Evaluation of a genus-specific rGroEL1-524 IgM-ELISA and commercial ELISA kits during the course of leptospirosis in Thailand

In the present study, we developed a genus-specific rGroEL1-524 IgM-ELISA assay for use in screening diagnosis of suspected leptospirosis among acute undifferentiated febrile illness patients during acute fever. The diagnostic accuracies of the rGroEL1–524 IgM-ELISA, commercial Panbio IgM-ELISA, and Virion-Serion Classic IgG-ELISA were evaluated using 133 Thai leptospirosis sera and 210 controls. Sensitivities were 91.7%, 59.6%, and 17.7% for acute infection, and the specificities were 92.6%, 90.2%, and 88.3% for the non-leptospirosis control, respectively. The rGroEL1-524 IgM-ELISA had high sensitivity, at 92.3% and 91.7%, among culture-positive and MAT-negative cases at 1–3 days post-onset of symptoms (DPO1–3), respectively. Impaired specificity on scrub typhus was found, possibly from antibody cross-reaction to ortholog GroEL. Commercial Panbio IgM-ELISA had sensitivities at DPO1–3 of 30.8% and 41.7% for culture-positive and MAT-negative cases whereas Virion-Serion IgG-ELISA showed sensitivities of 5.9% and 13.3%, respectively. The rGroEL1-524 IgM-ELISA could be useful as a screening test for early diagnosis. The performance of the commercial ELISA suggests the applicability of IgM-ELISA for diagnosis, while IgG-ELISA is useful for seroprevalence surveys. However, confirmation by reference tests is recommended.


Results
Verification of GroEL 1-524 sequence conservation within the genus Leptospira spp. and among orthologous GroEL proteins. The Leptospira GroEL 1-524 sequence had a high degree of homology, at 99% (522/524) sequence identity to other leptospiral serovars in the Leptospira spp. To evaluate their protein-sequence conservation across genera, GroEL orthologs among other tropical infectious diseases including leptospirosis, scrub typhus, melioidosis, and malaria were examined. Orthologous GroEL proteins of influenza A and dengue viruses were not found in the NCBI database. The orthologous GroEL proteins of L. interrogans, Burkholderia pseudomallei, Plasmodium vivax, and Orientia tsutsugamushi organisms demonstrated 60.5%, 51.3%, and 51.9% identity, respectively, compared with the cloned Leptospira GroEL 1-524 sequence. The GroEL 1-524 sequence was highly conserved in the genus Leptospira spp. and shared conserved peptides among orthologous GroEL proteins.
Prediction of GroEL 1-524 linear B-epitope peptides. Two B-epitope peptides of 30-LGPKGRN-36 (85.7% identity) and 404-AAVEEGIVPG-413 (100% identity) have been shown to be highly conserved among IgM sera reactivity of leptospirosis and controls to rGroEL 1-524 antigen by IgM-ELISA. The IgM antibody reactivities of leptospirosis paired sera in an optimized rGroEL  IgM-ELISA were presented as actual optical density (AOD) values with a range of 0-1.01. Median AOD values were 0.40 and 0.43 for leptospirosis paired sera, and were significantly higher than those of the non-leptospirosis control (P < 0.0001) and other febrile illness (P < 0.001). The IgM reactivities of leptospirosis sera versus scrub typhus sera were not significantly different (P = 0.15, P = 0.09, respectively) (Fig. 1). The median IgM reactivities of culture positive samples and seroconversion were 0.50 and 0.45 for acute sera and 0.25 and 0.34 for convalescent sera, respectively (Fig. 1).

Determination of optimal cut-off values.
To optimize cut-off values, receiver operating characteristic (ROC) curves were generated from leptospirosis paired sera and controls to define the optimal optical density (OD) of 0.15 to achieve an estimated sensitivity of 91.6% and 95.5% for the paired sera and a specificity of 87.5% for the controls.
