Exploring Of Differential Diagnostic Methods For Primary IgA Nephropathy And Non-IgA Nephropathy With Biochemical Marker's Assistance

Objective:IgA nephropathy(IgAN) was the predominant disease in the glomerulonephritis in China.Mass spectrometry and bioinformatics provided efficacious platforms for the research of disease's diagnosis.We aimed at the biomarkers for differential diagnosis of IgAN and non-IgAN through statistics,bioinformatics and MALDI-TOF MS technologies.Method:1.The biochemical parameters'differences were analyzed by statistical and bioinformatics tools between IgAN and non-IgAN.We used retrospective cohort study to screening 649 cases of kidney disease in 2007 with including criteria and excluding criteria.IgAN and non-IgAN were separated by biopsy-proved pathological diagnosis.CHISS software was adopted for analysis of biochemical parameters and SVM-RFE was introduced for classification and feature selection of IgAN and non-IgAN.2.The manipulation of serum protein with MB-WCX and the regulation parameters of MALDI-TOF MS and analytic softwares.The subjects of study were the patients with biopsy-proved primary IgAN and non-IgAN and without treatments.The adjustion of part parameters of MS and ClinProt tools software, the treatment of origin patterns and the selection of algorithm methods were performed for the best classified pattern and the most different peaks.3.On the basis of second experiment,differential diagnostic pattern of IgAN and non-IgAN would be set up.Search for the different peaks and predict the proteins.Results:1.145 cases were picked out in the first experiment and there were significant differences in 16 parameters between IgAN and non-IgAN.AFP,CA153,ALT and CK had statistic differences when ALB＜30g/L and however,LDH and NSE had significant differences when ALB＞30g/L.The results of logistic regression analysis showed LDH,NSE and CA153 had significant differences in the differential diagnosis for IgAN.The results of Bioinformatics by SVM-RFE classification and feature selection showed the prediction accuracy of two groups was 73.76%and top 3 features were TP,ALB and LDH.2.Offset value was changed into 30%and range into 70%of laser attenuator.Detection range was regulated from 720 to 17061 m/z.The origin fingerprinting data were all treated with"subtract baseline"and"smooth"by Flexanalysis software for efficacious peaks.Adding Savitsky Golay Smoothing and changing minimum range to 1800 m/z in recalibration setting,using intensity in calculation setting and selecting SNN algorithm method of ClinProt tools software,we got best classification pattern and most different peptides.3.Differential diagnostic pattern of IgAN and non-IgAN was set up with serum peptide mass fingerprint and 9 different peaks were found by using ClinProt tools.With the prediction of Biotools software,peak 5338 and peak 2082 could be a fragment of mucin 4 isoform andα1-typeⅡcollagen isoform.Conclusion:The technologies of bioinformatics and mass spectrometry provided a promising strategy for the study of kidney diseases.LDH,ALB et al parameters had the important roles in the classification of IgAN and non-IgAN.Biotools software could be facility to predict proteins with linear fingerprinting while we need further research to prove it.