Evaluation of a newly developed ELISA against Widal , TUBEX-TF and Typhidot for typhoid fever surveillance

Introduction: Typhoid fever is endemic in many parts of the world and represents a major cause of acute febrile illness (AFI). Rapid and accurate laboratory methods for diagnosis of this disease are needed for both patient care and surveillance situations. Methodology: Serum samples were collected from AFI patients and used to evaluate the performance of a newly developed ELISA assay that uses a mixture of somatic and flagellar antigens to detect the total antibody response against Salmonella enterica subspecies enterica serovar Typhi (S. Typhi) infection. The levels of Ig isotype response (IgG, IgM and IgA) were also evaluated, and results were compared to those of TUBEX-TF and Typhidot commercial kits. Results: Of 234 culture-confirmed typhoid patients, the total Ig ELISA diagnosed 93% compared to 71% using Widal test. This sensitivity level (93%) is higher than that observed for the individual Ig ELISAs (IgG 75%; IgM 79%; IgA 57%) and the commercial tests TUBEX-TF (75%), Typhidot IgM (63%) and Typhidot IgG (28%). An agreement of 78% was achieved between the total Ig ELISA and Widal test. The average specificity of the ELISA was 96%. Using ELISA, up to 200 samples can be tested per run with cost per test at US$0.20. Conclusions: The developed ELISA shows superior sensitivity and specificity, when compared to Widal, TUBEX-TF and Typhidot assays, is more cost effective and allows higher throughput. This method is highly recommended for active surveillance studies or outbreak investigations of typhoid fever.


Introduction
Typhoid fever, resulting from infection by Salmonella enterica subspecies enterica serovar Typhi (S.Typhi), is a major cause of morbidity and mortality worldwide.Recent surveillance studies have indicated that infection by S. Typhi causes 21 million illnesses and 200,000 deaths annually [1,2].In Egypt, the incidence rate of S. Typhi infections ranges between 13 and 59 per 100,000 persons per year [2,3].This range is among the highest reported incidence ranges in northern African countries (10-100 per 100,000 per year) [4].
Laboratory diagnosis of typhoid fever in developing or underdeveloped countries is primarily achieved either by bacterial culture, considered to be the gold standard for diagnosis, or detection of serum antibodies using the Widal test, which is currently the standard serological method for typhoid diagnosis [5].However, both methods are time-consuming with relatively low sensitivities and specificities, and culture methods require basic laboratory facilities [6][7][8].Efforts to develop faster and more sensitive and specific serological assays for the diagnosis of typhoid fever have led to the commercialization of the TUBEX-TF and Typhidot kits.Both kits have been evaluated in several studies with somewhat variable results [5,6,9,10].
The continued development and use of accurate diagnostic tests in disease surveillance studies provides critical information to public health decision makers with respect to clinical management, disease prevention, and infection control strategies [11].In response to this need, an in-house ELISA was developed for the diagnosis of acute typhoid fever in an endemic area during surveillance activities.The performance of this ELISA was evaluated against bacterial culture, the Widal test, and the commercialavailable serologic tests, TUBEX-TF and Typhidot (IgG and IgM).

Patient enrollment and sample collection
Patients presenting with acute febrile illness (AFI) from five Egyptian fever hospitals, located in distinct geographical regions, were enrolled in a surveillance study between June 2003 and May 2007.Informed consent was obtained from all adult participants and from parents or legal guardians of minors.
Admission was based on a standard AFI case definition per World Health Organization recommendations: any individual with fever lasting for at least two days or temperature on admission of ≥ 38.5°C; age 4 years with no identified cause of fever, such as diarrhoea or pneumonia; or clinically suspected of having typhoid fever or brucellosis [12,13].Blood for culture and serum samples were collected on admission and sent to the US Naval Medical Research Unit No. 3 (NAMRU-3) laboratory for diagnostic testing, confirmation, and archiving.