The results of IgM detection by Panbio IgM-ELISA and IgG detection by Virion-Serion IgG-ELISA on leptospirosis sera and controls are illustrated with adjusted cut-off values in supplementary Fig. S2. The optimal threshold for Panbio IgM-ELISA cut-off value was ≥ 7 Panbio units, and the optimized cut-off value for Serion IgG-ELISA was ≥ 0.35 OD ELISA to achieve higher sensitivities.
Diagnostic accuracy of rGroEL 1-524 IgM-ELISA. Thai blood samples were randomly selected to evaluate rGroEL  IgM-ELISA performance compared with reference methods. The diagnostic sensitivities were 91.7% for acute sera and 95.6% for convalescent sera ( Table 1).
Analysis of false-positive results among controls from rGroEL 1-524 IgM-ELISA. IgM positivity among controls by rGroEL 1-524 IgM-ELISA was 7.5% (supplementary Table S4). All false-positive samples were negative for anti-Leptospira IgM and IgG detection by commercial ELISA, suggesting that the false positives in the non-leptospirosis control may have arisen from non-specific binding or pre-existing antibodies (background antibody) in those samples.
The analysis of false-positive samples among the other febrile illness controls is summarized in supplementary Table S4. Melioidosis samples had 8.3% (1/12) false-positive results by rGroEL 1-524 IgM-ELISA. Serum showed anti-Leptospira IgG positivity using the commercial test, which suggested antibody cross-reactivity from previous exposure. The scrub typhus sera showed IgM reactivity ranging from 0.03 to 1.07; 45.5% (5/11) were deemed false positives. Two of the false positives had anti-Leptospira IgM positivity, which suggested antibody crossreactivity from recent or current exposure or leptospirosis-scrub typhus coinfection. Three of the false-positive results were likely caused by antigen cross-reactivity to ortholog GroEL antigen or pre-existing antibody from endemic leptospirosis.
Two of three false-positive malaria samples had anti-Leptospira IgG positivity by Virion-Serion IgG-ELISA detection. The false-positive reactions might be related to pre-existing antibodies from previous exposure. One malaria sample had false-positive result by an in-house IgM-ELISA, which suggested antigen cross-reactivity to ortholog GroEL antigen or pre-existing antibody. www.nature.com/scientificreports/ Two of ten influenza samples had false-positive results by IgM-ELISA (AOD, 0.16-0.2) at 1:100 dilution, but all were negative for anti-Leptospira IgM and IgG detection by the commercial tests. False positives were likely caused by non-specific reactions. False positives of dengue samples were 40% (4/10) by in-house IgM-ELISA. Three of the false-positive samples tested negative for anti-Leptospira detection, which suggests that false positives may arise from endemic background antibodies. One false-positive dengue sample had anti-Leptospira IgM positivity, which suggested recent or current infection and leptospirosis-dengue coinfection.

Discussion
Genus-specific antigen-based ELISAs using immunodominant outer membranes as antigens, such as LipL32, LipL41, Loa22, LigA, Lsa63, GroEL, and a combination (multiple antigens), have been widely developed for use as a screening test for leptospirosis [35][36][37][38] . The present study developed a prototype IgM-ELISA using a recombinant GroEL 1-524 formatted antigen as an early laboratory screening test for leptospirosis and evaluated its diagnostic accuracy in the context of disease outbreaks in Thailand compared with reference methods. We produced rGroEL 1-524 protein (C-terminal deletion of 22 amino acids) and used an ELISA antigen to detect anti-rGroEL  IgM antibody during the course of illness for early diagnosis of suspected cases among AUFI caused by other infections. The heat shock GroEL chaperonin has shown as a diagnostic potential in leptospirosis based on its upregulated expression during infection (temperature upshift). GroEL has been shown to be an immunodominant antigen and has less cross-reactivity with melioidosis and dengue hemorrhagic fever [33][34][35] . The immunoreactivity of severe leptospirosis, such as pulmonary involvement and renal failure, to recombinant GroEL has been shown to have 90.6% sensitivity and 94.9% specificity 39 . In addition, the GroEL 1-524 sequence is highly conserved within the genus Leptospira and shares lower sequence conservation with the orthologous GroEL. We evaluated the diagnostic performance of the rGroEL 1-524 IgM-ELISA using leptospirosis paired-sera derived from northeastern Thailand, i.e., Loei, Nakhon Ratchasima, Sakol Nakhon, and controls from non-endemic Bangkok and other febrile illnesses, compared to MAT and culture methods. Leptospirosis sera from Loei were collected from an outbreak in 2002. Samples were found to be positive by Leptospira isolation (30 culture-positive acute sera) and by seroconversion criteria. A small sample size of 28 single MAT ≤ 1:400 leptospirosis sera was obtained.
