| Objective:To investigate the relationship of contrast enhanced ultrasonography on hepatocellular carcinoma with metastasis and recurrence after microwave ablation,and screen serum differentially expressed proteins related to metastasis and recurrence of HCC after MWA.And to establish predicitive model combining characteristic parameters of CEUS and serum protein profiling for predicting metastasis and recurrence of HCC patients after MWA early and noninvasively.Materials and methods:(1)Prospective study was performed on 65 HCC patients underwent MWA.All the patients underwent CEUS examinations before MWA.The parameters of CEUS were quantitatively acquired by intensity-time curve of enhancement. Parameters of CEUS related to metastasis and recurrence of HCC were screened by using univariate and multivariate analysis.(2)Serum samples of 56 HCC patients were collected before and after MWA. The mass spectra of proteins were generated by using SELDI-TOF-MS, normalized and analyzed with Ciphergen ProteinChip Software.Intensity values for each peak were input into Biomarker Pattens Software for decision classification trees analysis and the best performing tree was chosen for testing. Then separate blinded samples were categorized for validation of the classification trees.Multivariate analysis was used to compare difference of serum protein abundance between pretherapy and posttherapy in recurrence group and non-recurrence group,to find characteristic parameters of serum protein profiling related to metastasis and recurrence of HCC patients after MWA.(3)Artificial neural network,combinng characteristic parameters of CEUS and serum protein profiling screened in the first two parts,was established to predict metastasis and recurrence of HCC patients after MWA.Results:(1)There was no significant difference in sex,age,HBV/HCV infected,AFP level before MWA and differentiation degree in recurrence group and non-recurrence group.(2)The univariate analysis showed that in the parameters of CEUS before treatment,p TIP,t-p TTP and t-p PI were significant different in recurrence group and non-recurrecne group.The multivariate analysis showed that t-p TTP and t-p PI were risk factors of recurrence.(3)According to serum protein profiling obtained from 40 HCC patients before treatment,7 significant differentially expressed proteins were detected in serum samples between recurrence group and non-recurrecne group.2 proteins with M/Z value of 5170 and 7688 were up regulated in recurrence group.While 5 proteins with M/Z value of 1103,6449,6646,6858 and 7787 were down regulated in recurrence group.Decision classification trees combined with 2 candidate proteins with M/Z value of 6858 and 1103 were produced by Biomarker Pattens Software with sensitivity of 96.30%and specificity of 76.92%.Then decision tree model were challenged by the double-blind test of 16 samlpes.As a result,a sensitivity of 50%and specificity of 75%were obtained.(4)According to serum protein profiling obtained from 40 HCC patients after treatment,26 significant differentially expressed proteins were detected in serum samples between recurrence group and non-recurrecne group.11 proteins were up regulated with M/Z value of 4426,6207,7161,9670,10028,12150,12222,24006, 28928,29021 and 29233 in recurrence group.While 15 proteins were down regulated with M/Z value of 3943,4100,4299,4657,5077,5348,6446,6644, 6855,7788,7992,8586,8708,8794 and 19966 in recurrence group.Decision classification trees combined with 2 candidate proteins with M/Z value of 4100 and 4657 were produced by Biomarker Pattens Software with sensitivity of 92.59%and specificity of 84.62%.Then decision tree model were challenged by the double-blind test of 16 samlpes.As a result,a sensitivity of 62.5%and specificity of 75%were obtained.(5)Logistic regression showed that 6 proteins with M/Z value of 28700,20628, 10868,2140,17300 and 2067 entered the regression equation stepwise,and concordance rate of the regression equation is 95.8%.The difference of serum protein abundance with M/Z value of 2140,17300 between pretherapy and posttherapy were protective factors,while difference of serum protein abundance with M/Z value of 28700,20628,10868 and 2067 between pretherapy and posttherapy were risk factors.(6)Artificial neural network combinng characteristic parameters of CEUS and serum protein profiling was in concordance with actual data,and the accuracy of training set was 100%.The predictive model could predict metastasis and recurrence with the accuracy of 83.3%in predictive set.Conclusions:(1)In the parameters of CEUS on HCC,t-p TTP and t-p PI are risk factors of recurrence,and are hopeful to be predictive index of HCC recurrence after MWA.(2)Serum differentially expressed peoteins screened by CM10 chip before and after treatment might be related with recurrence of HCC after MWA.The decision classification tree is valuable in predicting the prognosis of HCC patients after MWA.(3)Logistic regression derived of difference of serum protein abundance with M/Z value of 28700,20628,10868,2140,17300 and 2067 could reflect prognosis of metastasis and recurrence of HCC patients after MWA.(4)Artificial neural network combinng characteristic parameters of CEUS and serum protein profiling could predict metastasis and recurrence of HCC patients after MWA.
|
PDF Full Text Request