Construction Of A Prognostic Prediction Model For Patients With Pancreatic Ductal Adenocarcinoma Based On Stromal Microenvironment And Exploration Of Molecular Mechanism

BackgroundPancreatic cancer,characterized by desmoplasia,is well known for its aggressiveness and poor prognosis.In recent years,despite advances in various aspects of researches,survival of pancreatic cancer has not been improved significantly,which is closely related with the specific characteristic for richness of stroma.The limited understanding of underlying biological mechanism of stroma has further restricted the development of diagnosis and treatment in pancreatic cancer.The tumor microenvironment of pancreatic cancer is been payed more attention by researchers recently.The tumor cells are surrounded by dense desmoplasia tissue,which is an essential histological feature of PDAC.The main components of the stroma include extracellular matrix(ECM),immune cells,endothelial cells and cancer associated fibroblast(CAF).Infurther,the co spatial distribution mplicated interaction between cells of the microenvironment and tumor cells,which affects the biological behavior of tumor,is closely related to the prognosis of the patients.Therefore,it is very important to understand the biological behavior of PDAC with the knowledge of the biological features of tumor microenvironment.At present,there are disconsensus and even contradictory conclusions in the researches related with morphology of tumor microenvironment,which contributes to the heterogeneity of tumor and the discordance of evaluation methods.Therefore,our center apply Leeds pathology protocol to evaluate pancreatic neoplasms.On this basis,the database of digitalized whole-mount slide images(DWMSIs)is constructed and the tumor stroma ratio(TSR)can be quickly and conveniently evaluated.Therefore,TSR can indirectly reflect the relationship between tumor cells and tumor microenvironment,and then evaluate the value of TSR in predicting prognosis,so as to guide individualized and presion therapy.Moreover,CAFs are the main part of cells of tumor microenvironment.Studies related with RNA sequencing of single cells have confirmed the intratumoral and intertumoral heterogeneity of CAFs,and the functional classification of CAFs is also widely accepted.And,CAFs are mainly divided into myo CAFs and i CAFs.The representative markers of i CAFs are IL-6,CXCL12(SDF-1)and PDGFR-β,and those of myo CAFs are α-SMA、periostin and MMP-11.While FAP-α was the co-expression marker of the two kinds of CAFs.The heterogeneity for spatial distribution of CAFs is the hot topic recently and the related researches are still in the initial stage.Therefore,we try to evaluate the correlation between the spatial distribution of CAFs and prognosis by tissue microarray,and then screen out the spatial distribution characteristics of CAFs that can be used to predict prognosis.In order to clarify the molecular mechanism of CAFs’ function,we verified the feasibility of primary culture of CAFs in vitro,and confirmed the functional differences of two subgroups of CAFs by transcriptome sequencing.In conclusion,the purpose of this study is to clarify the role of tumor microenvironment in the progression of pancreatic cancer and its value as a stratified indicator for predicting prognosis.And in further we clarify the role of spatial distribution of CAFs in PDAC and investigate the molecular mechanism of function of CAFs by primary culture of CAFs in vitro.Part Ⅰ Establishment of the database for digitalized whole-mount slide images and the prognostic prediction of tumor stroma ratio in patients with PDACObjective Firstly,to construct the database of digitalized whole-mount slide images(DWMSIs)and clarify the relationship of tumor stroma ratio(TSR)and the prognosis in patients with PDAC.Secondly,to investigate the correlation of TSR evaluation based on DWMSIs and partial-mount images.Thirdly,to investigate the correlation of TSR evaluation by computor-aided