Vol. 32, No. 4JOURNAL OF CLINICAL MICROBIOLOGY, Apr. 1994, p. 964-9700095-1137/94/$04.00+0Copyright C) 1994, American Society for Microbiology

Diagnosis of Amebic Dysentery by Detection of Entamoebahistolytica Fecal Antigen by an Invasive Strain-Specific,

Monoclonal Antibody-Based Enzyme-LinkedImmunosorbent Assay

ARMANDO GONZALEZ-RUIZ,l.2* RASHIDUL HAQUE,3 TAYAB REHMAN,lt AURA AGUIRRE,4ANDREW HALL,3t FELIPE GUHL,4 DAVID C. WARHURST,1 AND MICHAEL A. MILES1

Department of Medical Parasitology, London School of Hygiene and Tropical Medicine, London WC1E 7HT, 1 andDepartment of Clinical Microbiology, University College London Hospitals, London WC1E 6AU, United Kingdom;

International Centre for Diarrheal Diseases Research, Bangladesh, Dhaka 1,000, Bangladesh3; andLaboratorio de Parasitologia y Microbiologia, Departamento de Ciencias Biol6gicas,

Universidad de los Andes, Santa Fe de Bogota, Colombia4

Received 20 September 1993/Returned for modification 28 November 1993/Accepted 17 January 1994

An invasive strain-specific monoclonal antibody against Entamoeba histolytica has been used in a captureenzyme-linked immunosorbent assay (ELISA) for the detection of invasive E. histolytica fecal antigen in clinicalspecimens and for the diagnosis of amebic dysentery in patients from Bangladesh. The fecal antigen captureELISA (FAC-ELISA) did not cross-react with other parasite species in the clinical specimens or withnoninvasive E. histolytica present in those specimens and in experimentally seeded stools. The limit of detectionof the assay for invasive E. histolytica crude antigen diluted in phosphate-buffered saline or in stools was 0.58and 3.9 ,ug/ml, respectively, which is the equivalent of approximately 72 and 487 E. histolytica trophozoites perwell, respectively. The sensitivity, specificity, and eficiency of the FAC-ELISA were 87, 100, and 98%,respectively, for the detection of invasive E. histolytica antigens and 100, 100, and 100%, respectively, for thediagnosis of amebic dysentery. The FAC-ELISA is a potential alternative for the field diagnosis of amebicdysentery and for epidemiological studies to define the distribution of invasive E. histolytica.

The protozoan parasite Entamoeba histolytica is the etiolog-ical agent of human amebiasis. Approximately 500 millionpeople are infected worldwide with E. histolytica, but only 10%develop invasive amebiasis (38). Like other diarrheal diseases,amebiasis is more prevalent in developing countries and pro-duces between 40,000 and 100,000 deaths per year (40). Thediscrepancy between the number of carriers and the relativelylow percentage of patients developing invasive amebiasis couldin part be explained by higher prevalence of infection withnoninvasive strains of the parasite, which are morphologicallyindistinguishable but genotypically distinct from invasivestrains (3, 4, 6). Failure to identify nonpathogenic amebae suchas Entamoeba coli and Entamoeba hartmanni may also increasethe apparent prevalence of asymptomatic E. histolytica infec-tion (17, 22, 29).

Microscopical examination of stools does not differentiatebetween invasive and noninvasive strains of E. histolytica, istime-consuming, and requires a skillful microscopist for thedetection and recognition of E. histolytica trophozoites withingested erythrocytes (22, 38). Differential diagnosis of amebicdysentery (AD) at present requires facilities for isolatingbacterial agents such as Shigella spp., Salmonella spp., Campy-

* Corresponding author. Mailing address: Department of MedicalParasitology, London School of Hygiene and Tropical Medicine,Keppel Street, London WC1E 7HT, United Kingdom. Phone: (71) 9272340. Fax: (71) 636 8739.

t Present address: Biomedical and Genetic Engineering Division,Dr. A. Q. Khan Research Laboratories, I & T Centre, G-9/1, Islam-abad, Pakistan.

t Present address: Department of Pure and Applied Biology, Impe-rial College of Science Technology and Medicine, London SW7 2BB,United Kingdom.

964

lobacter spp., or Yersinia spp. or for the recognition of clinicalconditions such as rectal infection by Chlamydia trachomatis,Treponema pallidum, and Neisseria gonorrhoeae (19). AD mustalso be differentiated from idiopathic intestinal inflammatorydiseases, namely, Crohn's disease and ulcerative colitis (28).The development of fecal antigen detection assays has been

designated a priority in amebiasis research by the WorldHealth Organization (40): such assays detect only currentinfections and can be very useful in epidemiological studies.The adaptation of an invasive strain-specific monoclonal anti-body (MAb) to a fecal antigen capture enzyme-linked immu-nosorbent assay (FAC-ELISA) would enhance the diagnosis ofintestinal amebiasis and differentiate that diagnosis from otherintestinal infections such as bacillary dysentery in carriers ofnoninvasive E. histolytica.We have produced an invasive strain-specific MAb (20/7D)

against E. histolytica which has been used in an indirectimmunofluorescence test to characterize E. histolytica isolatesfrom several geographical regions (15). Here we report theapplication of this MAb in a FAC-ELISA for the detection ofE. histolytica fecal antigen and for the diagnosis of AD inpatients from Bangladesh.