In the present study, a single IgM-ELISA was designed as a highly sensitive screening test. The cut-off was determined to be 0.15 AOD for single IgM-ELISA testing to achieve an estimated sensitivity of 91.6% and 95.5% for paired sera and specificity of 92.5% and 76.0% among the non-leptospirosis and febrile controls, respectively. One limitation of acute-phase IgM testing with a single specimen is that people in endemic areas are expected to have pre-existing antibodies causing impaired specificity. ELISA results give no indication of the infecting serovar, and a confirmatory diagnosis of leptospirosis should be performed. IgM antibody usually persists for 5 months 40 , ELISA can be used as a simple and rapid laboratory screening test for the diagnosis of leptospirosis for several months after the onset of symptoms.
The sensitivities of the rGroEL 1-524 IgM-ELISA were 91.7% and 95.6% for leptospirosis paired sera, and the specificity was 92.6% among the non-leptospirosis control. Lessa-Aquino et al. 35 reported GroEL IgM-ELISA sensitivities of 90% and 92.0% and specificities of 53.8% and 62.5% in paired sera. A systematic review and metaanalysis of the performance of Leptospira IgM-ELISA averaged 84% sensitivity and 91% specificity for acute infection 41 . The rGroEL 1-524 IgM-ELISA had higher diagnostic performance than previously reported 35,41 due to the ability of the refined rGroEL 1-524 molecule to encompass more antigenic moieties of the whole genus. The prototype IgM-ELISA had high sensitivities of 95.9% in culture-positive sera, 91.2% in seroconversion samples, and 88.2% in MAT-positive subgroups. With a cut-off of 0.2 AOD, expected test performance was 87.5% and 86.7% for sensitivity in paired sera and 81.5% specificity in the febrile control. The most prevalent serovars infecting patients in the sera used in this study were Bratislava, Autumnalis, Australis, New, Sarmin, and Bangkok 5 , while Autumnalis, Bratislava, and Pyrogenes were the most common serovars in Thailand in 2003-2012 42 . The in-house IgM-ELISA can detect IgM antibody as early as DPO1. The false negative results in the acute phase by in-house IgM-ELISA might be due to the long window by the dynamics of antibody production. Symptomatic patients may have no or low antibody levels at 1-2 weeks post-exposure, and the antibody titer will rise with time. We found two false negative convalescent sera, which might have been caused by a delayed response, which sometimes occurs over 30 days after infection 20 .
Leptospirosis infections are often under-reported due to false negatives among mild cases or those who have already received antibiotics, have suppressed immunity, or are in the very early or late phase of the immune response. In coinfection patients, weak or cross-reactions may occur. The false-positive rate among febrile patients is possibly caused by cross-reactivity, anti-GroEL 1-524 IgM antibody from leptospirosis co-infections, or pre-existing IgM antibody in patients with recent exposure in endemic areas.
Commercial ELISA tests have been used for the diagnosis of leptospirosis in Thai endemic settings, including the Panbio Leptospira IgM-ELISA and Virion-Serion classic Leptospira IgM/IgG. The performance of commercial ELISA tests varies by geographical setting, with the sensitivity of the Leptospira IgM-ELISA being 35-76% and specificity being 76-98% in different endemic settings [27][28][29][30][31][32]41 . These ELISA tests use whole-cell lysates from pathogenic L. interrogans, intermediate L. fainei, or saprophytic Leptospira biflexa antigens to detect genusspecific anti-Leptospira IgM/IgG antibodies. Heterogeneous native antigens in ELISA tests may not recognize the local serovars, so their sensitivities are frequently poor and have been limited by the heterogeneity of host immunological responses to native antigens.
