In clinical practice, it is very common and critical to distinguish TPE from MPE as the pathogenesis, treatment, and recovery of the two diseases are different. Thus early diagnosis is particularly important. In this study, comparing TPE with MPE, the patients of TPE were more likely to have a fever and the MPE was more bloody (p <0.01). It is commonly accepted by clinician that fever is a common symptom in patients with TPE. Other researchers proved that compared PE caused by malignant lymphoma, patients with TPE have a higher probability of fever (Fever >37.5 ℃, MPE=12%, TPE=48%, p <0.01) . Moreover, consistent with our results, other clinicians also observed that MPE is mostly bloody, which may be related to tumor invasion and destruction of capillaries leading to blood leakage.
The percentage of Rivalta test positive of TPE was higher than MPE in our results (p <0.01). It has reported that the positive rate of the Rivalta test is parallel to the amount of total protein in body cavity effusion . Some researchers showed that the protein level in TPE is higher than MPE (p <0.05) . Consistently, out results showed that the total protein (sTP) in TPE is higher than MPE (69.09±1.72 vs 64.01±0.67, p = 0.15) which could explain that the percentage of Rivalta test positive of TPE was higher than MPE.
In the study, the serum D-dimer were higher in the TPE group than the MPE group (2.69±1.19 vs 0.67±0.07, p = 0.37). It is proved by other researchers that the serum D-dimer level of TPE patients was higher than MPE patients. However, they found the difference of D-dimer in PE was more obvious. These show that D-dimer is a highly sensitive index in serum and PE, which helps to identify TPE and MPE.
We also observed the sLDH were higher in the MPE group than in the TPE group (221.54±31.01 vs 170.62±6.02, p = 0.12). It has been reported that sLDH level was positively correlated with lymphoma-associated malignant PE (L-MPE) (OR: 1.005, 95% confidence interval: 1.003-1.007, p <0.001). sLDH > 460U/L distinguishes L-MPE from TPE with a sensitivity of 76% and a specificity of 81%. Consistent with our results, other researchers observed that the sLDH in MPE is higher than in TPE( p = 0.08 ) . In multivariate logistic regression analysis, the ratio of sLDH: pleural fluid lymphocyte count (PELC) was positively correlated with MPE. The sensitivity and specificity of the ratio of sLDH: PELC were 0.63 (95% CI 0.51–0.73) and 0.85 (95% CI 0.68–0.94). So sLDH is an important indicator for distinguishing TPE from MPE.
High sensitivity C reactive protein (HsCRP) is widely used as a sensitive but non-specific marker of systemic inflammation [26, 27]. Increased sHsCRP levels have been reported in many lung diseases, including tumors and tuberculosis [28, 29]. In our study, the median levels of both pHsCRP and sHsCRP were both higher in the TPE group than in the MPE group (24.63±1.16 vs 8.07±0.87 and 50.30±2.67 vs 19.25±3.20, p <0.01). The AUC of pHsCRP and sHsCRP were 0.79 and 0.73. Consequently, HsCRP is an important reference indicator to differentiate TPE from MPE. A meta-analysis showed that the optimal critical value of pHsCPR was 21.9 mg/dL, which values above the critical value were classified as TPE and below the critical value were classified as MPE, the sensitivity was 0.91 (0.73 to 0.98), and specificity was 0.82 (0.7 to 0.9).
Although HsCRP is a valuable diagnostic indicator, the diagnosis efficiency is low, so the choice of multi-index joint analysis is conducive to improving the diagnosis efficiency and accuracy. Through logical analysis, we had selected six relevant indicators (pADA, pHsCRP, sD-dimer, sESR, sHsCRP, and sMONp), with the Logistic regression model, 3 variables of pADA, sMONp, and sHsCRP could better help distinguish patients with PE by tuberculosis from malignant tumor. The combined AUC of the three factors can reach 0.94 (95% CI: 0.91 ~ 0.97), higher than any single index, which has great significance for the clinical differentiation between TPE and MPE. Agreed with our data, there is a study analyzed 118 patients, including 84 patients with MPE (71.2%) and 34 patients with TPE (28.8%). They also found the pADA of TPE is higher than MPE (p <0.05) . Moreover, others have proved that the elevated levels of sHsCRP and pADA in PE were useful in distinguishing TPE from MPE. However, there is one study with a different result. After analyzing 17 patients with L-MPE and 216 patients with TPE. They found there was no statistically significant difference in sHsCRP and pADA levels between the two groups which could be related to number of MPE patients included in the study.
At present, to achieve better treatment efficacy in clinical practice, many researchers were interested in exploring the differentiation between TPE and MPE [32–34]. Some of them were focused on the inflammatory factors. It has been reported that the biomarkers of PE in 22 patients with MPE and 5 patients with TPE were compared. IL-1, IP-10, IL-13 and IFN-γ were significantly higher in TPE (p <0.05). The level of basic fibroblast growth factor in MPE was higher than that in TPE (p <0.05) . The highest AUC is IP-10 (AUC =0.95, 95% confidence interval, p <0.01), followed by IL-13 (AUC =0.86, 95% confidence interval, p <0.05) . However, though one of the indicators in this study has a high value of AUC, the detection is not a common clinical indicator, and the detection is complicated. And the sample size is small, the reliability is weak, and it is difficult to perform stratified analysis. Another study found that Fibronectin (FN) and cathepsin G (CTSG) in patients with MPE were significantly higher than in patients with TPE, while the leukotriene-a4 hydrolase (LTA4H) was lower than in patients with TPE. The AUC value was determined to be 0.285 for FN (95% CI: 0.174–0.396), 0.64 for LTA4H (95% CI: 0.518–0.762), 0.337 for CTSG (95% CI: 0.218–0.456), and 0.793 for a combination of these candidate markers (95% CI, 0.697–0.888). The AUC value is significantly lower than in our study.
In this study, our results has more significant advantages of high diagnostic accuracy (high AUC value, high sensitivity, and specificity) and large sample size which mean high data reliability. More favorably, pADA, sHsCRP and sMONp are all clinically common and easy-to-collect specimens, which are convenient and cheap to test, and will not increase the additional burden on patients. Because of the large sample size, hierarchical analysis can be performed and it is found that the diagnostic efficiency of pADA is different in various age groups, and as the age increases, the diagnostic efficiency of pADA gradually decreases. This phenomenon could be related to the percentage of tuberculosis decreased while the cancer diagnosis increased with the age growth. It suggests that the patients under 45 years old could choose the single indicator of pADA for diagnostic detection.
Of course, the gold standard for differentiating TPE and MPE in clinical practice still relies on pathological tissue biopsy, all cases in this study were examined by thoracoscopy, and pathological biopsy was completed in most cases, which ensured the accuracy of the diagnosis of the patients, but for some patients who refuse to accept the invasive examination, or whose constitution is difficult to bear invasive examination, the effective detection index of non-invasive provides a strong basis for timely diagnosis and accurate treatment. It is worth doing further research and exploration.