Clinical Value And Quantitative Imaging Evaluation Of Different Intratumoral Fiber Components In Pancreatic Ductal Adenocarcinoma

Part I The value of different intratumoral fiber components in prognosis and neoadjuvant chemotherapy efficacy of pancreatic ductal adenocarcinoma patientsObjective:To investigate the value of intratumoral collagen fiber(CF),reticular fiber(RF)and elastic fiber(EF)in prognosis and neoadjuvant chemotherapy efficacy of pancreatic ductal adenocarcinoma(PDAC)patients.Method:228 patients with surgically and pathologically confirmed PDAC at 3 centers from February 2014 to April 2021 were retrospectively collected,including 186 patients with resectable tumor and 42 patients with post-NAC tumor resection.Patients with post-NAC tumor resection were divided into remission group(n=13)and non-remission group(n=29)according to the classification criteria of tumor pathological regression recommended by the College of American Pathologists.Clinical and CT indicators of all patients were obtained,clinical indicators including:gender,age,body mass index(BMI),diabetes,hypertension,smoking,drinkink,preoperative lymphocyte ratio,monocyte ratio,neutrophil ratio,albumin,globulin,albumin/globulin ratio,CA19-9,CA12-5 and CEA levels,surgical method,surgical time,intraoperative blood transfusion or not,patients with NAC obtained additional indicators,including preoperative pre-NAC program,pre-NAC CA19-9,CA12-5 and CEA levels,and calculated change rate(%Δ).The pathology evaluation indexes included the degree of tumor differentiation,the presence or absence of nerve and microvascular invasion,T stage,N stage.CF,RF and EF in tumor tissues of all patients were stained by Masson,Gordon-Sweets and EVG staining.Then Image-Pro Plus 6.0 software was used to quantitatively analyze the ratios(CFR,RFR and EFR)of the above fiber components.CT evaluated indexes included maximum diameter,location and shape of the tumor,whether there is cystic,necrosis and calcification within the tumor,whether there is peritumoral retention cyst,infiltration,invasion of fat,blood vessels near the tumor,distally main pancreatic duct dilatation,intrahepatic and intrahepatic bile duct dilatation,whether there is atrophy of adjacent pancreatic parenchyma,suspected lymph node metastasis and ascites.At the same time,CT values of tumors were measured on 1mm reconstructed images from arterial phase(AP),portal venous phase(PVP)and delayed phase(DP).Multiple linear stepwise regression was used to analyze the clinical and CT indicators affecting CF,RF and EF in resectable PD AC.Independent sample t test,Mann-Whitney U test or Chi-square test were used to analyze the differences of clinical and CT indicators between the group in remission and non-remission post-NAC.ROC curve was used to evaluate the predictive efficacy of the above different indicators on the efficacy of NAC.Univariate and multivariate Cox regression analysis was used to identify independent factors associated with overall survival(OS)and disease-free survival(DFS)of patients with resectable and post-NAC.Results:1.Among 186 patients with resectable PDAC,multiple linear stepwise regression results showed that drinking and DP CT value had a significant positive influence on CFR(t=2.110,P=0.037,t=2.956,P=0.004),and there was no multicollinearity between them(VIF=1.004),preoperative monocyte ratio had a significant positive effect on RFR(t=2.477,P=0.015),preoperative CEA level had a significant positive effect on EFR(t=3.427,P=0.001).The results of multivariate COX regression analysis showed that preoperative CA19-9 level(P=0.004),atrophy of adjacent pancreatic parenchyma(P=0.008)and CFR(P=0.002)were independent influencing factors for OS in patients with resectable PDAC,among which preoperative CA19-9 level(HR=1.001)was risk factor,atrophy of adjacent pancreatic parenchyma(HR=0.559)and CFR(HR=0.981)were protective factors.CFR(P=0.001)was an independent and protective factor(HR=0.980)for DFS.2.Among the 42 patients with tumor resection post-NAC,the difference analysis of clinical and CT indexes showed that the absolute value of%Δ CA19-9(69.21%± 49.97)in the remission group was higher than that in the non-remission group(13.95%±74.67),and the difference was statistically significant(P=0.003).The PDAC in the head of pancreas in remission group(n=5)was less than that in non-remission group(n=21),and the difference was statistically significant(P=0.036).The CFR in remission group(77.78%± 12.22)was higher than that in non-remission group(53.00%± 14.83),and the difference was statistically significant(P<0.001).ROC curve was used to evaluate the predictive value of the above indexes for NAC efficacy.The results showed that the area under the curve(AUC)of%ΔCA19-9 was 0.790,95%CI:0.628～0.953,the AUC of tumor location was 0.670,95%CI:0.509～0.830,and the AUC of CFR was 0.902,95%CI:0.793～1.000,the accuracy of CFR was the highest.Multivariate COX regression analysis showed that%Δ CA19-9(P=0.033)and CFR(P=0.001)were independent influencing