Predictive Risk Factors of Pleural Tuberculoma Occurrence on Patients with Pleural Tuberculosis

Background: Pleural tuberculoma (PTM) had been observed more and more frequently on patients with pleural tuberculosis (PT). The occurrence of PTM is not favorable for the treatment outcome of PT. However, its predictive risk factors had not been clarified. This study was to explore the predictive risk factors of PTM occurrence on the patients with PT. Methods: From January 2017 to June 2018, patients diagnosed as PT were enrolled prospectively into the study. All clinical characteristics and pleural changes during the chemotherapy were recorded and adenosine aminase (ADA) lactate dehydrogenase (LDH)glucose and protein of pleural fluid were quantified. Interferon-gamma (IFN- γ) transforming growth factor -beta (TGF-β), plasminogen activator inhibitor type-1 (PAI-1) and tissue plasminogen activator (t-PA) of pleural fluid were tested before chemotherapy by enzyme-linked immunosorbent assay (ELISA). Results: A total of 162 patients with PT were enrolled. 82 patients (51.3%, 82/162) developed PTM during chemotherapy. We found the probability of PT to develop PTM reached 99.5% when the patients with pleural thickness, packaged plural effusion, LDH>461.5 IU/L , ADA>62.9 IU/L , GLU<4.75 mmol/L and age>32.5 years old. The probability to develop PTM was as low as 1.4% when these 6 indices were in the opposite trend to the conditions list above. Using tested cytokines in pleural fluid from 76 cases, the probability of PTM if patients satisfied 3 indexes including increased TGF-β(>15.235 μ g/L) , increased PAI-1(>180.720 μ g/L) and decreased t-PA(<2.875 μ g/L ) was 78%, and the probability to develop PTM was only as low as 2.7% when these 3 cytokines were in the opposite trend. Conclusions: Age, pleural thickness, packaged pleural fluid, increased ADA, LDH, TGF-β, PAI-1 and

indices were in the opposite trend to the conditions list above. Using tested cytokines in pleural fluid from 76 cases, the probability of PTM if patients satisfied 3 indexes including increased TGFβ(>15.235 μ g/L) , increased PAI-1(>180.720 μ g/L) and decreased t-PA(<2.875 μ g/L ) was 78%, and the probability to develop PTM was only as low as 2.7% when these 3 cytokines were in the opposite trend.
Conclusions: Age, pleural thickness, packaged pleural fluid, increased ADA, LDH, TGF-β, PAI-1 and decreased glucose and t-PA might be predictive risk factors of developing PTM for patients with PT, clinicians should take deep consideration if patients had above factors. Background Tuberculosis (TB) remains one of leading infectious diseases in the world. There were estimated 10.0 million new cases of TB worldwide in 2017 [1]. Pleural tuberculosis (PT) ranked the second most common form of extra-pulmonary tuberculosis [2,3]. In recent years, pleural mass or nodules had been observed in patients with PT, several published literatures reported its occurrence during the chemotherapy of PT and it was called as pleural tuberculoma (PTM) which was confirmed pathologically, bacterially or clinically [4][5][6][7]. PTM were observed by Chest CT with shown in figure 1(A, B, C and D). Figure 1A showed the chest CT with right pleural effusions when the patients admitted to hospital; Figure 1B showed the right pleura was clear after drainage of pleural fluid and chemotherapy; Figure 1C revealed the right pleural mass diagnosed as PTM after 4 months of standard first line anti-TB chemotherapy; Figure 1D indicated absorption of PTM after adjusting anti-TB regimen. In our study, PTM was referred as the pleural mass or nodules confirmed to be tuberculous granuloma by pathological examination with or without mycobacterium tuberculosis within lesions.
Hundreds of patients with similar radiological characteristic were found in clinics in China. However, published studies were limited in few case reports and all of them were reported in Asian countries or regions with high TB burden. There had no unified treatment regimen for patients with PTM currently.
In china, we found PTM occurrence frequently, predominant in the course of chemotherapy of PT [5,8]. We think that PTM might be the origin of anti-TB treatment failure or relapse because mycobacterium tuberculosis complex (MTC) were mostly found in the pleural mass proved pathologically although patients have finished part course of chemotherapy with first-line anti-TB drugs. However, few studies were published with regard to the reasons of its occurrence and there had no unified treatment regimen for PTM, how to get further treatment and prevent its occurrence remains elusive.
In this study, we tried to find out the predictive risk factors of PTM at the initiation of anti-TB chemotherapy when patients had pleural effusion without pleural mass or nodule in order to prevent its occurrence.
Because of few studies referred to the reason and mechanism of PTM, we hypothesized Tissue plasminogen activator (t-PA) and plasminogen activator inhibitor type-1 (PAI-1) might be involved into the development of PTM. t-PA and PAI-1 are vital cytokines of fibrinolytic system and played an important role in activation and inhibition in coagulation system [9]. Previous studies showed that imbalance of PAI-1 and t-PA in pleural spaces may lead to fibrin deposition and pleural inflammation [10,11]. Another cytokine, transforming growth factor -beta (TGF-β), a superfamily regulates tissue regeneration can also induce epithelial-mesenchymal transition (EMT) [12,13]. TGF-β might increase the levels of PAI-1 [14,15]. Therefore, we designed the prospected study including testing PAI-1, t-PA, TGF-β and Interferon-gamma (IFN-γ) in pleural effusion then observing the pleural changes during the chemotherapy, trying to find out the underlying predictive risk factors of PTM occurrence.

