Development and evaluation of a single-tube nested PCR with colorimetric assay for Mycobacterium tuberculosis detection

Abstract

Molecular techniques have revolutionized tuberculosis (TB) diagnosis by offering a faster and more sensitive approach, detecting Mycobacterium tuberculosis (Mtb) DNA directly from samples. Single-tube nested PCR (STNPCR) combines two PCR reactions with separate oligonucleotide sets in a single tube. Moreover, colorimetric methods in PCR products have been studied for pathogen detection. Thus, this study aimed to establish a novel system based on colorimetric STNPCR for Mtb detection using microtiter plates with IS6110-amplified fragments. The results showed a general colorimetric STNPCR detection limit of 1 pg/μl. Its general sensitivity and specificity were 76.62 and 60.53%, respectively, with kappa index agreement of 0.166.

Method summary

A total of 318 biological samples (urine, plasma, peripheral blood mononuclear cells, pleural fluid and sputum) from pulmonary/extrapulmonary TB and non-TB patients were used in this study. The colorimetric STNPCR assay using IS6110 as the target gene was developed and optimized for Mtb detection based on similar validated systems. Cut-off values based on receiver operator characteristic curve analysis were defined to determine the sensitivity and specificity for each sample type. The technique's performance was assessed according to kappa index calculations and interpretation.

Executive summary

• To improve rapid molecular tuberculosis (TB) diagnosis, a novel molecular test was developed and evaluated in different biological samples. This test is based on single-tube nested PCR (STNPCR) with colorimetric assay using IS6110 as the target gene for Mycobacterium tuberculosis (Mtb) detection.

Materials & methods

• A total of 318 biological samples from pulmonary/extrapulmonary TB and non-TB patients were used in this study.

• The colorimetric STNPCR assay used IS6110 as the molecular target optimized for Mtb detection based on validated similar systems.

Results & discussion

• The results showed a detection limit of 1 pg/μl according to the absorbance analysis.

• The colorimetric STNPCR assay general sensitivity and specificity were 76.62% and 60.53%, respectively, with kappa index agreement of 0.166.

• Its sensitivity and specificity for the biological samples were: plasma (55.5% and 80%), PBMC (72.3% and 54.3%), urine (61.3% and 51.5%), pleural fluid (75% and 100%) and sputum (64.7% and 66.7%).

Conclusion

• The study demonstrated a quick and straightforward test with promising results, suggesting the automated colorimetric STNPCR assay as a potential addition to laboratory routines for both pulmonary and extrapulmonary TB diagnosis.

Keywords: :

• IS6110
• Mtb complex
• rapid molecular test
• STNPCR
• tuberculosis diagnosis

Conventional laboratory diagnosis for tuberculosis (TB) is performed through Bacillus identification by microscopy or culture. However, these techniques have some limitations: they exhibit reduced sensitivity, especially for extrapulmonary TB samples, and cell culture is excessively time-consuming [1,2]. Alternatively, molecular techniques have been developed for the diagnosis of numerous infectious diseases, including TB [3–6].

When compared with conventional tests, molecular methods based on nucleic acid amplification (NAA), especially PCR, are faster, more sensitive and can detect Mycobacterium tuberculosis (Mtb) DNA directly from samples, giving a result in less than 2 h [5,7,8]. Although there are some molecular techniques recommended by the WHO, Xpert MTB/RIF and its variants currently stand as the primary diagnostic tests for TB worldwide and are the sole rapid molecular tests recommended by Brazilian guidelines [2,9]. They are fast and accurate methods, supporting the diagnosis of several pulmonary TB cases; however, Xpert MTB techniques have a high implementation cost, variable sensitivity for extrapulmonary samples and they are not validated for blood samples [2,10,11]. Therefore, new NAA tests are needed to improve the speed and accuracy of molecular TB diagnosis, mainly in resource-limited and high-burden countries [3].

Nested PCR (NPCR) consists of two PCRs with two sets of oligonucleotides in subsequent reactions. The amplified product in the first reaction is used as a template for the second reaction, offering more sensitivity and specificity for the technique [12–14]. Single-tube nested PCR (STNPCR) is a variant of NPCR and has the advantages of time optimization and presenting a low risk of cross-contamination since there is no need to open the tubes to add reagents [12,15]. In addition, the colorimetric detection technique is based on ELISA. This technique involves the use of a specific probe, complementary to the target DNA. The probe is immobilized on a plate and binds itself to the specific amplified sequence of the PCR products [16]. Some studies have demonstrated the utility of colorimetric methods in PCR products as a potential diagnostic tool for pathogen detection. This approach allows for adaptation of the molecular target in accordance with each infectious organism [17–19].