MATERIALS AND METHODSProduction of E. histolytica NP-40 protein extract. E. histo-

lytica Nonidet P-40 (NP-40) protein extract was produced fromthree axenic E. histolytica strains, HM-1:IMSS, formerly knownas ABRM (7); HK9 (13); and NIH:200 (32) (ATCC 30015,ATCC 30458, and ATCC 30459, respectively), which wereoriginally isolated from patients with invasive intestinal ame-biasis, and from E. histolytica polyxenic isolates cultured fromcases of symptomatic and asymptomatic amebiasis. The axenic

on January 21, 2021 by guesthttp://jcm

.asm.org/

Dow

nloaded from

FECAL ANTIGEN DETECTION FOR AMEBIC DYSENTERY 965

strains were cultured in Diamond's TPS-1 medium (9) at 37°C.The polyxenic isolates were initially isolated and maintained inbiphasic Robinson's medium at 37°C, and some were alsomass-cultured in liquid Robinson's medium (30) at 37°C. E.histolytica trophozoites were harvested from Diamond's andliquid Robinson's media by chilling the culture tubes inice-water for 15 min, washed three times in cold phosphate-buffered saline (pH 7.2) (PBS), and centrifuged at 250 x g for5 min at 4°C. The pelleted cells were lysed by the addition of3 volumes of lysis buffer (20 mM Tris-HCl [pH 8.0], 150 mMNaCl, 0.75% [wt/vol] NP-40 [BDH Chemicals, Poole, Dorset,United Kingdom]) containing protease inhibitors (1 mM phe-nylmethylsulfonyl fluoride [Sigma Chemical Co., Poole, Dor-set, United Kingdom], 1 mM iodacetamide [BDH]), and 1 ,ugof tosyl-L-lysine chloromethyl ketone per ml [Sigma]) withcontinuous stirring on ice for 15 min. The mixture wascentrifuged at 16,000 x g for 15 min at 4°C, and the superna-tant was recovered and stored at - 20°C. The protein contentof this supernatant and other preparations was determinedwith the bicinchoninic acid protein assay reagent (Pierce,Rockford, Ill.) with bovine serum albumin as standard.

Production of anti-E. histolytica hyperimmune serum. Twomale Half Lop rabbits were immunized with a mixed crude cellextract of trophozoites of three E. histolytica axenic strains(HM-1:IMSS, HK9, and NIH:200) as follows: trophozoiteswere harvested from Diamond's medium (above), they weredisrupted by three sequential cycles of freezing and thawing inliquid N2, and their protein content was determined (above).Rabbits were immunized with 50 ,g of extract emulsified 1:1with Freund's complete adjuvant, half of the volume beinginjected subcutaneously and half intramuscularly. Rabbits wereboosted every 6 to 8 weeks with extract 1:1 in Freund'sincomplete adjuvant. Antibody response was assayed beforeeach boost by indirect ELISA (36) against NP-40 proteinextract of E. histolytica. Rabbits were bled after the secondboost.

Affinity purification of rabbit antisera. A 10-ml affinity-chromatography column was prepared with E. histolyticaNP-40 protein extract dialyzed at 4°C against two changes ofcoupling buffer (100 mM sodium bicarbonate-0.5 M sodiumchloride [pH 8.3]) and coupled to cyanogen bromide-activatedSepharose 4B at a concentration of 5 mg of protein per ml ofgel (Pharmacia Fine Chemicals, Uppsala, Sweden), accordingto the manufacturer's instructions. The column was equili-brated with PBS and preeluted with 0.1 M glycine (pH 2.5),and after reequilibration with PBS, 5 ml of ammonium sulfate-precipitated hyperimmune rabbit serum was loaded onto thecolumn and cycled overnight at 4°C. After extensive washingwith PBS, bound immunoglobulin was eluted with 0.1 Mglycine (pH 2.5), and 3-ml fractions were immediately neutral-ized with 250 RI of 2 M Tris-HCl (pH 8). Fractions containingthe highest antibody activity, as assayed by indirect ELISA(36), were pooled, dialyzed overnight at 4°C against PBS, andconcentrated by means of negative-pressure ultrafiltration inan Amicon Ultrafiltration Cell, type YM 10 (10-kDa molecularmass cutoff; Amicon, Danvers, Mass.), and the protein contentof the concentrate was assayed. Concentrated antibodies werealiquoted and stored at - 20°C.

Stool samples. Stool samples were collected at the Interna-tional Centre for Diarrheal Diseases Research, Bangladesh(ICDDR,B), in Dhaka, from patients of all age groups seen atthe hospital and from children less than 14 years old living ina slum in Dhaka. Recently collected stools were examinedmacroscopically for the presence of blood and mucus, and asmear of feces in 0.9% saline was examined microscopically forthe presence of erythrocytes, leukocytes, and E. histolytica

trophozoites with or without ingested erythrocytes. Feces wereconcentrated by the formalin-ether concentration technique(23), and the concentrate was examined microscopically forcysts and ova. Feces were also inoculated into Robinson'smedium within 6 h of collection, and E. histolytica-positivecultures were subcultured every 48 h.

