A prospective study was conducted in the year 2010. Samples were collected from patient age between seven to 70 years old and without history of typhoid fever. Sera and saliva samples were collected from Hospital Universiti Sains Malaysia (HUSM) after approval by the Ethical Committee (Human) of Universiti Sains Malaysia (USM) accordance with Declaration of Helsinki. Eleven patients grouped under ‘typhoid fever cases’ constituted of patients with prolonged fever (fever more than three days) and were positive for S. Typhi by blood culture and Typhidot^® test. All 43 cases grouped under ‘non-typhoid fever cases’ were randomly selected cases of fever of more than three days duration, negative for S. Typhi culture and for Typhidot^® test. This group included cases of dengue, leptospirosis, infective diarrhea, urinary tract infection, and fever of unknown origin. All samples were collected before start of any therapy at the hospital within one to three days of registration. ‘Normal human control cases’ consisted of randomly selected subjects who did not have any history of typhoid fever or typhoid vaccination during their life time, and had no clinically detectable signs of any infection at the time of sample collection, and were negative for S. Typhi culture and Typhidot^® test. Informed consent and information of the duration of fever, antibiotic usage and previous history of typhoid fever or vaccination were obtained from the specimen donors. All participants’ blood were collected by routine venipuncture and saliva were collected as described by Nurkka A et al., [31]. The participants were asked to refrain from any food or liquid intake 30 minutes to an hour prior to collection of saliva to avoid the effects of food. About 3-5 ml of saliva was collected in sterile containers by drooling method over a period of 10-15 minutes after washing of the mouth by gargling at least three times with sterile distilled water. The collected samples were centrifuged at 2,000 rpm and the supernatant were stored in aliquots at -20ºC until ready for used.

The antigen was isolated from a clinical strain of S. Typhi (S1188/03) from the stool of a typhoid fever patient during an outbreak in year 2003 in Kota Bharu, Kelantan, Malaysia and stored in the bacterial bank of the Institute for Research in Molecular Medicine (INFORMM), USM, Kelantan, Malaysia. The growth and maintenance of bacteria was done as described by World Health Organization (WHO) [32].

The antigen was prepared as described by Choo KE et al., as used in the Typhidot^® assay with all the quality controls in place [13].

Test sera (n=107) were screened using Typhidot^® assays for both IgM and IgG antibodies and graded as per the instructions given in the test kit brochure. An in-house Dot-EIA method was used for detection of serum IgA antibodies. Paired saliva samples from all the three groups were tested for presence of IgA, IgM and IgG antibodies to the 50 kDa antigens. Initially, the concentration of antigen, saliva sample volume, dilution of secondary HRP-conjugated antibodies to be used and time of incubation were optimized by checkerboard titration. The optimum conditions as shown in [Table/Fig-1] were used for testing the test samples. Briefly, nitrocellulose strips dotted with the 0.03 µg/µl 50 kDa antigens, blocked with 1% bovine serum albumin overnight and air-dried at room temperature were incubated with 1 ml of neat saliva sample in sample troughs for one hour at room temperature on an orbital shaker (Heidolph Polymax 2040, Australia), followed by three times washing with 1xPhosphate-Buffered Saline (PBS), pH 7.4 for 15 minutes each prior to incubation with 1 ml of secondary antibody for one hour. The strips were separated into three troughs for incubation with HRP-conjugated rabbit anti-human IgA and IgM (Dako, Denmark) at 1:200 dilutions, and HRP-conjugated rabbit anti-human IgG (Dako, Denmark) at 1:400 dilution. On completion of the incubation period, the strips were washed three times with 1xPBS, pH 7.4 for 15 minutes each. The reaction was visualized by incubating with 1 ml HRP-substrate consisting of 0.01% hydrogen peroxide and chromogen (0.003% 4-chloro-1-naphthol in methanol) for 15 minutes at room temperature, and stoppage of the reaction by washing with distilled water. The colour intensity of the dots were recorded and graded on a scale from ‘-’ (negative-no staining), ‘1+’ (positive-weak staining), ‘2+’, ‘3+’ to ‘4+’ (positive-moderate to strong staining) by comparing the reaction colour given by a reference positive saliva. Each sample was tested in triplicates. The results were tabulated and evaluated for sensitivity, specificity, Positive Predictive Value (PPV) and Negative Predictive Value (NPV) using the criteria described by NCSSM Statistics Leadership Institute for Categorical Data Analysis in July 1999 [33].

The results of the Ig-class specific screening showed that number of paired samples positive for salivary anti-50 kDa IgA and IgM antibodies were higher compared to those in serum, suggesting that salivary IgA and IgM were more sensitive compared to its counterpart (serum) [Table/Fig-2]. Out of 11 typhoid fever samples, 10 samples (90.9%) were positive for salivary IgA anti-50 kDa antibody, whereas only four samples were positive for serum IgA (36.4%). For salivary IgM antibody, 8 samples (72.7%) were positive compared to 7 samples (63.6%) for serum IgM [Table/Fig-2]. In contrast, salivary IgG anti-50 kDa antibody was found in 8 samples (72.7%) compared to 10 serum samples (90.9%) [Table/Fig-2].

False-positive results were observed in the non-typhoid fever and normal control groups [Table/Fig-2]. For salivary IgA, 5 out of 43 non-typhoid fever (11.6%) and 5 out of 53 normal control (9.4%) samples showed false-positivity, whereas for salivary IgM, 4 out of 43 non-typhoid fever (9.3%) and 1 out of 53 normal control (1.9%) samples showed false-positivity. No false negative results (0.0%) were observed using serum and salivary IgG for diagnosis of typhoid fever [Table/Fig-2].