Several gene targets have been used in the molecular detection of Mtb. IS6110 stands out for its specificity to the mycobacteria within the Mtb complex, its high conservation and its presence in multiple copies [20]. Moreover, it is the main molecular target detected by WHO-recommended Xpert MTB techniques [11]. Given the paucibacillary nature of extrapulmonary TB samples, employing a multicopy gene is of paramount importance [20,21].

Thus, the aim of this study was to develop and evaluate a novel molecular protocol for Mtb detection using IS6110 as the molecular target, based on STNPCR with colorimetric assay in microtiter plates.

Materials & methods

Characterization of the study population

The present study included patients of both sexes with a wide range of ages and nonspecific symptomatology but suspected of having pulmonary or extrapulmonary TB. These patients were recruited from the Brazilian public health system in Pernambuco, Northeast Brazil, and grouped based on the following criteria:

• TB group: samples from patients who exhibited clinical and epidemiological evidence of TB, a positive tuberculin skin test and isolation of Mtb from clinical samples through direct detection tests or culture, or positive histopathological findings, or imaging results consistent with TB, or response to specific treatment, as recommended by the Brazilian guidelines [2].

• Non-TB group: patients with samples showing clinical and laboratory findings not compatible with TB. They had no contact with TB-confirmed adults and tested negative for a tuberculin skin test.

Selection of patients & sample collection

A total of 318 blood, urine, sputum, and pleural fluid samples were collected from 115 patients. The final TB diagnosis, following the Brazilian guidelines [2], was defined by the attending physician in the public health service. Patients already on TB treatment or with previous treatment lasting less than 1 year were excluded from the study. Therefore, 95 urine, 98 plasma, 100 peripheral blood mononuclear cells (PBMCs), 5 pleural fluid and 20 sputum samples were collected from 77 TB-positive (38 pulmonary and 39 extrapulmonary) and 38 TB-negative patients.

The study was approved by the Research Ethics Committee of IAM/FIOCRUZ (protocol no.: 45687215.2.0000.5190).

All patients received detailed information about the research and were invited to participate voluntarily. They completed standard questionnaires, gave written informed consent and provided biological samples for study analysis. For individuals under 18 years of age, their parents or guardians signed a consent form authorizing their participation in the research.

Sample processing & genomic DNA extraction

Biological samples were collected from the subjects as follows: blood in EDTA tubes (5 ml); urine in sterile tubes during three consecutive mornings (10 ml/day); sputum through spontaneous elimination (1–5 ml) in a single sterile tube during fasting; and pleural fluid in sterile tubes through puncture (2 ml) by a physician. PBMCs were separated from blood samples using Ficoll-Paque Plus (GE Healthcare, Uppsala, Sweden) density gradient centrifugation. Urine and sputum samples were decontaminated using Petroff's method, and then resuspended in 2 ml of sterile ultrapure water [2]. Decontaminated samples (urine and sputum) and pleural fluids (200 μl of each) were inoculated in Löwestein-Jensen (LJ) culture medium [2]. All biological samples (200 μl of each) underwent DNA extraction using a Qiagen Mini Prep Kit (Qiagen^®, USA), according to the manufacturer's instructions.

Single-tube nested PCR

The IS6110 insertion sequence of Mtb was the target used for the NPCR. Two sets of primers were employed. In the initial reaction, external primers TJ3 (5′-ATCCCCTATCCGTATGGTG-3′) and TJ5 (5′-CCGCAAAGTGTGGCTAAC-3′) were used, amplifying a 409-bp fragment. In the second reaction, the first PCR amplicons from the initial reactions were used as templates to amplify a 316-bp fragment, employing the internal primers: OLI5 (5′-AACGGCTGATGACCAAAC-3′) and STAN3 (5′-GTCGAGTACGCCTTCTTGTT-3′) [15].

The amplification reactions were performed in an automatic thermocycler (Eppendorf Gradient). The set of external oligonucleotides was used at a concentration of 0.5 pmol/μl in a final volume of 50 μl, containing: 200 mM Tris-HCl (pH = 8.4); 500 mM KCl (10× buffer); 2.5 mM MgCl₂; 2 mM DNTP × 5 μl; and 5U of Platinum Taq DNA Polymerase (Invitrogen Life Technologies, CA, USA). The first reaction consisted of 15 cycles (94°C – 1 min, 57°C – 1 min and 72°C – 1 min). The internal oligonucleotides were used at a concentration of 50 pmol/μl in the second PCR reaction, which consisted of 45 cycles (94°C – 1 min, 63°C – 1 min and 72°C – 1 min). A biotin molecule was added to the 5′ terminal region of OLI-5. For each STNPCR reaction, a negative control (10 μl of ultrapure water) and a positive control (2 ng of Mtb strain H37Rv DNA) were used. The PCR products were loaded in 1.5% agarose gel electrophoresis (AGE) with blue green loading dye staining and visualized under ultraviolet light. A Low DNA Mass Ladder (Invitrogen Life Technologies) was used as a molecular marker.

