Risk factors for tuberculous empyema in pleural tuberculosis patients

Tuberculous empyema (TE) is associated with high mortality and morbidity. In the retrospective cohort study, we aimed to find risk factors for TE among pleural tuberculosis (TB) patients. Between July 2011 and September 2015, all culture-confirmed pleural TB patients (474 cases) were enrolled in our study. Empyema was defined as grossly purulent pleural fluid. Demographic and epidemiological data were collected for further analysis. Multivariate logistic regression analysis was used to evaluate risk factors of TE in pleural TB, age–adjusted odds ratio (OR) and 95% confidence interval (CI) were calculated to show the risk. The mean age was 35.7 ± 18.1 years old, males comprised 79.1% of the participants (375 cases). Forty-seven patients (9.9%) were multidrug-resistant TB (MDR-TB), 29 (6.1%) had retreatment TB, 26 (5.5%) had diabetes mellitus. The percentage of empyema patients was 8.9% (42 cases). Multivariate analysis revealed that male (adjusted OR = 4.431, 95% CI: 1.411, 13.919), pleural adenosine deaminase (ADA, >88 U/L) (adjusted OR = 3.367, 95% CI: 1.533, 7.395) and white blood cell (WBC, >9.52 109/L) (adjusted OR = 5.763, 95% CI: 2.473, 13.431) were significant risk factors for empyema in pleural TB, while pulmonary TB (adjusted OR = 0.155, 95% CI: 0.072, 0.336) was the protective factor for the patients. TE remains a serious threat to public health in China. Male sex is a significant risk factor for TE while the presence of pulmonary TB is protective, and high levels of pleural ADA and WBC count could aid in early diagnosis of TE. This finding would help towards reducing the mortality and morbidity associated with TE.


Materials and Methods
The retrospective study was performed at the Shandong Provincial Chest Hospital (SPCH), located in Jinan city in eastern China. SPCH is a provincial referral TB hospital of approximately 800 beds. Each year, about 500 patients with pleural TB are diagnosed, most of them are diagnosed clinically, not confirmed by culture or pathological examination. In Shandong Province, which has a population of 90 million, 40, 000 new TB cases are found annually and about 10% isolates are MDR-TB 15 . This study was approved by the SPCH Ethics Committees. Under Article 39.1 of ethics guidelines enacted by National Health and Family Planning Commission of the PRC (http:// www.gov.cn/gongbao/content/2017/content_5227817.htm), this study was exempt from the need for written informed consent, as it used only secondary data.
Between July 2011 and September 2015, consecutive culture-confirmed pleural TB patients were enrolled in our study. Empyema was defined as grossly purulent pleural fluid 16 . Demographic and epidemiological data were collected using a questionnaire from electronic medical records, including sex, age, contact history of TB, smoking habit, treatment delays, underlying diseases and symptoms, and then were analyzed as described previously 17 .
Mycobacterium tuberculosis (M.TB) strains underwent drug susceptibility testing by using the absolute concentration method (isoniazid: 1 μg/ml, rifampicin: 50 μg/ml) on Lowenstein-Jensen medium 15 . PE samples obtained by thoracocentesis underwent biochemical and microbiological analysis. An automated chemistry analyzer (Advia 2400, Siemens Healthcare Diagnostics, Tokyo, Japan) was used to examine biochemical parameters, such as total protein, total bilirubin, glucose, lactate dehydrogenase (LDH) and adenosine deaminase (ADA). Hematological tests and flow cytometry analysis were performed on the EDTA whole blood, using a XT-1800i hematology analyzer (Sysmex Corporation, Kobe, Japan) and FACSCalibur Flow Cytometer (BD Biosciences, San Diego, USA), respectively.
Patient characteristics were summarized using means and standard deviations (or median and interquartile range) for continuous variables and counts/percentages for categorical variables. Categorical variables were compared using the χ2 test or Fisher exact test, and continuous variables compared using the Mann-Whitney U test or t test depending on the distribution of the data. Beyond descriptive statistics, associations between TE and clinical characteristics were analyzed by calculating the odds ratios (OR) and 95% confidence interval (CI), adjusted by age. Univariate logistic regression analysis was used to evaluate risk factors for TE, and significant variables (P < 0.1) were included in the multivariate logistic regression model. Prior to conducting multivariate logistic regression, continuous variables were transformed into categorical variables based on the cutoff points determined with receiver operating characteristic curve (ROC) analysis. Multivariate logistic regression analysis was then performed, including significant variables estimated in the univariate analysis or variables suggested in the literature. The Hosmer-Lemeshow goodness-of-fit test was performed to assess the overall fit of the model. A two-sided P value < 0.05 was considered significant for all analyses. Data analysis was carried out using SPSS 16.0 (IBM Corp., Armonk, United States). Table 1 shows the demographic characteristics of the 474 participants of this study. All 474 pleural TB patients (Additional File 1) enrolled were culture-confirmed, and 42 of them (8.9%) were diagnosed as TE because of an accumulation of pus in the pleural space. The mean age was 35.7 ± 18.1 years old, males comprised 79.1% of the participants. Forty-seven patients (9.9%) were MDR-TB. Forty-nine (10.3%) patients had contact of TB history. Smokers constituted 51.3% of the participants. The delay period of treatment was 94.3 ± 237.6 days. Among them, 344 (72.6%) participants had pulmonary TB, 70 (14.8%) had EPTB (excluding pleural TB), 29 (6.1%) had retreatment TB, 26 (5.5%) had diabetes mellitus, 4 (0.8%) had milliary TB. Cough (71.1%) was the most commonly reported symptom. Fever (70.7%), dyspnea (51.5%), chest pain (41.4%), sputum production (37.6%), and fatigue (8.2%) were the other symptoms. Chest X-ray revealed that 56.8% of the effusions were on the right, 33.8% on the left and 9.5% on the both.