Definition of diarrheal disease. Stool samples were classifiedinto several diagnostic categories according to microscopicalexamination and detection of E. histolytica by Robinson'sculture which was characterized as invasive or noninvasive byzymodeme analyses (14a). Briefly, AD was defined as bloodydiarrhea (DIA) with erythrophagocytic E. histolytica trophozo-ites seen by microscopy, a finding which is considered thediagnostic hallmark of the disease (37, 39); dysentery ofunknown etiology was defined as bloody DIA without evidenceof invasive amebic infection; DIA was defined as nondysentericloose stools with or without invasive E. histolytica infection;and asymptomatic carrier (AC) was defined as infection withE. histolytica detected in a patient with formed stools.Zymodeme identification of E. histolytica isolates. The zy-

modeme of the E. histolytica isolates was determined locally.At the ICDDR,B, this was done by cellulose acetate electro-phoresis plates (Zip Zone Chamber and Titan Plus Electro-phoresis Power Supply; Helena Laboratories, Beaumont, Tex.)as described previously (20); at Universidad de los Andes(UNIANDES), thin-layer starch gel electrophoresis was usedaccording to the methods of Sargeaunt and Williams (25) andSargeaunt et al. (26) for phosphoglucomutase (EC 2.7.5.1),L-malate:NADP+ oxidoreductase (oxaloacetate decarboxylat-ing) (EC 1.1.1.4.0), and glucose phosphate isomerase (EC5.3.1.9), and according to Farri et al. (10) for hexokinase (EC2.7.1.1). Briefly, E. histolytica trophozoites were harvestedfrom 2-day-old Robinson's cultures and washed twice withsterile 0.9% saline, and lysates were prepared by resuspendingpelleted cells in distilled water containing enzyme stabilizers (1mM EDTA, 1 mM dithiothreitol, and 1 mM Y.-aminocaproicacid [Sigma]) and by three sequential cycles of freezing andthawing. Lysates were centrifuged at 14,000 x g for 1 h at 4°C,and supernatants were aliquoted and stored in liquid N2.Isoenzyme profiles were visualized with an agar overlay (10,27).FAC-ELISA. Three sample sets were assayed for E. histo-

lytica antigen. Two were experimentally seeded (spiked) witheither E. histolytica NP-40 extract (at London School ofHygiene and Tropical Medicine [LSHTM]) or with intactpolyxenic E. histolytica trophozoites (at ICDDR,B), and oneset consisted of clinical stool samples from patients in Bang-ladesh. Spiked samples were prepared as follows: E. histolyticaNP-40 extract from several isolates was diluted at 50 ,ug/ml inPBS and stored at - 20°C. E. histolytica isolates were harvested(200 ill, at ICDDR,B) from 2-day-old biphasic Robinson'scultures yielding -2 x 103 trophozoites (24), mixed with 0.5ml of parasite-free feces (final protein, -2.3 ,ug/ml), frozen,and kept at - 20°C until assayed at LSHTM. Estimates of theFAC-ELISA detection limit were made with twofold serialdilutions of E. histolytica NIH:200 NP-40 extract in PBS andparasite-free stools at a starting concentration of 50 ,ug/ml.Aliquots of dilutions (100 [lI) were assayed by FAC-ELISA(below). To estimate the number of trophozoites per 100 RI, itwas assumed that 106 were equivalent to -0.8 mg of E.histolytica protein (8).

Clinical stool samples collected and examined at ICDDR,Bwere kept at -20°C and transported frozen to the LSHTM,where they were mixed 1:3 with fecal diluent (PBS-0.02%NaN3-0.3% Tween 20-50% fetal calf serum). Fecal superna-tants were prepared by freezing and thawing the fecal suspen-

VOL. 32, 1994

on January 21, 2021 by guesthttp://jcm

.asm.org/

Dow

nloaded from

966 GONZALEZ-RUIZ ET AL.

sion three times in liquid N2 and by centrifugation at 8,500 xg fdt 30 min and stored at - 20°C until assayed blindly atLSHTM.

Alternate rows of flat-bottomed polystyrene microplates(Immulon II; Dynatech Laboratories Ltd., Billingshurst, Sus-sex, United Kingdom) were coated overnight at 40C, eitherwith affinity-purified polyclonal rabbit antisera raised againstE. histolytica trophozoites or with nonimmune rabbit immuno-globulin G (Jackson Immunoresearch Inc., Philadelphia, Pa.),at 1 ,ug of protein per well in coating buffer (7.5 mMNa2CO3-17.4 mM NaHCO3-0.02% NaN3, pH 9.6). All subse-quent washes were with PBS-0.05% Tween 20. Workingdilutions of reagents were determined by checkerboard titra-tion. After three washings, plates were blocked (coating buff-er-2% casein) for 1 h at 37°C and again washed three times,and 100 ,ul of fetal calf serum was dispensed into all wells,followed by either 100 p.l of spiked PBS-parasite-free stools or100 ,ul of fecal supernatant dispensed into immune andnonimmune rabbit immunoglobulin G-coated wells. After in-cubation for 2 h at room temperature, plates were washed fivetimes and 100 ,ul of a 1/1,500 dilution of mouse ascitescontaining invasive strain-specific MAb 20/7D (15) was dis-pensed in PBS-Tween 20-2% casein. Plates were reincubatedfor 2 h at 37°C and washed five times, and 100 p.l of 1/10,000peroxidase-conjugated rabbit anti-mouse immunoglobulins(Jackson Immunoresearch Inc.) in PBS-Tween 20-2% caseinadded to each well. After further incubation for 1 h at 37°C,plates were washed five times and 100 p.1 of substrate solution(0.04% orthophenylenediamine [Sigma] in phosphate-citratebuffer, pH 5, with 0.012% H202) was added to all wells. Thechromogenic reaction was stopped after 30 min of incubationin the dark at room temperature with 50 p.1 of 2.5 M H2SO4 perwell.

