The utility of T-SPOT.TB for the diagnosis unconventional pleural tuberculosis is superior to ADA in high prevalence: A perspective analysis of 601 cases

BACKGROUND Recently, Interferon Gamma Release Assay (IGRA) is still controversial in differentiating tuberculous pleural effusion (TPE), through recommended by World Health Organization (WHO )for identication of latent tuberculosis infection. OBJECTIVES Aim to qualify the diagnostic ecacy for TPE by IGRA in comparison to Adenosine deaminase(ADA), to clarify its appropriate scene in clinical diagnosis. METHODS A prospective, single-centre study including all suspected pleural effusion patients consecutively from June 2015 to October 2018. Through receiver operating characteristic (ROC) curves, all enrolled participants were determined technical cut-off and the utility of IGRA(T-SPOT)for pleural uid(PF). Obtain the independent risk factors by logistic regression analysis for TPE, and evaluate the performance of T-SPOT stratied by risk factors, in comparison to ADA. RESULTS A total of 601 individuals were consecutively recruited. The maximum of early secretory antigenic target-6 (ESAT-6) and culture ltrate protein-10 (CFP-10) in PF T-SPOT had the best diagnostic eciency in our study, with a sensitivity of 83.0% and a specicity of 83.1%, corresponding cut-off value is 466, which was equal to ADA(0.885 vs 0.887, P=0.957) and superior than in PB; Among the TPE patients with low ADA(<40 IU/L), the sensitivity and specicity of PF T-SPOT was still 87.9%, 90.5% respectively. The utility of ADA was negative related to age ascents, but PF T-SPOT had steady performance at any age-stage. The age (<45 yrs; odds ratio (OR) 5.61), gender (male; OR 2.7) and body mass index (BMI)(<22; OR 1.93) was independently associated with the risk of TB by multivariate logistic regression analysis. Stratied by risk factors, notably the PF T-SPOT had superior sensitivity(76.5% vs. 23.5%,P=0.016) than ADA meanwhile had the non-inferior specicity(84.4% vs. 96.9%, P=0.370). the overall of assay for the TPE patients whose 40IU/L, and extremely superior to ADA in unconventional TPE patients(age>45yrs, female assay is an good choice to supplement ADA to diagnose TPE. only 4.65% of TPE subjects that main compose of elderly primarily had levels over 40 IU/L[21],our research fully demonstrated this phenomenon simultaneously. In addition, we observed that, the fuzzy boundary inuenced by age on ADA is opportunely concentrated in ADA indeterminate groups.among the patients of ADA indeterminate, there were 85.7%(18/21), 73.7%(19/26) and 47.7%(31/65) patients aging more than 45yrs in non-, conrm- and probable- TPE group, respectively. while the superior performance of PF T-SPOT between ADA indeterminate groups may be explained by its steady at all age stages, besides the interference for ADA by other inammatory etiologies. Still, ADA is a widely-use biomarker for screening TPE due to its simplicity, rapidity, and low nical costs,but above results proved that over-reliance on ADA differentiation may lead to missed diagnosis/misdiagnosis in clinical diagnosis, especially in indeterminate range. Meanwhile, PF T-SPOT as luck would ll this blank.

investigators, and tracked the treatment process and discharge diagnoses. All participants had followed up for 12 months to verify the nal diagnosis, and the patients with negative outcome of anti-TB treatment at the last 12 months were deemed as indeterminate diagnosis.
Clinical categories of pleurisy Patients were divided into three groups according to the composite reference standard(CRS),which was composed of clinical, laboratory, and radiological examinations and follow-up data of diagnostic treatment. (1) Bacteriologically con rmed TPE was de ned as any positive TB etiology testing from PE or sputum; (2) Probable TPE were lacking bacteriological con rmation but all were treated empirically for TB based on clinical suspicion(e.g. typical clinical symptoms, remarkable radiological imaging and positive outcome of anti-TB treatment during follow-up); (3) Non TPE indicated cases diagnosed de nitely as other diseases,such as malignant, empyema(Non-tuberculous disease) etc.