Internal control

To assess the DNA quality from biological samples, an internal control was created using genomic DNA from Mtb strain H37Rv, which was amplified by nested PCR. The amplified IS6110 product was subcloned into the pCR_4 TOPO vector using a TOPO TA Cloning Kit (Invitrogen Life Technologies), following the manufacturer's instructions. The recombinant plasmid (named pIS6110) was sequenced (ABI Genetic Analyzer 3100, Applied Biosystems, CA, USA), and its concentration was measured spectrophotometrically (514 ng/μ1) using an Ultraspec 3000 (Amersham Pharmacia Biotech, NJ, USA). Subsequently, a tenfold dilution curve (ranging from 10 ng to 1 fg) was carried out using the final cloning product. This curve served as a standard for each qPCR reaction, following the methodology described by [22].

Colorimetric assay

The colorimetric assay of the STNPCR products was optimized based on protocols from: Adikaram et al. [19], Michelon et al. [23], Scherer et al. [24], Verza et al. [25] and Wilson et al. [26]. The optimized protocol is presented in Box 1. The colorimetric assay involved two steps: plate sensitization with a probe and colorimetric detection. The probe's design (see Supplementary Material 1) featured the following sequence: 5′-TTTTTTTTTTTGGGTAGCAGACCTCACCTATGTGT-3′. Further details on the optimizations for the colorimetric STNPCR protocol are provided in Supplementary Material 1. The colorimetric STNPCR assay was performed in technical triplicates.

Box 1. Optimized protocol of the colorimetric single-tube nested PCR assay for Mycobacterium tuberculosis detection in biological samples.

Plate sensitization with the probe

1. Dilute the probe to 1 pmol/μl with carbonate buffer (pH 9.6)

2. Place 100 μl of the diluted probe solution to each plate well, cover with adhesive and store overnight (16 h–18 h) at 4°C

 Note: If using the plate after 18 h or more, store at -80°C until the next steps

3. Aspirate and discard the supernatant from each plate well

4. Add 300 μl of the blocking solution (1× PBS + 3% BSA) to each plate well and incubate the plate in a drying oven at 37°C for 60 min

5. Aspirate and discard the supernatant from each plate well

Colorimetric detection

1. Denature the STNPCR-amplified products in a thermocycler at 100°C for 5 min

2. Immediately place the STNPCR-denatured amplified products on ice to prevent renaturation

3. Add 100 μl of the hybridization solution (20× SSC + 0.1% Tween 20 + 0.5% BSA) to each plate well

4. Add 10 μl of each STNPCR-denatured amplified product to the appropriately labeled plate wells for each sample

5. Seal the plate and incubate for 2 h at 66°C in a hybridization incubator

6. Aspirate and discard the supernatant from each plate well

4. Add 300 μl of wash solution 2 (2× SSC + 0.1% Tween 20, pH 7.0) to each plate well. Aspirate and discard the supernatant from each plate well

 Note: Repeat the step 7 three times. This step may be performed by a microplate washer

8. Add 300 μl of heated (66°C) wash solution 2 to each plate well and incubate for 15 min at 66°C

9. Add 300 μl of wash solution 2 to each plate well. Aspirate and discard the supernatant from each plate well

 Note: Repeat the step 9 three times. This step may be performed by a microplate washer

10. Add 300 μl of blocking solution (1× PBS + 3% BSA) to each plate well and incubate for 1 h at 37°C

11. Add 300 μl of wash solution 2 to each plate well. Aspirate and discard the supernatant from each plate well

 Note: Repeat step 11 three times. This step may be performed by a microplate washer

12. Add 100 μl of streptavidin–peroxidase conjugate (1:2000 ratio in 1× PBS + 0.05% Tween 20)

13. Aspirate and discard the supernatant from each plate well

14. Cover the plate and incubate for 30 min at 37°C

15. Add 300 μl of wash solution 3 (1× PBS with 0.5% Tween 20, pH 7.2) to each plate well. Aspirate and discard the supernatant from each plate well

 Note: Repeat the step 15 five times. After the second and fourth washes, rest the plate for 30 s. After the third wash, rest the plate for 60 s. After first and fifth washes, there is no need to rest the plate. This step may be performed by a microplate washer

16. Add 100 μl of tetramethylbenzidine with hydrogen peroxide to each plate well and incubate for 15 min at room temperature protected from light

