Comparison of Stool Antigen Detection Kits to PCR for Diagnosis of Amebiasis

Abstract

The present study was conducted to compare two stool antigen detection kits with PCR for the diagnosis of Entamoeba histolytica infections by using fecal specimens submitted to the Department of Microbiology at St. Vincent's Hospital, Sydney, and the Institute of Medical and Veterinary Science, Adelaide, Australia. A total of 279 stool samples containing the E complex (E. histolytica, Entamoeba dispar, and Entamoeba moshkovskii) were included in this study. The stool specimens were tested by using two commercially produced enzyme immunoassays (the Entamoeba CELISA PATH and TechLab E. histolytica II kits) to detect antigens of E. histolytica. DNA was extracted from all of the samples with a Qiagen DNA stool mini kit (Qiagen, Hilden, Germany), and a PCR targeting the small-subunit ribosomal DNA was performed on all of the samples. When PCR was used as a reference standard, the CELISA PATH kit showed 28% sensitivity and 100% specificity. The TechLab ELISA (enzyme-linked immunosorbent assay) kit did not prove to be useful in detecting E. histolytica, as it failed to identify any of the E. histolytica samples which were positive by PCR. With the TechLab kit, cross-reactivity was observed for three specimens, one of which was positive for both E. dispar and E. moshkovskii while the other two samples contained E. moshkovskii. Quantitative assessment of the PCR and ELISA results obtained showed that the ELISA kits were 1,000 to 10,000 times less sensitive, and our results show that the CELISA PATH kit and the TechLab ELISA are not useful for the detection of E. histolytica in stool samples from patients in geographical regions where this parasite is not endemic.

Amebiasis is a parasitic infection caused by Entamoeba histolytica and is one of the most common parasitic infections world-wide, infecting about 50 million people and resulting in 10,000 to 40,000 deaths per annum (20). Manifestations of amebiasis include dysentery and extraintestinal invasive disease (18). The diagnosis of E. histolytica infection has traditionally relied upon microscopic examination of fresh or fixed stool specimens (5). However, microscopy has several limitations (4, 8, 16), most importantly, the inability to distinguish the pathogenic species E. histolytica from the morphologically identical nonpathogenic species E. dispar and E. moshkovskii (1, 3, 4, 9-11, 16). The sensitivity of microscopy is approximately 60% and is confounded with false positives due to misidentification of the other morphologically similar Entamoeba species (5, 9, 10, 16). It is important to correctly diagnose patients not only to reduce the morbidity and mortality of amebiasis but also to minimize the undue treatment of patients infected with E. dispar and E. moshkovskii with antiamebic therapy. The reference standard used to differentiate E. histolytica from E. dispar is amebic culture with isoenzyme analysis; however, this method is not widely available and is not practical for routine diagnostic laboratories (5, 11). In addition, the common occurrence of E. dispar and E. moshkovskii in human populations has led to the need for newer detection methods able to identify and detect several species of Entamoeba.

Several newer diagnostic tests are now available which surpass the microscopic detection of these parasites and facilitate a more accurate diagnosis. These approaches include PCR and antigen-based enzyme-linked immunosorbent assays (ELISAs) (7, 10, 12, 13, 16). Stool antigen assays have been reported to outperform microscopy and to be as sensitive (80 to 85%) and specific (99%) as culture with isoenzyme analysis for the detection of E. histolytica in areas of endemicity (10, 11). Recently, molecule-based PCR assays have been reported to demonstrate excellent sensitivity and specificity compared with microscopy (4, 5, 9, 15). In several evaluation studies, similar sensitivities and specificities were reported for PCR and ELISA (11, 14). As PCR techniques are not widely available and remain impractical tools in many developing countries, stool antigen assays are considered valid alternative diagnostic methods for the diagnosis of E. histolytica infections.

