Early postoperative recurrence prediction in pancreatic ductal adenocarcinoma: a nomogram based on dual-energy computed tomography and ¹⁸F-fluorodeoxyglucose positron emission tomography-computed tomography

Introduction

Pancreatic cancer, one of the most lethal malignancies of the gastrointestinal tract, poses a persistent and increasingly serious global health risk. Global Cancer Observatory’s 2022 update of worldwide cancer statistics revealed that pancreatic cancer contributes to approximately 511,000 new cancer cases worldwide, and among cancers, ranks 12th in terms of incidence and 6th in terms of the number of cancer-related deaths (1). Pancreatic ductal adenocarcinoma (PDAC), accounting for more than 90% of all pancreatic malignancies, is widely recognized for its poor clinical outcomes, as exemplified by a 5-year survival rate that is consistently less than 10% (2). Among patients with PDAC who undergo radical resection, 49% to 51.5% experience disease recurrence within 12 months after surgery (3), indicating the likely presence of preoperatively disseminated occult micrometastases (4). Notably, recent clinical studies have confirmed that adjuvant therapies combined with radical resection significantly improve patient survival (5,6). Therefore, accurate identification of patients at high risk of early recurrence (ER) is essential for individualized treatment.

In current clinical practice, the prediction of postoperative recurrence in patients with PDAC relies primarily on conventional indicators such as the tumor diameter and carbohydrate antigen 19-9 (CA19-9) level (7,8). However, these indicators are relatively simplistic and lack predictive efficacy (9). Interestingly, a previous study on dual-energy computed tomography (DECT) demonstrated that key parameters, including iodine concentration (IC) and normalized iodine concentration (NIC), are associated with tumor histological grade and angiogenesis (10) and can predict postoperative local recurrence and patient survival time (11,12). A recent study preliminarily demonstrated the potential of quantitative DECT parameters in predicting the risk of ER in patients with PDAC. However, this study did not incorporate all DECT-derived parameters (13). The maximum standardized uptake value (SUV_max) of ¹⁸F-fluorodeoxyglucose positron emission tomography-computed tomography (¹⁸F-FDG PET/CT), a well-established indicator of tumor glycolytic metabolic activity, is strongly correlated with the aggressive phenotype and chemotherapy resistance of PDAC (14,15). Thus, models integrating CT and PET imaging parameters are likely superior to simpler models in predicting the risk of postoperative recurrence in patients with PDAC. However, a review of the related literature indicates that the value of a multimodal prediction model including DECT and ¹⁸F-FDG PET/CT parameters has not been extensively assessed.

Therefore, in this study, we constructed a nomogram based on imaging and clinicopathological variables for predicting ER in patients with PDAC. A clinically actionable model was established to identify subgroups at high risk of postoperative recurrence. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1224/rc).

Methods

Patients

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments and was approved by the Ethics Committee of Zhejiang Provincial People’s Hospital (No. QT2025103). Due to the retrospective nature of our analysis, the requirement for informed consent was waived. Patients who underwent radical surgery (R0/R1 resection with standard lymphadenectomy) between November 2021 and July 2023 with postoperative pathological confirmation of PDAC were enrolled in this study (16).

The inclusion criteria for patients with PDAC were as follows: (I) completion of DECT and ¹⁸F-FDG PET/CT scans within 2 weeks before surgery and (II) completion of regular postoperative follow-up for at least 12 months with confirmed recurrence status according to imaging or histopathological examination. The exclusion criteria were as follows: (I) administration of neoadjuvant chemotherapy, radiotherapy, or immunotherapy prior to surgery; (II) a history of concurrent malignancies in other organs; and (III) severe motion or metal artifacts during preoperative imaging. A total of 80 patients were ultimately enrolled. The study flowchart is presented in Figure 1.

Figure 1 A diagram of patient inclusion and exclusion criteria. CT, computed tomography; DECT, dual-energy computed tomography; ER, early recurrence; PDAC, pancreatic ductal adenocarcinoma; PET, positron emission tomography.

Clinical and pathological data collection

Perioperative demographic and clinical data were extracted from the hospital’s electronic medical record system and included the following: age at diagnosis, sex, tumor location, maximum tumor diameter, pathological T stage (pTs), pathological N stage (pNs) (17), perineural invasion (PNI), lymphovascular invasion (LVI), and the levels of carcinoma embryonic antigen (CEA), carbohydrate antigen 125 (CA125) and carbohydrate antigen 19-9 (CA19-9). Postoperative follow-up included contrast-enhanced CT or magnetic resonance imaging and tumor marker assessments every 3 months. Recurrence was defined as radiologically or pathologically confirmed progression of local lesions (surgical bed or regional lymph nodes) or distant metastases (e.g., to the liver, lung, or peritoneum). In previous studies, ER has been conventionally defined as recurrence within 12 months postoperation, a cutoff widely used due to its strong association with patient prognosis (18,19). Therefore, in this study, ER was defined as recurrence within 12 months postoperation.