Interpretation of results and statistical analysis. Opticaldensities (ODs) were measured by spectrophotometry (492-nmDynatech plate reader). To allow for inter-ELISA plate vari-ation, OD values were normalized by using the ODs on eachplate of positive controls consisting of E. histolytica NP-40extract at 50 jig/ml in PBS. Negative controls (PBS and stools;conjugate and no second antibody) were also included on eachplate. Assay cutoff value was the average of ODs obtained withnonimmune rabbit immunoglobulin G plus three standarddeviations. After subtracting the OD of the nonimmune well,which represents nonspecific binding, from the OD of theimmune duplicate, samples with ODs of >0.035 were consid-ered positive by FAC-ELISA.

Predictive value of a positive and a negative assay and assayefficiency were calculated according to Vecchio (34) and Galenand Gambino (11), respectively. Statistical analysis of ODswith each group of stool samples was by the unpaired t test, andthat of demographic characteristics of patients was by the

TABLE 1. Specificity of MAb 20/7D in the FAC-ELISA withspiked stool samples'

No. of isolates and zymodemebFAC-ELISA Total

I II XIV

Positive 0 11 1 12Negative 10 4 1 15Total 10 15 2 27

a Stools spiked with either 50 p.g/ml of E. histolytica NP-40 extract per ml atLSHTM or with approximately 400 intact trophozoites (approximately 2.3 ,ug ofprotein per ml) at ICDDR,B.

b I, noninvasive zymodeme; II and XIV, invasive zymodemes.

TABLE 2. Distribution of results obtained in the FAC-ELISA withstool samples spiked with different amounts of E. histolytica protein

Antigen aldded No. of true No. of true No. of false No. of false Total(iLg/ml)a positives negatives positives negatives no.

2.3 9 6 0 4 1950.0 3 4 0 1 8Total 12 10 0 5 27

a Stools spiked with either 50.0 pg of E. histolytica NP-40 protein extract perml at LSHTM or approximately 2.3 ,ug/ml in the form of approximately 400 E.histolytica intact trophozoites at ICDDR,B.

unpaired t test and the chi-square test with the NANOSTATpackage developed at LSHTM.

RESULTS

MAb 20/7D is invasive strain specific in the FAC-ELISA.MAb 20/7D, an invasive strain-specific MAb against E. histo-lytica trophozoites (15), was adapted to the FAC-ELISA. Itsspecificity in the FAC-ELISA was tested by assaying a series ofPBS and parasite-free stools spiked with E. histolytica NP-40extract or E. histolytica trophozoites (Table 1). The FAC-ELISA was negative with the 10 samples spiked with noninva-sive E. histolytica antigen and positive with 12 of 17 samplesspiked with invasive E. histolytica antigen. Although four of fivefalse-negative samples were in the group of samples spikedwith the lowest amount of E. histolytica protein, there was aconsiderable number of true positives in this sample group(Table 2).

Detection limit of the E. histolytica FAC-ELISA. The detec-tion limit of the E. histolytica FAC-ELISA was estimated byassaying serial dilutions of E. histolytica protein in PBS andparasite-free stools. The ELISA could detect 0.58 jig of E.histolytica protein per ml in PBS (Fig. 1) and 3.9 ,ug/ml inparasite-free stools, equivalent to approximately 72 and 487 E.histolytica trophozoites, respectively, per well (100 ,ul).

Detection ofE. histolytica antigen in clinical stool samples byFAC-ELISA. Performance of the FAC-ELISA with clinical

O.D. 492 nm

-StoolsA PBS- - Cut-off point

-'----___----_--v.v50 25 12.5 6.25 3.12 1.5 0.78 0.39 0.19 0.097

E. histolytica NP40 protein extract (pg/ml)FIG. 1. Limit of detection of FAC-ELISA with twofold dilutions of

invasive E. histolytica antigen in PBS and in parasitologically negativestools.

J. CLIN. MICROBIOL.

on January 21, 2021 by guesthttp://jcm

.asm.org/

Dow

nloaded from

FECAL ANTIGEN DETECTION FOR AMEBIC DYSENTERY 967

TABLE 3. Intestinal parasites found in stool specimens with orwithout E. histolytica infection

No. of episodese (%)bParasite E. histolytica E. histolytica

present absent

Pentatrichomonas hominis 5 (10) 5 (8)Blastocystis hominis 3 (6) 7 (11)Entamoeba coli 4 (8) 1(1)Giardia intestinalis 0 (0) 5 (8)Endolimax nana 2 (4) 2 (3)Enteromonas hominis 0 (0) 2 (3)Chilomastix mesnili 1 (2) 0 (0)Ascaris lumbricoides 12 (25) 22 (36)Trichuris trichiura 9 (19) 12 (19)Hookworm sp. 2 (4) 5 (8)Strongyloides stercoralis 3 (6) 2 (3)Total 41 63

a One episode of parasitic infection is the presence of the parasite in a singlestool sample.

b Percentages were calculated from 47 stool samples with or without E.histolytica infection. In some stool samples, two or more parasites were presentsimultaneously.

samples was evaluated with stools collected at the ICDDR,B.The stool samples were classified on the basis of results ofmicroscopical examination, Robinson's culture, and zymo-deme analysis. The FAC-ELISA was first tested with three

A1.1.