Patients
Between January 2017 and June 2018, patients diagnosed as PT with pleural effusion were admitted to Shanghai Pulmonary Hospital. Inclusive criteria were patients newly diagnosed as PT, pleural fluid could be available, and had no pleural mass or nodules before chemotherapy. All individuals tested were negative for human immunodeficiency virus (HIV) and had no history of anti-TB treatment.
Patients were undergoing corticosteroid therapy, or had history of malignant tumor, or complicated by rheumatoid arthritis, or other pulmonary diseases were excluded.
Patients were diagnosed as PT according to the WHO guideline, if there was at least one of the following criteria [16,17] The diagnostic criteria of PTM were patients satisfied the following conditions: (1) patients were diagnosed as PT clinically; (2) Chest CT scan showed pleural effusion before chemotherapy, no pleural mass or nodule were found after drainage of pleural effusion before chemotherapy of TB and then solitary or multiple pleural nodule were found by CT scan after chemotherapy which was confirmed as granulomatous lesions pathologically through pleural biopsies.

Clinical Sample Selection and Data Collection
Suspected PT patients underwent a percutaneous transthoracic puncture to collect pleural effusion. 10 ml pleural effusion of 162 patients was sent to the clinical laboratory for clinical test and another 10 ml of 76 patients was frozen in the refrigerator at -20 °C. All clinical characteristics including name, included number, age, symptoms, complications, final diagnosis, pleural thickness, the diameter of pleural nodule, the history of diagnosis were recorded completely. Pleura thickness > 2 mm was considered as plural thickness [18]. Follow-up period was one year since starting the chemotherapy. Variables displayed as percentages were compared with a χ2 test. Differences with P <0.05 were considered significant. The diagnostical value of each biomarker was calculated with the area under ROC curve (AUC) of receiver operation curve (ROC), best cutoff value was found at the point where Youden Index is the largest (Youden Index = Sensitivity + Specificity -1). Pre-test and post-test possibility were used to evaluate the predictable ability of the biomarkers. The Fagan's nomogram was used to present pre-test probability and post-test probability which derived from Stata module (FAGAN: Stata module for Fagan's Bayesian nomogram). Alluvial Diagrams was used to present the study cohort's frequency at the study items with levels which derived from R package (ggalluvial:

Results
Clinical Characteristics 162 patients diagnosed as PT were finally enrolled in the study, 111 patients were male (68.5%), 51 were female (31.5%). The median age of them was 34 years old, range from 24-55 years old. 64.8%  Table 1 Patients with PTM were found to be younger (29 years old vs. 42 years old, p<0.05) than those who didn't developed PTM. The percentage of pleural thickness and packaged pleural effusion checked by ultrasound was significantly higher in PTM group compared to that in patients without PTM (80.5% vs.  Table 4. We use Fagan's nomogram to further evaluate the prognostic capacity of clinical characteristics, biochemical indexes and cytokines. From 162 patients, we found the probability of PT to develop PTM reached 99.5% when the patients with pleural thickness, packaged plural effusion, LDH>461.5IU/L, ADA>62.9IU/L, GLU<4.75mmol/L and age>32.5 years old. The probability to develop PTM was as low as 1.4% when these 6 indices were in the opposite trend to the conditions list above. Using tested cytokines in pleural fluid from 76 cases, the probability to occur PTM if patients had satisfied 3 indexes including increased TGF-β(>15.235μg/L), increased PAI-1(>180.720μg/L) and decreased t-PA(<2.875μg/L) was 78%, and the probability to develop PTM was as low as 2.7% when TGF-β, PAI-1 indexes were under their best cutoff value and t-PA were higher than its best cutoff value. The prior probabilities of these indexes were shown in Figure 3 and