By combining all isotypes (IgA, IgG or IgM), serum showed higher sensitivity (100.0%), specificity (100.0%), PPV (100.0%) and NPV (100.0%) compared to saliva (90.9%, 85.4%, 41.7% and 98.8%, respectively) [Table/Fig-3]. Overall, this study showed that salivary IgA was found to be more suitable for routine screening test due to its rapidness and high sensitivity (90.9%), whereas serum IgG was more suitable for confirmation test as it has higher specificity.

Saliva has been used by a number of investigators as a diagnostic fluid for various diseases [17-21]. However, very few studies have been carried out on typhoid fever patients using saliva. Herath et al., evaluated salivary IgA antibodies against Lipopolysaccharides (LPS) of S. Typhi with assay sensitivity of 83% and specificity of 97%, respectively [17]. A more recent study by Zaka-ur-Rab et al., found that S. Typhi LPS to have a sensitivity and specificity of 71.4% and 100%, respectively [34]. Also, a study by Chin KL et al., showed that salivary IgA against S. Typhi Haemolysin E protein can be used as a biomarker to differentiate acute typhoid, convalescent typhoid and other febrile diseases [35]. In this study, the potential of saliva as a diagnostic fluid and antibody against 50 kDa antigen were evaluated. Salivary IgA anti-50 kDa was more sensitive (90.9%) to diagnose typhoid fever than using salivary IgG (72.7%) and IgM (72.7%). This could be due to high concentration of IgA antibody in saliva. A similar observation was made by Romero MR et al., whereby a high concentration of IgA antibodies with traces of IgG and IgM present in parotid saliva. The mean concentration of salivary IgA in healthy subjects (81.0 µg/ml) was 400 times higher than salivary IgM (0.2 µg/ml) and five times higher than salivary IgG (14.8 µg/ml) [36]. This finding of an increased level of specific IgA in saliva in comparison to that in serum was also found in other studies [17,20,23]. This is consistent with the normal physiological occurrence of a higher IgA in saliva compared to serum. As observed in this study, the diagnosis of typhoid fever using levels of salivary IgG antibody as a diagnostic marker was found to be lower than that of serum. Even though, serum had high titres of IgG antibodies, the lower sensitivity of salivary IgG could be hypothesized to be due to a possible threshold level of antibodies are being required in the serum before they leak across capillaries and appear in the saliva as suggested by Nurkka A et al., [37].

The salivary antibodies showed false positivity in patients with two cases of leptospirosis, two cases of unknown fevers, one case of abdominal pain, one case of urinary tract infection and two cases of dengue among the non-typhoid fevers subjects, and six of 53 normal human controls. Thus, in this current study, the false positive rate for salivary antibodies against S. Typhi 50 kDa antigen in non-typhoid fever patients and normal individuals was 14.58% (14/96). Of these, the intensity of the reaction was high (4+) only in two cases of dengue fever and one case of leptospirosis. All the normal cases showed low intensity positivity (1+). The false negative result for salivary antibody detection of typhoid fever in this study was 9% (1/11 culture positive cases). The false positive and negative rates observed for saliva in this study were comparable to those in other studies. Herath HM et al., reported a false positive rate of 7.8% (4/51) and a false negative rate of 17% (4/29) [17], while Zaka-ur-Rab Z et al., reported a false positive rate of 10.8% (4/37) and false negative rate of 0% (0/60) in the evaluation of saliva for typhoid fever detection using anti-LPS antibody as the biomarker [34].

The occurrence of high IgA positivity in the saliva of culture negative non-typhoid fever cases could not be simply explained at this juncture. These results could not be simply dismissed as ‘false positive cases’ because they were S. Typhi culture-negative. Since the test subjects were from a typhoid endemic region, the frequency and level of anti-S. Typhi antibodies in this population could be higher. Also, non-specific binding may arise due to 3 factors: 1) HRP-conjugated anti-human IgA antibody binding to excess IgA found in saliva; 2) presence of cross-reactive antibodies in the test specimens which bind to both the 50 kDa antigen and other outer membrane proteins in Salmonella enterica species and; 3) low sensitivity of the test for detection of serum IgA antibodies which made it difficult to differentiate between antibody positive and negative samples. This study also found a high discrepancy in IgA results of the paired serum and saliva. This is due to the absent of a gold standard for assessing the predictive cut-off value for antibody level.

Using the salivary anti-50 kDa antibody detection assay developed in this study as biomarker for diagnosis of typhoid fever, approximately 9% of the results were false negative and 15% were false positive. As the sample size used in this study was small due to the low number of cases of typhoid fever cases reported in Kelantan in the year 2010 [38], a larger sample size is needed to increase the accuracy (sensitivity and specificity) of the study. This antigen also showed cross-reactivity with dengue and leptospirosis infections, and thus differential diagnosis should be performed to rule out the possibilities of co-infections. Also, serial samples should be obtained to test its accuracy to diagnose typhoid fever.

Overall, this study showed that saliva has the potential to be an alternative fluid for detection of typhoid fever due to its characteristics of: a) greater acceptability (non-invasive and painless); b) greater convenience (simple, rapid, and inexpensive collection with no requirement of specialized training); c) less hazardous (to subjects as well as to investigators, e.g., needle pricks); and d) better access and easy availability (in epidemic outbreaks, children, large populations and hard-to-reach high risk groups). The testing does not require sophisticated equipment and can be done in a rural clinic or even in the doctor’s consultation room for POCD. It is very much suitable in children, they being the highest infected group for this disease. Despite the modest PPV of the test, the salivary antibodies could be a good biomarker to detect immune responses to typhoid infection. When the positivity to any class of antibody against 50 kDa antigen was considered as the marker, the detection by salivary antibodies was found to have a high sensitivity (90.9%) and specificity (85.4%) for typhoid fever.