1.0.

0.9s

0.8

0.7-

0.6-

0.5-

0.4-

0.3-

02-

0.1-

0.0-

sample groups that differed in the microscopy result: 47 carriedE. histolytica either alone (25 samples) or with other parasites(22 samples), 33 contained other intestinal parasites but no E.histolytica (Table 3), and 28 had no parasitic infection. Therewere no significant differences in age and sex distribution ofpatients with and without parasitic infections (not shown).Among other parasitic protozoa found by microscopy, the mostcommon were the commensal flagellate Pentatrichomonashominis and Blastocystis hominis, which is associated with DIA(1). Ascaris lumbricoides and Trichuris trichiura, both recog-nized pathogens (2), were the most common helminths. A totalof 15 of the E. histolytica samples contained invasive strains,and 32 contained noninvasive strains according to zymodemeanalysis of cultured isolates. The FAC-ELISA detected E.histolytica antigen in 13 of 15 samples containing invasive E.histolytica and none of 93 samples from the other three groups(Fig. 2A). Sensitivity and specificity were accordingly 87 and100%, respectively. The FAC-ELISA was secondly tested withstool samples classified by diagnosis of diarrheal disease (14a).In this case, stool samples were separated into four groups: AD(9 samples), dysentery of unknown etiology (18 samples), DIA(68 samples), and asymptomatic carriers (AC) (7 samples).Samples in the DIA (five samples) and AC (one sample)groups which were included in the first test and known to befrom patients infected with invasive E. histolytica were ex-cluded from this analysis (see Discussion). The FAC-ELISAdetected all AD cases but was negative with dysentery ofunknown etiology cases and the remaining 77 samples in the

B1.1.

1.0.0S

0.9-

0.8.

0.7.

0.6

0.5

0.4

0.3-eS

02.

0.1-

0.0-

-0.1-0m uuu . aaaiia A B C D

-.1-IFIG. 2. Distribution of FAC-ELISA ODs obtained with stool samples infected with invasive E. histolytica (open circles), noninvasive E.

histolytica (closed circles), or other parasites (closed squares) and samples without parasitic infection (closed triangles). (A) Stool samples groupedaccording to parasitological diagnosis. (B) Stool samples grouped according to diagnosis of diarrheal disease: A, AD; B, dysentery of unknownetiology; C, DIA; D, AC.

VOL. 32, 1994

a0

OG

a

0

0

AAAA&O"AAAA"owes

on January 21, 2021 by guesthttp://jcm

.asm.org/

Dow

nloaded from

968 GONZALEZ-RUIZ ET AL.

TABLE 4. Mean ODs obtained by FAC-ELISA with different typesof parasitic infection and diagnosis of diarrheal disease

Type of infection or diagnosis Mean ODa P valueb

Invasive E. histolytica 0.51Noninvasive E. histolytica -0.00009 <0.0001Other parasites -0.0023 <0.0001No parasites -0.002 <0.0001

AD 0.635Dysentery of unknown etiology - 0.0036 <0.0001DIA -0.0009 <0.0001AC -0.001 <0.001

a ODs from wells coated with nonimmune rabbit antisera were subtractedfrom ODs obtained with wells coated with rabbit antisera against E. histolytica,resulting in negative values in some categories.

b Corresponding unpaired t test against invasive E. histolytica infection or AD.

DIA and AC groups (Fig. 2B). Accordingly, sensitivity andspecificity for AD were both 100% in this second test. Thedifferences between the means of the ODs obtained in theassay from the two assessments were statistically significant(Table 4).On the basis of the observed prevalence of invasive E.

histolytica infection (15 cases of 108 individuals; 14%), thepositive and negative predictive values of the FAC-ELISAwere 100 and 98%, respectively (Table 5). As the populationstudied was not representative of the general population andsince the true prevalence of invasive E. histolytica infectionmay be close to 1% (12), the effect of different theoreticalprevalences on predictive values was calculated (Table 5) (16).In the study population, efficiencies of FAC-ELISA for detect-ing fecal antigens of invasive E. histolytica and diagnosing ADwere 98 and 100%, respectively.

DISCUSSION

We have produced a MAb capable of distinguishing invasivefrom noninvasive strains of E. histolytica (15) and adapted it toa FAC-ELISA. MAb 20/7D showed a 100% specificity forinvasive E. histolytica in the FAC-ELISA with spiked stoolsand with clinical stool specimens containing other intestinalparasites, including noninvasive E. histolytica. The FAC-ELISA was also 100% sensitive and specific for diagnosis ofAD, which suggests a role for specific diagnosis of invasiveintestinal amebiasis in the field, where bacillary dysentery maybe endemic and there is no access to microbiological facilities.World Health Organization recommendations for treatment of

TABLE 5. Predictive values of positive and negative FAC-ELISAresults with various theoretical prevalences of invasive E. histolytica

infection when 1,000 individuals are screened

No. of No. of Positive NegativePrevalence . No. of false Pst(%)Xa infected noninfected teb predictive predictive

subjects subjects negatives value (%) value (%)

14 140 860 18 100 97.13 30 970 4 100 99.81.5 15 985 2 100 99.9

a Prevalence figures estimation. A total of 14% of the total number of samplestested by FAC-ELISA had invasive E. histolytica; the prevalence would be 3% if1/3 of the estimated 10% worldwide E. histolytica infection prevalence issecondary to invasive strains; the prevalence would be 1.5% if 1/2 of the 10%worldwide E. histolytica infection prevalence is E. hartmanni.

b Estimation based on 87% sensitivity obtained with this FAC-ELISA.

