responses. This outbreak in Namibia appeared to be an outbreak of bubonic plague; 76% of the patients had at least one bubo. Buboes were found in the inguinal (29% of cases), axillary (21%), femoral (18%), and cervical (11%) regions. However, buboes were not in evidence for nine (24/e) of the confirmed cases. Of these, seven were bacteraemic when first seen (Table 1; cases No. 1, 4, 6, 10, 11, 12 and 36) and were most appropriately classified as septicaemic plague. Thus, at least 18% (7/38) of the confirmed cases had Table 2. Antibody responses of patients in Ovambo, 1982, from whom Y. pestis was isolated Fl antibody titre Day in Response Case No.' hospitalb PHA 1gM IgG None 33 1 <1:16 <1:16 <1:16 14 5 <1:16 <1:16 <1:16 29 6 <1:16 <1:16 <1:16 9 15 <1:16 <1:16 <1:16 3 24 <1:16 <1:16 <1:16 PHA & IgM increase; 30 1 <1:16 < 1:16 < 1:16 no IgG 5 1:64 1:128 <1:16 PHA, IgM, & IgG 32 1 <1:16 <1:16 <1:16 increase 7 1:16 1:4096 1:1024 8 1 <1:16 <1:16 <1:16 12 1:128 1:512 1:256 PHA & IgM decline; 13 1 1 1:16 1:1024 1:4096 IgG unchanged 1 7 < 1:16 1:512 1:4096 No PHA; IgM & IgG 1 13 <1:16 1:2048 1:4096 present ' Case descriptions are in Table 1. bDay of admission = day 1. 748 ENZYME IMMUNOASSAYS FOR DIAGNOSIS OF PLAGUE Table 3. Patients in Ovambo, 1982, confirmed as plague by antibody responses Fl antibody titre Day in Response Case No.' hospital b PHA IgM IgG PHA increase; 16 1 < 1:16 < 1:16 < 1:16 ELISA incomplete 10 1:8192 No test No test PHA & IgM increase; 25 2 <1:16 <1:16 <1:16 no IgG 7 1:32 1:128 <1:16 27 1 <1:16 <1:16 <1:16 8 1:16 1:64 <1:16 PHA, IgM, & IgG increase 15 1 <1:16 <1:16 <1:16 8 1:1024 1:16384 1:256 23 2 <1:16 <1:16 <1:16 8 1:256 1:256 1:64 22 1 <1:16 <1:16 <1:16 9 1:1024 1:16384 1:4096 24 6 1:64 1:128 <1:16 11 1:256 1:256 1:64 20 1 <1:16 <1:16 <1:16 13 1:1024 1:2048 1:8192 18 4 <1:16 <1:16 <1:16 16 1:256 1:1024 1:512 PHA & IgM decline; 31 2 1:128 1:32768 1:8192 IgG increase 5 <1:16 1:16384 1:32768 PHAabsent;lgMdecline; 38 5 <1:16 1:512 1:1024 IgG increase 18 < 1:16 1:16 1:4096 PHAdecline;lgM&lgG 21 1 1:64 <1:16 <1:16 increase 26 1:16 1:2048 1:8192 PHA & IgM increase and 5 2 < 1:16 < 1:16 < 1:16 decline; IgG increase 8 1:256 1:8192 1:512 29 1:64 1:1024 1:4096 a Case descriptions are in Table 1. b Day of admission = day 1. septicaemic plague. This estimate was conservative, as another case without a bubo (case No. 19) had antigen in the blood and may have been septicaemic. Plague pneumonia was not observed in this outbreak. All patients were febrile when first seen, with temperatures of 38 IC to 41 °C. Most patients were females over 10 years of age (42%1), followed by males aged 10 years and under (29%), females of 10 years and under (13%), and males over 10 years of age (8%). Ages were not noted for three cases (8%). None of the patients had a history of plague vaccination. Acute sera in all bacteriologically confirmed cases were negative for antibody. Significant antibody responses were observed in the convalescent sera of four bacteriologically confirmed cases (Table 2; cases No. 8, 13, 30 and 32). However, two patients (cases No. 3 and 9), treated with antibiotics like all the others, remained seronegative in convalescence. Ouchterlony tests proved that the Y. pestis strains isolated from them were capable of producing Fl antigen. Isolates from all the other patients also were positive for Fl antigen. While acute sera from bacteraemic cases were always negative for antibody, significant titres were present in the sera of other patients when they arrived at the hospital (Table 3; cases No. 21 and 31). Antibody titres were found in 7 out of 42 (17%) patients from whom only one serum was collected for testing (Table 4), and these may have been cases of plague. Normally, a titre of > 1:16 in a single serum is considered to be presumptive evidence of plague infection (8) and, indeed, these patients were treated for plague. We did not consider them as confirmed for the purposes of this paper because at least some of the antibody titres may have resulted from an earlier exposure to Y. pestis. The serological data demonstrated patterns in the development of antibody titres during convalescence 749 J. E. WILLIAMS ET AL. Table 4. Seropositive suspect cases with unpaired sera in Ovambo, 1982 Fl antibody titre Date when Bubo Day in first seen Age Sex site' hospitalb