Establishment And Clinical Application Of Serum Proteome HPLC Fingerprint Of Rapid Detection Methods

The tumor markers (TM) we used commonly in clinical have low specificity and sensitivity, so people use serum proteomics technology to find tumor-specific serum protein markers to provide a new basis for the diagnosis or treatment of disease. Using this technology to screen serum TM of different cancer has remarkable achievements, so it has a great deal of research value and application prospect. This topic aimed to establish a high performance liquid chromatography (HPLC) fingerprint rapid detection method in the study of lung disease. After large-scale pre-processing and analysis of serum samples, we looked for the specific proteins in lung cancer patients compared with healthy people to find patients with suspected lung cancer rapidly, then diagnosed the lung cancer patients finally though further analysis of specific protein of suspected cases, clinical pathology and medical imaging detection analysis.We studied the separation conditions of HPLC and compared protein profiles with the different conditions of stationary phase and mobile phase. Finally we determined the separation conditions. The stationary phase was Lichrospher A85-3(300nm×4.6nm I.D.;7μm). The linear gradient elution conditions of mobile phase was from5%B to80%B during30min then to95%B during5min then continue95%B during10min. The composition of mobile phase was that phase A was water containing0.1%phosphoric acid and0.05%trifluoroacetic acid (TFA), phase B was acetonitrile containing0.05%phosphoric acid and0.05%TFA. The detection wavelength was214nm. The flow rate was1ml/min. The chromatographic separation temperature was22℃. The quantitative tube volume was20μl. Under this condition, all the protein peaks can be completely separated, protein profile contained more information, and the sensitivity of detection was great, not only the time was moderate but also the baseline was horizontal relatively.In addition the experiment also established the methanol precipitation method to remove of high abundance proteins (HAP) from human serum, this research has not been reported. The albumin was precipitated by mixing serum and methanol at ratio of1:2(v/v), the removal effects and repeatability were analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Statistical analysis showed that the average concentration of albumin decreased significantly from (47.65±0.35) g/L to (1.16±0.08) g/L after this precipitation method, the removal ratio of protein was over97%and the percentage of albumin in the total protein of serum was decreased from65.4%to49.3%. The significantly statistical differences were observed (P<0.001).The experiment established the HPLC fingerprint rapid detection method further. We determined the internal standard peak and screened12common peaks on the basis of healthy protein profiles. Using the relative retention time and relative peak area as parameter to analysis the precision, repeatability and stability of method, the relative standard deviation (RSD) were<0.29%and<3.53%respectively, so the method is scientific and applicable in clinical.Finally we found the lung cancer group has6different protein peaks compared with both the healthy group and benign lung disease group with a large number of samples. We analysis correlation between these six protein peaks with squamous cell carcinoma associated antigen (SCC-Ag), neuron-specific enolase (NSE), cytokerantin19fragment (Cyfra19) and carcino-embryonic antigen (CEA) these4TM used in lung cancer. The results showed that peak5has positive correlation with both SCC-Ag and NSE, peak8and peak9has negative correlation with Cyfral9respectively. Then ROC curve analysis proved that peak5can distinguish between patients with lung disease group and healthy group, its sensitivity and specificity were86%and98%respectively. Finally we use peak5, peak8and peak9to establish an analysis tree and a diagnostic model of lung cancer. Peak5as a parent node, if the relative peak area of peak5was≤2.22, then we judged as health people, otherwise we judged the one as patient with lung cancer or patient with benign lung disease. When the relative peak area of peak5was between2.23and3.96, or when it’s relative peak area was>3.96, but the relative peak area of peak8was≤0.26and the relative peak area of peak9was≤0.29, we judge people as the patient with lung cancer; the other cases were judged as patient with benign lungs disease. The diagnostic model was applied to diagnose,41cases were judged accurately in50patients with lung cancer,49cases were judged accurately in50healthy person,27cases were judged accurately in30patients with benign lung disease, the accuracy of model was90%(117/130), the sensitivity was82%(41/50), the specificity was98%(49/50), the positive predictive value was93%(41/44), the negative predictive value was86%(49/57).