The present study was designed to compare two commercially available stool antigen detection kits, namely, the Entamoeba CELISA PATH kit (Cellabs, Brookvale, Australia) and the TechLab E. histolytica II kit (TechLab, Inc., Blacksburg, VA), with conventional PCR amplification of small-subunit ribosomal DNA (rDNA) from fecal samples submitted by patients to St. Vincent's Hospital, Sydney (St. Vincent's), and the Institute of Medical and Veterinary Science (IMVS), Adelaide, Australia. Although the TechLab E. histolytica II kit has previously been evaluated, this is the first study to compare it with the Entamoeba CELISA PATH kit and PCR.

MATERIALS AND METHODS

Patient samples.

All patient fecal samples submitted to St. Vincent's and IMVS for parasitology (ovum, Cyst, and parasite program) investigation (note that some of the IMVS samples were from asymptomatic refugee patients) between March 2004 and August 2007 were screened for inclusion in the evaluation. A portion of the fecal samples submitted were fixed in sodium-acetate-formalin and permanently stained with a modified iron-hematoxylin stain (Fronine, Australia) according to the manufacturer's instructions. All microscopy-positive samples containing the E complex (n = 279) were then used for the evaluation of ELISA and PCR. A negative control group was also included in this study (see below).

ELISA.

Antigen-based testing with the ELISA kits was performed with fresh stool samples within 48 h of collection. The two ELISAs (TechLab E. histolytica II kit [TechLab, Inc., Blacksburg, VA] and Entamoeba CELISA PATH kit [Cellabs, Brookvale, Australia]) were performed according to the manufacturers' instructions. Both tests were designed to detect E. histolytica alone. Each test included positive and negative controls. The remaining fresh stool sample was stored at −20°C for PCR amplification at a later date.

PCR and DNA sequencing.

DNA extraction; PCR amplification of the small-subunit rDNA of E. histolytica, E. dispar, and E. moshkovskii; and sequencing were carried out as previously described (4, 17).To exclude inhibition from fecal inhibitors, all specimens were spiked with an equal volume of genomic DNA from E. histolytica strain HTH-56:MUTM controls and run in parallel with an unspiked specimen.

Limit of detection.

Xenic cultures of E. histolytica strain HTH-56:MUTM were grown in LE medium by standard procedures (2). Trophozoites were harvested and used as the antigen/DNA to determine the lower limits of detection of the ELISAs and PCR. Trophozoites were counted in a hemacytometer. Duplicate serial dilutions were prepared from suspensions of known concentrations of trophozoites. An aliquot of each sample underwent DNA extraction and was then suspended in 200 μl of a negative control stool sample and tested by PCR as described above. The duplicate sample underwent lysis by freeze-thawing in phosphate-buffered saline containing a mixture of various protease inhibitors as previously described (12). For ELISA, simulated stool samples were prepared by adding the serial dilutions of the trophozoite lysates to 400 μl of a negative control stool sample. The trophozoite lysate-seeded stool samples were then tested with the reagents and protocols of the two ELISA kits as described above. All testing was performed in duplicate at each dilution. The negative control stool sample used in both PCR and ELISA experiments was from a healthy volunteer and was negative for protozoan parasites by microscopy and E complex PCR (as described above).

Control group.

In addition, to confirm the absence of cross-reactivity, a control group comprising 100 stool samples that were negative by microscopy for E. histolytica were randomly selected. The samples were considered negative after examining approximately 200 to 300 oil immersion fields of view of the stained slides. These samples were further classified into specimens that were negative for all parasites (n = 50) and samples that were positive for one or more parasites, excluding the E complex (n = 50) (Table 1). All of these samples were included in the test protocol and underwent both ELISA and PCR testing as described above.

TABLE 1.

