correspondence Professor Julia Maria Costa-Cruz, Laboratório de Parasitologia, Departamento de Imunologia, Microbiologia e Parasitologia, Universidade Federal de Uberlândia. Av. Pará 1720, CEP 38400–902, Uberlândia, Minas Gerais, Brazil.
Summary A comparative study of total saline extract (SE) and cyst vesicular fluid (VF) of Taenia solium metacestodes by ELISA and Western blotting assay (WB) tests was conducted to detect IgG in sera for diagnosis of human cysticercosis. Sera were obtained and analysed by ELISA in 1 : 20 and 1 : 100 dilutions from 208 individuals: 22 confirmed neurocysticercosis (NC) (group 1), 101 suspected NC (group 2), 55 with various intestinal parasitosis (group 3) and 30 healthy individuals (group 4). The WB test was carried out on SE and VF extracts with and without reducing agent, 2-β-mercaptoethanol (2-ME) in 20 sera of each group. WB using extracts without 2-ME and ELISA at 1 : 100 dilution were compared in 20 sera from each group; sensitivity and specificity were calculated using samples from groups 1, 3 and 4. By ELISA, in the 1 : 100 sera dilution reactivity was reduced for both antigens without changes in the sensitivity of the test. By WB, antigens treated with 2-ME demonstrated low specificity. For SE and VF antigens, the proteins of 24, 39–42, 47–52, 56, 64–68, 126–155 kDa and 18, 24, 26–28, 32–36, 47–52, 75 kDa, respectively, were considered immunodominant markers, with high indices of specificity, suggesting a profile for NC patients. However, as the sensitivity was found to be low, it might still not be a definitive test for NC when used alone. These data suggest WB as an indicative test to determine exposure to T. solium. ELISA and WB together may supply reliable results for the diagnosis of human cysticercosis, since appropriate purified antigens are not available yet.
Cysticercosis is an important parasitic disease, common in developing countries, caused by the Taenia solium metacestode. When cysticerci settle in the brain, they can unchain serious polymorphic symptoms, which render definite neurocysticercosis (NC) diagnosis difficult ( Flisser et al. 1997 ). The Western blotting assay (WB) allows the identification of specific antigenic proteins which are recognized by sera through an electrophoresis of parasite crude extracts, and eliminates false-positive results that are more common by ELISA ( Larralde et al. 1989 ; Xu & Liu 1992; Simac et al. 1995 ; Ohsaki et al. 1999 ). In WB, purified antigens developed by Tsang et al. (1989) have been used in the diagnosis of NC ( Andriantsimahavandy et al. 1997 ; Jafri et al. 1998 ), but the crude antigens need more studies. Our aim was to conduct a comparative study of total saline extract and cyst vesicular fluid from T. solium metacestodes for diagnosis of NC by ELISA and WB in sera.
Materials and methods
The study received ethical approval from the Federal University of Uberlândia Ethical Committee.
T. solium metacestodes were obtained from muscles of naturally infected pigs by dissection, washed in saline solution (0.15 m NaCl) four times and stored at −20 °C.
Preparation of antigens
Total saline extract (SE) from 50 metacestodes of T. solium was prepared as described by Costa et al. (1982) . For cyst vesicular fluid (VF) extraction, the metacestodes were broken with scalpel blades and the fluid collected and centrifuged at 4800 × g for 15 min at 4 °C.
Serum samples were collected in the Laboratory of Clinical Analysis of the Hospital of the Federal University of Uberlândia, Minas Gerais, Brazil. Two hundred and eight sera were selected from patients based on clinical condition and analysed by ELISA. Sera were from 98 females and 110 males distributed in four groups:
Group 1: 22 patients with a diagnosis of established NC based on presence of clinical syndrome, epidemiological data, positive immunological test and evidence of the parasite by computerized tomography. The patients' ages ranged from 14 to 61 (mean = 38.53; sd = 13.10) Among these patients, 21 had been treated previously with praziquantel and/or albendazole and only one had been submitted to surgery.
Group 2: 101 patients with suspected NC diagnosis with neurological symptoms and positive epidemiology whose age ranged from 6 to 74 years (mean = 33.54, sd = 15.45). Patients from this group were not treated before this study.
Group 3: 55 patients with diagnosis of various intestinal parasitoses, listed in Table 1, aged 1–63 years (mean = 23.12; sd = 12.90).