 Note: Cover the plate with aluminum paper and/or turn off the lights

17. Add 100 μl of stop solution (0.1 M H₂SO₄) to each plate well

18. Analyze the plate using an ELISA reader at 450 nm with filter of 620 nm

Mtb DNA detection limit

To determine the detection limit of automated colorimetric STNPCR, we performed a DNA dilution curve using a 1:10 ratio (from 1 ng/μl to 10 fg/μl) of Mtb strain H37Rv with ultrapure water. Similar dilution curves were also carried out using biological samples (blood, urine, sputum and pleural fluid). For this purpose, a negative control (DNA sample from a patient without TB, confirmed by conventional and molecular laboratory tests) was spiked with a known concentration of Mtb strain H37Rv DNA and diluted at a 1:10 ratio (from 10 ng/μl to 1 fg/μl). In the case of blood samples, we conducted separate dilution curves for plasma and PBMC. All samples were tested in duplicate.

Cut-off definition for each clinical sample

The cut-off value was defined using IBM SPSS Statistics 20.0 and MedCalc program. This was achieved by generating a receiver operator characteristic (ROC) curve based on the absorbance values obtained from reading the ELISA plate. The ROC curve was used when dealing with one continuous and one dichotomous variable to determine an optimal cut-off point in the continuous variable. The analysis included samples from individuals with a confirmed TB diagnosis as well as those without TB.

Performance evaluation of automated colorimetric STNPCR assay

To evaluate the technique's performance, we individually calculated the kappa index for each clinical sample and overall for each patient. The kappa index was used to compare colorimetric detection with AGE for all samples, and with culture and bacilloscopy for sputum samples. Confidence intervals for the kappa index were calculated, and the results were interpreted based on the level of agreement, as follows: <0 (absent), 0.0–0.20 (weak), 0.21–0.40 (reasonable), 0.41–0.60 (moderate), 0.61–0.80 (substantial) and 0.81–1.0 (strong). This interpretation scale was adapted from [27].

Statistical analysis

Composite reference standard (CRS) was used to assess test reliability. The CRS consisted of TB final diagnosis that was based on the analysis of a result set: clinical–epidemiological criteria, tuberculin skin test, histopathological findings, imaging test, bacilloscopy, mycobacteria culture and treatment outcome. To assess the accuracy of the colorimetric STNPCR assay, ROC curves were employed to determine the cut-off point, the area under the curve (AUC), as well as test sensitivity and specificity with their respective 95% CIs. The performance of the molecular tests was evaluated using the kappa index to compare them. The kappa test can complement the results by providing a general perspective, unlike the specificity and sensitivity of the ROC curve. Statistical analyses were conducted using software programs including OpenEpi (version 3.01), IBM SPSS Statistics (version 20.0) and R program (version 4.2.0). All tests used a significance level (α) of 0.05 and were two-tailed.

Results & discussion

Mtb DNA detection limit through automated colorimetric STNPCR assay

The automated colorimetric STNPCR detection limit was determined based on the lowest DNA concentration of Mtb strain H37Rv. The results indicated a detection limit of 1 pg/μl as determined by absorbance analysis (Table 1). In the DNA dilution curve of the reference strain H37Rv spiked into urine, sputum and pleural fluid samples, the lowest detected DNA concentration was 1 fg/μl. However, in blood samples (plasma and PBMC), the detection limit was 10 pg/μl (Supplementary Table 1). It was already expected that the colorimetric STNPCR detection limit would be higher for PBMCs and plasma. This may be explained by the presence of Taq polymerase inhibitors in blood samples, which can persist even after the blood-separation process and influence PCR reactions [28]. In addition, nonsite-specific samples were used in our study based on evidence that urine could contain DNA fragments derived from cell-free nucleic acids in plasma and blood resulting from dying human cells and Mtb [21,29]. Some of these fragments pass though the kidney and are excreted in the urine as transrenal DNA. Therefore, Mtb DNA can be isolated from both blood and urine samples, regardless of clinical TB form.

Table 1. Absorbance readings (450 nm and 620 nm) of the colorimetric assay in the evaluation of DNA from reference strain H37Rv at serial dilutions.

DNA concentration of strain H37Rv	Colorimetric STNPCR
	Status	Absorbance (mean)
10 ng/μl	Positive	0.247
1 ng/μl	Positive	0.235
100 pg/μl	Positive	0.233
10 pg/μl	Positive	0.188
1 pg/μl	Positive	0.161
100 fg/μl	Negative	0.132
10 fg/μl	Negative	0.128
Negative control	Negative	0.150
Positive control	Positive	0.194
Plate control^†	Negative	0.155
Blank^‡	Negative	0.121

† Well without probe + STNPCR positive control.

‡ Ultrapure water (well without probe).

Bold: Mtb DNA detection limit of colorimetric STNPCR based on negative control.

