Evaluation Of Histological Tumor Necrosis Of Pancreatic Ductal Adenocarcinoma By Quantitative Parameters Of Enhanced CT And Its Relationship With Tumor Prognosis

Objectives: Contrast-enhanced CT(CE-CT)was used to evaluate the histological tumor necrosis(HTN)of pancreatic ductal adenocarcinoma(PDAC).Compared with pathological evaluation which was the gold standard,the value of CE-CT in the diagnosis of HTN of PDAC and its relationship with prognosis were discussed.Methods: Between January 2010 and December 2020,excluding patients with incomplete follow-up data and poor enhanced image,1116 patients with PDAC confirmed by pathology(our hospital,n=966;other hospitals,n=150)who underwent CE-CT studies and margin-negative(R0)pancreatectomy were recruited.Among the patients in our hospital,328 patients were reevaluated by 2 pathologists and reached an agreement through consultation.The volume registration of CE-CT image was performed by using the DEEDS method,the region of interest of tumor was obtained by automatic tumor segmentation based on 3D-nn UNet,and the subtraction map and attenuation difference(delta)of the portal vein phase and the scanning phase of the tumor were obtained by using the Matlab script.Necrosis detected by CT was defined as weak or no enhancement area on the subtraction image(delta difference: 10 Hounsfield [HU]～30 HU).In addition,tumor necrosis was evaluated independently and blindly by two abdominal radiologists based on CT images.In this study,the receiver operating characteristic(ROC)curve was used to evaluate the diagnostic performance and optimal threshold of CT diagnosis of necrosis for HTN,and the Cox proportional-hazards models was used to estimate the disease-free survival(DFS)and overall survival(OS).Results: Of the 328 patients who evaluated HTN,205 patients developed HTN(57.2%).The diagnostic efficacy of CT-defined necrosis(delta ≤15 HU)for HTN significantly outperformed the other delta cutoffs(10HU～30HU,area under the curve [AUC]:0.93 vs0.75-0.87,P<0.05)and radiologist-diagnosed necrosis(AUC: 0.93 vs 0.74 and 0.78;Delong test,all P<0.05).In the multivariate Cox model,CT-defined necrosis is an independent influence factor of DFS(our hospital,[preoperative model,hazard ratio[HR]=2.43(95% confidence interval [CI]: 2.07-2.86);comprehensive model,HR=2.13(95%CI: 1.78-2.53)];external hospital [preoperative model,HR=2.84(95% CI: 1.90-4.25);comprehensive model,HR=2.30(95% CI: 1.51-3.50)],P<0.001).In the Cox model for predicting OS,CT-defined necrosis is an independent influence factor of the preoperative model(HR=2.15;95% CI: 1.84-2.50)and the comprehensive model(HR=1.81;95% CI:1.54-2.14,P<0.001)in our hospital,and also an independent influence factor of the preoperative model(HR=1.98;95% CI: 1.35-2.92,P<0.001)in external hospitals.In the comprehensive model of external hospitals,CT-defined necrosis and OS had no independent correlation(HR=1.15,95% CI: 0.77-1.72,P=0.500).Conclusions: CT-defined necrosis can be used as an objective imaging biomarker to diagnose HTN and predict poor prognosis in PDAC preoperatively.