Xpert MTB/RIF Ultra enhanced tuberculous pleurisy diagnosis for patients with unexplained exudative pleural effusion who underwent a pleural biopsy via thoracoscopy: A prospective cohort study

Diagnostic accuracy of Xpert MTB/RIF Ultra for detecting pulmonary tuberculosis and rifampicin resistance: a systematic review and meta-analysis

BACKGROUND

Public health issues related to tuberculosis still exist. Because Xpert MTB/RIF Ultra is more effective than conventional TB diagnostic techniques are, it is now regarded as an emerging technology. The diagnostic accuracy of Xpert MTB/RIF Ultra for tuberculosis was assessed in this systematic study.

METHODS

We searched the PubMed, Cochrane, EMBASE, and Web of Science databases for pertinent literature published before January 18, 2024. The quality of the collected literature was assessed via Review Manager 5.3 software, which applies the quality assessment of diagnostic accuracy studies criteria. Using Metadisc 1.40 software, the sensitivity, specificity, and summary receiver operating characteristic curves were plotted and examined. Stata 12.0 was the program we utilized to assess publication bias in this investigation. The Prospero prospective register of systematic reviews included this study (reference number CRD42024569674).

RESULTS

Analysis of 187 fourfold tables from 72 studies revealed that Xpert MTB/RIF Ultra demonstrated an overall pooled sensitivity of 76% and specificity of 95% for detecting pulmonary tuberculosis. The positive likelihood ratio (PLR) was 14.91, and the negative likelihood ratio (NLR) was 0.23, with an area under the curve (AUC) of 0.9351 and a diagnostic odds ratio (DOR) of 73.39. For detecting rifampin resistance, the combined sensitivity and specificity were 94% and 97%, respectively. The pooled PLR was 24.94, the NLR was 0.07, and the DOR was 429.05. The area under the summary receiver operating characteristic (SROC) curve was 0.9868.

CONCLUSION

In conclusion, developing effective laboratory diagnostic tools for identifying Mycobacterium tuberculosis is essential for epidemiological research. This study demonstrated that Xpert MTB/RIF Ultra effectively diagnosed Mycobacterium tuberculosis infection, including pulmonary tuberculosis, as well as rifampin resistance, highlighting its potential as a valuable tool for both diagnosis and resistance detection.

A case of pleural Mycobacterium tuberculosis infection with reversion of Quantiferon Gold Plus results from positive to negative

Introduction. Mycobacterium tuberculosis (MTB) infections continue to have a high mortality and morbidity burden globally. Interferon-gamma release assays such as Quantiferon Gold Plus (QFG-Plus) aid in diagnosis of latent TB but diagnosis of pleural TB remains challenging. We present a case of active pleural MTB infection with reversion from positive to negative of IGRA result as well as negative Xpert MTB/RIF Ultra PCR result from tissues obtained from pleural biopsy. Case summary. A 52-year-old otherwise healthy male presented in August 2022 with a 2 week history of pleuritic chest pain associated with modest elevation in inflammatory markers. The patient had had a positive QFG-Plus result in 2018, however QFG-Plus during this admission was negative. Computed-tomography pulmonary angiogram and needle thoracocentesis showed an exudative left pleural effusion with predominant lymphocytes. The patient's symptoms failed to resolve with empiric antimicrobial therapy for community-acquired pneumonia. Broncho-alveolar lavage as well as biopsies of pleural tissues via video-assisted thoracoscopic surgery from the left lower lobe yielded negative results on routine microbiological culture as well as Xpert Ultra PCR. Growth of acid-fast bacilli was noted from mycobacterial cultures of pleural tissues which was identified as MTB. Conclusion. Despite significant technological advances, microbiological diagnosis of MTB infections remains challenging. We document QFG-Plus reversion during development from latent to active pleural TB. Decline in the ability of CD4+ and CD8+ T cells to produce interferon gamma in response to TB antigens (ESAT-6 and CFP-10) was likely associated with loss of host control of latent MTB. This case serves as a reminder that despite exhaustive testing with state-of-art diagnostic platforms, MTB infections can still elude discovery.

