Diagnosis of amoebic colitis by antigen capture ELISA in patients presenting with acute diarrhoea in Cairo, Egypt

Authors


Jonathan I. Ravdin Department of Medicine, University of Minnesota, 420 Delaware Street SE, Minneapolis, MN 55455, USA. Fax: +1 612 626 3055; E-mail: ravdi001@tc.umn.edu

Abstract

We studied 84 consecutive patients presenting with acute diarrhoea (less than 1 week in duration) at an outpatient tropical medicine clinic in Cairo, Egypt. The diagnosis of amoebic colitis was established by the presence of Entamoeba histolytica galactose-inhibitable lectin antigen and the presence of occult blood in stool. Controls were 182 healthy regional people and 64 patients complaining of prolonged diarrhoea lasting more than 1 week. Entamoeba histolytica infection was found more frequently in patients with acute diarrhoea (57.1%) than in healthy controls (21.4%) or patients with prolonged diarrhoea (25%) (P < 0.001). There was a higher prevalence of Entamoeba dispar infection in the two control groups (24.2 and 20.3%, respectively, P=0.004 and 0.061) compared with those with acute diarrhoea (8.3%). Of the 84 patients with acute diarrhoea 32 had amoebic colitis (38%), and of these, 31 (97%) had at least one positive assay for serum amoebic antibodies (P < 0.001 compared with control groups). In summary, as determined by antigen-detection enzyme-linked immunosorbent assay, there is an unexpectedly high prevalence of amoebic colitis among patients presenting with acute diarrhoea to a tropical disease clinic in Cairo, Egypt.

Introduction

Entamoeba histolytica and Entamoeba dispar are morphologically identical but genetically distinct species (Tannich et al. 1989; Tannich & Burchard, 1991): E. histolytica is capable of causing amoebic colitis and liver abscess, and E. dispar is a non-pathogenic commensal (Diamond & Clark 1993). The combined prevalence of both Entamoeba sp. infection varies depending on the region. In Mexico, it has been found to be 25% by Gutierrez et al. (1976) and 83.3% by Romero et al. (1997); on the Indian subcontinent, 38% (Meerovitch et al. 1978; Veerannan 1979; Kulkarni et al. 1987); in sub-Saharan and tropical Africa, 25% (Omer et al. 1981); in a Brazilian slum, 40% (Braga et al. 1996); and in an Egyptian village, 45.1% (Abd-Alla et al. 2000a).

Intestinal infection with E. histolytica is asymptomatic in up to 90% of those infected (Gathiram & Jackson 1985); but at least 100 000 people die worldwide each year of amoebic colitis and liver abscess. Misdiagnosis and delayed treatment are common among the estimated 50 million patients suffering from invasive amoebiasis each year; 10% of 469 tourists returning from developing countries with diarrhoea were found to have E. histolytica infection by faecal antigen detection (Jelinek et al. 1996). In another study, late diagnosis and ineffective therapy resulted in 91 patients with intestinal amoebiasis undergoing colonic surgery due to profuse haemorrhage, perforation of amoebic ulcers, gangrene and toxic dilatation, resulting in seven deaths (Dautov 1997).

Diagnosis of amoebic infection solely by faecal microscopy is an inaccurate (Krogstad et al. 1978) and time-consuming procedure. Isolation of amoebic trophozoites by stool culture and differentiation of E. histolytica from E. dispar by zymodeme analysis was considered the gold standard for diagnosing amoebic infection (Ravdin 1988). Recent advances in diagnosis have used antibody recognition of the galactose-inhibitable lectin of E. histolytica in faeces; the lectin is a highly conserved antigen (Ravdin et al. 1990; Abd-Alla et al. 1993, 1998; Soong et al. 1995; Abo-El-Maged et al. 1996) and a recombinant cysteine-rich section of the lectin produced in E. coli (designated LC3) is also highly antigenic (Song et al. 1995). The recombinant LC3 protein contains all the epitopes recognized by the monoclonal antibodies used in an antigen capture enzyme-linked immunosorbent assay (ELISA) (Abo-El-Maged et al. 1996). Detection of amoebic 170 kDa lectin antigen in faeces and saliva using epitope-specific monoclonal antibodies provides a quantitative method for differentiating E. histolytica from E. dispar (Abd-Alla et al. 1993, 2000b; Haque et al. 1995). We have repeatedly demonstrated the high sensitivity and specificity of this assay in field conditions in Cairo, Egypt (Abd-Alla et al. 1993; Abo-El-Maged et al. 1996); therefore, we conclude that confirmatory stool culture and zymodeme analysis is not indicated.