AD (41) rely on empirical treatment after failure to respond toantimicrobial agents against bacillary dysentery.The sensitivity of FAC-ELISA in this group of clinical

samples with invasive E. histolytica was 87% (13 of 15). Falsenegatives may be due to antigen degradation during storage,transportation, freezing, and thawing of the stools (33). TheMAb 20/7D epitope might also be masked by local intestinalimmune response. Control stools spiked in Bangladesh and inLondon gave approximately 25% false-negative results byFAC-ELISA (4 of 13 and 1 of 4, respectively). It is evident thatthere is some degree of antigen degradation during sampleprocessing. False negatives could also result from antigenicvariation between invasive E. histolytica strains, although this isunlikely as MAb 20/7D has recognized isolates from at leastthree diverse regions of endemicity (15).The limit of detection of the FAC-ELISA was lower with

stools than with PBS (Fig. 1). This may be due to backgroundin stools which decreases as stools are diluted (Fig. 1) or bystool interference with the FAC-ELISA. Part of this interfer-ence may be the proteolytic activity of feces, which are knownto degrade antibodies (35). Although fixation of stool sampleswith formalin has increased the performance of antigen detec-tion tests for Giardia intestinalis (14, 31), previous antigendetection assays for E. histolytica have produced unsatisfactoryresults with formalinized stool samples (21, 33).The detection limit of this FAC-ELISA is similar to that of

other E. histolytica antigen detection assays, and detection limitin stools even as low as 2.3 p.g of soluble E. histolytica proteinper ml could be below levels of microscopical detection.Detection of Giardia intestinalis antigen in the absence ofmicroscopically detectable organisms in feces is possible sev-eral days after treatment (18).The cutoff value of the FAC-ELISA was calculated with the

OD from all nonimmune wells in the microplate. This avoidsthe need for a group of parasitologically negative stools whichmay be difficult to obtain in regions of endemicity.As a crucial part of our study, we wanted to compare the

performance of the FAC-ELISA with a clearly defined groupofAD cases and other groups of stool samples without invasiveE. histolytica. So, six cases of known invasive E. histolyticainfection which did not fulfill the criteria for being classified asdysentery were excluded from the analysis of the FAC-ELISAagainst diagnosis of diarrheal disease (five in the DIA groupand one in the AC group; see Materials and Methods). Shouldthose cases not have been excluded, invasive E. histolyticaculture-positive stools in diagnostic categories other than ADwould have been labelled as false positives or false negatives.Furthermore, all the cases with invasive E. histolytica infection,regardless of diagnosis, were pooled in one group, and theresults obtained were compared with groups of samples con-taining noninvasive E. histolytica, other parasites, and noparasitic infection. Those results showed that with some non-dysenteric stools the FAC-ELISA gave false negatives butmaintained a high specificity for invasive E. histolytica.

In conclusion, the FAC-ELISA is a potential alternative tothe microscopical diagnosis of AD, although further assaydevelopment is required. This assay is more practical thanculture and zymodeme analysis. This rapid FAC-ELISA couldimprove clinical management of dysentery as the result isavailable on the same day and for ulcerative colitis and Crohn'sdisease (28), which may be clinically indistinguishable fromintestinal amebiasis. In regions where E. histolytica is notendemic, FAC-ELISA could distinguish carriers of invasiveand noninvasive E. histolytica, particularly for travellers return-ing from high-risk areas. In 1992, there were approximately 900cases of amebic infection in the United Kingdom, of which

J. CLIN. MICROBIOL.

on January 21, 2021 by guesthttp://jcm

.asm.org/

Dow

nloaded from

FECAL ANTIGEN DETECTION FOR AMEBIC DYSENTERY 969

around 60% were acquired abroad (5). An important epide-miological application could be the clarification of the distri-bution of invasive E. histolytica, with the possibility of targetingantiamebic treatment to areas of high endemicity or carriers ofinvasive strains (40), such as professional food handlers, andhousehold contacts of index cases of invasive amebiasis.

ACKNOWLEDGMENTS

A.G.-R. was a recipient of a scholarship from Consejo Nacional deCiencia y Tecnologia (CONACYT), Mexico; A.A. is partially sup-ported by a scholarship from the Sir Patrick Manson Bequest; andD.C.W. is supported by the Public Health Laboratory Service, UnitedKingdom. This work has been funded by the Nestle Nutrition ResearchGrant Programme, grant no. 87/34; a grant from the Commission ofthe European Communities, contract no. STD-2 0206-UK; the BritishCouncil in Colombia and the United Kingdom; the WHO Programmeon Intestinal Parasitic Infections; and Smith Kline Beecham.We are very grateful to the people who provided us with the

following parasite strains: E. histolytica HK9, NIH:200, and HM-1:IMSS, from L. S. Diamond; E. histolytica TE, SI, C29, and 8672, fromJ. P. Ackers; and E. histolytica SAW 1734, from D. Mirelman.

