Ethics statement

This study was approved by the Research Ethics Committee of the Faculty of Medicine, Chiang Mai University (No. EC0090/2568), and the Research Ethics Committee of the Affiliated Hospital of Youjiang Medical University for Nationalities (No. ECYYFY-2025–279). Written informed consent was obtained from all participants prior to enrollment in the study.

2.1 Serum samples

All samples were collected between September 2023 and October 2025 from the Clinical Pathology Unit, Maharaj Nakorn Chiang Mai Hospital (a tertiary hospital affiliated with Chiang Mai University), Chiang Mai, Thailand, and the Department of Clinical Laboratory, Affiliated Hospital of Youjiang Medical University for Nationalities (Fig 1). The samples were stored at -80 °C until use.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Flow diagram of the selection of 460 clinical samples included in this study.

The culture-confirmed T. marneffei infection group included 79 patients (Table A in S1 File). The non-T. marneffei control group consisted of 381 participants, including 139 patients with other microbial infections (Table B in S1 File), 170 patients without evidence of infection, and 72 healthy individuals from endemic areas. The 309 patients with or without additional isolated pathogens were epidemiologically similar to the 79 patients with T. marneffei. The serum samples were obtained from the Clinical Pathology Unit, Maharaj Nakorn Chiang Mai Hospital (a tertiary hospital affiliated with Chiang Mai University), Chiang Mai, Thailand, and the Department of Clinical Laboratory, Affiliated Hospital of Youjiang Medical University for Nationalities, China.

https://doi.org/10.1371/journal.pntd.0014314.g001

The diagnostic performance of the biotin-labeled 4D1 sandwich ELISA was evaluated using 460 clinical serum samples (79 patients with a positive T. marneffei culture were enrolled as the positive group, 309 patients with negative T. marneffei cultures, and 72 healthy individuals were enrolled as the negative control group) (Table 1). The non-T. marneffei control group consisted of 381 participants, including 139 patients with other microbial infections (Table B in S1 File), 170 patients without evidence of infection, and 72 healthy individuals from endemic areas. The 309 patients with or without additional isolated pathogens were epidemiologically similar to the 79 patients with T. marneffei. Positive T. marneffei cultures were confirmed by thermal dimorphic transition between 25 °C and 37 °C and microscopic analysis for phenotypic characterization. The cohort included both male (53) and female (26) patients, aged from 1 to 74 years (Table A in S1 File). The sample size was estimated based on the expected diagnostic performance of the sandwich ELISA. The parameters were set as follows: a type I error rate (α) of 0.05, a 95% confidence level, and a desired precision (margin of error) of 10%. The expected sensitivity and specificity were both set at 0.9, based on previously reported performance of similar immunoassays for T. marneffei [39,40]. Sample size estimation was performed according to the methods described by Buderer and Hajian-Tilaki [47,48]. Although no universal minimum sample size exists for diagnostic accuracy studies, previous work has suggested that studies with several dozen cases may provide initial evidence under certain conditions [49–51]. Accordingly, a minimum sample size of approximately 70 participants was considered sufficient for this study. Further details regarding samples from patients with infections caused by other pathogenic microorganisms are available in the supplementary materials (Table B in S1 File).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Baseline clinical characteristics of study participants.

https://doi.org/10.1371/journal.pntd.0014314.t001

2.2 T. marneffei cytoplasmic yeast antigen (TMCYA) preparation

Briefly, T. marneffei conidia were inoculated into brain heart infusion (BHI) broth (Becton Dickinson, Sparks, MD, USA) and cultured for 6 days at 37 °C, 150 rpm. After treatment with 0.02% (w/v) Merthiolate (Sigma-Aldrich, Poole, UK) at room temperature for 12 h, the yeasts were centrifuged for 10 min at 2,300 × g and the cell walls were mechanically broken with 0.5-mm glass beads (BioSpec, Bartlesville, OK, USA) in a bead beater homogenizer (BioSpec, Bartlesville, OK, USA). A cocktail of protease inhibitors including iodoacetic acid (IAA; 1 mM; GE Healthcare, Little Chalfont, Buckinghamshire, UK), phenylmethanesulfonyl fluoride (PMSF; 0.1 mM; Tokyo Chemical Industry Co., Ltd., Tokyo, Japan), and 1 mM ethylene diamine tetraacetic acid (EDTA; Merck, Darmstadt, Germany), were added to prevent fungal protein degradation. The mixture was centrifuged for 30 min at 4 °C at 10,000 × g, and the total antigen was collected [52]. The protein concentration of the antigen preparations was determined using a bicinchoninic acid (BCA) protein assay kit (Thermo Fisher Scientific, USA) according to the manufacturer’s instructions. The antigens of other fungi were prepared following the same standardized procedure. The fungal strains used in the current study are listed in Table 2.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Antigen preparations were derived from the diverse fungal isolates.