Image acquisition and analysis

CT images were acquired with a third-generation dual-source CT scanner in the Radiology Department, and ¹⁸F-FDG PET/CT scan was performed in the Nuclear Medicine Department (Appendix 1). Postprocessing and quantitative analysis were carried out on dedicated workstations, including measurement of triple-phase contrast-enhanced DECT parameters and SUV_max. Additionally, the invasion status of seven vessels (celiac trunk, common hepatic artery, superior mesenteric artery, portal vein, superior mesenteric vein, and splenic vein) was assessed on 40-keV monoenergetic images (Appendix 1).

All DECT parameters were independently measured by two experienced abdominal radiologists (L.S. and J.C., with 6 and 18 years of expertise in abdominal imaging, respectively). Interobserver agreement was assessed with the intraclass correlation coefficient (ICC) for continuous variables and the kappa statistic for categorical variables. For continuous variables, the mean value of the two measurements was adopted, while discrepancies in categorical variables were resolved through consensus discussions between the observers. A third senior radiologist was available for arbitration as needed. The results of interobserver agreement are provided in Appendix 1.

Statistical analysis

All the statistical analyses were conducted with IBM 25.0 (IBM Corp., Armonk, NY, USA), MedCalc 23.0 (MedCalc Software, Ostend, Belgium), and critical packages (“rms” 6.7-0 and “pROC” 1.18.4) in R v. 4.2.1 (The R Foundation for Statistical Computing). During the data preprocessing phase, all continuous variables were subjected to Shapiro-Wilk normality testing (α=0.10). Variables conforming to a normal distribution are expressed as the means and standard deviations and were analyzed with independent-sample t-tests, whereas nonnormally distributed variables are described as medians and interquartile ranges (IQRs) and were evaluated with the Mann-Whitney test. Categorical variables were compared via the χ² test or Fisher’s exact test, while ordinal variables were compared via the Mann-Whitney test. Univariate and multivariate logistic regression analyses were sequentially conducted to identify the covariates associated with ER. The nomogram was constructed with the “rms” R package. The model’s discriminatory ability, calibration accuracy, and clinical utility were evaluated with the area under the receiver operating characteristic curve (AUC), bootstrap-corrected calibration curves (1,000 resamples), and decision curve analysis (DCA), respectively (20). The DeLong test was applied to compare AUC values. Interobserver agreement was quantified via the ICC and Cohen kappa coefficient, with excellent agreement defined as an ICC >0.80 and κ>0.75. All the statistical analyses were two-tailed (α=0.05).

Results

Patient characteristics

A total of 125 patients who underwent radical resection for PDAC were initially enrolled in the study. After 45 patients were excluded on the basis of the predefined criteria (Figure 1), the remaining 80 patients were stratified into two groups: the ER group (n=54) and the non-ER group (n=26); ER was defined as recurrence within 12 months postoperation. Analysis of baseline characteristics revealed no significant differences between the groups in terms of sex, age, tumor location, PNI, or serum CA125 and CEA levels (all P values >0.05). The median CA19-9 level was significantly greater in the ER group (211.55 U/mL, IQR, 109.43–951.03 U/mL) than in the non-ER group (28.20 U/mL, IQR, 11.78–56.15 U/mL; P=0.015). The median tumor diameter was also greater in the ER group (3.50 cm, IQR, 2.68–4.50 cm) than in the non-ER group (2.50 cm, IQR, 2.00–3.00 cm; P<0.001). Furthermore, the ER group, as compared to the non-ER group, had a significantly greater proportion of patients with LVI (62.96% vs. 26.92%; P=0.003) and lymph node metastasis (79.63% vs. 7.69%; P<0.001) (Table 1).

Comparison of imaging parameters

Analysis of imaging characteristics revealed significant differences between the ER and non-ER groups. Among all three-phase spectral quantitative parameters, the portal venous phase NIC (PV-NIC) was the only parameter that was significantly different between the groups (ER group: 36.85±12.37; non-ER group: 57.62±16.43; P<0.001). PET metabolic analysis revealed a significantly greater median SUV_max in the ER group (11.40, IQR, 8.38–13.95) than in the non-ER group (4.85, IQR, 4.10–6.80; P<0.001). Additionally, the ER group had a greater median number of veins involved (1, IQR, 0–1) than did the non-ER group (0, IQR, 0–1; P<0.001). The interobserver agreement analysis indicated moderate-to-excellent consistency among quantitative variables, with dual-energy index (DEI) and fat fraction values demonstrating lower agreement (range of 0.51–0.58), while the PV-NIC value exhibited the highest consistency (ICC =0.91). The interobserver agreement on number of arteries and veins involved was graded as good and excellent, respectively (Appendix 1, Table S1).

