Research Of Differentiating Sarcoidosis From Atypical Tuberculosis

Objectives From nosazontology aspect, quantitative real-time PCR were used to detect Mycobacterium tuberculosis (MT) DNA from formalin-fixed and paraffin-embedded sections of biopsy samples based on our previous studies in order to evaluate the role of mycobacterial infection in the pathogenesis of sarcoidosis and the value of the quantitative real-time PCR in differentiating sarcoidosis from atypical tuberculosis. From the clinical aspect, scoring systems and discriminant functions in four groups of clinical, radiographic, clinical - radiographic and clinical– radiographic - histological manifestations were established to identify the two diseases. And we also compared the two methods and the four groups in order to find the best.Methods 1. Quantitative real-time PCR: 1). Genomes of MT DNA were counted by quantitative real-time PCR from formalin-fixed and paraffin-embedded sections of biopsy samples of lymph nodes and lung tissues which came from 76 patients in Shanghai Pneumology hospital from January 1998 to December 2003. Those patients included 31 sarcoidosis, 30 tuberculosis and 15 patients of other diseases ( as the normal samples). We took the 15 normal lung tissues of the unborn mice as the control group. 2). The univariateχ2 test was used to analyze the frequency of MT DNA in tuberculosis, sarcoidosis and the normal group. The differences of the MT DNA copies were analyzed by the Mann-Whitney U test. The kappa test were used to find the consistency between the copises and the real results in differentiating the two diseases. 2. Analysis was conducted retrospectively on patents who were initialed in Shanghai Pneumology Hospital and undertaken biopsy in this hospital from January 1998 to December 2004. 1). Univariate analysis: Choosing cases of 181 sputum negative TB and 117 SA pantients. The variables which were carried out multivariate statistical analysis were selected by univariate analysis of theχ2 test, combing with the clinical experiences and some literatures. 2). Establishing the scoring systems: The score of every variable were calculated using the coefficientβvalue dividing by the minimumβvalue which we got from logistic regression analysis in the four groups respectively. Then we can get sum score of different groups from adding up their point scores. Receiver operating characteristics (ROC) analysis helped to decide the optimal cutoff points for each scoring system based on their highest diagnostic accuracy. Then we can establish the four scoring systems. 3). Creating the discriminant functions: We carried out the canonical discriminant analysis on the final variables in four groups and established the discriminant functions. 4). Comparing the two methods and the four groups: the rates of cases correctly classified between tuberculosis and sarcoidosis in different combinations were tested for significance with partitions ofχ2 method to look for which one was the best. Then the McNemar test and the Kappa analysis was carried out on the two methods in order to find the different effects between scoring systems and discriminant functions.Results 1. Quantitative real-time PCR: Mycobacterial DNA was detected by PCR: 30 in 30 tuberculosis (100%), 6 in 31 sarcodosis (19.4%), 2 in 15 other diseases (13.3%) and 0 in the 15 fetus lung tissues of the mice. The frequency of tuberculosis samples with M. tuberculosis is higher than of sarcodosis (P=0.000) and normal samples (P=0.000) by theχ2 test. But that difference between sarcoidosis and normal samples has no statistical significance (P=0.928). By the Mann-Whitney U test, we find there are statistical significance in the differences of the abosulte and relative copies of the MT DNA between sarcoidosis and tuberculosis (P=0.000, P=0.000), between tuberculosis and normal samples (P=0.000, P=0.000). But there is no difference between sarcoidosis and normal samples (P=0.436, P=0.314). 2. Establishing the scoring systems: The scores in clinical, radiographic, clinical - radiographic and clinical– radiographic– histological groups have the statistical significances in distinguishing the two disases (P < 0.01) respectively. Using the ROC curve, the areas under the curve are all greater than 0.5 (the areas are 0.929, 0.987, 0.995, 0.999 respectively). 3. Establishing the discriminant functions: The rates of cases correctly classified in the four groups all have the statistical significances in distinguishing the two disases (the rates are 84.2%, 91.3%, 95.6%, 96.0% respectively). 4. Comparing the two methods and the four groups: There is the statistical significance between clinical and other groups not only in scoring system but also in discriminant function. But there are no statistical significances in other groups in distinguishing the two disases, and the rates are all smaller in clinical group than in other gruops. We found the scoring systems are better than the discriminant functions through the statistical analysis.Conclusions 1. The quantitative PCR study can't prove the infection of mycobacteria has specific relationship to the sarcoidosis. The quantitative PCR may be a reliable way to distinguish sarcodosis and proliferous tuberculosis. 2. The scoring systems and discriminant functions all have statistical significances in distinguishing the two disases. 3. The rates of cases correctly classified in clinical are lower than other groups not only in scoring system but also in discriminant analysis. We find the more variables the greater rates. 4. The scoring systems are better than the discriminant functions in each group.