T-SPOT.TB in PF and PB
The samples of PB (4 mL) and PF (45 mL) collected from each participant were tested within 6 hours. PB samples were diluted 1 fold and centrifuged at 900 g for 20 min and PF sample centrifuged at 500 g for 10 min respectively, both discarding plasma supernatant for T-SPOT.TB testing.
T-SPOT.TB was conducted following the manufacturer's instructions (Oxford Immunotec Ltd, Oxford, UK),which is identical for PB and PF samples.The pellets were resuspended in 8 mL AIM-V medium (GIBCO, Rockville, MD, USA) and the cell suspension was separated to obtain mononuclear cells. Setting an empty wells as negative controls, the T lymphocyte mitogen phytohaemagglutinin as positive control, and ESAT-6 and CFP-10 peptides in separate wells,respectively. Isolated peripheral blood mononuclear cells (PBMCs) and pleural uid mononuclear cell (PFMCs)were added to the wells (2.5 × 10 5 cells per well) that were pre-coated with a monoclonal antibody against IFN-γ, and incubated at 37 °C for 16 to 20 hours. The spot-forming cells (SFCs) were read from an automated enzyme-linked immunosorbent spot (ELISPOT) reader (CTL-ImmunoSpotS5 Versa Analyzer) .

Diagnosis
Smear, Acid fast bacilli(AFB) and Mycobacterial culture Specimens including the sputum and PF(5 mL) was prepared for direct smear and stained with auramine, examined by light-emitting diode microscopy. The smear was read and interpreted in accordance with the WHO guidelines [11]. The sputum and PF(5 mL) were preprocessed by N-acetyl-L-cysteine and sodium hydroxide (NALC-NaOH), centrifugated and discarding supernatant, the resuspended pellet was transferred to solid Lowenstein-Jensen medium (Encode Medical Engineering Co., Ltd, China) and liquid medium, and subjected to cultivating in mycobacteria growth indicator tube (MGIT) 960 system (Becton, Dickinson and Company, USA). The presence of Mtb complex in any medium represented the positive of MPT64 antigen testing.
Recording the positive events and time.

Gene-Xpert
The Gene-Xpert were performed as the manufacturer's instructions(Cepheid, Sunnyvale, CA, USA). Brie y, fully premixing the specimen(including sputum and concentrated PF) and sample reagent at the room temperature. Final 2 ml mixture was collected and transferred into the cartridge and loaded into the automatic GeneXpert instrument. Duplicated testing was performed on the sample with an invalid result.

Statistical analysis
Data were analyzed using IBM SPSS 25.0 (SPSS Inc. Chicago, IL, USA) GraphPad Prism 8.2.1 (GraphPad Software, Inc. La Jolla, USA). Quantitative variables were presented as mean ± standard deviation (SD) or median (interquartile ranges(IQR)), and categorical variables were presented as frequencies (percentages). To identify differences between two independent groups, Chi-square test was used to detect differences between categorical variables, Mann-Whitney U test and unpaired t-test was applied for continuous data in non-normal or normal distribution, respectively. A result was considered statistically signi cant when the P-value was < 0.05.
Receiver operating characteristic (ROC) curves were plotted to evaluate the diagnostic performance of ADA and T-SPOT.TB, respectively, obtaining optimal cut-off value and calculating corresponding areas under the ROC curve (AUCs), sensitivity, speci city, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (+ LR), negative likelihood ratio (-LR), diagnostic odds ratio(DOR) and accuracy were calculated.
Extracting From the clinical record, physical examination and PF biochemical data (Including age, gender,body-mass-index(BMI), onset-time, history of previous anti-TB, symptoms, PF total protein, glucose and LDH), independently risky factors of TB were obtained from regression models combining with multicolinearity assessment. The nal bioclinical scores for each factor were assigned through odds ratio(OR) in the nal multivariate model. Overall research were completed in keeping with the Standards for Reporting of Diagnostic Accuracy (STARD) template [12].
Among con rm TPE group, the sputum smear.AFB, culture, Gene-xpert were positive in 27.6%, 60.5% and 64.6%, respectively; while for PF corresponding positive in only 2.2%,37.4% and 25.4%, respectively, and higher total detection rate in sputum (68.3% vs. 46.9%) convinced that obtaining the direct proof of MTB infection from PF is more di cult than from sputum.