Study serum samples
A total of 2897 serum samples collected from AFI patients and healthy subjects were included in this work.Based on culture and/or serological results, specimens were divided into three groups: typhoidpositive (n = 332), typhoid-negative (n = 2525) and healthy controls (n = 40).Typhoid-positive samples included 234 culture-confirmed cases, and 98 cases with positive Widal test results (antibody titer > 320).The typhoid-negative group included samples collected from patients with AFI of known (n = 730) and unknown (n = 1795; negative for all tests) etiologies.Serum samples collected from AFI patients of known etiology included three subgroups: brucellosis (n = 290) confirmed by culture and/or serology, rickettsiosis (n = 290) diagnosed serologically by ELISA (antibody titer > 400) and leptospirosis (n = 150) confirmed by culture and/or the microscopic agglutination test (MAT).
To compare the newly developed typhoid immunoglobulin IgM, IgG and IgA ELISAs to the commercially available TUBEX-TF IgM and Typhidot IgM and IgG kits, a subset of serum samples (388/2897, 13%) was used.This subset included 165 typhoid-positive (67 culture-and 98 Widal test-confirmed cases) samples, 216 typhoidnegative samples from febrile patients with known and unknown etiology (73 and 143, respectively) and seven healthy patient control samples.

Laboratory methods
Blood cultures were performed using BACTEC culture media (Beckton-Dickenson, Cockeysville, MD, USA) per the manufacturer's instructions.Leptospira cultures were performed as previously described [13].Widal tube agglutination (TA) tests were performed for all cases enrolled in the study according to the manufacturer's instructions (Beckton-Dickenson and Company, Sparks, Maryland, USA) using BD-Difco Salmonella flagellar d (TH), BD-Difco Salmonella O group D and Brucella abortus antigens.Agglutination titers > 320 were considered positive [14,15].TUBEX-TF (IDL Biotech, Bromma, Sweden) and Typhidot (Malaysian Biodiagnostic Research, Bangi, Malaysia) serological tests were performed according to the manufacturers' recommendations.Other serological methods used in this study included ELISAs for diagnosis of brucellosis [15], rickettsiosis [16] and leptospirosis (Panbio Limited, Brisbane, Australia).

Typhoid ELISA development
Standard checkerboard titrations [17] were performed to determine the optimum conditions for loading S. Typhi somatic (TO) and flagellar (TH) antigens (the same ones mentioned previously for tube agglutination) onto ELISA plates using positive and negative anti-S.Typhi antisera.The TO and TH antigens were mixed in a 1:3 ratio and loaded onto the plates at 2.5 μl/ml in carbonate buffer.Flatbottom polystyrene ELISA plates (Thermo Labsystems, Franklin, MA, USA) were coated by adding 100 μl of the diluted antigen to each well and incubating overnight at 4°C.After decanting the antigen solution, the plates were washed three times with PBST (phosphate buffer saline [Sigma-Aldrich, St. Louis, MO, USA] containing 1% Tween 20).The coated plates were then blocked by adding 200 μl/well of blocking buffer (carbonate buffer with 1% w/v bovine serum albumin, BSA [Sigma-Aldrich, St. Louis, MO, USA]) and incubating for one hour at 37°C.After decanting the blocking solution, the plates were washed three times with PBST as before or stored at -20°C for future use (validated for up to one year).Each serum sample tested was diluted in PBST with 0.1% w/v BSA at 1:160 and 100 µl was added to each well (in duplicate).For each plate, positive and negative serum controls diluted in PBST buffer were also added.After incubation for one hour at 37°C, plates were washed as described above and 100 µl/well peroxidase-labeled goat anti-human (total Ig) conjugate (Sigma-Aldrich, St. Louis, MO, USA), diluted 1:20,000 in PBST with 0.1% BSA, was added.For detection of antibody isotypes IgG, IgM and IgA, the following respective conjugates were used: anti-human IgG and IgM (Sigma-Aldrich, St. Louis, MO, USA); anti-human IgA (Jackson Immuno Research Laboratories, West Grove, PA, USA).The optimum dilution for each conjugate was determined as 1:15,000 for IgG, and 1:10,000 for both IgM and IgA.Plates were incubated for 30 minutes at 37°C, then washed three times as described above.Ophenylene diamine (OPD) substrate solution (SigmaFast, Sigma-Aldrich, St. Louis, MO, USA) was prepared according to the manufacturer's instructions, 100 µl was added to each well, and the plates were incubated for 30 minutes at ambient temperature (approximately 21 to 24°C) in the dark.The reaction was stopped by adding 50 µl of stopping solution (1 to 2 M H2SO4) to each well.Color intensities were read using a BioTek microplate reader (Labsystems, Vantaa, Finland) at 490 nm.
Absorbance (A490) values obtained from the testing of serum samples from laboratory-confirmed non-typhoid and healthy control groups (18) were used to establish cut-off values: 0.3 for total Ig, IgG and IgM; and 0.25 for IgA.Thus any test sample that generated an A490 value greater than or equal to the cut-off value at a 1:160 dilution was considered initially reactive, while those that did not were considered non-reactive and reported as negative for anti-S.Typhi antibodies.
Initially reactive samples were serially diluted from 1:160 to 1:5,120 and tested to determine titer values.Samples giving titer values greater than 320 were considered reactive and reported as positive for anti-S.Typhi antibodies.
Sensitivities of the different methods used in this study were evaluated using culture-positive typhoid patients as true positives while the specificities were evaluated using typhoid-negative and healthy samples as true negatives.Assay sensitivity and specificity calculations were determined as reported previously [15,19].

