Summary

METHODS
We retrospectively analysed 18F-FDG PET/CT data, routine pre-operative blood tests (for GRIm Score), and post-surgical pathology (LVI, PNI, etc.) from 41 operable PDAC patients (2016-2024). The GRIm Score (0-3) was determined using LDH, Albumin, and NLR. Patients were categorized into low (0-1) and high (2-3) GRIm groups for statistical evaluation.

RESULTS
No significant correlation was found between the GRIm score and PET/CT or pathology features. Mean overall survival was 12.97 months. Patients with high GRIm scores showed significantly shorter survival (6.2 months vs. 16.7 months, p<0.001). Multivariate survival analysis confirmed that both the GRIm score (HR=10.258, p<0.001) and the presence of LVI (HR=14.899, p=0.019) were independent prognostic factors. Other parameters did not show a significant association with survival.

CONCLUSION
The GRIm score, easily and cost-effectively calculated from routine blood tests, holds significant and independent prognostic value in operable PDAC patients. Its independence from conventional tumor features suggests promising potential for assessing prognosis.

Introduction

The five-year survival rate is approximately 12%, and radical surgical procedures remain the only treatment option that can potentially offer a cure for the patient.[2,4]

Positron emission tomography/computed tomography (PET/CT) using 18-Fluorodeoxyglucose (18F-FDG) is currently noted in guidelines as being under development for PDAC, unlike for other solid tumours. However, studies over the past decade have demonstrated the potential utility of 18F-FDG PET/ CT imaging in the diagnosis and staging of pancreatic cancer. Beyond staging, 18F-FDG PET/CT images also provide valuable insights into tumour behaviour and aggressiveness through quantitative metabolic parameters derived from the primary tumour. These include the standardized uptake value (SUV), metabolic tumour volume (MTV), and total lesion glycolysis (TLG). Primary tumours exhibiting high maximum SUV (SUV_max), MTV, and TLG values are associated with a higher tumour grade and poorer survival outcomes.[4-7]

Traditional determinants of patient survival following pancreaticoduodenectomy include tumour margin, tumour type, tumour size, tumour differentiation, and regional lymph node status. Two other frequently reported, but less thoroughly analysed, parameters are perineural invasion (PNI) and lymphovascular invasion (LVI). PNI has been concluded to be highly significant in patient prognosis after pancreatic tumour resection. Furthermore, the presence of LVI has been reported to decrease survival in neuroendocrine tumours of the pancreas. It's hypothesized that PNI may be responsible for local failure due to tumour growth along the nerves that innervate the pancreas and ultimately form the periarterial neural plexus. Similarly, the presence of LVI may be responsible for regional or distant metastasis in lymph nodes or other organs like the lungs and liver. Pathological tumour size is also associated with poor overall survival.[8,9]

The Gustave Roussy Immune Score (GRIm score) was recently developed to improve participant selection for Phase I trials involving non-small cell lung cancer patients treated with immune checkpoint inhibitors. The GRIm score is a composite measure derived from serum lactate dehydrogenase (LDH) levels, serum albumin concentration (Alb), and the neutrophillymphocyte ratio (NLR). These parameters can provide insights not only into immunogenicity but also into components of the tumour microenvironment. Notably, elevated NLR and low albumin levels are indicative of high inflammation.[10] The other component of the GRIm score, lactate dehydrogenase (LDH), is associated with intracellular hypoxic reactions that contribute to the formation of a hypovascular microenvironment. A high GRIm score may help identify patients with aggressive disease and a poor survival prognosis in both localized and PDAC.[11,12]

This study aims to assess the relationship between the GRIm score, derived from pre-operative blood samples, and metabolic data from primary tumours on 18F-FDG PET/CT images, as well as histopathological findings from post-operative tumour specimens, in PDAC patients eligible for surgery. Furthermore, it seeks to determine the prognostic significance of these combined parameters.

Methods

Patient Selection Criteria
Patients included in this study met the following criteria: They were considered suitable for curative surgical resection due to a pancreatic mass and underwent either a Whipple operation or distal pancreatectomy with curative intent. Prior to surgery, specifically within 14 days, they had undergone 18F-FDG PET/ CT imaging for primary staging without any did not receive any preceding neoadjuvant treatment prior to surgery. Following successful surgical resection, all patients were routinely referred to the Medical Oncology department for standard adjuvant treatment protocols. Within 7 days before surgery, routine blood tests performed for surgical preparation provided accessible neutrophil and lymphocyte counts, as well as albumin and LDH levels. Furthermore, complete post-operative histopathological data and at least one year of followup data were available for these patients.