Discussion
Recent years PTM had been more and more observed in Chinese clinics most physicians failed to know how to get further treatment, especially for most cases without pleural nodule or mass on the initiation of anti-TB chemotherapy. These nodules or mass were mostly confirmed to be PTM which were granulomatous lesions, caseous necrosis with or without acid-fast bacillus (AFB) within the lesions. PTM might be single or multiple, it's occurrence will bring patients unfavorable outcome, high cost and might be origin of treatment failure or relapse [19]. Therefore, exploring the predicted factors of PTM is essential for physicians to prevent its occurrence.
For underlying clinical predicative factors, according to the results in present study, pleural thickness and packaged pleural effusion might be risk predictive factors of PTM. These severe statuses sometimes need surgical operation [20,21]. Pleural thickness was reported to be the predictor for PT, Ruiz reported [22] that PT with pleural thickness were more common in men, elderly people, however, other studies showed that age, symptoms and pleural thickness were not related with PT [18], therefore, although existed controversial conclusions, combined with above results, pleural thickness might be closely related to the occurrence of PTM because the location of PTM is accompanied by local pleural thickness [23].
It was been reported that the occurrence of PTM was related to nonstandard anti-TB treatment, delayed treatment and no use of corticosteroid [8]. However, this study did not find these risk factors associated with the occurrence of PTM. It showed that patients younger than 32.5 years old were easier to have PTM, the reason might be associated with stronger allergic reaction in pleura.
For biochemical markers in pleural effusion, our results showed ADA and LDH in pleural effusion were higher in PTM patients and glucose was lower than patients without PTM. It had been reported that LDH of plural fluid and the LDH ratio of pleural fluid to blood were higher in patients with pleural thickness than those in patients without pleural thickness [22]. Meanwhile another study found that pleural thickness may be related to lower glucose and higher ADA in pleural effusion [24][25][26].
Therefore, our findings in this study that higher ADA, LDH and lower glucose in pleural fluid associated with PTM might be reasonable.
For cytokines tested in pleural effusion, their change of expression had been associated with immune response in patients with PT. With the development of immunological reaction in host, the immune factors in pleural effusion might be predictors of pleural thickness, packaged pleural effusion which could be quantitatively analyzed [23,27,28]. According to Chung [9], it was reported that the fibrinolytic activity was depressed in packaged pleural effusions compared with free-flowing effusions.
We found PTM patients were more common with packaged pleural effusion. Other studies showed that high levels of IFN-γ, TGF-β and lower glucose in pleural effusion were correlated with the occurrence of pleural thickness in PT [29], and they thought that pleural thickness and loculation of pleural effusion were related to local pleural immune response associated with cytokines. Other researchers found pleural thickness was related to the imbalance of PAI-1 and t-PA in pleural effusion and PAI-1 increased, the t-PA decreased in PT patients with pleural thickness while decreased PAI-1, increased t-PA was found in patients with sufficient drainage of pleural effusions, therefore, PAI-1 and t-PA may be predictors of pleural thickness in PTM [9,18,27,29]. Our study had proved these changes of PAI-I and t-PA in PTM patients.
In order to figure out whether these indexes were valuable in predicting occurrence of PTM, we compared cytokines' level in PTM and non-PTM patients. Comparing four cytokines of pleural effusion in PTM and non-PTM groups, higher TGF-β (20.021μg/L vs 16.386μg/L),, higher PAI-1 (241.014μg/L vs 187.152μg/L) and lower t-PA (1.733μg/L vs 3.025μg/L) were found in PTM group. There was no statistical difference of IFN-γ between two groups. Increased TGF-β, PAI-1 and decreased t-PA were the predictors of PTM in PT patients. We further analyzed these indexes which were recognized to have statistical differences between PTM and non-PTM group using ROC curve to find the best cutoff value of these indexes, and calculated the posterior probability of these indexes. Our study showed that combination of six indexes including plural thickness, LDH, package of plural effusion, ADA, glucose and age can increase probability of developing PTM to 99.5%, on the contrast, the probability to develop PTM was as low as 1.4% if these indices were in the opposite trend. Estimating 3 cytokines indexes of pleural effusion, the probability to develop PTM was 78% when TGF-β, PAI-I were above the best cutoff value and t-PA was below the best cutoff value. The posterior probability was only 2.7%,

Availability of data and materials
The datasets used and/or analyzed during the present study are available from the corresponding author on reasonable request.     Fagan's nomogram for the 3 tests of 76 sub-cohort which take in sequence of t-PA, TGF-β, PAI-I.