E. histolytica negative control stool samples which contain one or more protozoa other than E. histolytica^a

Protozoa	No. of samples
Blastocystis hominis	5
Entamoeba coli	5
Entamoeba hartmanni	5
Giardia intestinalis	5
Endolimax nana	5
Iodamoeba bütschlii	2
Cryptosporidium sp.	2
Cyclospora sp.	1
Chilomastix mesnili	2
E. hartmanni, E. nana, Enteromonas hominis, B. hominis	2
E. nana, I. bütschlii, C. mesnili, B. hominis	2
Cryptosporidium sp., B. hominis	2
E. nana, I. bütschlii, C. mesnili, B. hominis	2
G. intestinalis, E. nana, B. hominis	1
E. coli, E. nana, I. bütschlii, B. hominis	1
E. coli, E. hartmanni, E. nana, I. bütschlii, B. hominis	1
E. hartmanni, E. nana	1
B. hominis, E. nana	4
C. mesnili, E. hominis, E. nana	1
G. intestinalis, E. nana, B. hominis	1

^aThe 50 fecal specimens listed in this table were all negative by PCR for E. histolytica but were positive by microscopy for a range of common human protozoa.

RESULTS

E. histolytica was detected in 6% (18/279) of the stool samples by PCR (Table 2). E. dispar was detected in 136 samples, while E. moshkovskii was detected in 73 samples (most of this information has been previously reported elsewhere [4, 15]). All of the DNA sequences revealed 99 to 100% homology with sequences stored in GenBank. PCR inhibition occurred in 1% (3/279) of the samples. All of the control fecal samples E. histolytica negative by microscopy were E. histolytica PCR negative, and there was no PCR product from any of the control samples which contained protozoa other than E. histolytica (Table 1). Similarly, E. histolytica antigen was not detected in microscopy-negative (including the control group) fecal samples with either ELISA kit. However, for samples that were microscopy positive for E complex, the TechLab E. histolytica II kit failed to detect any (0/18) of the PCR-positive samples. In addition, a false-positive result was obtained in 1% (3/261) of the samples E. histolytica negative by PCR. These three samples were found to be positive by PCR for non-E. histolytica species (E. moshkovskii [n = 2] and both E. dispar and E. moshkovskii [n = 1]). This was confirmed by DNA sequencing, which revealed 99 to 100% similarity to the E. dispar and E. moshkovskii 18S rDNA sequences deposited in GenBank (GenBank accession no. Z49256 and AF149906). Thus, these three samples were considered false positives. All 100 negative control group samples were negative by both ELISA kits. The Entamoeba CELISA PATH kit detected 28% (5/18) of the E. histolytica PCR-positive samples with no false-positive results (Table 2). Compared to PCR, the sensitivities were 0 and 28% for the TechLab E. histolytica II kit and the Entamoeba CELISA PATH kit, respectively. In contrast, the specificities of both stool antigen tests were similar at 99 and 100%, respectively.

TABLE 2.

Comparison of results obtained by testing 279 fecal samples positive for E complex by microscopy with the Entamoeba CELISA PATH and TechLab E. histolytica II ELISA kits compared with E. histolytica PCR

ELISA result	No. of samples PCR:
	Positive	Negative
CELISA
Positive	5	0
Negative	13	261
TechLab
Positive	0	3
Negative	18	258

Quantitative estimates with lysates produced from E. histolytica cultures revealed that the Entamoeba CELISA PATH kit was able to detect a 10-fold lower concentration of E. histolytica trophozoites per well (1, 000) compared with the TechLab E. histolytica II kit, which required lysate from 10,000 trophozoites for a positive reaction. In contrast, the E. histolytica PCR was able to detect a PCR product from a sample containing one trophozoite per reaction.

DISCUSSION

The E. histolytica PCR was found to be both sensitive and specific for the detection and differentiation of the E complex. In addition, the PCR was found to have a lower limit of detection of approximately one trophozoite per well. In contrast, both of the stool antigen kits (the Entamoeba CELISA PATH kit and the TechLab E. histolytica II kit) showed poor sensitivities of 28 and 0%, respectively, compared to PCR, with these results representing the first published standardized evaluation of the Entamoeba CELISA PATH kit compared with PCR.