Whole cell-based ELISAs (Panbio IgM-and Virion-Serion IgG-ELISA) demonstrated poor sensitivity against local Thai leptospirosis paired sera in the present study (Table 1). Another study found that the Panbio IgM-ELISA showed 90.8% positivity among samples from northeastern Thailand 32 . IgM antibodies appear earlier than IgG antibodies and remain detectable at low titers for months or even years. An IgG titer of 1:100 can be present due to past infection. Whole cell-based ELISA, which is affected by serogroup-specific antigens or whole-cell  45 . Another study examined the potential risk of a leptospirosis outbreak in Bangkok and Nakhon Pathom between 2011 and 2012, and found L. wolffii and intermediate L. licerasiae 46 .
Poor sensitivity can be attributed to several factors, such as acute serum being collected too early in the course of illness (less than DPO4-5), inadequate IgM antibody levels in the patient, second or subsequent episode of infection leading to IgG antibody production, and patient receiving antibiotic medication. To improve specificity due to high background antibodies among the seropositive population requires validation and adjustment of the cut-off. In this study, we optimized cut-offs for commercial ELISAs. An adjusted Panbio unit of ≤ 7 showed a sensitivity improvement to 54.8% for the DPO1-3 acute phase and specificity of 86.6%. The Panbio IgM-ELISA provided sensitivity and the ability to detect IgM antibodies as early as DPO1-3. Virion-Serion IgG-ELISA with an adjusted cut-off provided 17.7% and 48.9% sensitivity on leptospirosis paired sera and 81.7% specificity among the controls. False-positive IgG detection was 19% among the controls (10% for non-endemic samples and 25% for each infection, i.e., dengue, malaria, scrub typhus, and melioidosis). IgG seropositivity rates of 17.7% and 48.9% for paired sera suggest that IgG responses should be due to epidemic leptospirosis in Loei rather than background antibody, with 10% seropositivity among healthy and AUFI patients in low-prevalent leptospirosis areas such as Bangkok.
Several studies have reported that ELISA-based assays detect anti-Leptospira IgM antibodies earlier than MAT assay during the early course of disease 20,47,48 . Nicofa et al. 20 suggested that Leptospira-specific IgM antibodies appear 1-2 days earlier than the agglutinating antibodies detected in the MAT assay; therefore, earlier positive results could be expected from our genus-specific IgM detection. We found that the sensitivities in seronegative and culture-positive acute sera were 91.2%, and 95.9% for the prototype IgM-ELISA and 63.2% and 50.0% for commercial IgM-ELISA, respectively. An unvalidated diagnostic test with poor specificity may contribute to overdiagnosis of leptospirosis, because IgM antibodies from past infections are frequently detected among people living in endemic areas 20,48 .
The cross-reactivity of the rGroEL 1-524 IgM-ELISA was evaluated using sera from the local population and a non-leptospirosis febrile control group. The specificity of anti-Leptospira IgM detection is limited in pathogens expressing orthologous GroEL proteins, such as scrub typhus, malaria, and melioidosis, causing IgM crossreactivity in the rGroEL 1-524 IgM-ELISA. High anti-Leptospira IgM levels in sera collected from patients along the Thai-Myanmar border have been reported 49,50 . However, it should be noted that cross-reactivity with bacterial infections can occur when patients harbor co-infections or have cross-reactive antibodies, especially in the early phase of leptospirosis when the IgM-ELISA lacks full specificity 20 .
