| ObjectiveIn order to clarify the saliva-specific micro-features at the level of the saliva proteome in spleen and kidney deficiency;use bioinformatics tools to filtrate and optimize Protein composition and.then set up the combination of disease and symptom diagnosis model on spleen and kidney deficiency.Afterwards,conduct a preliminary study with a modern scientific basis on traditional Chinese medicine in different syndromes of the same diseases,and different diseases with same syndrome from the level of proteome.At the same time,set up with saliva proteome non-invasive diagnostic techniques of Chinese medicinal characteristics,to promote the development of micro-differentiation study and modernization of traditional Chinese medicine diagnosis.We used the technology matrix-assisted laser desorption ionization time of flight mass spectrometry(Matrix Assisted Laser Desorption Ionization Time of Flight Mass Spectrometry,MALDI-TOF-MS),and selected patients with Westem diseases such as:gastric cancer(GC),peptic ulcer(PU),and chronic gastritis(CG).They were grouped according to tongue coating and the principle combination of disease and symptom of traditional Chinese medicine. Then study groups were formed with diseases and normal controls of the same age on saliva proteome with a combination of disease and symptom.Methods95 cases of saliva specimens were collected,including 20 normal controls,34 with gastric cancer,20 with peptic ulcers,and 21 with chronic gastritis.All the patients have to undergo a gastroscopy and /or histopathological diagnosis after surgery.By the combination of diseases and symptoms,all objects were divided into the following three main groups:(1) grouping by different diseases and different syndromes,(grouping by the key link syndromes,in accordance with the deficiency syndrome standards all cases were divided into:spleen deficiency,kidney deficiency, and other syndromes to be comparative studied with the normal group)(2) grouping by the same disease and different syndromes,(The gastric cancer,in accordance with deficiency syndrome standards,were divided into spleen deficiency,kidney deficiency,and other syndromes to be comparative studied with the normal control group)(3) grouping by Western diseases,(to compare the differences of the protein expression spectrum between the normal controls and the three kinds of diseases as well as between diseases separately ) and(4) grouping by different tongue coating. (compare the differences between normal thin white tongue coating and three groups of pathological,including pathological thin fur,thick fur,and peeling fur,with the protein expression spectrum.Using the WCX beads and AutoflexⅢMALDI-TOF mass spectrometry that provided by Bruker company in German to detect the relative content of protein in saliva of the three major groups) Set all the saliva samples with collected protein molecular weight range of 1000～10000 Da.Proteins to be obtained by the form of peptide mass fingerprinting(PMF).Then use the Bruker's data analysis system(including a matching standard peak and peak calibration software FlexAnaly- sis3.0 and statistical analysis software ClinProTools2.1) to analyze the same mass/ charge and different protein content in each combination;finding that the combination of protein content have significant differences in the mass/charge.Set up a class prediction model by using the protein peaks that have significant differences(P＜0.05),select the appropriate conditions,and then use the statistical analysis software ClinProTools 2.1 to analyze protein peaks that have significant differences;therefore able to get correct grouping of specific protein markers and the use of specific protein markers in combination to establish a combination of disease and symptom diagnosis model.Using the random sampling methods(randomly selected 80%samples to build model,while using the remaining 20%as the verification sample,and then run ten times),verify the validity of the model(an average of specificity,sensitivity,and accuracy).Use the software SPSS13.0 to analyze the comparison of the baseline value in each group,the age comparison using one-way ANOVA,all data with(?)±s expression,and the chi-square test apply for the gender composition.Take P＜0.05 as statistically significant difference.Results1.Grouping by different diseases and different syndromes:all the patients were divided into deficiency,kidney deficiency,and other syndromes of 3 groups together with the normal group.A total 79 cases of samples were collected for mass spectrometry analysis and Comparison,a total of 175 protein peaks were obtained,10 significant differences protein peaks had been found by using of genetic algorithm, and the mass/charge(m/z) are:744.22Da,3963.26Da,7911.59Da,1390.85Da, 6574.69Da,1167.54Da,1230.73Da,7846.3Da,2724.41Da,and 2223.89Da.The class prediction model was set up based on the 10 protein peaks;identification rate is 89.15%,with a 38.63%prediction ability.Back to test the clinical results,19 cases of normal group cases were all accurately detected,16 of the 19 cases of spleen deficiency patients were accurately detected,18 of the 21 cases of kidney deficiency were accurately detected,and 17 of the 20 cases of patients with other syndromes were accurately detected.The accuracy of this model was 88.61%(70/79).2. Grouping by the same diseases and different syndromes:gastric cancer was divided into deficiency,kidney deficiency,and other diseases.A normal group total of 48 cases of samples were collected for mass spectrometry analysis and comparison,a total of 161 protein peaks were found,with the adoption of genetic algorithm 10 significantly different protein peaks between that mass/charge(m/z) are:8840.72Da, 3442.42Da,4120.81Da,3492.63Da,6558.43Da,5583.29Da,3286.1Da,4419.3Da, 5527.57Da,and 3923Da.The class prediction model was set up by analyzing the differences in protein;identification