STNPCR: Single-tube nested PCR.

DNA detection limit of Mtb via STNPCR with AGE

To confirm the results, we also observed the detection limit using 1.5% AGE (Supplementary Figure 1). As in the colorimetric assay, the detection limit of the H37Rv strain DNA dilution curve in agarose gel was also 1 pg/μl. Although a colony-forming unit assay to determine the volume of Mtb DNA was not performed, based on the literature this concentration could be equivalent to approximately 100–200 bacillus/ml of the sample [30–32]. Concerning observed variations in the electrophoresis gel figure (Supplementary Figure 1), it is essential to recognize that nonspecific bands could arise due to various factors, including low DNA concentration and the occasional hybridization of primers to unspecific regions [33]. Negative samples were derived from non-TB patients, so they still contained DNA. Primers may occasionally bind to nonspecific regions, possibly due to variations in annealing temperature. However, despite the presence of nonspecific bands, it is crucial to note that there was no amplification of the target gene, indicating that these samples remained negative. Concerning positive samples, regardless of the presence of smear, they consistently displayed the expected amplified fragments. Smears can be attributed to several factors, such as high levels of DNA fragmentation that can produce small or larger fragments prone to self-prime, or overloading the PCR with template DNA, increasing the chance of self-priming [33]. Degraded primers might also contribute to this issue. To mitigate smear formation, diluting the extracted DNA could be a solution. However, this approach may not be effective when dealing with DNA that is both highly fragmented and in a very low concentration, as was the case in our samples, which are considered paucibacillary. It is worth noting that despite the presence of smears and nonspecific bands, these PCR issues did not obscure the results in AGE, which the IS6110-amplified fragments were clearly discernible. In addition, it is also essential to acknowledge that as the DNA concentration decreased in the Mtb strain H37Rv curve (samples 1–4; Supplementary Figure 1), the intensity of nonspecific bands and smears slightly increased. As previously mentioned, this did not hinder the interpretation of the gel electrophoresis results.

Cutoff values via the ROC curve with CRS

The automated colorimetric STNPCR absorbances from biological samples were evaluated using ROC curve analysis (Table 2): plasma (AUC: 0.679; 95% CI: 57.7–76.9%); PBMC (AUC: 0.612; 95% CI: 51.0–70.8%), urine (AUC: 0.546; 95% CI: 44.1–64.9%), sputum (AUC: 0.549; 95% CI: 31.5–76.8%) and pleural fluid (AUC: 1.0; 95% CI: 100–100%); see Supplementary Figure 2 for more details. The ROC curve is an important tool for evaluating diagnostic tests since it determines sensitivity and specificity values based on a reference method, which in this case was the CRS [19,34,35]. After ROC curve analyses of the biological samples, we defined a cutoff based on the sensitivity and specificity values. Cutoff was established to strike a balance between sensitivity and specificity, effectively distinguishing positive and negative samples [19]. Among the analyzed biological samples, only plasma showed a statistically significant result (p = 0.001), with a sensitivity of 55.5% and a specificity of 80.0%, while there was a statistical trend for PBMC sample analyses. The detailed results are provided in Table 2. Although the AUC values were not particularly high, the test in plasma samples was able to distinguish between true-positive and false-positive samples. In the studies conducted by Lima et al. [12,36], they used only nested PCR or STNPCR and did not analyze the results based on ROC curve. Therefore, in the literature there are no studies regarding colorimetric STNPCR assays with blood samples for Mtb detection that we can compare our results with.

Table 2. Cutoff values for different biological samples by receiver operator characteristic curve analysis with the composite reference standard as the gold standard.

Sample	n	Cutoff	Sensitivity (%)	Specificity (%)	AUC (95% CI)	p-value
Plasma	98	0.110	55.5	80.0	0.679 (57.7–76.9)	0.001
PBMCs	100	0.090	72.3	54.3	0.612 (51.0–70.8)	0.0512
Urine	95	0.092	61.3	51.5	0.546 (44.1–64.9)	0.451
Pleural fluid	5	0.081	75.0	100.0	1.000 (100–100)	0.157
Sputum	20	0.164	64.7	66.7	0.549 (31.5–76.8)	0.794

Bold: Statistical significance (p-value < 0.05).

AUC: Area under the curve; CRS: Composed reference standard; PBMC: Peripheral blood mononuclear cell.

For urine samples, the results were not statistically significant, indicating that the colorimetric STNPCR test for this sample type had limited discriminatory power and struggled to differentiate effectively between true-positive and false-positive results. One possible explanation is the nature of the paucibacillary specimens and the challenge of establishing an ideal gold-standard test, given that culture-based methods have shown low sensitivity (~25%) for this sample type [2,7]. In fact, developing a diagnostic test for extrapulmonary TB cases is inherently more complex [6,10,28,37,38].