Tuberculous pleuritis: clinical presentations and diagnostic challenges

PURPOSE OF REVIEW

Tuberculous pleuritis (TBP) is one of the most common types of extrapulmonary tuberculosis. We highlight the latest epidemiology of TBP, the heterogeneity of its presentation and the performance of different diagnostic strategies.

RECENT FINDINGS

There are differential trends in the incidences of TBP worldwide. Its incidence increased in China but decreased in the United States in the past decade. The presentation of TBP is heterogeneous regarding clinical symptoms, radiological findings and pleural fluid analysis results. Conventional microbiological tests have low sensitivities to diagnose TBP. Recent research focused on various diagnostic tools with better yield. The sensitivity of nucleic acid amplification tests (NAAT) in pleural fluid, including the latest generation of PCR and sequencing-based techniques for detecting tuberculosis, remains suboptimal. Various pleural fluid biomarkers have been explored, but there is a lack of consensus on their clinical utility and cutoff levels.

SUMMARY

The heterogeneity of clinical presentation poses obstacles to diagnosing TBP. Further development of diagnostic tools, including more robust NAAT and biomarkers with additional validation, is needed before incorporation into routine clinical practice.

Pleural effusion guidelines from ICS and NCCP Section 1: Basic principles, laboratory tests and pleural procedures

Pleural effusion is a common problem in our country, and most of these patients need invasive tests as they can't be evaluated by blood tests alone. The simplest of them is diagnostic pleural aspiration, and diagnostic techniques such as medical thoracoscopy are being performed more frequently than ever before. However, most physicians in India treat pleural effusion empirically, leading to delays in diagnosis, misdiagnosis and complications from wrong treatments. This situation must change, and the adoption of evidence-based protocols is urgently needed. Furthermore, the spectrum of pleural disease in India is different from that in the West, and yet Western guidelines and algorithms are used by Indian physicians. Therefore, India-specific consensus guidelines are needed. To fulfil this need, the Indian Chest Society and the National College of Chest Physicians; the premier societies for pulmonary physicians came together to create this National guideline. This document aims to provide evidence based recommendations on basic principles, initial assessment, diagnostic modalities and management of pleural effusions.

The tuberculous pleural effusion

Pleural tuberculosis (TB) is a common entity with similar epidemiological characteristics to pulmonary TB. It represents a spectrum of disease that can variably self-resolve or progress to TB empyema with severe sequelae such as chronic fibrothorax or empyema necessitans. Coexistence of and progression to pulmonary TB is high. Diagnosis is challenging, as pleural TB is paucibacillary in most cases, but every effort should be made to obtain microbiological diagnosis, especially where drug resistance is suspected. Much attention has been focussed on adjunctive investigations to support diagnosis, but clinicians must be aware that apparent diagnostic accuracy is affected both by the underlying TB prevalence in the population, and by the diagnostic standard against which the specified investigation is being evaluated. Pharmacological treatment of pleural TB is similar to that of pulmonary TB, but penetration of the pleural space may be suboptimal in complicated effusions. Evidence for routine drainage is limited, but evacuation of the pleural space is indicated in complicated disease.

Educational aims

To demonstrate that pleural TB incorporates a wide spectrum of disease, ranging from self-resolving lymphocytic effusions to severe TB empyema with serious sequelae.To emphasise the high coexistence of pulmonary TB with pleural TB, and the importance of obtaining sputum for culture (induced if necessary) in all cases.To explore the significant diagnostic challenges posed by pleural TB, and consequently the frequent lack of information about drug sensitivity prior to initiating treatment.To highlight the influence of underlying TB prevalence in the population on the diagnostic accuracy of adjunctive investigations for the diagnosis of pleural TB.To discuss concerns around penetration of anti-TB medications into the pleural space and how this can influence decisions around treatment duration in practice.