Predictive marker screening and model construction

Univariable logistic regression analysis identified seven variables significantly associated with ER (screening threshold P<0.1): CA19-9 [odds ratio (OR) =1.001; 95% confidence interval (CI): 1.000–1.002], tumor diameter (OR =2.590; 95% CI: 1.468–4.570), lymph node metastasis (OR =46.909; 95% CI: 9.592–229.398), LVI (OR =4.614; 95% CI: 1.651–12.894), PV-NIC (OR =0.908; 95% CI: 0.869–0.948), SUV_max (OR =1.754; 95% CI: 1.369–2.248), and the number of veins involved (OR =4.634; 95% CI: 1.892–11.350). A comprehensive multivariate model incorporating all seven predictors further examined these associations, with lymph node metastasis, PV-NIC, SUV_max, and the number of veins involved emerging as robust independent predictive markers. These four parameters were ultimately integrated into a nomogram to stratify individualized recurrence risk. Univariate logistic regression analysis revealed that CA19-9 (OR =1.001; 95% CI: 1.000–1.002), tumor diameter (OR =2.590; 95% CI: 1.468–4.570), lymph node metastasis (OR =46.909; 95% CI: 9.592–229.398), LVI (OR =4.614; 95% CI: 1.651–12.894), PV-NIC (OR =0.908; 95% CI: 0.869–0.948), SUV_max (OR =1.754; 95% CI: 1.369–2.248), and the number of veins involved (OR =4.634; 95% CI: 1.892–11.350) were significant predictors of ER risk (all P values <0.1). Subsequent multivariate analyses were conducted in two domains. In the clinicopathological domain (CA19-9, lymph node metastasis, and LVI), lymph node metastasis emerged as the sole independent predictor of ER risk (OR =37.527; 95% CI: 7.072–199.134). Among the imaging parameters [maximum tumor diameter, parenchymal phase NIC (P-NIC), SUV_max, and number of veins involved], multivariate analysis identified three independent predictors of ER: P-NIC (protective effect; OR =0.924; 95% CI: 0.867–0.985), SUV_max (OR =1.722; 95% CI: 1.243–2.386), and the number of veins involved (OR =4.508; 95% CI: 1.077–18.879). A comprehensive multivariate analysis incorporating all seven parameters further revealed lymph node metastasis, P-NIC, SUV_max, and the number of veins involved to be robust independent predictors of ER risk, and these parameters were subsequently integrated into a nomogram (Table 2).

Model predictive performance

In predicting ER, the comprehensive model achieved an AUC of 0.979 (95% CI: 0.918–0.998), whereas the clinicopathological model and imaging model yielded AUC values of 0.860 (95% CI: 0.764–0.927) and 0.949 (95% CI: 0.876–0.986), respectively. Both the comprehensive model and the imaging model outperformed the clinicopathological model (P<0.001 and P=0.049, respectively). However, no significant difference was observed between the predictive performance of the comprehensive model and that of the imaging model (P=0.142) (Figure 2A).

Figure 2 Development and validation of a nomogram integrating clinicopathological and imaging models. (A) The ROC curves for clinicopathological model, imaging model and combined model. (B) Calibration curve of the combined model. (C) The nomogram of combined model. *, P<0.05; **, P<0.01. LN status: 0, negative; 1, positive. LN, lymph node; NIC, normalized iodine concentration; PV, portal venous phase; ROC, receiver operating characteristic; SUV_max, maximum standardized uptake value.

Nomogram model performance

The nomogram model demonstrated high predictive accuracy (AUC =0.979) and good calibration according to the Hosmer-Lemeshow test (calibration slope =0.91). DCA further validated the nomogram model’s clinical utility, demonstrating significant net benefit across clinically relevant risk thresholds (Figures 2B,2C,3). Figures 4,5 include representative cases from the ER and non-ER groups, respectively.

Figure 3 DCA of the nomogram based on combined model. The orange line, blue line, and black line represent the net benefit of the nomogram, treat-all strategy, and treat-none strategy, respectively. DCA, decision curve analysis.