https://doi.org/10.1371/journal.pntd.0014314.t002

2.3 Purification of mAb 4D1

The murine-derived hybridoma cell line 4D1 was cultured in hybridoma serum-free medium (Gibco, Grand Island, NY, USA). The culture supernatant of mAb 4D1 was concentrated with a Vivaspin 20 ultrafiltration device (MWCO: 30 kDa; GE Healthcare, Little Chalfont, Buckinghamshire, UK). The supernatant was filtered through a 0.45 µm membrane filter (Corning Incorporated, Corning, NY, USA). After that, the concentrated 4D1 supernatants were purified by ÄKTA start chromatography system (Cytiva, Uppsala, Sweden) with HiTrap protein G affinity chromatographic column (GE Healthcare, Uppsala, Sweden). Protein G was covalently linked to Sepharose bead for capturing mAb 4D1, which belongs to the mouse IgG1 subclass. MAb 4D1 was eluted by 0.1 M glycine-HCl pH 2.7. Then, the purified antibody solution was immediately neutralized by adding 60 µL of 1 M Tris-HCl pH 9.0. Eluted fractions were screened by indirect ELISA. The eluted fractions containing mAb 4D1 were pooled and dialyzed in PBS pH 7.4. Final mAb 4D1 concentrations were measured by Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Rockford, IL, USA).

The purity of mAb 4D1 was validated using 10% SDS-PAGE. To confirm the mouse IgG subclass of mAb 4D1 by immunoblotting, SDS-PAGE gels containing separated polypeptides were transferred electrophoretically to nitrocellulose membranes (Hybond extra; Amersham, Little Chalfont, Buckinghamshire, UK). The membranes were washed with PBS and then blocked overnight at 4 °C with PBS-0.05% Tween 20 containing 5% skim milk (Sigma-Aldrich, St. Louis, MO, USA). Later, the membranes were washed with PBS-0.05% Tween 20 and incubated for 1 hour with HRP-conjugated goat anti-mouse IgG (Jackson, West Grove, PA, USA) diluted 1:3000 in PBS-0.05% Tween 20. The membranes were washed, and the bound conjugates were visualized by incubation with TMB/H₂O₂ chromogenic substrate (BioFX Laboratories, Surmodics, Eden Prairie, MN, USA). The reactions were stopped by submersion of the membrane in distilled water.

2.4 Biotin-4D1 IgG conjugation

The conjugation of mAb 4D1 IgG with biotin was performed according to the manufacturer’s (Thermo Fisher Scientific, Waltham, MA, USA) instructions. Briefly, the antibody was buffer-exchanged to remove amine-containing components by dialysis against 0.1 M sodium phosphate, 0.15 M sodium chloride (pH 7.2) for 1 week. The dialyzed antibody solution was concentrated using 10 kDa Vivaspin 6 centrifugal concentrators, and the protein concentration determined by BCA assay. Then, 2 mg/mL of purified 4D1 IgG was reacted with freshly prepared 10 mM Sulfo-NHS-LC-Biotin (Thermo Fisher Scientific, Waltham, MA, USA) in ultrapure water. For conjugation, 27 µL of biotin solution was added per 1 mL of antibody solution and incubated on ice for 2 hours. Following conjugation, 10 mM Tris-HCl was added and incubated for 1 hour to quench unreacted NHS esters. The biotinylated antibody was then dialyzed extensively for 1 day to remove unbound biotin before storage.

2.5 Verification of biotin-labeled 4D1 conjugation

Validation of the biotin-labeled 4D1 conjugation was performed using 10% SDS-PAGE and Western blotting. A total of 12 protein antigens from 10 fungal species were analyzed by SDS-PAGE, including cytoplasmic antigens of T. marneffei obtained from its conidial, mold, and yeast phases. The same set of proteins was loaded onto five identical gels at a final concentration of 10 μg per lane. After electrophoresis, the separated polypeptides were transferred onto nitrocellulose membranes. One gel was stained with Coomassie Brilliant Blue to visualize protein bands, whereas the remaining four gels were used for membrane transfer.