The varied sensitivities likely reflect different case definitions and control groups, timing of collection, local prevalent serovar distribution, and the platform and protocol used in detection. A significant limitation of the rGroEL 1-524 IgM-ELISA was poor specificity for leptospirosis-endemic areas. The test specificity was affected by ortholog GroEL antigen cross-reactivity, antibody cross-reactivities caused by previous exposure, and coinfections. Co-infections with leptospirosis were not assessed in samples from other febrile illnesses. The use of the rGroEL 1-524 IgM-ELISA as a screening test for leptospiral infection would facilitate the difficult reference and differential tests. However, the test should not be used as the sole criterion for diagnosing leptospirosis. The ELISA results must be confirmed by convalescent serum. MAT is still recommended for disease confirmation and epidemiological study, and Leptospira isolation and molecular characterization should be performed for confirmation of the infecting serovars 51 .

Conclusion
Our data demonstrated that IgM-ELISA using rGroEL 1-524 antigen has sufficiently high sensitivity to screen for anti-GroEL 1-524 IgM antibodies in the early leptospirosis diagnosis of suspected cases and among high-risk groups during leptospirosis epidemics. However, diagnostic specificity needs to be improved for implementation in areas with high levels of infectious tropical diseases. The commercial ELISA performance data suggest the applicability of IgM-ELISA for early diagnosis during disease outbreaks in low-prevalence areas for leptospirosis. IgG-ELISA is useful for seroprevalence surveys; however, confirmation by reference tests is recommended.

Methods
Ethics and biosafety. The protocol for using achieved sera and patient data was performed in concordance with the recommendation of the Declaration of Helsinki. Documentary Proof of Exemption Review was obtained from the Ethics Committee of the Faculty of Tropical Medicine, Mahidol University (MUTM-EXMPT2017-005). The written informed consents were obtained from participants. Sample anonymity was maintained, and all samples were re-coded without name and hospital ID. Biosafety was approved by the institute's Biosafety Committee (MU2019-002).

Study design.
A retrospective study was carried out to assess the diagnostic performance of rGroEL  IgM-ELISA and to evaluate commercial whole-cell antigen-based ELISA performance using local Thai blood samples compared with culture and MAT methods. The laboratory investigations were conducted at the Faculty of Tropical Medicine, Mahidol University, in Bangkok. Leptospirosis patients and sera. A suspected leptospirosis case was clinically diagnosed based on WHO criteria, i.e., AUFI in patients (fever ≥ 38 °C) with headache and myalgia and a history of exposure to animal reservoirs or flooded environments 17 . A confirmed leptospirosis case is defined as a clinically diagnosed, suspected leptospirosis case combined with positive laboratory diagnosis by the culture method or MAT assay. Leptospirosis sera (n = 133) were obtained from patients during an epidemic outbreak at Loei Provincial Hospital (n = 95) and sporadic cases in the Nakhon Ratchasima and Sakhon Nakhon provinces (n = 38) 5,36 (Fig. 2). Leptospirosis sera (n = 133) were acute sera (n = 52) and classified as DPO1-3 (n = 34), DPO 4-10 (n = 12), and convalescent sera (n = 51). Among the sera, samples with a single MAT titer of 1:100-1:200 (n = 26) were excluded. The most prevalent serogroups (serovars) among the MAT-positive sera were Autumnalis (Autumnalis, New), Australis (Australis, Bangkok, Bratislava), Icterohaemorrhagiae (Copenhageni), Sarmin (Sarmin), and Sejroe (Sejroe) 5 .

Control samples.