Figure 4 A 63-year-old male patient with PDAC. (A) Portal-venous-phase mixed-energy CT image demonstrated a tumor (arrow) in the pancreatic uncinate process. (B) Iodine map revealed an NIC value of 37.5% in the tumor (arrow). (C) CT MPR image showed tumor invasion into the portal vein (*) and superior mesenteric vein (arrowhead), with luminal narrowing. (D) PET/CT image displayed intense ¹⁸F-FDG uptake in the tumor (arrow), with an SUV_max of 10.6. (E) In the nomogram, vertical lines drawn for each variable correspond to a total of 172 points, indicating a recurrence risk of 99.9%. *, P<0.05; **, P<0.01. LN status: 0, negative; 1, positive. CT, computed tomography; FDG, fluorodeoxyglucose; LN, lymph node; MPR, multiplanar reconstruction; NIC, normalized iodine concentration; PDAC, pancreatic ductal adenocarcinoma; PV, portal venous phase; SUV_max, maximum standardized uptake value.

Figure 5 A 67-year-old female patient with PDAC. (A) Portal-venous-phase mixed-energy CT image demonstrated a tumor (arrow) in the pancreatic body and tail. (B) Iodine map revealed an NIC value of 48.4% in the tumor (arrow). (C) MPR image indicated that the tumor was adjacent to the portal vein (*) and superior mesenteric vein (arrowhead) but no vascular invasion. (D) PET/CT image showed ¹⁸F-FDG uptake in the tumor (arrow), with an SUV_max of 4.10. (E) In the nomogram, vertical lines drawn for each variable correspond to a total of 79 points, with a recurrence risk of 1.7%. *, P<0.05; **, P<0.01. LN status: 0, negative; 1, positive. CT, computed tomography; FDG, fluorodeoxyglucose; LN, lymph node; MPR, multiplanar reconstruction; NIC, normalized iodine concentration; PDAC, pancreatic ductal adenocarcinoma; PV, portal venous phase; SUV_max, maximum standardized uptake value.

Discussion

Despite recent advancements in surgical techniques and optimized systemic treatment regimens that have significantly improved survival outcomes in patients with solid tumors (1), the 5-year survival rate for patients with PDAC remains ≤9% (2). Moreover, even in patients who achieved R0 resection, the overall recurrence rate remains alarmingly high at 73.7% (21), and ER has emerged as a critical indicator of poor prognosis. This underscores the urgent need to establish a robust framework for predicting the risk of ER. In this study, we integrated PV-NIC, the SUV_max, lymph node metastasis, and the number of veins involved to develop a multimodal imaging-based nomogram model for predicting the risk of postoperative ER in patients with PDAC.

In this study, ER occurred in 54 (54/80, 67.5%) cases. Notably, 20 (37.04%) of these patients with ER presented with liver metastasis as the first evidence of recurrence, suggesting the potential presence of subclinical micrometastases at the time of surgery (22). The serum CA19-9 level, currently the most widely used clinical biomarker for PDAC, reflects the tumor burden and may indicate the presence of micrometastases (23). In our cohort, the median preoperative CA19-9 levels were significantly elevated in the ER group (211.55 U/mL, IQR, 109.43–951.03 U/mL) as compared with the non-ER group (28.20 U/mL, IQR, 11.78–56.15 U/mL; P=0.015); this aligns with findings from Daamen et al. (24), who identified CA19-9 >150 U/mL as an independent predictor of poor overall survival [hazard ratio (HR) =1.779; P=0.002] in a large-scale study of 836 patients. However, CA19-9 was excluded from the final predictive model in our study, which may be attributed to limitations in sample size, potential cohort selection bias, and false-positive results due to obstructive jaundice (25).

Our study also confirmed that lymph node metastasis (pN1–2) is as an independent predictor of ER risk (OR =19.610; 95% CI: 1.211–340.406; P=0.032), a finding corroborated by multicenter data reported by Honselmann et al. (26), further underscoring the pivotal role of lymph node metastasis in the malignant biological behavior of pancreatic cancer. Once lymph node metastasis occurs, tumor cells may disseminate to distant organs via the circulatory system, thereby driving ER (27). In our cohort, the incidence of lymph node metastasis in the ER group was 33.33%, second only to liver metastasis, indicating that the lymphatic system acts as a critical bridge for distant tumor dissemination.