Diagnostic value of nucleic acid amplification tests for tuberculous pleural effusion

Diagnosing tuberculous pleural effusion (TPE) is challenging for pulmonologists and laboratory scientists. The gold standards for TPE diagnosis are pleural fluid Ziehl-Neelsen staining, Mycobacterium tuberculosis (Mtb) culture and pleural biopsy. These tools have limitations, including low sensitivity, long turnaround time and invasiveness. The nucleic acid amplification test (NAAT) is a rapid and minimally invasive tool for diagnosing TPE. This review summarizes the diagnostic accuracy of available NAATs for TPE, with a focus on the evidence from systematic reviews and meta-analyses. The NAATs summarized in this review include in-house NAATs, GeneXpert-MTB/RIF, GeneXpert-MTB/RIF Ultra, simultaneous amplification and testing-tuberculosis, FluoroType MTB and loop-mediated isothermal amplification.

Diagnosis of extrapulmonary tuberculosis by GeneXpert MTB/RIF Ultra assay

INTRODUCTION

Diagnosis of extrapulmonary tuberculosis (EPTB) is an arduous task owing to different anatomical locations, unusual clinical presentations, and sparse bacillary load in clinical specimens. Although GeneXpert® MTB/RIF is a windfall in TB diagnostics including EPTB, it yields low sensitivities but high specificities in many EPTB specimens. To further improve the sensitivity of GeneXpert®, GeneXpert® Ultra, a fully nested real-time PCR targeting IS6110, IS1081 and rpoB (Rv0664) has been endorsed by the WHO (2017), wherein melt curve analysis is utilized to detect rifampicin-resistance (RIF-R).

AREA COVERED

We described the assay chemistry/work design of Xpert Ultra and evaluated its performance in several EPTB types, that is, TB lymphadenitis, TB pleuritis, TB meningitis, and so on, against the microbiological reference standard or composite reference standard. Notably, Xpert Ultra exhibited better sensitivities than Xpert, but mostly at the compensation of specificity values. Moreover, Xpert Ultra exhibited low false-negative and false-positive RIF-R results, compared with Xpert. We also detailed other molecular tests, that is, Truenat MTBTM/TruPlus, commercial real-time PCR, line probe assay, and so on, for EPTB diagnosis.

EXPERT OPINION

A combination of clinical features, imaging, histopathological findings, and Xpert Ultra are adequate for definite EPTB diagnosis so as to initiate an early anti-tubercular therapy.

The diagnostic performance of endobronchial ultrasound with Xpert MTB/RIF Ultra in smear-negative pulmonary tuberculosis

OBJECTIVE

This study investigated the diagnostic performance of endobronchial ultrasound with Xpert MTB/RIF Ultra (Ultra) for detecting smear-negative pulmonary tuberculosis (TB).

METHODS

143 patients suspected of sputum smear-negative pulmonary tuberculosis were enrolled in this study in Shanghai Pulmonary Hospital, China. These patients underwent endobronchial ultrasound with a guide sheath (EBUS-GS) or endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) based on their chest CT manifestations. We assessed the sensitivity and specificity of tissue specimens with Ultra in the TB group and non-TB group. Culture and clinical diagnosis were used as gold-standard for TB.

RESULTS

Among these 143 patients, 11 patients were culture-positive TB, 85 patients were diagnosed with culture-negative TB and 47 were with the non-TB diseases. Direct testing with microscopy (Acid-Fast Bacilli smear, AFB), liquid culture, pathology, Xpert MTB/RIF(Xpert) test and Ultra had a sensitivity of 8.3%, 11.5%, 42.7%, 64.6%, and 78.1% individually among all the TB patients. Ultra had a higher sensitivity than Xpert (P = 0.011). But Ultra had a specificity of 59.6% (95% CI 44.3-73.3), lower than that of Xpert (89.4%, 95% CI 76.1-96.0, P = 0.001). Ultra had the same sensitivity on specimens from EBUS-TBNA and EBUS-GS (P = 0.975). Ultra's positive predictive value and negative predictive value were 79.8% and 57.1% respectively.