rate was 80.97%,with a 33.47%prediction ability.Back to test the clinical results,17 of the 19 cases of the normal group were accurately detected,10 deficiency in patients with gastric cancer were all accurately detected,7 of the 11 cases of gastric cancer kidney patients were accurately detected, and 6 of the 8 other diseases group were accurately detected.The accuracy of this model was 83.33%(40/48).3.Grouping by Western diseases:①41 cases of samples of the normal control group and the gastric cancer group were collected for mass spectrometry analysis.The two groups were compared,and 74 of protein peaks were obtained with 14 statistical significant differences between the protein peaks(P＜0.05), and select mass/charge(m/z) are:1472.78Da,2936.49Da,6556.81Da and 7081.17Da with these four protein peaks for modeling.The 1472.78Da peak in patients with gastric cancer was significantly higher than the normal group.Through these four peaks identification,set up a class prediction model.The identification rate was 97.83%,with a 79.82%prediction ability.Back to test the clinical results,22 of the 23 cases of gastric cancer patients were accurately detected.Out of the 18 cases of normal group,all of 18 cases were identified as non-gastric cancer.The results show that the accuracy of this model was 97.56%(40/41),sensitivity was 95.65%(22/23) and specificity was 100%(18/18).②32 cases of samples from the normal control group and the chronic gastritis group were collected for mass spectrometry analysis and compared,74 protein peaks were obtained between the two groups,and five statistical different significant protein peaks(P＜0.05) were found;the mass/charge(m/z) are:5502.36Da,1441.75Da,and 3442.47Da.These three-correlation peaks,through the analysis of differences in protein peaks, established a class prediction model;with a recognition rate was 91.67%,and a 73.33%prediction ability.Back to test the clinical results,14 cases of patients with chronic gastritis were all accurately detected,and 15 of the 18 cases of normal controls were correctly detected.The accuracy of this model was 90.63%(29/32), with a sensitivity of 100%(14/14),and the specificity was 83.33%(15/18).③35 cases samples from the normal group and the peptic ulcer group were collected for mass spectrometry analysis and comparison,66 protein peaks were obtained between the two groups,10 of which have significant statistical differences(P＜0.05),the select mass/charge(m/z) are:2934.36Da,5502.38Da and 3472.94Da.With these three-correlation peaks,and through the analysis of different protein expression spectrum,set up a class prediction model.The identification rate was 88.40%,with a 80.35%prediction ability.Back to test the clinical results,14 of the 17 cases of the peptic ulcer group were accurately detected,and 17 of the 18 cases of the normal group were correctly detected.The accuracy of this model was 88.57%(31/35),the sensitivity was 82.35%(14/17) and specificity was 94.44%(17/18).④37 cases of samples of the chronic gastritis group and the gastric cancer group were collected for mass spectrometry analysis and comparison.A total 77 protein peaks were found,of which,there is one statistical significant difference between the protein peaks;the mass/charge is 6021.72Da.By analyzing the differences in protein,a class prediction model was set up,with an identification rate of 83.54%,and a 60.23%prediction ability.⑤the chronic gastritis group and the peptic ulcer group collected a total of 29 cases of samples for compared analysis,in two groups 79 protein peaks were obtained, using statistical analysis software ClinProTools to get the most appropriate distinguish model.The model included four quality-related peaks the mass/charge that are:3369.79Da,3472.13Da,2900.2Da and 3489.29Da.Through the analysis of these differences in protein expression spectrum,set up a class prediction model,with a recognition rate of 92.86%,and a 56.87%prediction ability.⑥42 cases of samples of the peptic ulcer group and the gastric cancer group were collected for mass spectrometry analysis and compared.Between the two groups,77protein peaks were obtained.With a different protein analysis and set up modele.4.Grouping by tongue coating:a total of 75 cases of samples of normal thin white tongue coating and with pathology were collected for analysis and comparison.A total of 187 protein peaks were obtained,four statistically significant different protein peaks were found between that(P＜0.05),and the mass/charge are:6447.39Da,2938.47Da,1472.34Da, and 1451.77Da.Through the analysis of different in protein spectrum,set up a class prediction model,with a identification rate of 85.31%,and a 39.91%prediction ability.Back to test the clinical results,18 of the 19 cases of the normal group were accurately detected,and 40 of the 56 cases of the pathological group were accurately detected.The accuracy of this model was 85.33%(64/75).ConclusionUsing MALDI-TOF-MS technology to study the saliva protein expression spectra for the detection of the normal group,the gastric cancer group,the peptic ulcer group,and the group of chronic gastritis patients;the preliminary study results of saliva proteins can be used in clinical diseases of the specific combination of diagnostic biomarkers,and explore the establishment of the relevant combination of disease and syndromes diagnosis model.The method,in comparison with the traditional disease biomarker detection methods,its feature are non-invasive,simple, rapid,sensitivity,and high specificity.The selected biomarkers make for rapid and accurate diagnosis of disease,and identification of syndromes,to determine prognosis have important clinical significance.Therefore,this study for the clinical disease diagnosis and differential diagnosis of early stage has opened up a new way,and micro-differentiation study of TCM research study has exploited new ideas and fields.
|
PDF Full Text Request