Our examination of DECT parameters revealed that the NIC derived from the portal venous phase serves as an independent predictor of ER following PDAC resection (OR =0.908; 95% CI: 0.869–0.948; P=0.028). Previous studies have suggested that alterations in the NIC may be correlated with tumor vascularization and extracellular matrix (ECM) remodeling (12,28). From an oncological perspective, the excessive deposition of hyaluronic acid and ECM hyperplasia in PDAC can induce stromal hypertension (29), leading to compromised perfusion. Furthermore, pathological changes such as aberrant tumor neovascularization and ECM reorganization (30) impede contrast agent penetration into the tumor parenchyma, resulting in reduced iodine uptake during the early enhancement phases. These mechanisms collectively account for the superior predictive performance of PV-NIC in the portal venous phase compared with that in the arterial or parenchymal phase.

In patients with PDAC, the assessment of vascular involvement via multidetector CT serves as a critical determinant for predicting tumor resectability and survival outcomes (31,32). Tran Cao et al. (31) demonstrated that tumor-vein interface characteristics effectively predict surgical feasibility and oncological prognosis. In our study, we optimized vascular visualization by reconstructing 40-keV virtual monoenergetic images, which significantly improved the delineation of the pancreatic tumor parenchyma and peritumoral vasculature compared with conventional imaging protocols (33,34). The results revealed that the number of veins involved was independently associated with early postoperative recurrence (OR =8.660; 95% CI: 1.083–110.245; P=0.043), a finding consistent with previous studies (24,35).

The multivariate analysis in this study identified the SUV_max derived from FDG-PET to be an independent predictor of ER risk following PDAC resection (OR =1.739; 95% CI: 1.091–4.142; P=0.027), establishing a metabolic basis for the association between ER and aggressive tumor biology. The SUV_max is a direct quantification of glucose metabolic activity within tumor tissue (36), with elevated values indicating hypermetabolic states that correlate with increased invasiveness and the propensity for early relapse. A recent study by Lancellotti et al. similarly identified SUV_max to be a predictor of early postoperative recurrence in patients with PDAC (37), although the median SUV_max values they reported for patients with disease-free survival <12 months were lower than those in our cohort. These discrepancies may stem from variations in PET/CT equipment, image reconstruction algorithms, population characteristics, or patient preparation protocols across different institutions (38). For instance, the larger tumor volume in patients with ER could lead to interobserver variability in region of interest delineation; furthermore, differences in glycemic levels between the two study populations might also have influenced the SUV_max values. Nevertheless, both our own and Lancellotti et al.’s study were consistent in confirming an elevated SUV_max to be an independent prognostic factor for ER. Moreover, Jeong et al. (39) reported a positive correlation between the SUV_max and preoperative CA19-9 level, further supporting the value of SUV_max as an independent risk factor for postoperative hepatic metastasis and peritoneal recurrence. Finally, the SUV_max has potential value in predicting long-term PDAC prognosis (40).

The multimodal data integration strategy significantly improved the predictive performance of the models. The integrated model had the highest AUC; it was superior to the clinicopathological model (P<0.001) but showed comparable performance to the imaging model (P=0.1419). This may be attributed to the fact that the quantitative DECT parameters and the SUV_max of ¹⁸F-FDG PET/CT indirectly or directly provide information on tumor metabolism. The integrated model, which was based on preoperative and postoperative data, further overcomes the limitations of single-modality approaches, improving the comprehensiveness and accuracy of postoperative ER evaluation.

There are several limitations to this study which should be acknowledged. First, as a retrospective single-center study, the exclusion of patients who did not undergo dual-source CT and PET/CT examinations might have introduced selection bias. Second, not all the clinical and imaging data were incorporated in the analysis; for example, other tumor metabolism-related parameters obtained via PET/CT were not included, immunohistochemistry results were not considered among the pathological data, and the extent of vascular involvement was not classified. Third, the number of patients included in the study was relatively small, and the findings of this study need to be validated in external cohorts to further confirm the model’s stability. Fourth, we did not perform in-depth analyses via radiomics or texture analysis, and further research is needed to explore their unique advantages.

Conclusions

This study developed a nomogram integrating multiparametric DECT and PET/CT imaging data with clinicopathological parameters to predict early postoperative recurrence in patients with PDAC. The model demonstrated robust discriminatory capability in identifying cohorts at high-risk of recurrence and may serve as a practical clinical decision-support tool for guiding therapeutic interventions and personalizing treatment strategies.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1224/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1224/dss