Patients were excluded from the study if they had a second malignancy other than PDAC, did not undergo 18F-FDG PET/CT imaging for pre-curative surgical staging, or if there was a gap of more than 14 days between imaging and operation. Additionally, patients were excluded if pre-operative routine blood test results were unavailable, or if more than 7 days elapsed between the blood tests and the operation date, or if there was missing data for at least one of the neutrophil or lymphocyte counts, albumin, or LDH levels. Patients with inaccessible post-operative histopathological data or a follow-up period of less than 1 year were also excluded.

18F-FDG PET/CT Imaging Protocol
Before 18F-FDG PET/CT imaging, patients fasted for at least 6 hours. Their blood glucose levels were confirmed to be below 180 mg/dL before 18F-FDG injection. For eligible patients, 18F-FDG was injected intravenously at a dose of 0.1 mCi/kg, and images were acquired approximately 60 minutes post-injection. PET images underwent attenuation correction using CT images. First, a CT scan was performed. Immediately following the CT scan, a standard PET imaging protocol was applied in 3D mode, with an acquisition time of 2 minutes per bed position, extending from the vertex to the mid-thigh. All PET images were acquired in 3D mode. CT images were obtained without intravenous contrast administration, at 70 mA, 120 kV, and an axial slice thickness of 2.5 mm. The spatial resolution of the PET camera system was 5 mm. BT and PET images were coregistered and fused into transaxial, coronal, and sagittal views. For each patient, axial PET slices containing the primary pancreas tumour were processed into DICOM (Digital Imaging and Communications in Medicine) format with a 128×128 matrix. The data were then transferred to a processing workstation (AW Volume Share5 GE Medical Systems S.C.S, France) via the DICOM protocol.

18F-FDG PET/CT Image Analysis
18F-FDG PET/CT images were visually and quantitatively assessed by two experienced nuclear medicine specialists. For the PET images, an adaptive threshold of 42% of the maximum lesional metabolic activity was applied, and a volume of interest (VOI) was positioned to encompass the primary tumour. This relative threshold method was selected to ensure reproducible and accurate calculation of the MTV for the primary tumour, which helps in minimizing the partial volume effect inherent PET quantification.[13] The SUV_max, SUV_mean, and MTV values for the primary tumour were automatically calculated from the PET images using the PET REV software on the workstation. Total Lesion Glycolysis (TLG) was then computed by multiplying MTV by SUV_mean.

Histopathological Evaluation
Postoperative histopathological examinations of the patients were retrospectively reviewed and re-evaluated according to the 8th edition of the American Joint Committee on Cancer.[14] From existing histopathology results, the maximum pathological dimension of the primary tumour (Dpat), tumour grade (G), pathological tumour stage (pT), pathological nodal stage (pN), and the presence of lymphovascular invasion (LVI) and perineural invasion (PNI) were recorded. For patients with identified pT and pN stages, the TNM stages were also defined.

GRIm Score Evaluation
For each patient, routine blood tests were retrospectively evaluated from the hospital information system, specifically examining hemogram and biochemistry panels conducted within 7 days prior to surgery. From the hemogram panel, neutrophil count (NS) and lymphocyte count (LS) were recorded. Albumin (Alb) and LDH levels were recorded from the biochemistry panel. Each patient was then scored as follows: LDH>upper limit of normal received 1 point, Alb<35 grams/liter received 1 point, and NS/LS (NLR)>6 received 1 point. The sum of these points defined the patient"s total GRIm score, which could range from 0 to 3. Patients were subsequently categorized into two groups based on their GRIm scores, as originally defined by Bigot et al.:[10] A low-score group (GRIm score: 0 and 1) and a high-score group (GRIm score: 2 and 3).

