Identifying predictors to gain sensitivity of nucleic acid amplication tests for Mycobacterium tuberculosis

Background: Tuberculosis (TB) remains one of the primary threatening of human health and its diagnosis remains unsatisfactory in clinic. Nucleic acid amplication tests (NAAT) showed higher sensitivity than culture for the diagnosis of pulmonary TB (PTB). However, NAAT are expensive and not easily deployable at the peripheral level. To improve the sensitivity of NAAT for the PTB diagnosis, the predictive factors that might be utilized to give the optimized choice of NAAT were investigated. Methods: A total of 1263 PTB suspects were enrolled for evaluation. The sensitivity, specicity and accuracy of Mtb detection in sputum and bronchoalveolar lavage uid (BALF) were compared. Odds ratios and 95% condence intervals were used to assess variables that associated with positive NAAT in sputum and BALF of PTB suspects. Results: An signicantly enhanced sensitivity was observed when performed on NAAT (61.1%) compared with smear (9.0%) and culture of Mtb (47.8%). We found that erythrocyte sedimentation rate (ESR) (+), cavities (+) and IFN-γ release assay (IGRA) (+) are involved in the positivity of Mtb detection through NAAT. Moreover, those who are ESR (+), cavities (+) and IGRA (+), showed 86% diagnostic positivity of Mtb by NAAT. Conclusions: Our study suggested that combination of the results of ESR and IGRA and the presence of pulmonary cavity is helpful to predict the positivity of Mtb detection through NAAT. Those who are ESR (+), cavities (+) and IGRA (+), should perform NAAT for Mtb detection, because they are most likely to be bacteriologically conrmed as TB. tuberculosis; ESR, erythrocyte sedimentation rate; IGRA, interferon-γ release assays.

Background TB, caused by Mycobacterium tuberculosis (Mtb), remains one of the primary threatening of human health in the world. There were approximate 10 million new TB cases and caused 1.45 million TB deaths in 2018 (1). Given that Bacille Calmette-Guerin (BCG), the only licensed vaccine, has variable e cacy that gives protection against childhood TB but is not valid in adult pulmonary TB (PTB) (2), the timely and rapid diagnosis of PTB is crucial for individual anti-TB treatment as well as could e ciently prevent the transmission of TB.
Culture of the bacteria is the gold standard for TB diagnosis, but it has low sensitivity, high contamination rate and needs long growth period and thus results in the delay of anti-TB treatment and increases the exposure of the TB patients to the healthy population (3). Acid-fast bacilli smear is fast and cheap, while has low sensitivity and speci city (4,5). Mtb-antigen speci c IGRA has higher sensitivity, but it can not discriminate active TB (ATB) from dormant TB infections (6,7). Thus, it is eager to identify optimal methods for PTB diagnosis.
As a newer method, NAAT that has been developed during the last decades, is a promising diagnostic technique of PTB that shows higher sensitivity and speci city. For example, the widely used and WHO-endorsed Xpert®MTB/RIF assay (Cepheid, Sunnyvale, CA, USA) is an automated, single cartridge-based NAAT and can rapidly detect TB and rifampicin resistance (8,9). Although providing the possibility to increase the positive rate and save time for the diagnosis of PTB, NAAT is costly and not easily deployable at the peripheral level. Thus, ascertaining factors to speculate the yield of NAAT for PTB diagnosis is necessary to obtain the maximized bene t from NAAT but minimizing the expense to TB control programs. Here, the sensitivity and speci city between culture and NAAT for PTB diagnosis were compared and clinical factors involved in the bene t of NAAT in discriminating PTB were also con rmed.

Methods And Materials
Study population This study that included a total of 1263 PTB suspects collected from February 2011 to June 2015 was retrospectively performed at Shenzhen Third People's Hospital (Guangdong, China). Iatrical data were collated including age, sex, premonitors, previous TB history, lab tests and imaging examination. Lab experiments used in the present study contained sputum smear, NAAT, culture, IGRA, and CD4 and CD8 cell calculation. Patients with a follow-up lacking 6 months were precluded. The study was agreed by the Institutional Review Board of Shenzhen Third People's Hospital. Written informed consent were gained from every patients.