Funding: This study was supported by Zhejiang Medical and Health Science and Technology Project (No. 2021KY508).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1224/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of Zhejiang Provincial People’s Hospital (No. QT2025103). The requirement for informed consent was waived due to the study’s retrospective nature.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, Jemal A. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Klein AP. Pancreatic cancer epidemiology: understanding the role of lifestyle and inherited risk factors. Nat Rev Gastroenterol Hepatol 2021;18:493-502. [Crossref] [PubMed]
3. Groot VP, Gemenetzis G, Blair AB, Rivero-Soto RJ, Yu J, Javed AA, Burkhart RA, Rinkes IHMB, Molenaar IQ, Cameron JL, Weiss MJ, Wolfgang CL, He J. Defining and Predicting Early Recurrence in 957 Patients With Resected Pancreatic Ductal Adenocarcinoma. Ann Surg 2019;269:1154-62. [Crossref] [PubMed]
4. Barhli A, Cros J, Bartholin L, Neuzillet C. Prognostic stratification of resected pancreatic ductal adenocarcinoma: Past, present, and future. Dig Liver Dis 2018;50:979-90. [Crossref] [PubMed]
5. Moaven O, Clark CJ, Russell GB, Votanopoulos KI, Howerton R, Levine EA, Shen P. Optimal Adjuvant Treatment Approach After Upfront Resection of Pancreatic Cancer: Revisiting the Role of Radiation Based on Pathologic Features. Ann Surg 2021;274:1058-66. [Crossref] [PubMed]
6. Pu N, Wu W, Liu S, Xie Y, Yin H, Chen Q, He T, Xu Z, Wang W, Yu J, Liu L, Lou W. Survival benefit and impact of adjuvant chemotherapy following systemic neoadjuvant chemotherapy in patients with resected pancreas ductal adenocarcinoma: a retrospective cohort study. Int J Surg 2023;109:3137-46. [Crossref] [PubMed]
7. Liu W, Tang B, Wang F, Qu C, Hu H, Zhuang Y, Gao H, Xie X, Tian X, Yang Y. Predicting early recurrence for resected pancreatic ductal adenocarcinoma: a multicenter retrospective study in China. Am J Cancer Res 2021;11:3055-69.
8. Matsumoto I, Murakami Y, Shinzeki M, Asari S, Goto T, Tani M, Motoi F, Uemura K, Sho M, Satoi S, Honda G, Yamaue H, Unno M, Akahori T, Kwon AH, Kurata M, Ajiki T, Fukumoto T, Ku Y. Proposed preoperative risk factors for early recurrence in patients with resectable pancreatic ductal adenocarcinoma after surgical resection: A multi-center retrospective study. Pancreatology 2015;15:674-80. [Crossref] [PubMed]
9. Leonhardt CS, Gustorff C, Klaiber U, Le Blanc S, Stamm TA, Verbeke CS, Prager GW, Strobel O. Prognostic Factors for Early Recurrence After Resection of Pancreatic Cancer: A Systematic Review and Meta-Analysis. Gastroenterology 2024;167:977-92. [Crossref] [PubMed]
10. Chen XH, Ren K, Liang P, Chai YR, Chen KS, Gao JB. Spectral computed tomography in advanced gastric cancer: Can iodine concentration non-invasively assess angiogenesis? World J Gastroenterol 2017;23:1666-75. [Crossref] [PubMed]
11. Chen W, Lin G, Li X, Feng Y, Mao W, Kong C, Hu Y, Gao Y, Yang W, Chen M, Yan Z, Xia S, Lu C, Xu M, Ji J. Dual-energy computed tomography for predicting histological grading and survival in patients with pancreatic ductal adenocarcinoma. Eur Radiol 2025;35:2818-32. [Crossref] [PubMed]
12. Mathy RM, Fritz F, Mayer P, Klauss M, Grenacher L, Stiller W, Kauczor HU, Skornitzke S. Iodine concentration and tissue attenuation in dual-energy contrast-enhanced CT as a potential quantitative parameter in early detection of local pancreatic carcinoma recurrence after surgical resection. Eur J Radiol 2021;143:109944. [Crossref] [PubMed]
13. Yu SY, Shu YP, Bai XH, Yu J, Lu ZP, Jiang KR, Xu Q. Efficiency evaluation of dual-energy CT to predict the postoperative early recurrence of pancreatic ductal adenocarcinoma. Pancreatology 2024;24:1123-32. [Crossref] [PubMed]