In the pleural fluid sample, despite the ideal AUC value (1.0) the result was not statistically significant because of the very limited number of samples analyzed. The decision to include a pleural fluid sample in the colorimetric STNPCR analyses was driven by the fact that pleural TB was the most common form of extrapulmonary TB (20.5%). The difficulty associated with obtaining specimens from specific infection sites in extrapulmonary TB cases [1,6,10,28] is evidenced by the small number of pleural fluid samples available. Nevertheless, out of the eight pleural TB cases diagnosed in this study, only five pleural fluid samples were obtained for analysis. With a sensitivity of 75%, the colorimetric STNPCR test demonstrated higher sensitivity in detecting Mtb DNA compared with the bacilloscopy method, which has a sensitivity of approximately 5% for pleural TB cases [39,40].

Regarding sputum samples (AUC: 0.549), there was no statistical significance indicating a low capacity of the tests to discriminate between positive and negative samples. In a study by Michelon et al. [23], 301 induced sputum samples and 175 spontaneous sputum samples were analyzed using colorimetric detection in a microtiter plate, following a similar protocol to our study. In their work, the authors obtained a cutoff of 0.250 with higher sensitivity (94%) and specificity (100%) values compered with our results, probably due to limited sample numbers. These results could potentially be improved with a larger sample size, allowing for more reliable colorimetric STNPCR analyses with better sensitivity and specificity results.

Colorimetric STNPCR & STNPCR-AGE system performance with CRS

Considering the CRS as a comparison standard, the general colorimetric STNPCR assay sensitivity and specificity were 76.62% (95% CI: 65.6–85.5) and 60.53% (95% CI: 43.4–75.9), respectively, with weak kappa index agreement (κ = 0.166). For STNPCR with AGE, the sensitivity was 46.67% (95% CI: 35.1–39.6) and the specificity was 39.47% (95% CI: 24.1–56.6), with a kappa index of 0.062, also indicating weak agreement (Table 3). These weak agreements by kappa index may occur due to the divergence in results comparisons.

Table 3. Colorimetric single-tube nested PCR assay and single-tube nested PCR with agarose gel electrophoresis performance between tuberculosis and nontuberculosis patients for all sample type using composite reference standard as the comparison standard.

	TB^†	Non-TB	Sensitivity		Specificity		Kappa index
			%	95% CI	%	95% CI
Colorimetric STNPCR
Positive	59	15	76.62	65.6–85.5	60.53	43.4–75.9	0.166
Negative	18	23
Total	77	38
STNPCR with AGE
Positive	35	23	46.67	35.1–58.6	39.47	24.1–56.6	0.062
Negative	40	15
Total	75	38

† Group formed by patients with pulmonary or extrapulmonary tuberculosis.

STNPCR: Single-tube nested PCR; TB: Tuberculosis.

Detecting Mtb DNA in nonsite-specific samples, such as urine and blood, by the NAA technique is a challenging target since there are many variables that could influence the sensitivity and specificity of diagnostic tests [4,7,21]. In addition, the reduced sensitivity of a molecular test, even using a multicopy gene (IS6110) as a target for TB diagnosis, can also be attributed to the inherent low bacilli load in nonsite-specific TB samples [6]. This factor may considerably impact the performance of the colorimetric STNPCR test.

Moreover, in high-burden countries for TB, such as Brazil, many individuals come into contact with Mtb but do not develop an active form of the disease; instead, they have what is called latent tuberculosis infection (LTBI) [4,41]. This means that there is an increased likelihood of finding Mtb DNA fragments in biological samples from LTBI individuals. Therefore, false-positive results reducing the specificity of the colorimetric STNPCR test can occur due to possible LTBI cases and nonspecific reactions caused by the primers binding. In our study, we identified 15 biological samples that presented false-positive results in the colorimetric STNPPCR assay when compared with the CRS as the gold standard. It is worth noting that all these patients displayed symptoms that could be related to TB. However, none of the clinical tests confirmed TB, including smear microscopy, culture, tuberculin skin test and imaging results, leading to a negative TB diagnosis. In addition, the limited sample size of pleural fluid (5 samples) and sputum (20 samples) might hinder more consistent result analysis of the colorimetric STNPCR assay.