Diagnosis of PTB
The clinical premonitor, radiological sign indicative of PTB, culture of Mtb, microscopy, NAAT, and response to anti-TB therapy were used for the diagnosis of PTB. Both sputum and BALF collected from every patients were used to detect Mtb by microscopy, culture of Mtb and NAAT. Patients who were sputum or BALF culture (+) for Mtb, or NAAT (+) for Mtb, plus clinical premonitor and radiological sign indicative of PTB, were con rmed as de nite TB. Patients who were culture (-), NAAT (-) and AFB (-) for Mtb both in sputum and BALF, plus clinical premonitor, radiological sign indicative of PTB and response to anti-TB therapy, were con rmed as probable TB. The PTB patients who had nished standard anti-TB therapy for no less than 6 months without subsequent proof of ATB were con rmed as cured TB. Non-TB lung disease (non-TB) was demonstrated according to having no microbiological or no histological proof of PTB or demonstration of an alternative diagnosis or a lack of progression to PTB more than 6 months of followed-up without PTB therapy (10). Non-tuberculosis mycobacteria (NTM) infection was de ned according to the con rmation of NTM from the same species in sputum and/or BALF.

Specimen acquisition, processing and detection
To collect BALF, isotonic saline was used for bronchial washing of the related sub-segment. All sputum and BALF samples were performed on microscopy, NAAT and culture of Mtb in Lowenstein-Jensen medium and BACTEC 9000 MB liquid medium for the con rmation of Mtb. The IGRA was determined as reported previously using an in-house IFN-γ enzyme-linked immunoblot assay, which the sensitivities and speci cities was similar to that of commercial available kits (6).

Statistical analysis
For identi cation of potential variables involved in positivity of NAAT (positivity in Mtb examination in sputum and BALF through NAAT), single variable analysis with Pearson's χ 2 test was carried out. Every variable with P < 0.05 in single variable analysis were subjected to the unconditional multiple logistic regression analysis and the associations were evaluated by odd ratios (OR) and 95% con dence intervals (CIs) and those variables with P < 0.05 were subjected to strati ed analysis for identi cation of the superimposed roles on positivity of NAAT. All analysis were conducted by SPSS 26.0.

Study population
A total of 1263 patients were enrolled in this study. 1173/1263 patients were diagnosed with PTB, 18/1263 patients were cured TB without TB relapse, 16/1263 cases were infected with NTM, and 56/1263 patients were discriminated with non-TB lung diseases ( Table 1). The average age was 35.96 and 57.4 %were males.
NAAT enhanced the diagnostic accuracy of PTB suspects As shown in Table 2, sensitivity and speci city etc. of NAAT, culture and smear for the diagnosis of PTB patients were determined. Among the 1263 patients, NAAT showed signi cant higher positive detectable rate for the diagnosis of PTB suspects in comparison to that of smear and culture of Mtb, with the positivity of 58.0%, 8.5% and 44.4%, respectively (Table 2). An signi cantly enhanced sensitivity was observed when performed on NAAT (61.1%) compared with smear (9.0%) and culture of Mtb (47.8%) ( Table 2). Similarly, the accuracy of NAAT (62.6%) was signi cantly higher than that of smear (15.4%) and culture of Mtb (51.5%) ( Table 2, Table S1). Taken together, these results suggested that the method of NAAT could better contribute to the diagnosis of PTB suspects.

Factors involved in positive Mtb detection through NAAT in PTB
Consistent with previous reports(11), a signi cantly increased Mtb detection rate in PTB suspects assayed on NAAT was observed. 717/1173 (61.1%) patients were positive for Mtb detection performed on NAAT. Although greatly contributed to the PTB diagnosis and drug susceptibility testing, NAAT test is costly and not easily deployable at the peripheral level. For identi cation of the patients most probably to pro t from NAAT test, factors involved in Mtb examination of PTB through NAAT were analyzed. We found that lose weight (P = 0.020), bilateral lung infection (P = 0.018), cavity (P < 0.001), ESR (P < 0.001), IGRA results (P < 0.001) were all markedly involved in Mtb examination of PTB through NAAT test when performed univariate analysis (Table 3) (Table 4). Further analysis suggested that the combined results of ESR, cavity and IGRA gained the highest positivity of Mtb examination of PTB by NAAT test (Table 5).