14. Mohamed E, Needham A, Psarelli E, Carroll M, Vinjamuri S, Sanghera B, Wong WL, Halloran C, Ghaneh P. Prognostic value of 18FDG PET/CT volumetric parameters in the survival prediction of patients with pancreatic cancer. Eur J Surg Oncol 2020;46:1532-8. [Crossref] [PubMed]
15. Abdelrahman AM, Goenka AH, Alva-Ruiz R, Yonkus JA, Leiting JL, Graham RP, et al. FDG-PET Predicts Neoadjuvant Therapy Response and Survival in Borderline Resectable/Locally Advanced Pancreatic Adenocarcinoma. J Natl Compr Canc Netw 2022;20:1023-1032.e3. [Crossref] [PubMed]
16. Tol JA, Gouma DJ, Bassi C, Dervenis C, Montorsi M, Adham M, et al. Definition of a standard lymphadenectomy in surgery for pancreatic ductal adenocarcinoma: a consensus statement by the International Study Group on Pancreatic Surgery (ISGPS). Surgery 2014;156:591-600. [Crossref] [PubMed]
17. Tempero MA, Malafa MP, Al-Hawary M, Behrman SW, Benson AB, Cardin DB, et al. Pancreatic Adenocarcinoma, Version 2.2021, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw 2021;19:439-57. [Crossref] [PubMed]
18. Palumbo D, Mori M, Prato F, Crippa S, Belfiori G, Reni M, Mushtaq J, Aleotti F, Guazzarotti G, Cao R, Steidler S, Tamburrino D, Spezi E, Del Vecchio A, Cascinu S, Falconi M, Fiorino C, De Cobelli F. Prediction of Early Distant Recurrence in Upfront Resectable Pancreatic Adenocarcinoma: A Multidisciplinary, Machine Learning-Based Approach. Cancers (Basel) 2021;13:4938. [Crossref] [PubMed]
19. Tang TY, Li X, Zhang Q, Guo CX, Zhang XZ, Lao MY, Shen YN, Xiao WB, Ying SH, Sun K, Yu RS, Gao SL, Que RS, Chen W, Huang DB, Pang PP, Bai XL, Liang TB. Development of a Novel Multiparametric MRI Radiomic Nomogram for Preoperative Evaluation of Early Recurrence in Resectable Pancreatic Cancer. J Magn Reson Imaging 2020;52:231-45. [Crossref] [PubMed]
20. Alba AC, Agoritsas T, Walsh M, Hanna S, Iorio A, Devereaux PJ, McGinn T, Guyatt G. Discrimination and Calibration of Clinical Prediction Models: Users' Guides to the Medical Literature. JAMA 2017;318:1377-84. [Crossref] [PubMed]
21. Choi SH, Kim HY, Hwang HK, Kang CM, Lee WJ. Oncologic Impact of Local Recurrence in Resected Pancreatic Cancer and Topographic Preference in Local Recurrence Patterns According to Tumor Location. Pancreas 2020;49:1290-6. [Crossref] [PubMed]
22. Neoptolemos JP, Stocken DD, Bassi C, Ghaneh P, Cunningham D, Goldstein D, et al. Adjuvant chemotherapy with fluorouracil plus folinic acid vs gemcitabine following pancreatic cancer resection: a randomized controlled trial. JAMA 2010;304:1073-81. [Crossref] [PubMed]
23. Daamen LA, Groot VP, Heerkens HD, Intven MPW, van Santvoort HC, Molenaar IQ. Systematic review on the role of serum tumor markers in the detection of recurrent pancreatic cancer. HPB (Oxford) 2018;20:297-304. [Crossref] [PubMed]
24. Daamen LA, Dorland G, Brada LJH, Groot VP, van Oosten AF, Besselink MG, et al. Preoperative predictors for early and very early disease recurrence in patients undergoing resection of pancreatic ductal adenocarcinoma. HPB (Oxford) 2022;24:535-46. [Crossref] [PubMed]
25. Tsen A, Barbara M, Rosenkranz L. Dilemma of elevated CA 19-9 in biliary pathology. Pancreatology 2018;18:862-7. [Crossref] [PubMed]
26. Honselmann KC, Pergolini I, Castillo CF, Deshpande V, Ting D, Taylor MS, Bolm L, Qadan M, Wellner U, Sandini M, Bausch D, Warshaw AL, Lillemoe KD, Keck T, Ferrone CR. Timing But Not Patterns of Recurrence Is Different Between Node-negative and Node-positive Resected Pancreatic Cancer. Ann Surg 2020;272:357-65. [Crossref] [PubMed]
27. Chon HY, Lee HS, Sung YN, Tae YK, Park CH, Leem G, Kim SJ, Jo JH, Chung MJ, Park JY, Park SW, Hong SM, Bang S. Uncovering the clinicopathological features of early recurrence after surgical resection of pancreatic cancer. Sci Rep 2024;14:2942. [Crossref] [PubMed]
