Encouraging results with a specificity of 85.7% and a sensitivity of 83.3% did indicate that the B-Raf assay model can effectively discriminate active TB from CRD and HC (Table 3). The demonstration elsewhere suggested that this classification tree model could be a potential diagnostic tool for active TB. Similar research of TB has been performed in the recent years, which set up a diagnostic model containing 20 peaks that can distinguish

TB from other inflammatory diseases and healthy controls [5]. However, the model we established is based on recruitment of several pulmonary diseases with clinical manifestations or laboratory indices that can overlap those of active TB. Apparently, the latter one is more appropriate for clinical utility, but a second dataset, which is prospectively obtained

from patients with respiratory symptoms as Agranoff et al. did [5] should be used to further confirm the model’s specificity and sensitivity for diagnosing Opaganib datasheet active TB. Although we tried our best to rule out patients with latent TB from the non-TB group, some patients or healthy controls with latent TB might still be recruited. As no similar research has been performed between latent TB and active TB, we cannot decide whether latent TB affects the performance of the model or not and this should be further explored. Also, HIV/TB, multidrug TB and ETB restrain the management of TB so strongly that related classification tree models should be set up. Some studies reported that different biomarkers might exist in diverse situations of sputum smear microscopy of patients with

TB [27], while others considered it results from the bias of quality control. To investigate this interesting phenomenon, comparison among peaks of SPP-TB, SNP-TB and non-TB has been performed. There were 54 proteins that can discriminate these three groups (Table 4). Forty of the 54 proteins also showed up in the differential expressed proteins between active TB and non-TB, which suggested that these proteins not only play an important role in the pathogenesis of active TB but also regulate the status of active TB. Surprisingly, both 8561 and 8608 m/z showed up in this Florfenicol analysis, which further highlighted that the importance of these two peaks and further identification of them are needed. Comparing to the prior study that only recruited 10 patients with pneumonia and three patients with COPD in the non-TB group [28], none of their differential expressed peak was found in our research. Inherent complexity of active TB, technological difference between magnetic beads and protein chips and different composition of non-TB group might result in this inconsistent condition. As we know, identification of meaningful peaks is necessary for understanding the pathogenesis of TB. Furthermore, Agranoff et al. [5] identified two of their differently expressed peaks to be serum amyloid A protein and transthyretin.