Group 4: 30 healthy undergraduate students aged 19–41 years (mean = 23.57; sd = 4.50).
Table 1. Cross-reactivity, using the ELISA test, at 1 : 20 and 1 : 100 dilution in 85 serum samples (55 from patients with various intestinal parasitoses and 30 from healthy individuals) from the total saline extract (SE) and cyst vesicular fluid (VF) of Taenia solium metacestode
The individuals from groups 3 and 4 had three negative assays for T. solium and T. saginata by Lutz parasitological test ( Lutz 1919) and no epidemiological data for cysticercosis. For the WB, 80 sera (20 samples from each group) were used. To determine the sensitivity and specificity of both tests, only groups 1, 3 and 4 were used. For comparisons between dilutions in the ELISA, the sample number was 22 in group 1, 55 in group 3 and 30 in group 4. To compare sensitivity and specificity between ELISA and WB, 20 samples each were analysed in group 1, 3 and 4.
Polystyrene microplates (Interlab, Brazil) were for a final assay volume of 50 μl/well. Between each step of the reaction, microplates were washed three times, 5 min each with PBS containing 0.05% Tween 20 (PBS-T). For microplate coating, SE and VF were diluted at 10 μg/ml in 0.06 m carbonate-bicarbonate buffer, pH 9.6 and incubated overnight at 4 °C. Serum samples were diluted 1 : 20 and 1 : 100 in PBS-T and incubated for 45 min at 37 °C. The conjugate (rabbit IgG antihuman IgG, heavy chain specific, labelled with peroxidase prepared as described by Wilson & Nakane 1978) was diluted at 1 : 200 for both antigens in PBS-T and incubated for 45 min at 37 °C. The assay was developed by adding the enzymatic substrate consisting of H2O2 and o-phenylenediamine (OPD) in 0.1 m citrate phosphate buffer pH 5.5 for 15 min followed by 25 μl/well of 1 m H2SO4 to stop reaction. The absorbance was determined at 492 nm in an ELISA reader (Titertek Multiskan Plus, Flow Laboratories, USA). The cut-off was established using the mean of nonreactive samples plus two standard deviations, as described by Bassi et al. (1991) .
Electrophoresis and electrophoretic transfer
SE and VF antigens were diluted (v/v) in sample buffer, and after boiling for 3 min at 100 °C, all antigen preparation and molecular weight markers (Sigma Chem. Co., St. Louis, USA) were submitted to electrophoresis in SDS-PAGE at 12% under reducing or nonreducing conditions according to Laemmli (1970). After SDS-PAGE, the gels were transferred to nitrocellulose membranes (0.45 μm; Hybond-C Amersham, Life Science) as described by Towbin et al. (1979) using a semidry transfer apparatus (Multiphor II, Pharmacia-LKB).
Western Blotting assay
Nitrocellulose strips containing fractions of the SE and VF antigens treated or not with 2-β-mercaptoethanol (2-ME, Merck) were blocked with 5% nonfat milk in PBS-T for 2 h at room temperature and incubated overnight with serum samples diluted at 1 : 50 in 1% nonfat milk in PBS-T (PBS-TM) at 4 °C. After several washes with PBS-TM, the strips were incubated for 2 h at room temperature with peroxidase-labelled goat IgG antihuman IgG conjugate, heavy chain specific (Sigma), diluted 1 : 1500 in PBS-TM. The strips were washed in PBS and developed for 3 min in substrate solution containing hydrogen peroxide and 3,3′-diaminobenzidine tetrahydrochloride (DAB-Sigma) in PBS. The reaction was stopped by washing with distilled water and positive reactions were determined by the appearance of clearly defined bands. The relative molecular mass (Mr) of each recognized band was determined by comparison with standard molecular markers (Sigma).
A graphic method was used to measure the frequency of antigen recognition (f) from 0 to 1 of each specific antibody in serum samples from the studied groups ( Larralde et al. 1989 ).
The sensitivity and specificity were determined using binomial distribution and the differences between two or more proportions calculated by chi-square test. The data was obtained using the Statistics for Windows software (Stat soft, Inc. 1993). The null hypothesis was rejected when P < 0.05. For the WB test all proteins recognized were analysed individually, but in order to design the profile of NC patients for both extracts (SE and VF), the frequency of the immunodominant markers among the individual groups was analysed.