Following transfer, the membranes were blocked overnight at 4 °C in 3% (w/v) BSA in 20 mM Tris-HCl (pH 7.4) containing 0.15 M NaCl and 0.05% Tween-20. After blocking, all membranes were washed with PBS-0.05% Tween-20. The first membrane was incubated with mAb 4D1 for 2 hours, followed by incubation with HRP-conjugated goat anti-mouse IgG (Jackson ImmunoResearch, West Grove, PA, USA) diluted 1:3000 in PBS-0.05% Tween-20 for 1 hour. The second membrane was incubated with biotin-labeled 4D1 (20:1, diluted 1:1000 in 1% (w/v) BSA prepared in 10 mM Tris-HCl, pH 7.4) for 2 hours and subsequently incubated with streptavidin-HRP diluted 1:6000 in 1% (w/v) BSA prepared in 10 mM Tris-HCl (pH 7.4) for 1 hour. The remaining two membranes served as negative controls. For these controls, mAb 4D1 and biotin-labeled 4D1 were omitted and replaced with either PBS-0.05% Tween-20 or 1% (w/v) BSA in 10 mM Tris-HCl (pH 7.4), respectively, prior to incubation with HRP-conjugated goat anti-mouse IgG (1:3000) or streptavidin-HRP (1:6000) for 1 hour. After the final washing steps, bound conjugates were visualized using a TMB/H₂O₂ chromogenic substrate. The reactions were stopped by rinsing the membranes in distilled water.

2.6 Biotin-labeled 4D1-based sandwich ELISA procedure

Fig 2 illustrates the principle of the biotin-labeled 4D1-based sandwich ELISA. Briefly, 100 µL of purified 4D1 monoclonal antibody (50 μg/mL in 0.06 M carbonate buffer, pH 9.6) was coated onto MaxiSorp 96-well plates (Greiner Bio-One GmbH, Frickenhausen, Germany) and incubated at 37 °C for 90 minutes. The wells were then washed six times with phosphate-buffered saline containing 0.1% Tween-20 (PBST) and blocked with 200 µL of 3% (w/v) BSA in 20 mM Tris-HCl (pH 7.4) containing 0.15 M NaCl and 0.05% Tween-20 for 2 hours at room temperature. After blocking, 100 µL of serum samples (diluted 1:10 in PBS containing 0.1% BSA and 0.05% Tween-20) was added to each well in triplicate and incubated for 2 hours at room temperature.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Schematic representation of the biotin-labeled 4D1-based sandwich ELISA for the detection of TMCYA.

(1) Capture mAb 4D1 is coated on the solid phase. (2) TMCYA antigen binds to the immobilized antibody via its Asn(Asparagine)-containing epitope. (3) Biotin-labeled 4D1 detection antibody binds to a second epitope on TMCYA, forming the sandwich complex. (4) Streptavidin-HRP associates with the biotin-labeled 4D1. (5) Addition of chromogenic substrate yields an HRP-catalyzed colored product for detection.

https://doi.org/10.1371/journal.pntd.0014314.g002

Following washing, 100 µL of biotin-labeled 4D1 antibody (1:1000 dilution in 10 mM Tris-HCl containing 3% BSA) was added and incubated for 90 minutes. Then, the wells were washed and incubated with 100 µL of streptavidin-HRP (1:6000 dilution in 10 mM Tris-HCl with 3% BSA) for 1 hour at room temperature.

After a final wash step, 100 µL of 3,3’,5,5’-tetramethylbenzidine (TMB; Surmodics IVD, Eden Prairie, MN, USA) substrate was added and allowed to develop for 15 minutes before stopping the reaction with 50 µL of 2 M H₂SO₄. Optical density was measured at 450 nm (ELISA reader, Shimadzu model UV-2401PC, Kyoto, Japan).