To assess the specificity of the ELISA tests, a panel of control samples (n = 210) consisting of 60 non-leptospirosis plasma (seronegative and negative for leptospirosis IgM detection) and 150 laboratoryconfirmed infectious diseases other than leptospirosis were used (Fig. 2). Non-leptospirosis control were 60 plasma were collected from healthy volunteers and febrile plasma (n = 60) at the Hospital for Tropical Diseases,  ELISAs. Confirmed leptospirosis sera (n = 133) and control samples (n = 210) were subjected to an assessment of the diagnostic sensitivity and specificity of the following tests: (i) rGroEL 1-524 IgM-ELISA, and commercial (ii) Panbio Leptospira IgM-ELISA, and (iii) Serion-Virion classic Leptospira IgG-ELISA. Leptospirosis sera were acute sera collected on the date of admission (n = 52) and convalescent-sera collected in the later 2 weeks (n = 51). Acute-phase sera were classified according to days post-onset of symptoms (DPO) into DPO1-3 (n = 34), and DPO4-10 (n = 12) acute serum. Control samples were 51 non-leptospirosis plasma (seronegative and negative for leptospirosis IgM detection), 88 serum or plasma samples and 50 whole blood of laboratoryconfirmed infectious diseases other than leptospirosis. Of 343 sera, 26 leptospirosis sera and 21 control samples were excluded from study. Positive results of the commercial ELISAs were considered using the recommended cut-offs. www.nature.com/scientificreports/ eases, Bangkok, from 2013 to 2015. The samples included laboratory-confirmed murine typhus (n = 15), dengue (n = 30), and bacterial sepsis (n = 5) caused by E. coli, Streptococcus agalactiae, Salmonella Typhi, and Viridans Streptococci infections (Fig. 2) 14 . The samples were collected in microtubes and stored at − 70 °C. A soluble fraction from the IPTG-induced bacteria containing rGroEL 1-524 protein was prepared in phosphatebuffered saline (1× PBS, pH 7.4) using a French pressure cell press at 30 kilo-pounds per square inch, repeated four times. The rGroEL 1-524 protein was purified from the soluble proteins by native affinity chromatography using Ni 2+ -sepharose (GE Healthcare, Uppsala, Sweden). The purified rGroEL 1-524 protein was concentrated in 1× PBS (pH 7.4) using a 3-kDa cut-off Amicon Ultra filter (Merck Millipore, MA, USA), and the protein concentration was determined using Bradford assay (Thermo Fisher Scientific, MA, USA). Aliquots of the protein (1 mg/mL) were lyophilized using the Labcono Freeze Dry system and then kept at − 70 °C. In-house rGroEL 1-524 IgM-ELISA. Recombinant GroEL 1-524 (1 μg) immobilized ELISA strips (Jet Biofil, Guangzhou, China) were prepared as follows: rGroEL 1-524 protein in 100 µL of carbonate-bicarbonate buffer (pH 9.6) was immobilized on ELISA wells at 37 °C for 24 h, and the antigen-coated wells were washed using washing buffer (300 µL/well of PBST; 0.05% Tween 20 in 1× PBS, pH 7.4). Washing was conducted by an automated microplate washer (Tecan Trading AG, Switzerland) three times to remove unbound material. The coated wells were then incubated with blocking reagent (300 µL of 1% BSA in 1× PBS) for 1 h at 37 °C, followed by incubation of the pre-blocked wells with 2% sucrose solution (300 µL) at 25 °C for 1 h. The ELISA wells were washed after each incubation step, as described above, and then air-dried. The pre-blocked rGroEL 1-524 ELISA strips were packed with desiccant in press-seal bags and stored at − 20 °C until use.

SDS-PAGE and Western blotting.
To detect anti-GroEL 1-524 IgM antibody, serum dilution (1:100, 100 µL) in a serum diluent (1× PBS containing 0.2% gelatin, 0.2% BSA), along with an internal positive control (pooled MAT-positive patient sera, where the adjusted AOD exceeded 0.2) and a reagent control (serum diluent) were incubated in pre-blocked antigen-coated wells at 37 °C for 1 h, followed by washing three times with PBST. Thereafter, HRP-conjugated goat anti-human IgM antibody (100 μL, 1:2,000) (Southern Biotechnology, AL, USA) was added to ELISA wells at 37 °C for 1 h incubation. ABTS chromophore diammonium salt (EMD Millipore, Germany) substrate solution (1 mg/mL ABTS tablet in 0.1 M sodium citrate buffer) was added (100 µL), and the plate was incubated for 15 min at 37 °C, after which 1% SDS solution (100 µL) was added to stop the reaction. The OD was measured at a wavelength of 410 nm against the reference at 650 nm (OD 410nm/650 nm ) using a microplate reader (Bio-Tek Instruments, VT, USA). Sample AOD was calculated by subtracting the OD of the reagent blank. The IgM-ELISA assay is valid when the OD of the reagent blank is < 0.2 and the positive AOD control is ≥ 0.2. A rGroEL 1-524 IgM-ELISA protocol was optimized, and the optimal concentration of rGroEL 1-524 was 1 µg/well; serum dilution was 1:100 and secondary antibody dilution was 1:1000-1:3000 dilutions.