CONCLUSIONS

Tissue specimens from interventional bronchoscopy combined with Ultra provide a sensitive method for diagnosing smear-negative pulmonary tuberculosis, but its specificity was lower than Xpert.

Use of Pleural Fluid Digital PCR Analysis to Improve the Diagnosis of Pleural Tuberculosis

The diagnosis of pleural tuberculosis (TB) remains difficult due to the paucity of Mycobacterium tuberculosis in pleural fluid (PF). This study aimed to improve pleural TB diagnosis using highly sensitive digital PCR (dPCR) technique. A total of 310 patients with evidence of PF were consecutively enrolled, 183 of whom suffered from pleural TB and 127 from non-TB. PF samples were prospectively collected and total DNA was extracted. The copy numbers of M. tuberculosis insertion sequence (IS) 6110 and IS1081 in DNA were quantified using dPCR. The overall area under the curve of IS6110-dPCR was greater than that of IS1081-dPCR (0.85 versus 0.79). PF IS6110 OR IS1081-dPCR (according to their cut-off values, "positive" was defined as either of them was positive, while "negative" was defined as both of them were negative) had higher sensitivity and equal specificity compared with single target-dPCR. The sensitivity of PF IS6110 OR IS1081-dPCR for total, definite, and probable pleural TB was 59.0% (95% CI = 51.5% to 66.2%), 72.8% (95% CI = 62.6% to 81.6%), and 45.1% (95% CI = 34.6% to 55.8%), respectively. Its specificity was 100% (95% CI = 97.1% to 100.0%). PF IS6110 OR IS1081-dPCR showed a higher sensitivity than smear microscopy (57.4% versus 7.1%), mycobacterial culture (55.3% versus 31.8%), and Xpert MTB/RIF (57.6% versus 23.0%). Long antituberculosis treatment time (>1 month) was found to be associated with negative dPCR results in pleural TB patients. This study indicates that PF IS6110 OR IS1081-dPCR is an accurate molecular assay, which is more sensitive than routine etiological tests and has the potential to enhance the definite diagnosis of pleural TB. IMPORTANCE Pleural TB is one of the most frequent causes of pleural effusion, especially in areas with high burden of TB. Due to the paucibacillary nature of the disease, the diagnostic sensitivities of all available bacteriological and molecular tests remain poor. There is an urgent need to develop new efficient methods. Digital PCR (dPCR) is the third generation of PCR that enables the exact quantification of trace nucleic acids in samples. This study evaluates the diagnostic performance of pleural fluid (PF) dPCR analysis for pleural TB, and shows that PF IS6110 OR IS1081-dPCR has a higher sensitivity than routine etiological tests such as smear microscopy, mycobacterial culture, and Xpert MTB/RIF. This work provides a new choice for improving the definite diagnosis of pleural TB.

The Clinical Experience of Mycobacterial Culture Yield of Pleural Tissue by Pleuroscopic Pleural Biopsy among Tuberculous Pleurisy Patients

Background and Objectives: Tuberculous pleurisy is a common extrapulmonary TB that poses a health threat. However, diagnosis of TB pleurisy is challenging because of the low positivity rate of pleural effusion mycobacterial culture and difficulty in retrieval of optimal pleural tissue. This study aimed to investigate the efficacy of mycobacterial culture from pleural tissue, obtained by forceps biopsy through medical pleuroscopy, in the diagnosis of TB pleurisy. Materials and Methods: This study retrospectively enrolled 68 TB pleurisy patients. Among them, 46 patients received semi-rigid pleuroscopy from April 2016 to March 2021 in a tertiary hospital. We analyzed the mycobacterial culture from pleural tissue obtained by forceps biopsy. Results: The average age of the study participants was 62.8 years, and 64.7% of them were men. In the pleuroscopic group, the sensitivity of positive Mycobacterium tuberculosis (M. TB) cultures for sputum, pleural effusion, and pleural tissue were 35.7% (15/42), 34.8% (16/46), and 78.3% (18/23), respectively. High sensitivities of M. TB culture from pleural tissue were up to 94.4% and 91.7% when pleural characteristic patterns showed adhesion lesions and both adhesion lesions and presence of micronodules, respectively. Conclusions: M. TB culture from pleural tissue should be considered a routine test when facing unknown pleural effusion during pleuroscopic examination.