Key assay parameters, including coating antibody concentration, serum dilution, biotin-labeled 4D1 dilution, and streptavidin-HRP dilution, were systematically optimized through preliminary serial dilution experiments. For the optimization experiments, TMCYA-spiked serum samples were prepared at an initial concentration of 100 μg/mL. Serial dilutions were performed stepwise using negative serum or appropriate buffer to obtain dilution ratios ranging from 1:2–1:100. Each subsequent dilution was prepared from the preceding solution. The performance of each condition was evaluated based on both the P/N ratio and the difference between the mean OD values of positive and negative samples (ΔOD), allowing identification of optimal assay parameters with maximal signal discrimination and minimal background interference. The selected conditions (5 μg/well coating antibody, 1:1000 biotin-labeled 4D1 (20:1), 1:6000 streptavidin-HRP, and 1:10 serum dilution) were determined based on these optimization experiments. The corresponding optimization results are presented in the Supplementary Materials (Tables C-E and Figs D-F in S1 File).

2.7 Application of the ELISA: standard curve, specificity, and sample testing

To evaluate the performance of the biotin-labeled 4D1 sandwich ELISA, serum samples were collected from ten healthy volunteers residing in T. marneffei-endemic areas. The sera were pooled to generate a reference sample. This pooled serum was used for two purposes: (1) preparation of a standard curve via serial dilution to establish assay linearity and dynamic range, and (2) evaluation of assay specificity by comparing the biotin-labeled 4D1 sandwich ELISA signal with clinical sera from patients with culture-confirmed T. marneffei infection. All samples were derived from the same cohort described in Section 2.1 and were collected under the same institutional ethics approvals.

2.7.1 Standard curve construction.

For standard curve construction, pooled serum samples were diluted 1:10 and spiked with TMCYA at concentrations ranging from 0.781-1000 μg/mL. The biotin-labeled 4D1 sandwich ELISA was performed as described in Section 2.6 to generate a standard curve and assess assay linearity and dynamic range.

The standard curve was fitted using a four-parameter logistic (4-PL) regression model of the form:

where A and D represent the asymptotic minimum and maximum responses, B is the slope factor, and C is the inflection point (EC50) [53,54]. This model was used to interpolate antigen concentrations from OD values.

2.8 Statistical analysis

Statistical analyses and graphing were performed using GraphPad Prism (version 10.0; GraphPad Software, San Diego, CA, USA) and MedCalc Statistical Software (version 23.4; MedCalc Software Ltd., Ostend, Belgium). GraphPad Prism was used to generate four-parameter logistic (4-PL) standard curves, receiver operating characteristic (ROC) curves, and to calculate the area under the curve (AUC), limit of detection (LOD), and optimal cut-off values. Diagnostic accuracy parameters, including Cohen’s kappa (κ), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and their 95% confidence intervals (CIs), were calculated using MedCalc. Positive (LR⁺) and negative likelihood ratios (LR⁻) were derived to evaluate the diagnostic performance. P < 0.05 was considered statistically significant.

3.1 Purification of mAb 4D1

The purity of the mAb 4D1 fractions was assessed at the elution step by monitoring the UV absorbance peak (Fig BA in S1 File). Subsequently, the collected fractions were analyzed by SDS-PAGE (Fig BB in S1 File). The IgG subclass of mAb 4D1 was determined by Western blot with HRP-conjugated goat anti-mouse IgG, confirming that mAb 4D1 belongs to the IgG1 subclass (Fig BC in S1 File).

3.2 Verification of biotin-4D1 conjugation

Successful conjugation of biotin-labeled 4D1 was confirmed by SDS-PAGE and Western blotting (Fig 3). In the SDS-PAGE gel stained with Coomassie Brilliant Blue, all ten fungal cytoplasmic antigen preparations were clearly resolved, displaying a broad molecular weight distribution (Fig 3A). When the membrane was probed with the mAb 4D1 followed by HRP-conjugated goat anti-mouse IgG, a distinct immunoreactive signal was observed exclusively for TMCYA, which migrated between approximately 50–150 kDa, whereas no detectable reactivity was observed for other fungal antigens (Fig CA in S1 File). This result is consistent with the findings reported by Pruksaphon et al. [34]. Similarly, incubation with the biotin-labeled 4D1 followed by streptavidin-HRP resulted in a detectable signal exclusively for TMCYA, whereas all other fungal antigens remained non-reactive (Fig 3B). As negative controls, membranes incubated with HRP-conjugated goat anti-mouse IgG or with streptavidin-HRP (without addition of mAb 4D1 or biotin-labeled 4D1) showed no detectable signal for any of the fungal antigens (Fig CB and CC in S1 File). Collectively, these results confirm successful biotin labeling of mAb 4D1 and demonstrate that the conjugate retains high specificity for TMCYA, supporting its use in subsequent experiments.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Verification of biotin-labeled 4D1 antibody conjugation and its binding specificity for TMCYA.