Panbio Leptospira IgM-ELISA. The diagnostic performance of the commercial Panbio Leptospira IgM-ELISA (Abbott Diagnostics, Illinois, USA) (Lot no. 02P10E001), using Leptospira genus-specific antigen, was assessed in Thai blood samples. The Panbio IgM-ELISA protocol was performed per the manufacturer's instructions, measuring absorbance at OD 450nm/650 nm. An index value was calculated by dividing the sample absorbance by the cut-off value. The result was expressed as Panbio units (index value multiplied by 10). Interpretation of the validity results was as follows: Panbio units (anti-Leptospira IgM) < 9 was a negative result, suggesting no Virion-serion classic Leptospira IgG-ELISA. Institute Virion-Serion ELISA Classic Leptospira IgG (Institut Virion/Serion GmbH, Warburg, Germany) (order no. ESR 125 G) was used to detect anti-Leptospira IgG antibody from serum or plasma using a crude membrane extract of L. biflexa serovar Patoc strain Patoc I, which contains genus-specific epitopes for all Leptospira spp. The Virion-Serion IgG-ELISA procedure was performed per the manufacturer's instructions, with absorbance measured at OD 405nm/650 nm. To interpret the qualitative results, the upper and lower cut-off range was calculated according to parameters provided with the kit. Actual OD (AOD) value (anti-Leptospira IgG) lower than the cut-off was considered a negative result suggesting no evidence of past exposure, an AOD value in the cut-off range was a borderline result suggesting possible past exposure, and an AOD value higher than the upper cut-off was positive by IgG detection, suggesting previous exposure. A borderline value was considered a positive result. The diagnostic performance of the Virion-Serion Leptospira IgG-ELISA was 96.7% sensitivity and 99.8% specificity.
Evaluation of diagnostic accuracy. The Standards for Reporting of Diagnostic Accuracy studies (STARD 2015) checklist for reporting diagnostic accuracy is provided in supplementary Table S6.
Sample size was estimated as a minimum of 35 cases and control samples to achieve 90% sensitivity and specificity at a 95% confidence interval (CI) and 7% precision. All sera were tested as anonymous samples. Leptospirosis sera and controls ( Fig. 1) were randomly selected to evaluate the performance of the following tests: (i) rGroEL 1-524 IgM-ELISA, (ii) commercial Panbio Leptospira IgM-ELISA, and (iii) Virion-Serion Classic IgG-ELISA. The estimated diagnostic sensitivity and specificity with 95% CI were calculated by a 2 × 2 crosstabulation table.
Bioinformatics. Conservation of selected GroEL sequences in the genus Leptospira and among GroEL orthologs was determined using Clustal Omega multiple sequence alignment program interface 53 , and the results were analyzed using the BioEdit sequence alignment editor tool. Linear B-cell epitopes of the L. interrogans serovar Icterohemorrhagiae GroEL sequence were computationally predicted using a Bepipred-1.0 Linear Epitope Prediction tool 54 Statistical calculations. Data were collected in Microsoft Excel and analyzed using MedCalc Statistical Software version 19.2.5 (MedCalc Software Ltd, Ostend, Belgium; https:// www. medca lc. org; 2020). Diagnostic parameters were calculated as follows: sensitivity = [(true positive (TP)/(TP + false negative (FN))] × 100; specificity = [(true negative (TN))/(TN + false positive (FP))] × 100. Normal distribution was tested using the Kolmogorov-Smirnov test. The Mann-Whitney test was used in non-normal distributed data. P < 0.05 was considered statistically significant.

Data availability
The datasets used and analyzed from the current study are available from the corresponding author on reasonable request.