Entamoeba histolytica lectin antigenemia occurs during invasive amoebiasis (Abd-Alla et al. 1993) and clears within 1 week after treatment of amoebic colitis or liver abscess (Abo-El-Maged et al. 1996). Serum antilectin immunoglobulin G (IgG) antibodies are present within 1 week after onset of symptoms in over 95% of patients with amoebic colitis or liver abscess (Ravdin et al. 1990; Abd-Alla et al. 1992) and may persist for years (Mirelman et al. 1980). In endemic areas, population-wide seropositivity rates of 20–25% limit the use of antiamoebic IgG antibodies as a diagnostic test.

The purpose of our study was to determine the incidence of amoebic colitis among 84 consecutive patients presenting to a tropical disease clinic in Cairo, Egypt, with acute diarrhoea, as determined by monoclonal antibody-based ELISA for detection of faecal 170 kDa lectin antigen, combined with a stool Hemoccult test. Over 99% of patients with acute amoebic colitis will have occult blood in their stools (Adams & MacLleod 1977). ELISAs to detect a number of different antiamoebic antibodies (serum anti-LC3 IgM, IgG and IgA, and faecal antilectin IgA) were performed to confirm the diagnosis of amoebic colitis and to compare the sensitivity of serology with the faecal antigen detection assay.

Materials and methods

Human subjects and study samples

Sera and stool samples were obtained from 84 consecutive patients presenting over 1-month period with acute diarrhoea of less than 1 week in duration at the outpatient clinic of the Tropical Medicine Department at El-Hussein University Hospital, Cairo, Egypt. There were 30 females and 54 males, with an age range of 9–60 years. Egyptian controls included 182 healthy individuals and 64 patients with prolonged diarrhoea lasting longer than 1 week. Subjects had no history of antiamoebic therapy before entering the study. Control sera were obtained from 56 healthy employees of the University of Virginia who did not have any history of amoebic infection. Control faeces (56) were obtained from the University of Minnesota employees having no history of amoebic infection or recent travel to endemic areas. Egyptian controls infected with E. histolytica or E. dispar were not excluded before entry into the study.

Detection of serum and faecal lectin antigen by ELISA

The ELISA for detection of 170 kDa lectin antigen was performed as described (Abd-Alla et al. 1993). Briefly, 96 flat-bottomed microtiter polystyrene ELISA plates (Costar, Corning, NY, USA) were coated with monoclonal antibody 3F4, which recognizes epitopes present in both E. histolytica and E. dispar lectin, or the 8C12 antibody, which is specific for epitopes present only in E. histolytica lectin epitopes (Petri et al. 1990). Faeces were mixed in an equal volume of phosphate-buffered saline (PBS) containing 2 mM phenylmethylsulphonyl fluoride (PMSF) (USB, Cleveland, OH, USA), serum was diluted 1 : 100 in PBS-Tween with 1% bovine serum albumin (BSA). Serum or faecal samples were added at 100 μl per well and incubated for 2 h at room temperature or overnight at 4 °C. Alkaline phosphatase-conjugated antilectin monoclonal antibodies 8A3 (recognizing both E. histolytica and E. dispar lectin) or 1G7 (specific for E. histolytica) (Petri et al. 1990) were added at 1 : 1000 dilution and incubated in developing buffer for 2 h at room temperature. Plate readings were corrected for non-specific background by subtracting the optical density (OD) from that of paired wells not coated with monoclonal antibodies but exposed to the identical procedure. The cutoff point was calculated as mean +2 standard deviations (SD) of OD reading of endemic area controls (Ravdin et al. 1990).