(A) Coomassie Brilliant Blue-stained 10% SDS-PAGE of ten fungal cytoplasmic antigen preparations (10 μg/well), demonstrating successful protein extraction and clear band separation across a broad molecular weight range (11-180 kDa). (B) Western blot of the same antigen panel probed with biotin-labeled 4D1 followed by streptavidin-HRP detection. A positive signal was observed exclusively in the TMCYA lane (lane 2), producing a broad immunoreactive band spanning approximately 50-150 kDa. No reactivity was detected against cytoplasmic antigens derived from T. marneffei conidia or mold forms, or from other fungal species tested under the experimental conditions, supporting the high specificity of the 4D1 conjugate for TMCYA. Lane assignments are as follows: lane 1, molecular weight marker; lane 2, T. marneffei yeast; lane 3, T. marneffei conidia; lane 4, T. marneffei mold; lane 5, C. albicans; lane 6, P. kudriazevii; lane 7, Geotrichum sp.; lane 8, S. apiospermum; lane 9, Trichosporon sp.; lane 10, S. schenckii (yeast); lane 11, C. neoformans; lane 12, P. citrinum; lane 13, A. fumigatus.

https://doi.org/10.1371/journal.pntd.0014314.g003

3.3 Standardization of the biotin-labeled 4D1 sandwich ELISA

A standard curve for the biotin-labeled 4D1 sandwich ELISA was established using pooled serum spiked with TMCYA at concentrations ranging from 0.781 to 1000 μg/mL (Fig 4). Serial dilutions were assayed with two replicate wells per concentration, and the mean OD₄₅₀ values from three independent experiments were used to generate the curve. The four-parameter logistic (4-PL) regression demonstrated excellent reproducibility and a strong concentration-dependent response. The coefficient of determination (R²) for the fitted curve was 0.997, indicating that 99.7% of the variability in OD₄₅₀ values could be explained by the TMCYA concentration. This high R² value reflects a highly reliable model with minimal deviation between the experimental data points and the fitted curve. Based on the standard curve, the detection limit for T. marneffei antigen spiked into pooled serum was determined to be approximately 19.398 μg/mL.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Standard curve of the biotin-labeled 4D1 sandwich ELISA for the detection of TMCYA.

A standard curve was generated using serial dilutions of pooled serum spiked with TMCYA at concentrations ranging from 0.781 to 1000 μg/mL. Each concentration was assayed in triplicate, and the results are presented as mean ± standard deviation (SD) from three independent experiments. The curve was fitted using a four-parameter logistic (4-PL) regression model and demonstrated excellent reproducibility. Within this concentration range, the coefficient of determination (R²) between the OD values and TMCYA concentrations was 0.997, indicating a highly reliable fit. The detection limit of the assay for T. marneffei antigen spiked in pooled serum was approximately 19.398 μg/mL.

https://doi.org/10.1371/journal.pntd.0014314.g004

3.4 Analytical performance of the assay (LOB, LOD, and LOQ)

To further evaluate the analytical performance of the assay, the LOB, LOD, and LOQ were determined. The analytical LOB was calculated to be 0.168, and the LOD was determined to be 0.175 based on low-concentration samples. The OD value corresponding to the LOD was converted to antigen concentration using the four-parameter logistic (4-PL) standard curve, yielding an LOD of 19.398 μg/mL. The LOQ was defined as the lowest concentration that could be quantified with acceptable precision (CV < 20%). Based on the observed variability of low-concentration samples, the LOQ was estimated to be 25.0 μg/mL. The determination of LOB, LOD, and LOQ is illustrated in Fig G in S1 File.

3.5 Specificity of the biotin-labeled 4D1 sandwich ELISA

To assess the specificity of the biotin-labeled 4D1 sandwich ELISA, we tested serial concentrations (0.781-1000 μg/mL) of cytoplasmic antigens from T. marneffei (yeast, mold, and conidia) and 9 other medically relevant fungi. As shown in Fig 5, only TMCYA produced a strong and proportional increase in OD₄₅₀ values across the tested concentration range, while T. marneffei mold and conidia generated only baseline-level signals similar to the non-T. marneffei fungi. Two-way ANOVA demonstrated a highly significant interaction between fungal species and antigen concentration (P < 0.0001), confirming that the observed dose-dependent response is unique to TMCYA. These findings indicate that the biotin-labeled 4D1 sandwich ELISA is highly species- and phase-specific, recognizing only the TMCYA without cross-reactivity to other fungal species.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. The specificity of the biotin-labeled 4D1 sandwich ELISA.