Comparison of ELISA to the Widal test
Using sera from culture-positive patients, the newly developed total Ig ELISA showed improved sensitivity (93%, Table 1a) compared to that of the Widal test (71%, Table 1b) or previously published ELISAs [5,18,20].Additionally, among these culturepositive samples, the total Ig ELISA detected 75% (50/67, Table 1b) of those not detected by the Widal test, while the Widal test detected only 1% (1/67, Table 1b) of culture positive samples testing negative for the total Ig ELISA.
This difference in sensitivity may be due to the detection limit of each assay: ELISA may be sensitive to antibody concentrations as low as 0.05 ng, while agglutination tests such as Widal may require concentrations greater than 500 ng [20].
Within the probable typhoid patient samples (culture-negative, Widal-positive), the total Ig ELISA detected 78% (Table 1a and 1b).This sensitivity level, although lower than that observed for cultureconfirmed typhoid samples, is still comparable to or higher than those previously reported for the Widal- diagnosed typhoid [18,22,23].Alternatively, these observed data may be interpreted to indicate poor specificity of the Widal test, rather than poor sensitivity of the total Ig ELISA.To investigate this, the Widal-positive samples were further analyzed according to their titer values and evaluated against the total Ig ELISA.A direct correlation was seen between Widal test antibody titer and total Ig ELISA positivity.ELISA total Ig detected 58%, 79%, 89% and 100% of Widal-positive typhoid patients with antibody titers of 320, 640, 1280 and 2560, respectively.The Widal test has been modified over the years, particularly regarding the determination of antibody titers considered to be clinically relevant [24].Titers against the somatic (TO) and flagellar (TH) antigens were set at values ranging from 1:20 [25], 1:40 [26], 1:80 [27] to 1:160 [28].Accordingly, we may recommend that interpretation of Widal test results with low antibody titers (< 320) should be made with caution.For example, studies using the Widal test in typhoid endemic regions may be problematic due to repeated exposure to S. Typhi or cross-reactivity with other diseases (e.g., malaria), making it difficult to establish a steady-state baseline titer for the population [29].
The specificity values of the total Ig ELISA, using sera from patients with laboratory-confirmed brucellosis, rickettsiosis, leptospirosis, fever with unknown etiology and healthy subjects, were 97%, 93%, 96%, 96% and 100%, respectively (Table 1a).Taken together, the specificity of the total Ig ELISA (96%) was superior to that of the Widal test {76% [6]; 80% (this study; data not shown); 85% [9]} and previously published ELISAs [5,18,20,22].Although both tests utilize the Salmonella somatic (TO) and flagellar (TH) antigens, antigen-antibody interactions in the ELISA are more specific, compared to the whole cell agglutination reactions detected in the Widal test.In addition, the difficult subjective nature of reading agglutination reactions may contribute to the Widal test's lower specificity.