28. Wen Y, Song Z, Li Q, Zhang D, Li X, Yu J, Li Z, Ren X, Zhang J, Liu Q, Huang J, Zeng D, Tang Z. Development and validation of a model for predicting the expression of Ki-67 in pancreatic ductal adenocarcinoma with radiological features and dual-energy computed tomography quantitative parameters. Insights Imaging 2024;15:41. [Crossref] [PubMed]
29. Vohra R, Wang YN, Son H, Totten S, Arora A, Maxwell A, Lee D. Non-Invasive Monitoring of Increased Fibrotic Tissue and Hyaluronan Deposition in the Tumor Microenvironment in the Advanced Stages of Pancreatic Ductal Adenocarcinoma. Cancers (Basel) 2022;14:999. [Crossref] [PubMed]
30. Houg DS, Bijlsma MF. The hepatic pre-metastatic niche in pancreatic ductal adenocarcinoma. Mol Cancer 2018;17:95. [Crossref] [PubMed]
31. Tran Cao HS, Balachandran A, Wang H, Nogueras-González GM, Bailey CE, Lee JE, Pisters PW, Evans DB, Varadhachary G, Crane CH, Aloia TA, Vauthey JN, Fleming JB, Katz MH. Radiographic tumor-vein interface as a predictor of intraoperative, pathologic, and oncologic outcomes in resectable and borderline resectable pancreatic cancer. J Gastrointest Surg 2014;18:269-78; discussion 278. [Crossref] [PubMed]
32. Zaky AM, Wolfgang CL, Weiss MJ, Javed AA, Fishman EK, Zaheer A. Tumor-Vessel Relationships in Pancreatic Ductal Adenocarcinoma at Multidetector CT: Different Classification Systems and Their Influence on Treatment Planning. Radiographics 2017;37:93-112. [Crossref] [PubMed]
33. Liang HW, Zhou Y, Zhang ZW, Yan GW, Du SL, Zhang XH, Li XY, Lv FJ, Zheng Q, Li YM. Dual-energy CT with virtual monoenergetic images to improve the visualization of pancreatic supplying arteries: the normal anatomy and variations. Insights Imaging 2022;13:21. [Crossref] [PubMed]
34. Nagayama Y, Tanoue S, Inoue T, Oda S, Nakaura T, Utsunomiya D, Yamashita Y. Dual-layer spectral CT improves image quality of multiphasic pancreas CT in patients with pancreatic ductal adenocarcinoma. Eur Radiol 2020;30:394-403. [Crossref] [PubMed]
35. Xiang F, He X, Liu X, Li X, Zhang X, Fan Y, Yan S. Development and Validation of a Nomogram for Preoperative Prediction of Early Recurrence after Upfront Surgery in Pancreatic Ductal Adenocarcinoma by Integrating Deep Learning and Radiological Variables. Cancers (Basel) 2023;15:3543. [Crossref] [PubMed]
36. Ariake K, Motoi F, Shimomura H, Mizuma M, Maeda S, Terao C, Tatewaki Y, Ohtsuka H, Fukase K, Masuda K, Hayashi H, Takadate T, Naitoh T, Taki Y, Unno M. 18-Fluorodeoxyglucose Positron Emission Tomography Predicts Recurrence in Resected Pancreatic Ductal Adenocarcinoma. J Gastrointest Surg 2018;22:279-87. [Crossref] [PubMed]
37. Lancellotti F, Patel A, Tsaramanidis S, Edy E, Satyadas T, Filobbos R, Jamdar S, Barker S, Siriwardena AK, de Liguori-Carino NHPB-MRI Collaborative group. The role of (18)F-fluorodeoxyglucose positron emission tomography/computed tomography-derived parameters as prognostic factors in patients undergoing resection for pancreatic ductal adenocarcinoma. Surgery 2025;181:109271. [Crossref] [PubMed]
38. Boellaard R. Standards for PET image acquisition and quantitative data analysis. J Nucl Med 2009;50:11S-20S. [Crossref] [PubMed]
39. Jeong B, Oh M, Lee SS, Kim N, Kim JS, Lee W, Kim SC, Kim HJ, Kim JH, Byun JH. Predicting Recurrence-Free Survival After Upfront Surgery in Resectable Pancreatic Ductal Adenocarcinoma: A Preoperative Risk Score Based on CA 19-9, CT, and 18F-FDG PET/CT. Korean J Radiol 2024;25:644-55. [Crossref] [PubMed]
40. Im HJ, Oo S, Jung W, Jang JY, Kim SW, Cheon GJ, Kang KW, Chung JK, Kim EE, Lee DS. Prognostic Value of Metabolic and Volumetric Parameters of Preoperative FDG-PET/CT in Patients With Resectable Pancreatic Cancer. Medicine (Baltimore) 2016;95:e3686. [Crossref] [PubMed]