method and pathologists.Methods1.A total of 440 consecutive patients with a final histopathological diagnosis of PDAC who underwent primary pancreatic resection at the Department of Hepatobiliary Pancreatic Surgery in Changhai Hospital(Shanghai,China)were enrolled for this study.TSR was evaluated by computor-aided method and pathologists based on the DWMSIs.And the core area,transitional area and peripheral were also defined.TSR >1 denoted low stromal content,whereas TSR ≤1 indicated high stromal content.2.Clinicopathological data were retrospectively collected by using clinical research database.3.Agreement between two different methods for TSR evaluation was measured using weighted Cohen kappa coefficient(κ).Analyses were performed using SPSS version 25.0.For all analyses,a two-tailed P<0.05 was considered statistically significant.Results1.The database of digitalized whole-mount slide images(DWMSIs)was firstly contructed in a high volume ceter of China mainland.Of the 440 consecutive patients in our study,the developing cohort comprised 207 patients,whereas the validation cohort consisted of 193 patients.Relevant baseline variables were similarly distributed in the developing and validation cohorts.2.In logistic regression analyses,two independent variables associated with low stromal content were identified in the developing cohort—namely,intratumoral necrosis(OR,3.530;95% confidence interval [CI],1.953–6.379;P<0.001)and R1(OR,2.281;95% CI,1.219–4.265;P=0.01).Both variables were validated in the validation cohort(intratumoral necrosis:OR,3.890;95% CI,2.097–7.217;P<0.001 and R1: OR,2.034;95% CI,1.059–3.910;P=0.033).3.Multivariate COX analysis confirmed that TNM stage(II vs.I: HR,2.584;95% CI,1.386–4.819;P=0.003;III vs.I: HR,4.384;95% CI,2.285–8.411;P<0.001),stromal content(low vs.high:HR,1.876;95% CI,1.227–2.870;P=0.004),tumor grade(G3 vs.G1/2: HR,2.124;95% CI,1.419–3.179;P<0.001),and perineural invasion(with vs.without: HR,2.147;95% CI,1.187–3.883;P=0.011)were independent prognostic factors in the developing cohort(Table 5).The abovementioned independent prognostic factors were also validated in the validation cohort.Moreover,stromal content categories could classify patients into subgroups and that high stromal content could predict good prognosis within TNM stages I,II,and III,which were also validated in the validation cohort.4.Of the 41 consecutive patients,the weighted kappa value for TSR evaluation between core area and the whole mount was 0.854(95% CI,0.699-1.000),that for transitional area and the whole mount was 0.951(95% CI,0.849-1.000)and that for peripheral area and the whole mount was0.606(95% CI,0.361-0.806).Therefore,the transitional area is the most suitable representative of the whole mount.5.Of the 41 consecutive patients,the weighted kappa value for categorical assessments between the pathologists’ evaluation and computer-aided evaluation was 0.804(95% CI,0.573–0.951),suggesting strong agreement.Conclusion1.TSR evaluation in PDAC according to the assessment method that we first used and validated provides independent prognostic information complementary to the TNM staging system.Moreover,we demonstrate the robustness and potential of a simple,standardized,inexpensive,and reliable scoring system,which may facilitate routine TSR documentation in histopathology reports of patients with PDAC.2.The heterogeneity of TSR in different part of the whole mount was clarified and the repretative part can replace the whole mount in TSR evaluation.3.The computor-aided method may replace the pathologists for TSR evaluation.Part Ⅱ Construction of a Predictive Model for Prognosis in Patients with Pancreatic Ductal Adenocarcinoma Based on Markers of CAFsObjective To construct a nomogram for prognosis prediction by combining the current PDAC staging system and the markers of CAFs.Methods1.The FFPE tissues of 380 consecutive patients with a final histopathological diagnosis of PDAC who underwent primary pancreatic resection at the