Detection of faecal antilectin IgA antibodies by ELISA

Galactose-inhibitable lectin protein (Petri et al. 1987) was purified as described. ELISA for detection of faecal antilectin IgA antibodies was performed as described (Abo-El-Maged et al. 1996). Briefly, flat-bottomed microtiter ELISA plates were coated with lectin protein (0.1 μg/well) and non-reactive sites were blocked with 1% BSA. Faeces was mixed with equal volumes of PBS-2 mM PMSF, added at 100 μl/well and incubated for 2 h at room temperature or overnight at 4 °C. Alkaline phosphatase-conjugated goat antihuman IgA antibodies (Sigma, St Louis, MO, USA) were added (1 : 5000) in PBS–Tween containing 1% BSA for incubation over 2 h at room temperature. Correction of ELISA OD readings and calculation of the cutoff point were performed as described for lectin antigen detection.

Detection of serum anti-LC3 IgG, IgM and IgA antibodies using ELISA

This procedure was performed as described previously (Ravdin et al. 1990; Jelinek et al. 1996; Abd-Alla et al. 1998). Recombinant 52 kDa LC3 protein (Song et al. 1995) was purified as described. Briefly, 96-well microtiter flat-bottomed polystyrene ELISA plates were coated with LC3 protein and non-reactive sites were blocked with 1% BSA. Serum samples were studied at 1 : 1000 dilution for IgG and at 1 : 500 for IgM or IgA ELISAs, all in PBS–Tween (1% BSA) and incubated for 2 h at room temperature. Alkaline phosphatase-conjugated goat antihuman IgG (Sigma, St Louis, MO, USA), IgA or IgM antibodies ICN Biomedicals (Costa Mesa, CA, USA) were diluted (at 1 : 5000 for IgG, 1 : 3000 for IgM, and 1 : 2000 for IgA) in PBS-Tween (1% BSA) for incubation in a 100-μl well for 2 h at room temperature. Correction of non-specific background binding and calculation of the cutoff point were performed as described.

Statistics

Results were expressed as the mean (+ 3 SD, of percent positive and percent negative). The Z-test (converted to P-value) and unpaired Student's t-test were used to determine the significance of difference (Sox 1986).

Results

Clinical characteristics

Acute diarrhoea was of rapid onset and short duration, an average of 2 days in 95% of cases. Diarrhoea was associated with abdominal pain, distention, dysentery, tenesmus and offensive flatus in most cases. Abdominal examination revealed tenderness mainly along the course of the colon and occasionally centrally. Faeces were loose to fluid in most cases. Prolonged diarrhoea was characterized mainly by increased frequency of defecation, a dull ache along the course of the colon, and semiformed to loose stools for at least 1 week (average 12 days).

Prevalence of E. histolytica and E. dispar intestinal infection

Table 1 summarizes the results of ELISA for detection of faecal lectin antigen in the two control groups and in Egyptian patients with acute or prolonged diarrhoea. Of 182 Egyptian controls, 39 (21.4%) had E. histolytica antigen and 44 (24.2%) had E. dispar antigen in their faeces (P < 0.005 compared with patients with acute diarrhoea). Of 84 cases with acute diarrhoea, 48 (57%) had E. histolytica antigen in their faeces and only seven (8.3%) had E. dispar antigen. The low prevalence of E. dispar infection detected in this group may be partly because of the fact that the ELISA method used cannot detect mixed infections if E. histolytica is present (Abd-Alla et al. 1992). None of the subjects infected with E. dispar had occult blood in their stools; in comparison, 32 of 48 patients with acute diarrhoea and E. histolytica antigen in their faeces were Hemoccult positive (P=0.01). There was no difference in the prevalence of E. histolytica or E. dispar infection between subjects with prolonged diarrhoea and regional controls. Of 56 American controls, two were positive for E. dispar faecal antigen and one for E. histolytica (P < 0.001 compared with Egyptian patients with acute diarrhoea).