Clinic, demographic and biochemical data
The demographic and clinical characteristics of all participants were summarized in Table 1. The patients in TPE group is more younger than that in non TPE group(42.15 ± 19.78 vs. 57.59 ± 15.36, p < 0.001), and there were more male subjects in TPE group (75.3%,299/397 vs. 59.2%,103/174, p < 0.001), and predominant in unilateral PE(83.9%, 333/397), all features were in accordance with a case series from Qatar [13]. The TPE group had signi cantly thinner than the non TPE group (21.70 ± 4.24 vs. 23.23 ± 3.37, p < 0.001). Patients with TPE more frequently presented with fever (74.8% vs. 39.7%, p < 0.001),but had less chest tightness (58.6% vs. 71.8%, p = 0.013). the probable TPE that inferred by clinicians had more obvious clinical symptoms relating to TB infection, for instance more night sweat (21.4% vs. 10.3%, p = 0.009), more wight lose (34.1% vs. 23.6%, p = 0.042)and less hemoptysis(l.2% vs. 7.5%, p < 0.001). However, there were no signi cant differences in various characteristics between the con rm and probable TPE(P > 0.05).  Diagnostic utility of T-SPOT.TB assay for PB and PF As shown in Fig. 2, the nal ESAT-6, CFP-10, and Max SFCs (represents the larger of ESAT-6 and CFP-10) for PB and PF respectively were a rmed to discriminate TPE from non TPE, and no signi cant differences were observed between con rm-and probable-TPE. In Table 3,We exhibited that cut-off value of PB derived from receiver operating characteristic curve (ROC) analysis between con rm TPE group and non TPE group (Fig. 2), which is extremely close to the positive cutoff value(24 SFCs/10 6 mononuclear cells) provided by the manufacturer. Overall, when taken the same cut-off value = 22 SFCs/10 6 mononuclear cells, the performance of ESAT-6 was slightly better than CFP-10 in PB, with AUC of 0.840 vs 0.796(P = 0.055), as well as a sensitivity of 82.1% vs 75.2% (P = 0.123)and a speci city of 75.3% vs 77.9% (P = 0.847); However, when considering Max SFCs (cut-off value = 22 SFCs/10 6 mononuclear cells), of AUC, sensitivity and speci city was 0.83(95% CI 0.794-0.884), 90.3% and 67.2% respectively, no better than ESAT-6(P = 0.954). (Table 3).  Age < 45 yrs, male and BMI < 22 scored 6,3,2 points, respectively; §Comparisons were performed for sensitivity between ADA and PF T.SPOT using chi-square test. ¶ Comparisons were performed for speci city between ADA and PF T.SPOT using chi-square test.
a Values are expressed as IU/L for ADA, and as SFCs/10 6 mononuclear cells for T-SPOT.Data are presented as percentage.
As expected, the performance of PF T-SPOT was distinctly improved in contrast to PB T-SPOT. ESAT-6 and CFP-10 speci c cells were more highly concentrated in PF than in PB by median ratio of 12.13(IQR 3-29.4) and 9.30(IQR 1.22-30.15) in con rm TPE group, and median ratio of 11.87(IQR 3.96-35.15) and 10.60(IQR 2.63-32.21) in probable TPE group, and no signi cance were observed between any groups. Based on the ROC analysis, the optimal cut-off point was 170 for PF ESAT-6, which produced a sensitivity of 86.9% and speci city of 78.2%; and 142 for PF CFP-10, which produced a sensitivity of 85.5% and speci city of 73.6%. While, of that Max SFCs in PF exhibited the best diagnostic e ciency which was equal to ADA(0.885 vs 0.887, P = 0.957), with a sensitivity of 83.0% and a speci city of 83.1%, corresponding cut-off value is 466 SFCs/10 6 mononuclear cells (Fig. 2, Table 3).
Comparison of diagnostic utility of ADA and T-SPOT.TB assay strati ed by bioclinical score The median of ADA levels in non-,con rm-and probable-TPE group were 11.8 U/L(IQR8.25-18.65), 50 U/L(IQR37.85-62.15), and 45 U/L(IQR31.875-57.9), respectively (Table 1), con rming that low ADA level was satisfactory in excluding the TPE, but the ADA level in probable-TPE group was slighter lower than that in con rm-TPE group(P = 0.055).
Each participant was grouped by scoring of logistic regression coe cient [14]. As shown in Table 5, when score = 11, meeting three risky factors(< 45 yrs,male and BMI < 22), the performance of PF T-SPOT was non-inferior to ADA; comparing to the stable utility of PF T-SPOT, the sensitivity of ADA was positive related to score decline, while speci city negatively,suggesting that PF T-SPOT performed better than ADA in the unconventional patients, especially when score = 0(represents the female patients whose age more than 45 yrs and BMI ≧ 22), the PF T-SPOT had the non-inferior speci city(84.4% vs. 96.9%, P = 0.370) meanwhile had superior sensitivity(76.5% vs. 23.5%,P = 0.016) than ADA.
Comparison of diagnostic utility of ADA and T-SPOT.TB strati ed by age In Fig. 3A, the scatter plot showed that the distribution of ADA levels in the TPE group had shifted downwards(P < 0.05) from age of 40+, especially the median ADA level in patients over than 60 years old were lower than clinical diagnostic point (40 U/L), intended that that ADA activity level had signi cantly negatively correlated with age-ascent (P < 0.001); notably, the performance of PF T-SPOT at different age stage is steady, which had no signi cant differences among all groups(P = 0.604 ) (Fig. 3B).The above results showed that the performance of T-SPOT.TB was superior to ADA in older patients.
Diagnostic utility of T-SPOT.TB assay in patients with ADA indeterminate (ranging from 20 to 40) In our study, the cut-off value of ADA derived from ROC analysis was 22.4 U/L, which had higher sensitivity(89.0%); conversely, when ADA more than 40 U/L recognized as positive (International Recommends), it produced higher speci city(93.1%). Therefore,we de ned 112 patients (19.6%) with ADA value ranging from 20 to 40(21 in non TPE, 26 in con rm TPE, 65 in probable TPE) as the ADA indeterminate. In Fig. 4, the scatter plot exhibited the diagnostic utility of PF T-SPOT in the indeterminate,the sensitivity, speci city, PPV and NPV was 87.9%,90.5%,97.6% and 63.3%, respectively. Youden index was 0.784; and the sensitivity, speci city, PPV and NPV of PB T-SPOT was 83.5%,76.2%,93.8% and 51.6%,respectively.Youden index was 0.597. These data intended that T-SPOT.TB assay could discriminate the TPE patients with ADA ranging from 20 to 40, and the utility of T-SPOT in PF was superior to that in PB.