PHA 1gM IgG 19Oct 40 F F 11 <1:16 1:256 1:256 5 Nov 20 F I 1 1:16 <1:16 <1:16 8 Dec 13 F A 13 <1:16 <1:16 1:4096 9 Dec 13 F A+C 1 <1:16 1:256 <1:16 16 Dec Adult F A 1 1:16 1:1024 1:128 20 Dec 49 F F 1 <1:16 1:512 1:64 26 Dec Adult F F 1 <1:16 1:512 1:16 Axillary (A), cervical (C), femoral (F), inguinal (I). b Day of admission = day 1. Table 5. IgM and IgG antibody titres versus days in hospital, Ovambo 1982 No. of cases Days in Group hospital IgM>IgG IgG>IgM Case No. Seroconversion in hospital 5-9 8 0 5, 15, 22, 23, 25, 27, 30, 32 12-16 2 1 8,18,20 26-29 0 2 5, 21 Time of seroconversion 2-13 1 4 1,13,24,31,38 not established" 17-18 0 2 13,38 ' Seroconversion possibly occurred prior to hospitalization. Table 6. Concentrations of Fl antigen in the blood of plague patients in Ovambo, 1982 Fl antigen concentration (ng/ml) Day in Case No.' hospital" Serum Clot 17 1 64 32 19 1 256 Notest 26 1 64 16 28 1 128 <16 34c 1 Notest 16 35 1 1024 1024 37 1 16 < 16 a Case descriptions are in Table 1. b Day of admission = day 1. c Y. pestis was isolated from the blood. from plague (Tables 2 and 3). PHA and IgM anti- bodies appeared shortly before IgG antibody, with the IgM titre exceeding the PHA titre. Early in con- valescence, all three types of antibody usually were present, with the IgM titre higher than the IgG titre. Patients who experienced seroconversion while in hospital displayed these patterns from days five to nine (Table 5). Later, IgM and PHA titres declined, often to a point where a PHA titre was undetectable. In contrast, the IgG titre did not decline and con- tinued, in some cases, to increase. Sera collected from patients who had been in the hospital for 17 or more days always had an IgG titre that exceeded the IgM and PHA titres. Fl antigen was found in the blood of seven patients, including one from whom Y. pestis was isolated (Table 6; case No. 34). Another patient (case No. 35) had an F1 antigenaemia of 1 tg/ml. Antigen- aemia was found only on the day of admission to hospital. The concentration of Fl measured in the serum generally exceeded that measured in the clot from the same blood sample. All six acute sera that contained a detectable amount of Fl antigen (Table 6) lacked antibody titre when tested by PHA and ELISA. Unfortunately, none of these sera were paired with a convalescent ENZYME IMMUNOASSAYS FOR DIAGNOSIS OF PLAGUE serum, so it was not possible to directly relate antigen- aemia to subsequent antibody responses. However, antigen was not present in any of the 31 sera that had antibody. Also, the data in Table 6 must not be construed as evidence for antigenaemias in the absence of bacteraemias. Of the patients listed, a blood culture for isolation was prepared only for case No. 34, and Y. pestis was isolated from it. The data indicate that antigenaemia was occasionally present when bacteraemia was not suspected. The acute sera and corresponding clots from seven patients with bacteraemia (Table 1; cases No. 1, 8, 9, 10, 30, 32 and 33) were tested for Fl antigen. All were negative, proving that absence of a detectable anti- genaemia does not signify that bacteraemia is totally absent. DISCUSSION This is the first study in which ELISA for the detection of both plague antigen and antibodies has been applied with traditional laboratory diagnostic techniques to define a plague outbreak. As more refined technologies are utilized, greater success in diagnosing diseases is to be expected. Indeed, that is what occurred in the present investigation. Using only bacteriological isolation and PHA, 31 cases would have been confirmed as plague in the 1982 plague outbreak in Namibia. The addition of ELISA resulted in another seven confirmations, or a 23% increase in confirmed diagnoses without any im- provement in specimen acquisition. Most of the increase was from application of the ELISA to detect antigen in cases where cultural isolation was not attempted; the rates for the detection of antibody were very similar using ELISA and PHA. PHA is the procedure currently recommended by WHO for the serodiagnosis of plague (5, 6). The results from this study support the view that the PHA test is a specific and useful test for the detection of plague antibody, and it remains a simpler and less expensive procedure than ELISA. However, several advantages of ELISA