Table 1.   Prevalence of Entamoeba histolytica and E. dispar infection as detected by ELISA for faecal lectin antigen in subjects with acute diarrhoea (< 1 week) compared with prolonged diarrhoea (> 1 week) and healthy controls Thumbnail image of

Occurrence of amoebic colitis

Of 84 patients with acute diarrhoea, 32 had amoebic colitis as defined by a positive ELISA for E. histolytica faecal antigen and the presence of occult blood in stool. The prevalence of serum anti-LC3 IgM (53.1%) and faecal antilectin IgA (56.3%) antibodies were both significantly higher in patients with acute amoebic colitis than in Egyptian controls (P < 0.001), subjects with prolonged diarrhoea (P < 0.01), and patients with acute diarrhoea without E. histolytica infection (P < 0.01) (Table 2). The prevalence of serum anti-LC3 IgG antibodies (56.3%) was significantly greater in amoebic colitis patients than in Egyptian controls and patients with prolonged diarrhoea (P < 0.001 and 0.03, respectively). Patients with acute diarrhoea not caused by E. histolytica, those with prolonged diarrhoea, or healthy subjects all had a similar prevalence of serum anti-LC3 IgM or IgA antibodies (P=0.171–0.929). The prevalence of serum anti-LC3 IgG antibodies was higher in subjects with acute diarrhoea without E. histolytica infection, compared with those with prolonged diarrhoea or controls (P < 0.01).

Table 2.   Prevalence of antilectin antibodies in patients with amoebic colitis compared with subjects with acute diarrhoea due to other causes, prolonged diarrhoea, and healthy Egyptian controls Thumbnail image of

Discussion

Symptomatic invasive amoebiasis, manifest as amoebic colitis or liver abscess, occurs in approximately 10% of E. histolytica intestinal infections, the remainder are asymptomatic (Gathiram & Jackson 1985). We determined the prevalence of E. histolytica infection among consecutive patients presenting to a Tropical Disease Clinic in Cairo, Egypt with acute diarrhoea and found that 38% had amoebic colitis as defined by occult blood and E. histolytica antigen in faeces. Occult blood is found in the faeces of 99% of patients with acute amoebic colitis (Adams & MacLleod 1977). By contrast, amoebiasis was not found to be a cause of prolonged diarrhoea in this region. Entamoeba histolytica infection was no more common in this group of patients than in regional asymptomatic controls. The results of ELISA for detection or serum antilectin IgM, IgA and IgG antibodies positively correlated with the diagnosis of acute amoebic colitis, as 97% of colitis patients had at least one positive antiamoebic antibody test. However, no individual antibody test was sensitive enough for exclusive use for diagnosis of amoebic colitis in this endemic region.

Entamoeba dispar is not known to cause symptoms of diarrhoea (Ravdin 1988; Sather et al. 1990; Abd-Alla et al. 1993; Diamond & Clark 1993), as was the case in this study given its higher prevalence in healthy Egyptian controls. However, the prevalence of asymptomatic E. histolytica and E. dispar infection among healthy controls in Egypt was higher than expected.

Acute amoebic colitis is known to be associated with a positive ELISA for serum anti-LC3 IgM antibodies (Petri et al. 1987; Abd-Alla et al. 1998) and intestinal antilectin IgA antibodies (Abo-El-Maged et al. 1996). Serum antilectin or anti-LC3 IgG antibodies (Ravdin et al. 1990; Abd-Alla et al. 1992) persist for years following E. histolytica infection. Therefore, in a highly endemic area it is not surprising that ELISA for serum anti-LC3 IgG antibodies did not correlate well with the occurrence of acute amoebic colitis. The ELISA used in this study detects the same lectin antigen as the Tech lab assay (Haque et al. 1995) (Blacksburg, VA, USA), but our results are not directly applicable to that commercially available test.

This study establishes that E. histolytica is one of the most prominent pathogens causing acute diarrhoea in Cairo, Egypt, but does not have a substantial aetiologic role in prolonged diarrhoea. In endemic areas, faecal antigen detection, combined with testing for occult blood in stool, is a more sensitive and effective diagnostic strategy than serology.

Acknowledgements

This work was supported by grants UO1-AI35840-01 (ICIDR) and PO1-AI 36359-01, Project 1, from the National Institute of Health and a grant from the AVICENNE Program, Commission of European communities (AVI*-CT93-0008). All studies were approved by the Institutional Review Boards for human subjects research. The authors thank Joan M. Portel and Shana Brooks for excellent secretarial assistance.

Ancillary