Kappa index: colorimetric STNPCR detection assay × STNPCR-AGE

The kappa index values for each sample type were determined based on the cutoff. Positive and negative samples were defined in terms of the kappa index, and the results obtained in the colorimetric assay were compared with those found in AGE. Among the biological samples, sputum showed the highest level of agreement (kappa = 0.633, p-value = 0.001), which is considered substantial (Table 4). Blood sample (plasma + PBMC) and combined blood + urine samples were also evaluated, considering them positive when at least one sample type (plasma, PBMC or urine) tested positive. Moreover, kappa index analyses were also performed for patient status. A TB-positive patient was defined when any sample type (plasma, PBMC, urine, sputum and pleural fluid) tested positive using molecular testing (colorimetric STNPCR assay and STNPCR with gel agarose electrophoresis). However, when it came to patient status, the kappa index indicated week agreement (kappa = 0.184, p = 0.018; Table 4).

Table 4. Kappa index analyses between colorimetric single-tube nested PCR detection assay and single-tube nested PCR with agarose gel electrophoresis in biological samples.

Sample	N	Kappa index	Agreement	p-value
Plasma	98	0.026	Weak	0.662
PBMCs	100	0.132	Weak	0.044
Urine	95	0.224	Reasonable	0.030
Pleural fluid	5	-0.176	Absent	0.171
Sputum	20	0.633	Substantial	0.001
Blood^†	100	0.126	Weak	0.072
Blood^† and urine	107	0.077	Weak	0.273
Patient^‡	115	0.184	Weak	0.018

† Plasma + PBMCs.

‡ Analyses considering a TB-positive patient when any sample was positive.

Bold: Statistical significance (p-value < 0.05).

PBMC: Peripheral blood mononuclear cell.

Comparison of results or reproducibility evaluation between diagnostic tests can be performed by estimating the kappa index. In this study, the kappa index was used to analyze the agreement between colorimetric STNPCR assay and STNPCR with AGE for Mtb detection. Electrophoresis is a technique that has already been standardized for the visualization of STNPCR products [12].

The kappa index estimation for plasma (kappa = 0.026, p = 0.662) and PBMC (kappa = 0.132, p = 0.044) samples revealed a weak level of agreement between the two techniques. This finding corroborates with the results obtained by Lima et al. [12], who also observed only a slight agreement between the two methods. The lower agreement may be attributed to the sensitivity in plasma (55.5%) and specificity in PBMC (54.3%) of the colorimetric STNPCR assay. In addition, it is worth noting that plasma and PBMC samples displayed a higher detection limit (10 pg/μl) compared with other samples (1 fg/μl). This difference can also influence agreement between the tests, as blood samples are inherently paucibacillary in pulmonary and extrapulmonary TB forms.

The methods comparison for urine samples showed a reasonable agreement (kappa = 0.224) with statistical significance (p = 0.030). The colorimetric STNPCR sensitivity in urine was also considered reasonable (61.3%). These results differ from those observed by Lima et al. [12], as their study showed absent agreement in the methods comparison when evaluating urine samples. However, when considering all samples together, including blood (plasma + PBMC) and blood + urine (plasma + PBMC + urine) samples, the agreement remained weak and lacked statistical significance.

Based on the kappa index, the results of the methods for pleural fluid samples were discordant (kappa = -0.176). This discordance can be attributed to the limited sample size, which was insufficient to carry out a statistically significant comparison between the two methods.

With kappa index analyses demonstrating a substantial (0.633) and statistically significant (p = 0.001) agreement, the sputum sample exhibited the best results when comparing the methods, which was consistent with the findings of Lima et al. [12]. Sputum contains the highest concentration of bacilli and, consequently, it has elevated Mtb DNA levels. For this reason, it has been commonly used as a sample for colorimetric system analyses in other studies [23,24,42]. Despite the sensitivity of the colorimetric STNPCR for sputum sample not being exceptionally high (64.7%), possibly due to the relatively small sample size, a considerable level of agreement was observed between the colorimetric STNPCR assay and STNPCR with AGE for Mtb detection.

Regarding kappa index analyses per patient, the agreement level was considered weak yet statistically significant. This result can also be attributed to the overall sensitivity of the colorimetric STNPCR, which did not display high values in the biological samples. It is important to note that this is a relatively ‘new’ molecular diagnostic system that is still under evaluation and requires further analysis. Moreover, gene sequencing would have been a better reference method for comparison with the colorimetric STNPCR assay. However, the only available technique was gel electrophoresis, a method already standardized for STNPCR product visualization [12]. Gene sequencing is a costly and not readily accessible technique for TB diagnosis [3].

Kappa index: colorimetric STNPCR detection assay × culture & bacilloscopy tests

The kappa index values for sputum samples were also determined by comparing the colorimetric STNPCR assay with culture and bacilloscopy methods (Table 5).

Table 5. Kappa index analyses comparing colorimetric single-tube nested PCR detection assay with culture and bacilloscopy tests in sputum samples.