over the PHA test for measuring antibodies are defined and illustrated by our investigation. While PHA responds more like IgM than IgG, PHA titres are typically lower and more transient. Thus, the probability for success in detecting antibody in a patient can be greater with ELISA, especially in late convalescence. In addition, the relationship between IgM and IgG titres, as deter- mined by ELISA, could suggest whether a patient was in early or late convalescence; PHA titres cannot do this. In a serological survey, where paired sera may not be obtained, such information is very valuable. An absence of sera illustrating patterns characteristic of early convalescence would suggest that the poten- tial for new infections has subsided in an area, and vice versa. Applying this logic to our study, it appears that four patients, each with a single serum dating between 9 and 26 December, were in early conval- escence, while one patient with a serum dated 8 December either was in late convalescence or had been infected previously (Table 4). Useful as the ELISA may be, a clear definition of early or late con- valescence will not be possible for every seropositive patient. Six cases from the 1982 outbreak were confirmed by using the ELISA to detect antigen when other serological methods were ineffective. The data from these patients support earlier work (2) suggesting that Fl antigen and F1 antibody are never found together in the blood, and that antigenaemias in the micro- grams-per-ml range occur in plague. The ELISA for antigen supplements bacteriological isolation for identifying the disease while in the acute phase, when signs and symptoms are nonspecific. Unlike iso- lation, which is too slow to benefit patients or their contacts, the assay for antigen can be done rapidly for information to assist in determining appropriate therapy and courses of action. On the other hand, blood culture can be more sensitive than ELISA for retrospective diagnosis. During a low-grade bac- teraemia, antigenaemia may be undetectable or insig- nificant (i.e., < 16 ng/ml of Fl antigen), if present at all, whereas only a few Y. pestis organisms present in the blood will multiply in culture. Indeed, failure of the ELISA to detect antigen in the blood of seven bacteraemic patients may have been because the bacteraemias were of such low levels. The cases of plague described in this paper were not reported previously in the Weekly epidemiological record (9) but accurately reflect the incidence of human plague in northern Namibia late in 1982. ACKNOWLEDGEMENTS The authors thank Dr 0. van Niekerk, the then Secretary for Health and Welfare, Government Service of Ovambo; Dr S. S. Grove', Regional Director, South African Institute for Medical Research (SAIMR), Namibia; and the physicians and staff of the Oshakati and Onandjokwe Hospitals for their interest and collaboration in these investigations. The trans- port of clinical specimens was accomplished with the kind assistance of the South African Medical Service under direction of Surgeon General N. J. Nieuwoudt. Mr Vincent Palmer assisted in the laboratory. 752 J. E. WILLIAMS ET AL. RtSUMt APPLICATION DE METHODES IMMUNOENZYMATIQUES POUR LA CONFIRMATION DU DIAGNOSTIC DE PESTE DANS DES CAS CLINIQUEMENT SUSPECTS EN NAMIBIE, EN 1982 Une 6pidemie de peste est survenue en Ovamboland, au nord de la Namibie, entre octobre et decembre 1982. Des hemocultures, des serums et des caillots sanguins ont e recueillis dans des hopitaux r6gionaux et transportes a Johannesburg (Afrique du Sud) pour examen. L'isolement de Yersinia pestis a partir d'h6mocultures, la mise en 6vi- dence d'anticorps par des epreuves d'h6magglutination passive, le titrage avec immunoadsorbant lie a une enzyme (ELISA) en vue de determiner les titres des anticorps IgM et IgG, ainsi que de la methode ELISA pour determiner les concentrations de l'antigene Fl sp6cifique de Y. pestis dans les serums et les caillots ont ete employ6s en vue d'obtenir au laboratoire la confirmation des cas cliniquement suspects. Toutes les epreuves ELISA reposaient sur l'emploi d'un anticorps monoclonal r6agissant specifiquement avec l'anti- gene Fl. La methode d'inhibition d'ELISA a e utilis6e pour d6finir les valeurs limites des epreuves et confirmer la sp6cificit6. Trente-huit cas ont W confirm6s dont 