Several ELISA kits have been developed and reported to possess high sensitivity and specificity (7, 10, 12, 16). However, this evaluation has found that both ELISA kits performed poorly compared with PCR when testing routine microscopy-positive stool samples submitted to two diagnostic parasitology laboratories in Australia. The quantification from cultured lysates revealed that the Entamoeba CELISA PATH kit was the more sensitive, with the ability to detect approximately 1,000 trophozoites per well compared to the TechLab E. histolytica II kit, which required a 10-fold greater load, at approximately 10,000 trophozoites per well for a positive test. Both kits use the same target, a monoclonal antibodies against the Gal/GalNAc-specific lectin (adhesin molecule) of E. histolytica. The differences in performance between the two ELISAs may be attributed to the amounts of antibody used to coat the wells of the ELISA plates. The level of detection observed with the antibody-based systems was >1,000-fold less sensitive than that which can be attained by PCR amplification targeting the rDNA. As none of the PCR-positive samples were quantitated, it is unclear whether this is the only reason for the lower detection level. However, it may explain the difference in performance between the two ELISA kits.

The TechLab antigen kit has been used over several years in different laboratories for the detection of E. histolytica in regions of the world where it is endemic or nonendemic. The results obtained with the TechLab antigen kit are in conflict with those of studies conducted in countries where E. histolytica is highly endemic that reported high sensitivities between 95 and 100% (9, 11, 14). However, a recent study conducted in a region of northern Ecuador where E. histolytica is highly endemic found that the TechLab E. histolytica II test performed poorly, with a reported sensitivity of 14.3% and a specificity of 98.4% compared to isoenzyme analysis (6). In low-endemicity settings, the TechLab ELISA has been documented to have a sensitivity lower than that of microscopy (7). Similar results were obtained when the TechLab ELISA was compared to real-time PCR as a reference test in a low-endemicity setting (19). Mirelman et al. (1997) were able to quantitate the difference in sensitivity, with the TechLab kit >100 times less sensitive than PCR (12). These previous findings are all supported by our results, which showed that the TechLab ELISA kit was not as sensitive or specific as PCR. In addition, the TechLab ELISA kit was 1,000 times less sensitive than PCR. Gatti et al. proposed that the poor performance of ELISA kits could be due to the fact that the assays recognize the antigens on the vegetative forms only, which are generally found in diarrheal stool samples during an acute amebic infection and not in the cystic stage of the parasite (6). In our study, at least half of the patients were symptomatic and in the majority of the cases both trophozoites and cysts were present, as proven by microscopy, yet the ELISA kits still performed poorly. It should be noted, however, that this study used cell lysates of E. histolytica to calculate the analytical sensitivity of the ELISAs, and it is not clear if the ELISAs have comparable limits of detection of trophozoites, cysts, and cell lysates.

In conclusion, antigen detection by ELISA is technically simple to perform, rapid, and cheaper than molecular methods; however, in view of the poor performance of both commercial ELISA kits, it can be argued that they should not be used as the mainstay in the diagnosis of E. histolytica. Furthermore, if these tests are used they should first undergo extensive local evaluation compared with PCR as the “gold standard” to determine the level of false-negative results expected in that population when using ELISA for the diagnosis of E. histolytica. Both ELISA kits were specific and therefore may still have a place in the differentiation of species of the E complex when large numbers of cysts and/or trophozoites are detected by microscopy. This study clearly demonstrates the advantages of PCR over ELISA-based kits in both sensitivity and specificity. In addition, PCR has the advantage of specifically targeting and detecting E. histolytica, E. dispar, and E. moshkovskii in clinical samples. Given the improvements in the cost of PCR and the advent of automation and simplification of PCR protocols, we believe that all detection and differentiation of Entamoeba spp. should be performed by PCR.