factors for OS in patients with PDAC after NAC,among which,%Δ CA199(HR=1.907)was risk factor and CFR(HR=0.952)was protective factor.N stage(P=0.032)and CFR(P=0.015)were independent influencing factors for DFS,among which,N stage(HR=3.036)was risk factor and CFR(HR=0.969)was protective factor.Conclusions:CF is a favorable prognostic indicator for resectable and NAC PDAC patients,as well as the best predictor of NAC efficacy in patients with PD AC,which provides useful information for preoperative risk stratification and individualized precision treatment for patients with this disease.Part II:The value of CECT radiomics in predicting CF in resectable PDACObjective:To investigate the predictive value of the radiomics model based on three-phase CECT for collagen fiber(CF)in resectable pancreatic ductal adenocarcinoma(PDAC).Method:A total of 186 patients with pathologically confirmed PDAC after surgical resection in 3 centers from February 2014 to April 2021 were retrospectively collected,clinical and CT data were obtained.Among them,a total of 139 patients in center 1 and center 2 were randomly divided into training set(n=97)and internal validation set(n=42)according to a ratio of 7:3,and patients in center 3 were used as external validation set(n=47).CF staining and CFR calculation in tumor are the same as in the part I.Restricted cubic spline(RCS)analysis of overall survival(OS)of all PDAC patients was used to determine the optimal cutoff value for CFR,and all patients in the training set,internal validation set,and external validation set were divided into low and high CFR groups.Univariate and multivariate logistic regression was used to analyze the clinical and CT indicators affecting different CFR.3D Slicer software was used to delineate the full-level areas of interest for PDAC lesions on 1mm reconstructed images of arterial phase(AP),portal venous phase(PVP)and delayed phase(DP)respectively,extract the radiomics features,and incorporate the three phase radiomics features into the combined phase(CP).The variance threshold method and selectKbest method were used to screen the radiomics features.The least absolute shrinkage and selection operator(Lasso)was used to construct the radomics score of each phase.Binary logistic regression was used to construct AP,PVP,DP and CP radiomics models.The area under the curve(AUC)was used to evaluate the predictive performance of different models.The DeLong test was used to compare the predictive performance of different models.The consistency between the model predictions and the actual results was evaluated by the calibration curve,and the clinical benefit of the models was evaluated by the decision curve analysis(DCA).Results:1.The optimal cut-off value of CFR in PD AC determined by RCS method was 57.76%,94 patients in the low CFR group(CFR≤57.76%)and 92 patients in the high CFR group(CFR>57.76%).In the training set,49 patients in the low CFR group and 48 patients in the high CFR group,in the internal validation set,21 patients in the low CFR group and 21 patients in the high CFR group;in the external validation set,24 patients in the low CFR group and 23 patients in the high CFR group.2.There were no significant differences between CFR and clinical and CT indexes(P>0.050).After dimension reduction by Lasso,6 AP,5 PVP,11 DP and 16 CP radomics features were screened to construct radomics scores.The AUC of the final AP radiomics model in the training set,internal validation set and external validation set were 0.704(95%CI:0.621～0.652),0.657(95%CI:0.520～0.765)and 0.614(95%CI:0.449～0.780),respectively,the AUC of PVP radiomics model in the training set,internal validation set and external validation set were 0.710(95%CI:0.612～0.795),0.567(95%CI:0.419～0.657)and 0.592(95%CI:0.427～0.758),respectively,the AUC of DP radiomics model in training set,internal validation set and external validation set were 0.854(95%CI:0.801～0.921),0.800(95%CI:0.651～0.860)and 0.690(95%CI:0.580～0.799),respectively,the AUC of CP radiomics model in the training set,internal validation set and external validation set were 0.879(95%CI:0.817～0.934),0.821(95%CI:0.695～0.905)and 0.739(95%CI:0.607～0.825),respectively.Delong test showed that the predictive efficiency of CP model in the training set was significantly better than that of AP(P=0.007)and PVP(P=0.008),but there was no statistical difference between CP model and DP model(P=0.619),internal validation set showed that the prediction efficiency of CP model was significantly better than that of AP(P=0.002)and PVP(P=0.006),but there was no statistical difference between CP model and DP model(P=0.779),external validation set showed that the prediction efficiency of CP model was significantly better than that of AP(P=0.019),PVP(P=0.008)and DP(P=0.028),which indicated that CP radiomics model had the best prediction value.The calibration curve shows that the prediction results of CP radiomics model are in good agreement with the actual situation.DCA showed that the CP radiomics model provided