Xpert MTB/RIF Ultra outperformed the Xpert assay in tuberculosis lymphadenitis diagnosis: a prospective head-to-head cohort study

OBJECTIVES

Xpert MTB/RIF Ultra (Xpert-Ultra) has shown better sensitivity in comparison with Xpert MTB/RIF (Xpert) in extra-pulmonary tuberculosis, while the head-to-head comparison of these methods in tuberculosis lymphadenitis had barely been performed.

METHODS

Patients with undiagnosed lymphadenopathy were recruited prospectively and consecutively, and fine-needle aspirates biopsy (FNA) or lymph node tissue was collected. The specimen was subjected to smear, culture, Xpert and Xpert-Ultra assays. Culture and/or smear for AFB or AFB observed on histopathology were performed as a reference.

RESULTS

106 participants were recruited, including 41 confirmed TB, 33 probableTB and 32 non-TB lymphadenopathy. The head-to-head comparison for Mtb detection showed that Xpert-Ultra produced the highest sensitivity when compared with smear, culture and Xpert (75.7% vs. 5.4 %, 13.5% and 48.7%).When Xpert-Ultra outcomes were integrated for diagnosis, the percentage of confirmed TB lymphadenitis case increased from 55.4% (41/74) to 85.1% (63/74). The sensitivities of Xpert-Ultra and Xpert on tissue were 73.6% (95%CI,59.4-84.3) and 39.6% (95%CI,26.8-54.0), respectively. The sensitiviyt of Xpert-Ultra on FNA samples (81.0%,95%CI 57.4-93.7) was higher than that of Xpert (71.4%,95% CI 47.7-87.8).

CONCLUSIONS

Xpert-Ultra detected significantly more TB lymphadenitis cases than Xpert or culture. This superiority was particularly distinct using lymph node tissue than FNA detection.

Xpert MTB/RIF Ultra versus Xpert MTB/RIF for diagnosis of tuberculous pleural effusion: A systematic review and comparative meta-analysis

Objective

We compared diagnostic accuracy of pleural fluid Xpert MTB/RIF (Xpert) and Xpert MTB/RIF Ultra (Ultra) assays for diagnosing tuberculous pleural effusion (TPE), through systematic review and comparative meta-analysis.

Methods

We searched PubMed and Embase databases for publications reporting diagnostic accuracy of Xpert or Ultra for TPE. We used bivariate random-effects modeling to summarize diagnostic accuracy information from individual studies using either mycobacterial culture or composite criteria as reference standard. We performed meta-regression through hierarchical summary receiver operating characteristic (HSROC) modeling to evaluate comparative performance of the two tests from studies reporting diagnostic accuracy of both in the same study population.

Results

We retrieved 1097 publications, and included 74 for review. Summary estimates for sensitivity and specificity for Xpert were 0.52 (95% CI 0.43–0.60, I² 82.1%) and 0.99 (95% CI 0.97–0.99, I² 85.1%), respectively, using culture-based reference standard; and 0.21 (95% CI 0.17–0.26, I² 81.5%) and 1.00 (95% CI 0.99–1.00, I² 37.6%), respectively, using composite reference standard. Summary estimates for sensitivity and specificity for Ultra were 0.68 (95% CI 0.55–0.79, I² 80.0%) and 0.97 (95% CI 0.97–0.99, I² 92.1%), respectively, using culture-based reference standard; and 0.47 (95% CI 0.40–0.55, I² 64.1%) and 0.98 (95% CI 0.95–0.99, I² 54.8%), respectively, using composite reference standard. HSROC meta-regression yielded relative diagnostic odds ratio of 1.28 (95% CI 0.65–2.50) and 1.80 (95% CI 0.41–7.84) respectively in favor of Ultra, using culture and composite criteria as reference standard.