Discussion
According to our grasping, through there were some publications on evaluating the utility of T-SPOT for diagnosis of TPE [9][10], the majority of subjects recruited small sample population(n < 100) for evaluation and the results was con icting.Our study provided the largest cohort(TPE:non TPE = 397:145) so far, con rmed that the value of T-SPOT assay for diagnosis of TPE with high con dence, and providing speci c reference suggestions for clinical application of T-SPOT.
In this study, all suspected PE participants had been consecutively, unselectedly enrolled, and all TPE patients had the de nite diagnosis by bacteriological con rmation or positive outcome for anti-TB therapy, which highly re ecting the demographic epidemiological characteristics of tuberculosis-prone areas. The recruited samples covered from adolescents to elderly patients aging 90+, con rmed again that the TPE patients had the clinical characteristics of younger male, combining with higher ADA activity and higher T-SPOT response for PF and PB. It's similar to other studies when IGRA are applied to PB, the unsatisfactory speci city was common as a result of ine ciency to discriminating active TB from latent TB infection, reversely the speci city of T-SPOT for PF is much higher with approximate sensitivity, thus the diagnostic cutoff obtained from PB is not available for PF; in our study the cut-off value derived from ROC analysis was higher than that in low prevalence areas,majority of which had taken as equal as PB cutoff in Europe [15],and some subjects obtained the cutoff by ROC analysis also is very low, such as 30SFCs/10 6 mononuclear cells in London[16]and 300 SFCs/10 6 mononuclear cells in Korea [17],respectively, which was in line with the expect of high-burden settings low threshold would compromise speci city, therefore it's crucial for tuberculosis-prone areas that to chose a proper threshold to apply PF T-SPOT.
Another biomarker ADA, a value of more than 40 IU/L in lymphocyte dominate PE carries a PPV of 98% in high TB endemic region [3][4]; while an retrospective analysis of assessing ADA in 1637 subjects con rmed less than 15 IU/L can get a NPV of 100% [18].As same as in our study, we found that recognizing > 40 IU/L as the solo indicator of TPE maybe not the most suitable, as it got the high speci city meanwhile sacri cing sensitivity, 35.5% (141/397) TPE patients' ADA level was lower than 40 IU/L in this study, 29.8% of which had de nite etiological basis. In addition, the utility of cutoff 22.4 IU/L derived from ROC analysis was better than > 40 IU/L(Youden index 0.729 vs. 0.641). However, comprehensively considering to the non-speci c elevation of ADA level caused by the non tuberculosis in ammatory PE[18](particularly complicated parapneumonic effusions and empyemas) and lymphomas, the patients whose ADA more than 20 IU/L and less than 40 IU/L could be classi ed as ADA indeterminate appreciatively. Reversely, PF T-SPOT showed the excellent diagnostic utility between ADA indeterminate groups, whose accuracy is higher to 90.2%, predicted that the result of PF T-SPOT could be a considerable indicator for the highly-suspected TPE patients with indeterminate ADA.
Many previous studies indicated that [19][20], the performance of diagnosing TPE for patients aging more than 45yrs is still an open question, a recent study about only 4.65% of TPE subjects that main compose of elderly primarily had levels over 40 IU/L [21],our research fully demonstrated this phenomenon simultaneously. In addition, we observed that, the fuzzy boundary in uenced by age on ADA is opportunely concentrated in ADA indeterminate groups.among the patients of ADA indeterminate, there were 85.7%(18/21), 73.7%(19/26) and 47.7%(31/65) patients aging more than 45yrs in non-, con rm-and probable-TPE group, respectively. while the superior performance of PF T-SPOT between ADA indeterminate groups may be explained by its steady at all age stages, besides the interference for ADA by other in ammatory etiologies. Still, ADA is a widely-use biomarker for screening TPE due to its simplicity, rapidity, and low nical costs,but above results proved that over-reliance on ADA differentiation may lead to missed diagnosis/misdiagnosis in clinical diagnosis, especially in indeterminate range. Meanwhile, PF T-SPOT as luck would ll this blank.
Besides age,we screened another two high risk factors that were signi cantly related to TPE, gender and BMI. There were barely 5 non TPE patients scoring 11(simultaneously ful ll three high risk factors:age < 45yrs,male and BMI < 22), it has directly demonstrated these clinical characters could be the effective reference index for discern TPE from other PEs. We often de ned these patients as the high incidence population to TPE, also called as typical population. However, it' s notably that the utility of ADA uctuated distinctly by strati ed analysis, and if not satis ed any one condition(de ned as the unconventional population) it is inferior to that in T-SPOT assay. The unconventional population frequently are the focal points and have di culties in diagnosing, then PF T-SPOT can provide powerful identi cation evidence for those.
There are several limitations on current study. First, The whole study were performed in a single center which specialized in TB. The geography, relative single control composition and etiological attribution error and/or bias was incalculable. Second, we got the optimal cut-off of PF T-SPOT.TB from ROC analysis in this training cohort, and its de nite accuracy would need the further validation. Finally, 38.1% of the clinical diagnosis patients lack the etiological basis due to the objective factors such as no sputum or sputum unsatis ed with the detection standard, which may bias sputum detection rate.