Comparison of ELISA to other commercially available tests
The newly developed ELISA was further optimized for detection of the individual antibody isotypes (IgG, IgM and IgA) produced against S. Typhi infection and evaluation of the distribution of the different antibody isotypes among acute typhoid patients.Among the individual isotype ELISAs, IgM was found to be the most sensitive (79%) while IgA the least (57%).Our results indicate that no antibody isotype is predominant, with a combination of two or three antibody isotypes detected in more than 85% of the samples tested (Table 2).Previous reports for other AFI of bacterial origin (e.g., brucellosis) indicate that the textbook immune response (initial IgM response followed by class-switching to IgG) may not occur.Also, acute and chronic stages of bacterial AFI may not be distinct, leading to the presence of more than one antibody isotype at the time of testing [15,30].The performance of the total Ig and each individual isotype ELISA was then compared to the commercially-available TUBEX-TF and Typhidot IgM and IgG test kits (Table 2) using a selected subset of serum samples (388/2897, 13%; see Methodology for subset details).Considering first the culture-confirmed samples, the total Ig ELISA showed superior sensitivity (93%) when compared to all other tests.
In addition, sensitivity values of the individual isotype ELISAs were comparable to TUBEX-TF and Typhidot IgM and higher than that of Typhidot IgG.When results from the IgG, IgM and IgA ELISAs were combined, the sensitivity of detecting antityphoid antibodies improved significantly, approaching that of the total Ig ELISA and congruent with that of a previous study [31].
The specificities of the different methods are also indicated in Table 2. TUBEX-TF showed the lowest specificities (85% and 88% for AFI of known and unknown etiologies, respectively), while the other assays varied between 94% and 99%.All assays were 100% specific when testing healthy control samples.The specificity levels of the different methods evaluated in this study are encouraging when compared to that of the Widal test.
The total Ig ELISA showed strong correlation (r = 0.85) with culture isolation.In comparison, TUBEX-TF and Typhidot (combined IgG and IgM) showed more moderate correlation (r = 0.65 and 0.69, respectively).Typhidot correlated moderately with ELISA and TUBEX-TF (r = 0.43 for each), while   *** In one ELISA run up to 200 samples could be tested and their results could be reported.For Typhidot, each kit has reagents enough for 30 samples and the total test time is almost one hour.However, using TUBEX-TF and Typhidot kits, we found difficulties running more than10 samples per run correlation between ELISA and TUBEX-TF was poor (r = 0.2).
Past studies have shown that the use of TUBEX-TF and Typhidot yields highly variable sensitivity and specificity profiles, depending on the country and/or geographical region, study population, and nature of the study (Table 3).This has created difficulties in comparing results between studies and setting worldwide standards for typhoid fever diagnosis.One study revealed that the TUBEX-TF and Typhidot tests were no better than the Widal test [9], underlining the need for a more efficient diagnostic test for typhoid fever, especially during the acute stage of the disease.

Practical aspects
Practical aspects of operating each of the assays in this study were also considered as part of the evaluation (Table 4).The total Ig ELISA was relatively easy, fast and cost-effective, costing 11 and 30 times less per sample compared to the cost of the TUBEX-TF and Typhidot kits, respectively.Of the three assays, the total Ig ELISA was also capable of the highest throughput; this is of particular value when processing large batches of samples during surveillance studies or outbreak events.For example, each operator can test up to 200 ELISA samples per run compared to less than 15 samples per run for the other assays.We also believe results interpretation to be superior for ELISA, as digital formats with preset cut-off values are used to differentiate between positive and negative reactions, not requiring subjective evaluations for color intensities of spots or bands produced.

Conclusion
Typhoid fever continues to be a major public health problem worldwide.Surveillance studies provide critical information for guiding public health decisions related to the disease.Due to the nature of these surveillance studies, participating laboratories receive large numbers of samples that are tested in batches.Though current clinical diagnostic tests such as Widal, TUBEX-TF and Typhidot are more suitable and convenient than ELISA for bedside testing and in small-scale, inadequately equipped or field laboratories, an effective ELISA is superior in routine surveillance studies or outbreak investigations where high throughput is important.This study demonstrates the development of a total Ig ELISA, with superior sensitivity and specificity rates to existing tests, that is more cost-effective when testing large numbers of samples.

Table 1a .
Sensitivity and specificity of the Typhoid ELISA (total Ig).Patient samples tested were culture-confirmed typhoid, probable typhoid (by Widal test), febrile patients with known non-typhoidal etiology (brucellosis, rickettsiosis and leptospirosis), febrile patients with unknown etiology and healthy controls.
*N/A: Not applicable

Table 1b .
ELISA total Ig and Widal test results for typhoid patients confirmed by culture (n = 234) and by Widal only (n = 98).

Table 3 :
Sensitivity and specificity of the Typhidot and TUBEX-TF tests for different sample populations from different geographical areas * Typhidot IgM testing ** Rates using confirmed typhoid cases