Conclusion

Ultra provides superior diagnostic accuracy over Xpert for diagnosing TPE, mainly because of its higher sensitivity.

Rapid Molecular Diagnosis of Extra-Pulmonary Tuberculosis by Xpert/RIF Ultra

Rapid detection of Mycobacterium tuberculosis complex and determination of drug resistance are essential for early diagnosis and treatment of tuberculosis (TB). Xpert MTB/RIF Ultra (Xpert Ultra), a molecular test that can simultaneously identify M. tuberculosis complex and resistance to rifampicin directly on clinical samples, is currently used. Xpert Ultra represents a helpful tool for rapid pulmonary TB diagnosis, especially in patients with paucibacillary infection. The aim of this review is to provide an overview of the diagnostic performance of Xpert Ultra in detection of extra-pulmonary tuberculosis.

Evaluating the efficacy of stool sample on Xpert MTB/RIF Ultra and its comparison with other sample types by meta-analysis for TB diagnostics

Precise and timely detection of tuberculosis (TB) is crucial to reduce transmission. This study aims to assess the accuracy of Xpert MTB/RIF Ultra on stool samples and systematically review the performance of Xpert MTB/RIF Ultra with different sample types by meta-analysis. Stool samples of smear-negative pulmonary TB (PTB), cervical lymph node TB, and abdominal TB patients were tested on the Xpert MTB/RIF Ultra system. Meta-analysis was performed on a set of 44 studies. Data were grouped by sample type, and the pooled sensitivity and specificity of Xpert MTB/RIF Ultra were calculated. The sensitivity of Xpert MTB/RIF Ultra with stool samples was 100% for smear-negative PTB, 27.27% for cervical lymph node TB, and 50% for abdominal TB patients, with 100% specificity for all included TB groups. The summary estimate for all PTB samples showed 84.2% sensitivity and 94.5% specificity, and EPTB samples showed 88.6% sensitivity and 96.4% specificity. Among all sample types included in our meta-analysis, urine showed the best performance for EPTB diagnosis. This pilot study supports the use of stool as an alternative non-invasive sample on Xpert MTB/RIF Ultra for rapid testing, suitable for both PTB and EPTB diagnosis.

Head-to-head comparison of the efficacy of Xpert MTB/RIF Ultra and Xpert MTB/RIF for the diagnosis of tuberculous pleurisy: A systematic review and meta-analysis

BACKGROUND

The aim of this study was to evaluate the diagnostic accuracy of Xpert MTB/RIF Ultra (Xpert Ultra) and Xpert MTB/RIF (Xpert) for the diagnosis of tuberculous pleurisy (TBP) head-to-head using meta-analysis method.

METHODS

On May 12, 2021, we searched multiple databases for reports that used Xpert Ultra and Xpert for TBP diagnosis head-to-head and screened eligible studies for inclusion. Accuracy of Xpert Ultra and Xpert were compared to that of the composite reference standard (CRS) and culture. When heterogeneity was evident, sources of heterogeneity were explored using subgroup analyses, sensitivity analysis, and meta-regression analyses.

RESULTS

Five articles met the inclusion criteria for meta-analysis. When results from different specimens or different reference standards were reported in the same article, we analyzed them as separate studies. Thus, 6 studies compared Xpert Ultra and Xpert with CRS, 5 studies compared Xpert Ultra and Xpert with culture. Pooled sensitivity and specificity of Xpert Ultra were 52% and 98% compared to CRS, and 82% and 77% compared to culture. Pooled sensitivity and specificity of Xpert were 22% and 99% compared to CRS, and 48% and 94% compared to culture. Significant heterogeneity in sensitivity was observed compared to CRS.