The concentrations of protein and polysaccharide found in the antigenic extracts were, respectively, 1.0 mg/ml and 0.9 mg/ml for SE, and 9.7 mg/ml and 9.1 mg/ml for VF. The ELISA sensitivity at 1 : 20 and 1 : 100 sera dilutions was, respectively, 90% and 85% for SE, and 95% and 90% for VF. The ELISA specificity at 1 : 20 and 1 : 100 dilution was, respectively, 52% and 88% for SE and 74% and 96.5% for VF. No statistical difference for the sensitivity among the extracts and the dilutions was found (P > 0.05). However, the 1 : 100 dilution was significantly more specific for all antigens (P < 0.05). Between extracts, VF was significantly more specific than SE (P < 0.05).
Table 1 shows the cross-reactivity in ELISA tests for 85 sera (55 from patients with intestinal parasitosis and 30 from healthy individuals) for both extracts from metacestodes of T. solium, at 1 : 20 and 1 : 100 sera dilution.
Twenty-one proteins (8, 12–14, 18, 24, 26–28, 32–36, 39–42, 47–52, 56, 64–68, 72, 75, 80, 85, 88, 92, 98, 105, 110, 115–120 and 126–155 kDa) were recognized by IgG antimetacestode of T. solium antibodies in 80 sera by WB, with extracts with and without 2-ME.
Presence and/or intensity of the proteins at 24, 32–36, 39–42, 47–52, 56, 64–68, 80, 105 and 126–155 kDa (SE) and 8, 18, 24, 26–28, 32–36, 39–42, 47–52, 64–68, 72, 75, 80, 85 and 88 kDa (VF) for groups 1 or 2 were significant (P < 0.05) compared to groups 3 and/or 4. Figure 1 shows a representative WB with all antigenic preparations.
Figure 2 (panels A, B, C and D) shows the frequency of antigenic proteins recognized by antibodies from sera of groups 1, 3 and 4, in the SE and VF extracts, without treatment with 2-ME. The recognition of two out of the six immunodominant markers was used as positivity criterion for WB test. When using SE without 2-ME, the immunodominant markers among sera from NC patients were: 24, 39–42, 47–52, 56, 64–68 and 126–155 kDa ( Fig. 2A). For this antigen, the recognition of immunodominant markers by group 1 was statistically significant (P < 0.00001) compared to other groups. There were no statistical differences between groups 2, 3 and 4. For the VF without 2-ME the markers 18, 24, 26–28, 32–36, 47–52 and 75 kDa were considered immunodominants ( Fig. 2B). For this antigen group 1 was statistically significant: P < 0.0001 compared to group 2 and P < 0.00001 against groups 3 and 4. It was possible using the VF extract, to separate group 2 from group 4 (P ≤ 0.005). Between groups 2 and 3, and between groups 3 and 4 no statistical differences were observed. Regarding immunodominant markers for antigens treated with 2-ME, we found no statistical differences among groups for SE; for VF groups 1 or 2 could be statistically separated from groups 3 and 4 (P < 0.05).
The sensitivity and the specificity of ELISA (1 : 100 sera dilution) and WB (extracts without 2-ME) were calculated using 20 samples each from groups 1, 3 and 4. With 2-ME, the extracts gave high indices of false-positive results. There was no statistical difference concerning sensitivity between the ELISA (80% for both extracts) and WB (60% for SE and 65% for VF) (P > 0.05). Regarding specificity, the WB using SE (97.5%) was statistically significant (P < 0.05) in comparison with ELISA (82.5%); for ELISA, VF (97.5%) was more specific (P < 0.05) than SE (82.5%). Higher predictive values for positivity were encountered in ELISA with VF and 1 : 100 dilution (94%) and in WB with SE without 2-ME (92%). Cross-reactions when using these two tests occurred mainly with sera of group 3.
Positive and negative concordant indices between WB and ELISA in sera of group 1 were 50% and 10%, respectively, for both extracts; for this group, the two antigens gave 40% discordant results. In the 20 sera of group 2, 45% (SE) and 40% (VF) of cases agreed for both tests, giving 15% and 30% of the positive concordant indices, and 30% and 10% of the negative concordant indices, respectively, for SE and VF extracts. 55% of results were discordant for SE and 60% for VF ( Table 2).