Discussion
The early diagnosis and timely treatment are necessary for e cient TB control. The failure to rapidly recognize and cure affected patients caused improved mortality, secondary resistance, and sustaining transmission (11). The faster NAAT test has brought about a shorter time for TB diagnosis and, thus, rapider anti-TB treatments. In the present study, the sensitivity of NAAT test was 61.1% (717/1173) in sputum and BALF of PTB patients, while the sensitivity of smear and culture was 9.0% (106/1173) and 47.8% (561/1173) respectively. Besides, 16 NTM infected patients were determined through NAAT tests.
Together, about 50% and 15% patients bene ted from NAAT tests compared with smear and culture, respectively. NAAT test improved the positive detectable rate of PTB diagnosis as well as contributed to the effective anti-TB therapy by identifying species and drug susceptibility testing (12).
Although fast, safe, convenient and accurate, NAAT test are expensive and not easily deployable at the peripheral level, such as MTB/RIF test (13). The MTB/RIF test employs sophisticated technology to obtain great simplicity of use, which is expensive to manufacture. Moreover, the need for annual calibration was di cult to implement at peripheral laboratories, particularly in rural areas. Thus, we identi ed factors correlated with the positive detection of Mtb of PTB through NAAT test trying to maximize the pro t of PTB patients who performed NAAT test. We found that risk factors related to positivity of Mtb detection through NAAT tests were ESR, cavity and IGRA results. Those factors play an overlapping role in predicting positivity of Mtb examination through NAAT test with the highest positivity of 86% (a combination of ESR (+), cavity (+) and IGRA (+)) for Mtb detection.
ESR gets the highest predictive value among them. The ESR test is widely used as nonspeci c test for many pathological conditions such as infections and in ammatory diseases. Infections could elevate the value of ESR and are said to be one of the primary reasons of elevated values ≥ 100 mm/h (14)(15)(16). A previous study suggested that ATB was mainly involved in extremely high ESR values (≥ 100 mm/h) (16). Another study also found that TB patients without HIV infection had the mean and median values more than 100 mm/h (17). In line with these studies, our results suggested that the elevated values of ESR were associated with ATB and therefore, associated with Mtb detection through NAAT test. Mtb infections could often result in the formation of cavities in the lungs of patients and the bacilli grows on the cavity wall and enters into the bronchial tree, thus enabling them spread to other parts of the lung and also infect other people (18). Positive IGRA results was also predictive factors for Mtb detection through NAAT test. Our recent result found that the degree of IFN-γ response in BALF of PTB determined by IGRAs was positively associated with bacteria load (19). Moreover, other researchers reported that the degree of IFN-γ are involved in sputum positivity in immune competent PTB patients (12,20). Consistent with these results, our result showed that IGRA results gives a fatidic odds ratio of 2.651 (95% CI: 1.861-3.776) for positivity of Mtb detection through NAAT test. In summary, we predicted that patients with higher ESR are more likely ATB enabling to mount a robust Mtb-speci c IFN-γ response and develop cavities with positive detection in both sputum and BALF for Mtb through NAAT test.

Conclusions
Our study suggested that combination of the results of ESR and IGRA and the presence of pulmonary cavity is helpful to predict the positivity of Mtb detection through NAAT. Those who are ESR (+), cavities (+) and IGRA (+), should perform NAAT for Mtb detection, because they are most likely to be bacteriologically con rmed as TB.

None
Author's contributions X-FH, Q-LG, QL, JB, JL, J-JZ, Z-HL, X-FZ and G-WZ performed the study. X-FH, Q-LG and QL analyzed the data and wrote the manuscript. T-SY, KY and YZ provided clinical data. X-DF, H-JZ and G-LZ supervised the study and applied for grants. X-FH, Q-LG, QL and G-LZ were identi ed as the guarantor of the paper, taking responsibility for the integrity of the work as a whole, from inception to published article. All authors read and approved the nal manuscript.

Availability of data materials
The dataset used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
The study was approved by the Institutional Review Board of Shenzhen Third People's Hospital, and informed consent was obtained from each participant.    NAAT, nucleic acid ampli cation test; ESR, erythrocyte sedimentation rate; IGRA, interferon-γ release assays.

Consent for publication
a Controlling for variables with P < 0.05 in univariate analysis Table 5. Strati ed analysis of factors correlated with NAAT positivity in pulmonary TB patients.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. TableS1.docx