Characteristics of patients.
The results of PE biochemical tests, flow cytometry analysis and hematological tests were also summarized in Table 2 and Supplementary Table 1. For comparisons between groups in terms of continuous variables, Mann-whitney U tests showed that differences in glucose (P < 0.001), LDH (P < 0.05), ADA (P < 0.05), red blood cell count (P < 0.01), hemoglobin (P < 0.01), hematocrit (P < 0.01) and erythrocyte sedimentation rate (P < 0.05) were significant when comparing the two groups. The t tests showed that age (P < 0.05) and white blood cell count (WBC, P < 0.05) were significantly different between the two groups. The other continuous variables did not reach significance (all P > 0.05). For dichotomous variables, chi-square analysis of these data showed that the two groups were significantly different in terms of pulmonary TB (P < 0.001), dyspnea (P < 0.05) and effusion site (right, P < 0.05). The other categorical variables did not reach significance (all P > 0.05).

Discussion
Currently, the pathogenesis of empyema is not well understood 18 . TE, is less common and represents an infection of the pleural space by M.TB that provokes the accumulation of purulent pleural fluid 19 . TE usually occurs in younger and middle-aged patients 20 . However, the accurate incidence has not been properly evaluated recently. In a previous study, TE has been reported that accounts for 3%-6% of all cases of empyema in South Korea 21 . In this study, we conducted a retrospective cohort study to evaluate the risk factors of TE in pleural TB patients. To our best knowledge, this is the first report detailing risk factors for the incidence of TE among pleural TB patients in China.
Our results indicated that pulmonary TB is a protective factor for TE. TE usually develops due to chronic pulmonary TB 22 , and rarely as a result of untreated tuberculous pleural effusion (TPE) 23 . Prior to the diagnosis of TE, patients often have suffered from chronic pulmonary TB for more than 10 years 6 . This is due to the fact that most of TE patients are asymptomatic 24 . Nevertheless, sputum culture positive for M.TB is also considered as a gold standard criteria for TPE diagnosis 25 . Therefore, the TE secondary to pulmonary TB would be more easily detected, because of the additional specimen (sputum) examined. Then, anti-TB drugs would be given timely, in other words, this may have a positive effect on the prognosis of TE patients. In addition, Ornstein GG et al. suggested that, when the underlying pulmonary TB is under control, the prognosis of TE is good; nevertheless, when the associated pulmonary TB not under control, the prognosis is not good 26 . This also indicates that an increased association between pulmonary TB and TE outcomes is plausible.
In the study, we found that male sex was a risk factor for TE. A possible explanation for it is that sex imbalance exists in TPEs 27,28 . Das DK et al. analyzed the sex distribution of TPE patients included in the previous published literatures, and most of studies (3/20) showed that female patients constituted the majority of TPEs (>50%) 28 . In the United States, from 1993 through 2003, 7,549 cases of pleural TB were reported, males comprised 67% and females 33% of the patient population. A similar result was reported in a retrospective study of 254 pleural TB patients 29 .
High levels of pleural ADA and peripheral WBC count predict increased TE risk. The findings suggest that the two biomarkers may reveal the progression of TE. Several reports have discovered the diagnostic role of the two in patients with empyma. Li R et al. reported that PE ADA could be used as an alternative biomarker for early and quick discrimination of Gram-negative from Gram-positive bacterial infections of the pleural space, which is useful for the selection of antibiotics 30 . A retrospective study conducted by Porcel JM et al. concluded that a high level of ADA is a general characteristic of lymphocytic and neutrophilic TB effusions, and an extremely high ADA activity is usually considered in empyema or lymphoma 31 . For the treatment of empyema, the stage of its development is the main parameter to be considered. A variety of treatments, including appropriate antibiotic therapy, pleural drainage, decortications, thoracotomy and video-assisted thoracoscopic surgery, must be selected and combined together in the most appropriate way with optimal timing 32 . However, currently there is no biomarker available that would help to choose appropriate interventions 33 . Thus, careful consideration should be given to the indications of each therapy for critical patients. In the next study, we would examine whether the pleural ADA level and peripheral WBC count could help to identify stage II-III empyema thoracis.
The study was conducted in a large population, and the findings represent a valuable contribution to the field. However, several limitations in the study are worth noting. First, retrospective nature could not control patients' baseline characteristics, so the results might be subject to selection bias. Second, as you know, the stage of empyema is considered as a main factor for selection of appropriate treatment approach. If an association between empyema stages and these clinical-pathological characteristics is observed, it may aid to control the development of TE. However, we failed to investigate it, this was because of the lack of information on the empyema stage of each patient. Third, because of a single-center design, the findings may be limited which may not accurately reflect the general characteristics of TE in China. In addition, although several risk factors were identified, further analysis must be performed to validate our findings.

conclusions
Our study found that male sex is a significant risk factor for TE while the presence of pulmonary TB is protective, with high levels of pleural ADA and WBC count could aid in early diagnosis of TE. Given the high proportion of TE in pleural TB was reported, TE remains a serious threat to public health in China. Table 3. Age-adjusted OR for risk factors associated with TE in pleural TB. TE, tuberculous empyema; TB, tuberculosis; OR, odds ratio; CI, confidence interval; ADA, adenosine deaminase; WBC, white blood cell.