REFERENCES1. Ash, L. R., and T. C. Orihel. 1990. Protozoans, p. 82-88. In L. R.

Ash and T. C. Orihel (ed.), Atlas of human parasitology. AmericanSociety of Clinical Parasitologists, Chicago.

2. Beaver, P. C., R. C. Jung, and E. W. Cupp. 1984. Helminths andhelminthic infections, p. 240-244 and 307-319. In P. C. Beaver,R. C. Jung, and E. W. Cupp (ed.), Clinical parasitology, 9th ed.Lea & Febiger, Philadelphia.

3. Blanc, D. S. 1992. Determination of taxonomic status of patho-genic and nonpathogenic Entamoeba histolytica zymodemes usingisoenzyme analysis. J. Protozool. 39:471-479.

4. Clark, C. G., and L. S. Diamond. 1991. Ribosomal genes of"pathogenic" and "nonpathogenic" Entamoeba histolytica aredistinct. Mol. Biochem. Parasitol. 49:297-302.

5. Communicable Disease Surveillance Centre. 1993. Other gastro-intestinal tract infections, England and Wales: laboratory reports,weeks 92/52-93/01. CDR Weekly 3:11.

6. Cruz-Reyes, J. A., W. M. Spice, T. Rehman, E. Gisborne, and J.Ackers. 1992. Ribosomal DNA sequences in the differentiation ofpathogenic and non-pathogenic isolates of Entamoeba histolytica.Parasitology 104:239-246.

7. De la Torre, M., L. Landa, and B. Sepulveda. 1970. Avances en losmetodos para el cultivo de Entamoeba histolytica. Arch. Invest.Med. 1(Suppl.):S9-S14.

8. Diamond, L. S. 1968. Techniques of axenic cultivation of Entam-oeba histolytica Schaudinn, 1903 and E. histolytica-like amebae. J.Parasitol. 54:1047-1056.

9. Diamond, L. S. 1983. Lumen dwelling protozoa: Entamoeba,trichomonads and Giardia, p. 65-109. In J. B. Jensen (ed.), In vitrocultivation of protozoan parasites. CRC Press, Inc., Boca Raton,Fla.

10. Farri, T. A., P. G. Sargeaunt, D. C. Warhurst, J. E. Williams, andR. Bhojnani. 1980. Electrophoretic studies of the hexokinase ofEntamoeba histolytica groups I to IV. Trans. R. Soc. Trop. Med.Hyg. 74:672-673.

11. Galen, R. S., and S. R. Gambino. 1975. How to determine thepredictive value and efficiency of a test, p. 29-40. In R. S. Galenand S. R. Gambino (ed.), Beyond normality-the predictive valueand efficiency of medical diagnosis. John Wiley & Sons, New York.

12. Gathiram, V., and T. F. H. G. Jackson. 1985. Frequency distribu-tion of Entamoeba histolytica zymodemes in a rural South Africanpopulation. Lancet i:719-721.

13. Geiman, Q. M., and C. E. Becker. 1953. In vitro growth andmetabolism of Endamoeba histolytica. Ann. N.Y. Acad. Sci.56:1048-1056.

14. Goldin, A., E. Apt, X. Aguilera, I. Zulantay, D. C. Warhurst, andM. A. Miles. 1992. A capture ELISA detects Giardia lambliaantigens in formalin-treated faecal samples. Trans. R. Soc. Trop.Med. Hyg. 86:164-165.

14a.Gonzailez-Ruiz, A., R. Haque, A. Aguirre, G. Castafi6n, A. Hall, F.Guhl, G. Ruiz-Palacios, M. A. Miles, and D. C. Warhurst. 1994.Value of microscopy in the diagnosis of dysentery associated withinvasive Entamoeba histolytica. J. Clin. Pathol. 47:236-240.

15. Gonzalez-Ruiz, A., R. Haque, T. Rehman, A. Aguirre, C. Jaramillo,G. Castafi6n, A. Hall, F. Guhl, G. Ruiz-Palacios, D. C. Warhurst,and M. A. Miles. 1992. A monoclonal antibody for distinction ofinvasive and noninvasive clinical isolates of Entamoeba histolytica.J. Clin. Microbiol. 30:2807-2813.

16. Gonzalez-Ruiz, A., M. A. Miles, and D. C. Warhurst. 1993.Predictive value of diagnostic tests and prevalence of invasiveEntamoeba histolytica infection. J. Infect. Dis. 168:513-514.

17. Gonzalez-Ruiz, A., and G. Ruiz-Palacios. 1991. Entamoeba hart-manni: missing or misidentified. J. Infect. Dis. 164:612-613.

18. Green, E. L., M. A. Miles, and D. C. Warhurst. 1985. Immunodi-agnostic detection of Giardia antigen in faeces by a rapid visualenzyme-linked immunosorbent assay. Lancet ii:691-693.

19. Guerrant, R. L. 1990. Inflammatory enteritides, p. 870-880. InG. L. Mandell, R. G. Douglas, and J. E. Bennett (ed.), Principlesand practice of infectious diseases, 3rd ed. Churchill Livingstone,New York.

20. Haque, R., A. Hall, and S. Tzipori. 1990. Zymodemes of Entam-oeba histolytica in Dhaka, Bangladesh. Ann. Trop. Med. Parasitol.84:629-632.

21. Jain, U., M. T. Patel, and P. K. Desai. 1990. Development ofsimple and stable ELISA for detection of antigens in intestinalamoebiasis. Indian J. Exp. Biol. 28:671-675.