CONCLUSION

The sensitivity of Xpert Ultra was moderate but better than that of the Xpert; however, its specificity was lower. The role of Xpert Ultra and Xpert in the early and rapid diagnosis of TBP was limited.

Accuracy of Xpert MTB/RIF assay for the diagnosis of tuberculous pleural effusion

Background

Tuberculosis poses a severe threat to human health. At present, compared with the traditional diagnostic methods for tuberculosis pleural effusion, such as Löwenstein–Jensen culture, pleural biopsy, and Ziehl–Neelsen smear microscopy, Xpert MTB/RIF was regarded as an emerging technology for its efficiency. The Xpert MTB/RIF accuracy for tuberculous pleural effusion diagnosis was evaluated in this systematic study.

Materials and methods

We searched the relevant literature published before January 2021 in PubMed, Cochrane, EMBASE, and Web of Science databases. Utilizing Review Manager 5.3 software, the quality of the included literature was evaluated based on the Quality Assessment of Diagnostic Accuracy Studies criteria. Sensitivity, specificity, and the summary receiver operating characteristic curves were plotted and analyzed with Metadisc 1.40 software. We used Stata 12.0 software to evaluate the publication bias of this study.

Results

Eighteen articles were identified in total. The sensitivity of Xpert MTB/RIF in the pleural effusion was 0.24, and specificity was 1.00, respectively. The area under the summary receiver operating characteristic curve was 0.9737, which indicated that the overall accuracy of the Xpert MTB/RIF was high. In addition, based on the Deeks funnel plot, no publication bias of the study was found.

Conclusion

Xpert MTB/RIF is a rapid method with high specificity but relatively low sensitivity for detecting Mycobacterium tuberculosis in pleural effusion. Its less sensitivity made it difficult to be used clinically, but the high specificity suggests that it can be used as a specific diagnostic method for tuberculous pleural effusion.

Diagnostic accuracy of adenosine deaminase for pleural tuberculosis in a low prevalence setting: A machine learning approach within a 7-year prospective multi-center study

OBJECTIVE

To analyze the performance of adenosine deaminase in pleural fluid combined with other parameters routinely measured in clinical practice and assisted by machine learning algorithms for the diagnosis of pleural tuberculosis in a low prevalence setting, and secondly, to identify effusions that are non-tuberculous and most likely malignant.

PATIENTS AND METHODS

We prospectively analyzed 230 consecutive patients diagnosed with lymphocytic exudative pleural effusion from March 2013 to June 2020. Diagnosis according to the composite reference standard was achieved in all cases. Pre-test probability of pleural tuberculosis was 3.8% throughout the study period. Parameters included were: levels of adenosine deaminase, pH, glucose, proteins, and lactate dehydrogenase, red and white cell counts and lymphocyte percentage in pleural fluid, as well as age. We tested six different machine learning-based classifiers to categorize the patients. Two different classifications were performed: a) tuberculous/non-tuberculous and b) tuberculous/malignant/other.

RESULTS

Out of a total of 230 patients with pleural effusion included in the study, 124 were diagnosed with malignant effusion and 44 with pleural tuberculosis, while 62 were given other diagnoses. In the tuberculous/non-tuberculous classification, and taking into account the validation predictions, the support vector machine yielded the best result: an AUC of 0.98, accuracy of 97%, sensitivity of 91%, and specificity of 98%, whilst in the tuberculous/malignant/other classification, this type of classifier yielded an overall accuracy of 80%. With this three-class classifier, the same sensitivity and specificity was achieved in the tuberculous/other classification, but it also allowed the correct classification of 90% of malignant cases.

CONCLUSION

The level of adenosine deaminase in pleural fluid together with cell count, other routine biochemical parameters and age, combined with a machine-learning approach, is suitable for the diagnosis of pleural tuberculosis in a low prevalence scenario. Secondly, non-tuberculous effusions that are suspected to be malignant may also be identified with adequate accuracy.