Conclusion
In conclusion, whether ADA or IGRA, they both can't be de ned as the de nite proof of ruling in or out, but as a powerful reference tool for TPE diagnosis.
The overall potency of PF T-SPOT assay is equal to ADA for diagnosing TPE in our study. In addition, PF T-SPOT can effectively discriminate the TPE patients with low ADA, and extremely superior to ADA in unconventional TPE patients(age > 45yrs, female or BMI < 22), intending applied population are wider than ADA. PF T-SPOT assay is an extremely good choice to supplement ADA to diagnose TPE.

Consent for publication
We have obtained consent to publish from all the participants.

Authors' contributions
Conception and design of the study: XTY, XYC, ZDZ; Sample and data collection, reviewing manuscript:QTL,LPP,HFD,YY,HL,CG,JZ; Experiments performing: HYJ,BPD,RRW,AYX; Data management and statistical analysis: JZ. Drafting the manuscript: JZ,XTY. All authors were involved in preparation and review of the manuscript and approved the nal version to be submitted.

Availability of data and materials
The datasets analyzed during the current study are not publicly available, but are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.  The receiver operating characteristic (ROC) curves of adenosine deaminase (ADA) and T-SPOT assay on the pleural uid (PF) or peripheral blood(PB) for diagnosing tuberculous pleural effusion (TPE). Samples were obtained from enrolled participants that included 145 with con rm TPE and 174 with non