Diagnostic test	n	Kappa index	Agreement	p-value
Culture	20	0.304	Reasonable	0.127
Bacilloscopy	20	0.144	Weak	0.294

When comparing the colorimetric STNPCR assay with culture and bacilloscopy tests in sputum samples, we observed a reasonable (kappa = 0.304) and weak (kappa = 0.144) agreement level, respectively, but with no statistical significance (Table 5). Culture remains the gold-standard test for TB diagnosis, consistently demonstrating better results in our analyses. Although it is a very time-consuming Mtb-detection technique, culture is more sensitive than bacilloscopy and can identify Mycobacterium sp. [2].

In general, when compared with other TB diagnostic systems, the colorimetric STNPCR demonstrates notable cost–effectiveness. The qPCR and Xpert MTB systems are faster than the colorimetric STNPCR because the detection of amplified products occurs in real time, eliminating the additional time required for post-PCR amplicon visualization. However, these real-time PCR systems employ high-cost equipment and reagents, posing implementation challenges in routine laboratories, particularly in resource-limited countries [3,4,16]. On the other hand, colorimetric STNPCR systems require the use of equipment already available in diagnostic routine laboratories. It has been observed that the cost of Xpert MTB/RIF per test is approximately US$25, whereas for colorimetric STNPCR this decreases to approximately US$12, demonstrating greater cost–effectiveness compared with real-time PCR systems [43,44]. Regarding PCR with AGE, despite its lower cost and shorter processing time, it is typically limited to research institutions and demands skilled professionals for execution and interpretation. It is also important to note that colorimetric STNPCR benefits from spectrophotometric results analysis, minimizing potential interpretation errors. Moreover, although the colorimetric STNPCR assay requires a full day for execution, it has demonstrated superior cost–effectiveness, sensitivity and specificity in comparison to conventional TB diagnostic methods, such as culture (an excessively time-consuming method) and bacilloscopy, and PCR with agarose gel electrophoresis, which can be completed in half a day [16,44].

Conclusion

Despite the limitations primarily related to limited sample size, this study demonstrated a quick and straightforward test with promising results. In fact, the technical expertise required for colorimetric STNPCR execution is common in a routine laboratory, as it only utilizes a conventional thermocycler, a washer and an ELISA reader. It also offers an easy and objective interpretation. Furthermore, the colorimetric STNPCR test revealed notably low Mtb DNA detection limits and significant sensitivity and specificity values across different biological samples. These findings suggest that the automated colorimetric STNPCR method holds promise as a valuable molecular diagnostic test to potentially integrate into laboratory routines, supporting TB diagnosis. However, more studies are needed to thoroughly evaluate and validate colorimetric STNPCR tests for each clinical sample type, both in pulmonary and extrapulmonary TB diagnosis

Future perspective

TB diagnosis remains challenging, particularly for extrapulmonary forms due to nonspecific symptomatology and paucibacillary samples that hamper Mtb detection by conventional methods. Xpert MTB techniques are rapid and sensitive molecular tests that support TB diagnosis but their high implementation cost and variable sensitivity for most extrapulmonary samples, as well as their ineffectiveness for blood samples, pose limitations. Various molecular methods, mainly PCR based, have shown promise in enhancing TB diagnosis. Among these, the colorimetric STNPC assay stands out as a potential addition to laboratory routines for both pulmonary and extrapulmonary TB diagnosis. However, as mentioned before, additional studies are encouraged to comprehensively evaluate and validate colorimetric STNPCR tests across various clinical sample types for Mtb detection.

Author contributions

MP Salazar and JFCLS Monteiro designed the study, conducted the experiments, and wrote the manuscript. LML Montenegro performed some of the bench work. WHV Carvalho-Silva analyzed the data. GTN Diniz performed the statistical analyses. RP Werkhauser and HC Schindler supervised the project. All authors were involved with the manuscript review and editing.

Financial disclosure

The authors have no financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Writing disclosure

No writing assistance was utilized in the production of this manuscript.

Ethical conduct of research

The authors state that they have obtained appropriate institutional review board approval (protocol number: 45687215.2.0000.5190) and have followed the principles outlined in the Declaration of Helsinki for all human or animal experimental investigations. In addition, for investigations involving human subjects, informed consent has been obtained from the participants involved.

Acknowledgments

The authors are grateful to Aggeu Magalhães Institute (IAM/FIOCRUZ-PE) for all the support provided during the development of this study.

Supplementary data

To view the supplementary data that accompany this paper please visit the journal website at: www.future-science.com/doi/suppl/10.2144/btn-2023-0080

Competing interests disclosure

The authors have no competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, stock ownership or options and expert testimony.

Additional information

Funding

The authors are grateful to Aggeu Magalhães Institute (IAM/FIOCRUZ-PE) for all the support provided during the development of this study.