Objective   To compare the serum differential proteins of Han and Tibetan patients with pulmonary tuberculosis

and those of their corresponding gender- and age-matched non-tuberculosis controls with surface enhanced laser desorption/ionization time-of-flight mass spectrometry (SELDI-TOF MS) in order to establish a diagnositic model for tuberculosis differential proteins in these 2 nationalities.  Methods   A total of 87 sera samples from patients with pulmonary tuberculosis

67 matched non-tuberculosis controls from Han and Tibetan people

54 health individuals

10 acquaired pneumonia and 3 chronic obstructive pulmonary disease were analyzed with SELDI-TOF MS. A training set of spectra derived from analysis of sera from 53 patients with pulmonary tuberculosis (Han 23

Tibetan 30) and 42 non-tuberculosis controls (Han 20

Tibetan 22) were analyzed with a machine learning algorithm called as decision tree boosting. The discovered decision trees were then used to identify an independent testing set of 59 masked samples: 34 pulmonary tuberculosis (Han people 15

Tibetan 19)

and 25 controls (Han people 16

Tibetan 9).  Results  The decision tree model of m/z 3193.61 and 4592.11 detected 20 of 23 Han people pulmonary tuberculosis (sensitivity 86.59%) and 17 of 20 Han people controls (specificity 85.00%)

and the model of m/z 4821.45 detected 28 of 30 Tibetan pulmonary tuberculosis (sensitivity 93.33%) and 19 of 22 Tibetan controls (specificity 86.36%) in the training set. However

the decision tree model of m/z 4091.98

3398.27

7970.44

4965.51 and 7970.44 gave an accuracy of 92.45% for identifying Tibetan patients from Han people patients. Validation of independent blinded testing set detected 70.96%

71.42%

and 67.92% accuracy

respectively. There were 16 differential protein/peptide peaks between Han and Tibetan patients with pulmonary tuberculosis.  Conclusion  The above decision model suggests a potential application of SELDI proteomic pattern analysis as a rapid and accurate method to diagnose individuals with pulmonary tuberculosis in both Han people and Tibetan

and there are proteomic differential between pulmonary tuberculosis patients from the 2 nationalities.