50% par isolement bact6riologique, 34% d'apres les r6ponses en anticorps et 16% par d6tection de l'antig6n6mie. Les 6preuves d'Ouch- terlony ont montre que toutes les souches de Y. pestis isol6es 6taient capables de produire I'antigene Fl et, dans les cas confirm6s bact6riologiquement, 1'examen de paires de s6rum a revele de notables r6ponses en anticorps. Les r6sultats de cette etude viennent a I'appui de l'opinion selon laquelle l'epreuve d'h6magglutination passive (HAP) est un test specifique et utile pour le diagnostic serologique de la peste; toutefois les avantages de la methode ELISA sur 1'epreuve d'hemagglutination passive en vue de mesurer les anticorps ont ete d6finis et illustr6s. L'6volution des titres en hemagglutination passive 6tait plus proche de celle des IgM que de celle des IgG, mais les titres en HAP &taient plus faibles et plus transitoires. Ainsi, la d6tection de l'anticorps etait plus fr6quente avec ELISA, en particulier vers la fin de la convalescence. En outre, la relation entre les titres d'IgM et d'IgG pourrait indiquer si un malade se trouve au debut ou a la fin de la convalescence. Les s6rums du debut de la convalescence contiennent generalement des anticorps HAP, IgM et IgG titrables, le titre des IgM etant le plus eleve. Dans les serums de la fin de la convalescence, les titres des IgG etaient superieurs aux titres des IgM ou HAP et ce dernier n'etait plus decelable chez certains malades. Ces caracteristiques des anticorps antipesteux pourraient avoir des applications pratiques pour les enquetes serologiques et permettre de diff6rencier les infections recentes d'infections plus anciennes par des epreuves sur des serums uniques. Dans l'epidemie de 1982, 6 cas ont ete confirmes par detection de l'antigene au moyen de l'epreuve ELISA, alors que les autres methodes serologiques etaient inefficaces; un autre malade presentait une bacteriemie et une antigen6mie. Les donn&es obtenues a partir de ces cas viennent confirmer les resultats d'etudes anterieures, a savoir que l'antigene Fl et l'anticorps anti-Fl ne se trouvent jamais en meme temps dans le sang et qu'on observe dans la peste des antigenemies de l'ordre du microgramme par ml. Dans le sang d'un malade, la concentration d'antigenes Fl etait de 1 Fg/ml. La detection de l'antigene par la m6thode ELISA completait l'isolement bact6riologique pour l'identification de la maladie au cours de la phase aigue. L'hemoculture s'est revel6e plus sensible que la m6thode ELISA pour le diagnos- tic retrospectif de sept cas qui pr6sentaient une bacteriemie en l'absence d'antigen6mie. L'application des methodes ELISA a augmente de 23% les confirmations en laboratoire par rapport au nombre qui aurait ete obtenu par hemoculture et epreuve HAP. L'isole- ment et la r6action d'hemagglutination passive ont permis de confirmer le diagnostic de peste dans 31 cas; la methode ELISA a permis d'en confirmer sept de plus. REFERENCES 1. WILLIAMS, J. E. ET AL. Comparison of passive haemagglutination and enzyme-linked immunosorbent assay for serodiagnosis of plague. Bulletin ofthe World Health Organization, 60: 777-781 (1982). 2. WILLIAMS, J. E. ET AL. Use of an enzyme-linked immunosorbent assay to measure antigenaemia during acute plague. Bulletin of the World Health Organ- ization, 62: 463466 (1984). 3. SHEPHERD, A. J. ET AL. A comparison of serological techniques for plague surveillance. Transactions of the Royal Society of Tropical Medicine and Hygiene, 78: 771-773 (1984). 4. ISAACSON, M. & JAMISON, J. R. Epidemiological and clinical aspects of plague in Southern Africa. In: Gear, J. H. S., ed. Proceedings of the International Sym- posium on Medicine in a Tropical Environment, South Africa, 1976. Rotterdam, A. A. Balkema, 1977, pp 479-486. 5. BAHMANYAR, M. & CAVANAUGH, D. C. Plague manual. Geneva, World Health Organization, 1976. 6. WHO Technical Report Series, No. 447, 1970 (WHO Expert Committee on Plague: fourth report). 7. WILLIAMS, J. E. & ROBINSON, D. M. Requirement to confirm the specificity of ELISA reactions. Trans- actions of the Royal Society of Tropical Medicine and Hygiene, 76: 280-281 (1982). 8. BARNES, A. M. & POLAND, J. D. Plague in the United States, 1983. Morbidity and mortality weekly report, 33: 15SS-21SS (1984). 9. Human plague in 1984. Weekly epidemiological record, 60: (39): 297-298 (1985).