Serial concentrations (0.781-1000 μg/mL) of cytoplasmic antigens from T. marneffei (yeast, mold, and conidia) and 9 non-T. marneffei fungi were evaluated using the biotin-labeled 4D1 sandwich ELISA. Each concentration was assayed in duplicate across three independent experiments, and data are presented as mean ± SD. A clear dose-dependent increase in OD₄₅₀ was observed exclusively for TMCYA, whereas T. marneffei mold and conidia as well as all other fungal antigens produced only minimal background signals without a detectable response curve. Two-way ANOVA revealed a highly significant interaction between fungal species and antigen concentration (P < 0.0001), demonstrating that only TMCYA is specifically recognized by the assay.

https://doi.org/10.1371/journal.pntd.0014314.g005

3.6 Detection of serum samples by the biotin-labeled 4D1 sandwich ELISA and diagnostic performance

In total, 460 serum samples were subjected to the biotin-labeled 4D1 sandwich ELISA, including 79 samples from patients with T. marneffei infection and 381 samples from the control group without T. marneffei infection. The OD₄₅₀ values of the T. marneffei infection group and the control group were significantly different (Fig 6A). This result implied that mAb 4D1 was specifically reactive to TMCYA in serum, allowing the biotin-labeled 4D1 sandwich ELISA to distinguish patients with T. marneffei infection from controls. The ROC curve was constructed by plotting sensitivity (true positives) versus the corresponding 1-specificity (false positives) at various OD cut-off values (Fig 6B). The area under the ROC curve (AUC) is an important index of accuracy of the curve [57].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Diagnostic performance of the biotin-labeled 4D1 sandwich ELISA for detecting TMCYA in serum.

(A) Comparison of OD₄₅₀ values between culture-confirmed T. marneffei-positive patients and the control group. Each point represents an individual serum sample. The median OD₄₅₀ in the culture-positive group was significantly higher than that of controls (Mann-Whitney test, P < 0.0001). A diagnostic cut-off value of 0.268 (dashed blue line) was determined based on ROC curve analysis. (B) Receiver operating characteristic (ROC) curve of the assay. The area under the curve (AUC) was 0.983 (95% CI: 0.972-0.994, P < 0.0001), indicating excellent diagnostic accuracy. The optimal cut-off point was identified at the maximal Youden index.

https://doi.org/10.1371/journal.pntd.0014314.g006

In this study, the AUC was 0.983 (95% CI: 0.970-0.993, P < 0.0001), which indicated that the biotin-labeled 4D1 sandwich ELISA offered excellent detectability of talaromycosis (Fig 6B). The operating point in the upper left corner of the curve indicated by the arrow was chosen as the cut-off with an OD₄₅₀ value of 0.268 (Fig 6A). This cut-off was determined based on the maximum Youden index derived from ROC curve analysis, providing the optimal balance between sensitivity and specificity (Youden index = sensitivity + specificity − 1) [58]. The corresponding diagnostic performance was illustrated in Tables 3 and 4. The biotin-labeled 4D1 sandwich ELISA correctly diagnosed T. marneffei infection in 70 of 79 (88.61%) cases (95% CI, 79.47%-94.66%), suggesting no significant difference compared to the culture assay (P > 0.05). The test correctly excluded T. marneffei infection in 366 of 381 (96.06%) control individuals (95% CI, 93.59%-97.78%). The accuracy was 94.78% (95% CI, 92.34%-96.63%). The positive likelihood ratio (LR) was 22.51, indicating an association with disease. The negative LR was 0.12, suggesting an association with the absence of disease. Overall, there was a practically perfect agreement between the biotin-labeled 4D1 sandwich ELISA and the culture assay (kappa = 0.824, 95% CI: 0.739-0.885, P < 0.001).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Diagnostic performance of the biotin-labeled 4D1 sandwich ELISA compared with culture (reference standard).

https://doi.org/10.1371/journal.pntd.0014314.t003

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Diagnostic performance of the biotin-labeled 4D1 sandwich ELISA when testing serum samples from individuals with or without T. marneffei infection.

https://doi.org/10.1371/journal.pntd.0014314.t004