a greater net benefit than other models in most of the threshold probability ranges.Conclusions:The radiomics model based on CP combined with three-phase CECT has the best predictive performance for different CFR in PDAC,providing more accurate reference information for preoperative diagnosis and treatment of PD AC,so as to improve the diagnosis and treatment level of PD AC.Part III:The value of quantitative parameters of functional MRI in evaluating CF in resectable PD ACObjective:To investigate the value of quantitative parameters of functional magnetic resonance imaging(fMRI)in evaluating collagen fiber(CF)in resectable pancreatic ductal adenocarcinoma(PDAC).Method:55 patients with pathologically confirmed PD AC after surgical resection in center 1 from November 2020 to December 2022 were prospectively collected,specimens were taken from the margin and/or center of the resected lesion,a total of 99 specimens were collected,the CF staining and CFR calculation of the specimen were the same as in the part I.All patients underwent routine MRI and fMRI scans before surgery,the latter included:native transversal relaxationtime quantitative imaging(T1 mapping),transverse relaxation time quantitative imaging(T2 mapping,T2*mapping),intravoxel incoherent motion diffusionweighted imaging(IVIM-DWI),diffusion kurtoid diffusion-weighted imaging(DKI-DWI),dynamic contrast enhancement(DCE)imaging and enhanced T1 mapping imaging.Two radiologists with 6 and 8 years of experience in image diagnosis used Siemens workstation and MITK software to outline region of interest in the above fMRI images that were roughly consistent with the specimens taken during surgery for three times,generate corresponding functional images,and obtain corresponding parameter data,including:native T1 mapping values,T2 mapping values,T2*mapping values,IVIM-DWI related parameters(f values,D values,and D*),DKI-DWI related parameters(MD values and MK values),DCE related parameters(Ktrans values,Kep values,and Ve values),and extracellular volume(ECV)value based on native and enhanced T1 mapping value,at the same time,fMRI parameters of surrounding normal pancreatic(NP)tissue were measured as controls.The consistency of fMRI parameter values measured by two physicians was tested by intraclass correlation coefficient(ICC).Independent sample t test or Mann-Whitney U test were used to compare the differences of fMRI parameter values between PD AC and NP.Pearson correlation test was used to analyze the correlation between fMRI parameters and CFR in PD AC.Results:1.In PDAC,the ICC values of the native T1 mapping value,T2 mapping value,T2*mapping value,ECV value,f value,D value,MD value,Ktrans value,Ve value,and enhanced T1 mapping value are 0.821,0.901,0.865,0.847,0.915,0.811,0.772,0.846,0.845,and 0.836,respectively,which are greater than 0.750,indicating that the results measured by the two physicians have good consistency.The ICC values of the D*value,MK,and Kep value are 0.689,0.711,and 0.726,which are greater than 0.400,indicating that the results are generally consistent.In NP,the ICC values of native T1 mapping,T2 mapping,T2*mapping,ECV,f value,D value,D*value,MK value,MD value,Ktrans value,Kep value,Ve value,and enhanced T1 mapping value are 0.794,0.801,0.789,0.839,0.835,0.907,0.846,0.841,0.831,0.839,0.867,0.840,and 0.865,all greater than 0.750,indicating that the results measured by the two physicians have good consistency.2.In PDAC,the native T1 mapping value(666.1±134.70 vs 458.85±52.82)and ECV value(0.57 ± 0.34 vs 0.30 ± 0.19)were significantly higher than those of NP(P<0.001,P<0.001);the T2 mapping value(68.80±12.60 vs 86.60±38.85),T2*mapping value(32.15 ± 9.33 vs 46.03 ± 24.02),f value(2.49 ± 6.10 vs 11.94±12.13),MD value(1.42±0.68 vs 8.45±2.85),Ktrans value(0.19±0.11 vs 0.52 ± 0.22),and enhanced T1 mapping value(314.40 ±81.90 vs 337.00 ±54.13)in PDAC were significantly lower than those in NP(P<0.001,P<0.001,P<0.001,P<0.001,P<0.001,P=0.009);However,there was no significant difference between PD AC and NP in the values of D value,D*value,MK value,Kep value,and Ve value(P=0.166,P=0.251,P=0.827,P=0.696,P=0.105).3.In PDAC,there is a pole-strong positive correlation between CFR and ECV(r=0.851,P<0.001)and initial T1 mapping(r=0.805,P<0.001).there was a strong negative correlation with MD value(r=-0.742,P<0.001)and D value(r=-0.679,P<0.001),there was a moderate negative correlation with f value(r=-0.516,P<0.001)and T2,mapping value(r=-0.447,P<0.001),there was a weak negative correlation with Ktrans value(r=-0.397,P<0.001)and enhanced T1 mapping value(r=-0.213,P=0.034),there was no correlation with D*value(r=0.060,P=0.555),MK value(r=-0.147,P=0.145),Kep value(r=-0.122,P=0.230),T2 mapping value(r=0.111,P=0.273)and Ve value(r=-0.111,P=0.273,P=0.274).Conclusions:Among all fMR1 parameters,ECV has the highest correlation with CFR in PDAC,showing great value in the evaluation of CF in PDAC,which lays a foundation for preoperative prognostic judgment and treatment selection of patients with PD AC.