Department of Hepatobiliary Pancreatic Surgery in Changhai Hospital(Shanghai,China)from January 2016 to December2017 were retrieved.2.Tissue miacroarray(TMA)were constructed and the juxtatumoral stroma(s-jux),peripheral stroma(s-per)and the septal stroma(s-sep)were accurately defined.3.The markers,including CK-19,FAP-α,SPARC,α-SMA,periostin(POSTN),IL-6,SDF-1 and PDGFR-βwere analysed by using IHC.The α-SMA was also analysed by immunofluorescence.4.Analyses were performed using SPSS version 25.0.For all analyses,a two-tailed P<0.05 was considered statistically significant.R 3.5 was used for construction of nomogram.Results1.By analyzing 13 slides of patients with PDAC,we found that CAFs markers FAP-α,SPARC,α-SMA,POSTN,MMP-11,IL-6,SDF-1 and PDGFR-βwere differently expressed in the three parts of stroma.2.Based on the TMA study,the developing cohort comprised 190 patients,whereas the validation cohort consisted of 79 patients.In the analysis of developing cohort,the intensity of α-SMA score were significantly higher in the s-jux than that of s-per and the coefficient of intensity of α-SMA between s-per and s-jux is relatively low as 0.648,comapred with that of other markers.The patients with α-SMA expression intensity ≥2 in s-jux and <2 in s-per showed a better prognosis compared with the other patients(HR=2.873,95%CI 1.768-4.669,P<0.001).And the patients with α-SMA expression intensity more than SDF-1 expression intensity in s-jux also showed a better prognosis compared with the other patients(HR=1.809,95% CI 1.198-2.731,P=0.0042).In further,both the patients with α-SMA expression intensity ≥2 in s-jux and <2 in s-per(P=0.022)and the patients with α-SMA expression intensity more than SDF-1 expression intensity(P=0.042)in s-jux were significantly less likely with positive lymph node.3.Multivariate COX analysis confirmed that TNM stage(II vs.I: HR,2.112;95% CI,1.270-3.514;P = 0.004;III vs.I: HR,3.990;95% CI,2.256-7.054;P < 0.001),α-SMA expression intensity(Others vs.s-jux≥2&s-per <2: HR,2.819;95% CI,1.733-4.587;P< 0.001)and tumor grade(G3 vs.G1/2: HR,1.582;95% CI,1.024–2.442;P = 0.039)were independent prognostic factors in the developing cohort.4.The nomogram for prognosis prediction was constructed based on the independent prognostic variables.And the AUC was 0.742 in the developing cohort,so as 0.765 in the valiadtion cohort.The dynamic nomogram was showed at https://jamesyin.shinyapps.io/Dyn Nomapp/.Conclusions1.The expression of CAFs markers were characterized by spacial heterogeneity.2.The feature of α-SMA expression intensity in different area of stroma can effectively predict the prognosis of PDAC patients.3.The nomogram with a variable related with CAFs for prognosis prediction can facilitate the development of precision medicine.Part Ⅲ Exploration for molecular classification of CAFs in transcriptomicsObjective To explore molecular classification of CAFs in transcriptomics with primary cultured CAFs.Methods1.The primary culture of CAFs or PSCs was performed by using samples from surgical resection.2.The IHC staining of CAFs climbing slices were used to classify CAFs as subtypes.3.Transcriptomics sequencing data was obtained from CAFs subtypes and used for bioinformatics analysis.Results1.The success rate of primary culture for CAFs was 70%,and that for PSCs was 50%.2.Based on the IHC staining of CAFs climbing slices,CAFs were classified as CAF-Ⅰ and caf-Ⅱ,PSCs as CAF-Ⅲ.3.The CAFs transcriptome sequencing data showed that molecular function of CAF-Ⅱwere relatively stable,and the gene expression of CAF-Ⅰ and CAF-Ⅱ was significantly different.Furthermore,the up-regulated signal pathway of CAF-Ⅱ mainly focused on regulation of neutrophil migration and lymphocyte differentiation,while the up-regulated signal pathway of CAF-Ⅰ was mainly related to the contraction and development of muscle cells.ConclusionsThe gene expression of CAF-Ⅰ and CAF-Ⅱ was significantly different and the primary cultured CAFs can be used in researches in vitro.