22. Krogstad, D. J., H. C. Spencer, G. R Healy, N. N. Gleason,D. J. Sexton, and C. A. Herron. 1978. Amebiasis: epidemiologicstudies in the United States, 1971-1974. Ann. Intern. Med. 88:89-97.

23. Ritchie, L. 1948. An ether sedimentation technique for routinestool examinations. Bull. U.S. Army Med. Dep. 8:326.

24. Sargeaunt, P. G. 1988. Zymodemes of Entamoeba histolytica, p.370-387. In J. I. Ravdin (ed.), Amebiasis. Human infection byEntamoeba histolytica. John Wiley & Sons, New York.

25. Sargeaunt, P. G., and J. E. Williams. 1978. Electrophoreticisoenzyme patterns of Entamoeba histolytica and Entamoeba coli.Trans. R. Soc. Trop. Med. Hyg. 72:164-166.

26. Sargeaunt, P. G., J. E. Williams, and J. D. Grene. 1978. Thedifferentiation of invasive and non-invasive Entamoeba histolyticaby isoenzyme electrophoresis. Trans. R. Soc. Trop. Med. Hyg.72:519-521.

27. Sargeaunt, P. G., J. E. Williams, and R. A. Neal. 1980. Acomparative study of Entamoeba histolytica (NIH:200, HK9, etc.),"E. histolytica-like" and other morphologically identical amoebaeusing isoenzyme electrophoresis. Trans. R. Soc. Trop. Med. Hyg.74:469-474.

28. Schleupner, C. J., and A. S. Barritt III. 1988. Differentiation andoccurrence of amebiasis in inflammatory bowel disease, p. 582-593. In J. I. Ravdin (ed.), Amebiasis. Human infection by Entam-oeba histolytica. John Wiley & Sons, New York.

29. Spencer, H. C., J. J. Sullivan, H. M. Mathews, M. Sauerbrey, M.Bloch, W. Chin, and G. R. Healy. 1981. Serologic and parasiticstudies of Entamoeba histolytica in El Salvador, 1974-1978. Am. J.Trop. Med. Hyg. 30:63-68.

30. Spice, W. M., and J. P. Ackers. 1990. Large-scale production ofEntamoeba histolytica trophozoites in polyxenic culture. Trans. R.Soc. Trop. Med. Hyg. 84:693-694.

31. Stibbs, H. H. 1989. Monoclonal antibody-based enzyme immuno-assay for Giardia lamblia antigen in human stool. J. Clin. Micro-biol. 27:2582-2588.

32. Tobie, J. E. 1949. Experimental infection of the rabbit withEndamoeba histolytica. Am. J. Trop. Med. Hyg. 29:859-870.

33. Ungar, B. L. P., R. H. Yolken, and T. C. Quinn. 1985. Use of amonoclonal antibody in an enzyme immunoassay for the detectionof Entamoeba histolytica in fecal specimens. Am. J. Trop. Med.Hyg. 34:465-472.

34. Vecchio, T. J. 1966. Predictive value of a single diagnostic test inunselected populations. N. Engl. J. Med. 274:1171-1173.

35. Viscidi, R, B. E. Laughon, M. Hanvanich, J. G. Bartlett, and R H.Yolken. 1984. Improved enzyme immunoassays for the detectionof antigens in fecal specimens. Investigation and correction of

VOL. 32, 1994

on January 21, 2021 by guesthttp://jcm

.asm.org/

Dow

nloaded from

970 GONZALEZ-RUIZ ET AL.

interfering factors. J. Immunol. Methods 67:129-143.36. Voller, A., and D. Bidwell. 1986. Enzyme-linked immunosorbent

assay, p. 99-109. In N. R. Rose, H. Friedman, and J. L. Fahey(ed.), Manual of clinical laboratory immunology, 3rd ed. AmericanSociety for Microbiology, Washington, D.C.

37. Walker, E. L., and A. W. Sellards. 1978. Experimental entamoebicdysentery, p. 121-141. In B. H. Kean, K. E. Mott, and A. J. Russell(ed.), Tropical medicine and parasitology. Classic investigations.Amoebiasis, vol. 1. Cornell University Press, London. [Repro-duced from Philipp. J. Sci. B. Trop. Med. 8:253-330, 1913.]

38. Walsh, J. A. 1986. Problems in recognition and diagnosis ofamebiasis: examination of the global magnitude of morbidity and

J. CLIN. MICROBIOL.

mortality. Rev. Infect. Dis. 8:228-238.39. Wenyon, C. M., and F. W. O'Connor. 1917. The characters and

diagnosis of the various intestinal protozoa of man in Egypt, witha description of three new forms, p. 41-95. In C. M. Wenyon andF. W. O'Connor (ed.), Human intestinal protozoa in the NearEast. The Wellcome Bureau of Scientific Research, London.

40. World Health Organization. 1991. WHO/PAHO informal consul-tation on intestinal protozoal infections. 21 to 23 October 1991,Oaxtepec, Mexico, p. 5-10.

41. World Health Organization, Programme for Control of Diar-rhoeal Diseases. 1992. Update. Towards rational use of drugs inthe management of diarrhoea in children. No. 11 (September):1-4.

on January 21, 2021 by guesthttp://jcm

.asm.org/

Dow

nloaded from