Table 2. Comparative analysis of ELISA and Western blotting (WB) in 20 serum samples from group 1 (panel A) and 20 samples from group 2 (panel B) using total saline extract (SE) and cyst vesicular fluid (VF) of Taenia solium metacestodes
The ELISA test standardized for crude antigens has been used in the diagnosis of most infectious and parasitic diseases, including human cysticercosis ( Diwan et al. 1982 ), but due to the frequency of cross-reactions observed in endemic areas for cysticercosis, as well as to other helminthic infections, differences of sensitivity and specificity have been found ( Zini et al. 1990 ; Garcia et al. 1998 ).
Similar to the results described by Costa (1986), our use of different antigenic fractions from T. solium metacestodes in the ELISA resulted in differences of sensitivity and specificity. Cross-reaction indices are high in sera ( Andriantsimahavandy et al. 1997 ), but we used the 1 : 100 dilution and observed a decrease in cross-reactions among patients from group 3 without changing the sensitivity of the test for both antigens (SE and VF).
All 21 antigenic markers recognized by our serum samples were already described in the literature. The WB results for antigens of T. solium metacestodes without 2-ME confirm the human immune response heterogeneity for this parasite ( Flisser et al. 1980 ).
The proteins of 64–68 kDa (SE) and 47–52 kDa for both antigens were the most frequently recognized markers among all epitopes, when the serum samples from group 1 were tested. It is important to point out that sera from groups 1, 2 and 3 also recognized the 80 kDa protein, suggesting that this epitope might be considered as a common marker of intestinal parasitosis.
The 2-ME was used in the extracts from T. solium metacestodes with the purpose of improving the sensitivity of the WB test. In general, the recognition of proteins by antibodies from serum samples of group 1 was analogous for extracts independently of the use of 2-ME. However, when antigens were treated with 2-ME, specificity of the WB in sera from control groups was low, possibly because 2-ME breaks disulphide-bonds, altering the conformational structure of epitopes and probably removing the specific reactivity of the proteins.
We tested two extracts (SE and VF) by ELISA and WB. ELISA using VF as antigen was better, while WB results were better using SE as antigen. Between the tests, ELISA was more sensitive, but the WB (SE) was more specific. These data showed the importance of the suitable antigen for each technique, because false negative and false positive results are frequent in endemic areas for cysticercosis and other helminthic infections.
Similar to cysticercosis, the diagnostic sensitivity and specificity of the tests for echinococcosis change according to the nature and quality of the antigen, and the methodological sensitivity of the selected technology ( Poretti et al. 1999 ). Other studies have demonstrated extensive cross-reactivity between serum from patients with cysticercosis and cystic and alveolar echinococcosis ( Gottstein et al. 1986 ; Ito et al. 1993 ; García et al. 1998 ). The proteins to which specific antibodies were encountered more often in sera of the echinococcosis patients were: 16 and 18 kDa ( Ito et al. 1993 ); 8, 12, 17 and 32 kDa ( García et al. 1998 ); 8, 29 and 34 kDa ( Poretti et al. 1999 ). Using WB, we demonstrated cross-reactivity between cysticercosis and other helminthic infections such as S. stercoralis, A. lumbricoides, S. mansoni and E. vermicularis. Gottstein et al. (1986) achieved similar results.
The protein of 26 kDa has been described in the literature as sensitive and specific for NC in crude antigens ( Gottstein et al. 1987 ; Larralde et al. 1989 ; Xu & Liu 1992). We found 26 kDa protein to be highly specific for NC, but due to the low sensitivity in the sera studied, the detection of this protein alone was not viable for NC diagnosis. Therefore we could determine a profile of immunodominant markers that could be used for the NC diagnosis. Using this model, the WB test showed high sensitivity and specificity for both tested antigens without treatment with 2-ME.
It can be concluded that the high indices of the specificity demonstrated by immunodominant markers in serum samples characterize qualitatively the WB test, and suggest a profile of the NC patient. However, as the sensitivity was low, it may not still be a definitive test for NC for sole use, but only an indicative test to determine exposure to T. solium. Thus, ELISA and WB together may supply reliable results for the diagnosis of human cysticercosis, since appropriate purified antigens are not available yet.
This study was supported by Fundação Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Brazil. The authors thank Dr Marcelo Simão Ferreira for the referral of patients from the Department of Internal Medicine, Universidade Federal de Uberlândia and to Dr Janeth Deolinda de Oliveira Pena for reviewing the manuscript.