Statistical Analysis
Data obtained from our study were analysed using SPSS 26.0. The Shapiro-Wilk test was employed to assess the normality of data distribution. For normally distributed data, Student's t-test was used for group comparisons. If data did not follow a normal distribution, the Mann-Whitney U test was applied for comparisons between two independent groups, while the Kruskal-Wallis test was utilized for comparisons involving more than two independent groups. When using ANOVA for comparisons with more than two groups, Tukey's HSD test was applied to identify differing groups when the assumption of homogeneity was met, and Tamhane's T2 test was used when this assumption was violated. The Chi-square test was employed for evaluating count data. The effects of parameters on survival were examined using the log-rank test. Survival rates were calculated using Kaplan- Meier survival analysis. Additionally, to identify independent prognostic factors, multivariate analysis was performed using the Cox proportional hazard regression model, including variables found to be significant in the univariate analysis and other clinically relevant factors. A significance level of 0.05 was set.

Results

The mean (± standard deviation (s.d.)) values for the quantitative metabolic parameters of the primary tumour were calculated as follows: SUV_max was 5.85 (±1.84), SUV_mean was 3.36 (±1.03), MTV was 19.59 (±12.95), and TLG was 62.51 (±41.84).

The mean (± s.d.) maximum histopathological dimension of the primary tumour was calculated as 36.77 mm (±14.36 mm). Histopathologically, primary tumours were reported as 7.3% (n=3) Grade 1, 87.8% (n=36) Grade 2, and 4.9% (n=2) Grade 3. PNI was observed in 95.1% (n=39) of cases, with only 4.9% (n=2) showing no PNI. LVI was positive in 80.5% (n=33) of cases, while it was not detected in 19.5% (n=8). Regarding pathological T stages, 4.9% (n=2) of cases had T1 tumours, 65.9% (n=27) had T2, 22% (n=9) had T3, and 7.2% (n=3) had T4 tumours. The pathological N stages of the patients were determined as N0 in 17.1% (n=7), N1 in 31.7% (n=13), and N2 in 51.2% (n=21). All patients were non-metastatic. When examining the TNM stages, 4.9% (n=2) were classified as Stage 1A, 9.8% (n=4) as Stage 1B, 2.4% (n=1) as Stage 2A, 29.3% (n=12) as Stage 2B, and 53.6% (n=22) as Stage 3.

The GRIm score was calculated as low (GRIm score: 0 and 1) in 65.9% (n=27) of patients, while it was high (GRIm score: 2 and 3) in 34.1% (n=14).

Relationship Between GRIm Score, Quantitative 18F-FDG PET/CT Data, and Histopathological Features
No significant relationship was observed between the low and high GRIm score patient groups concerning either the quantitative 18F-FDG PET/CT parameters or the histopathological features (p>0.05). The quantitative data pertaining to these findings are summarized in Table 1 and Table 2.

Table 1 Relationship between quantitative 18F-FDG PET/ CT and quantitative histopathological parameters with GRIm score

Table 2 Relationship between categorical histopathological features and GRIm score (n=41)

Survival Analysis of Variables
Before performing survival analysis, continuous numerical variables were characterized by their median (min - max) values. The median values for SUVmax, SUV mean, MTV, TLG, and Dpat were determined to be 5.4 (3.37 -11.0), 3.06 (2.0-6.32), 18.02 (3.5-54.84), 52.67 (11.94-195.23), and 30 (6.0-85.0), respectively. These variables were then categorized into two groups: Those equal to or below the median value, and those above the median value. Survival analysis was conducted after all data were organized as categorical variables.

For the entire cohort, the mean overall survival was calculated as 12.97 months (95% CI: 9.5-16.3 months), and the median overall survival was 10.6 months (95% CI: 6.2-14.9 months).

When all variables were evaluated, a significant relationship was found between GRIm score, LVI status, and survival. The low GRIm score group comprised 27 patients; 21 of them died during the follow-up period, while 6 were still alive. In contrast, the high GRIm score group consisted of 14 patients, all of whom died within the follow-up period. The estimated mean overall survival for the low GRIm score group was 16.7 months (95% CI: 12.3-21.2 months), with a median survival of 16.0 months (95% CI: 10.3-21.7 months). Conversely, the high GRIm score group had an estimated mean overall survival of 6.2 months (95% CI: 2.9-9.4 months), and a median survival of 3.3 months (95% CI: 0?12.1 months). The Log-Rank (Mantel-Cox) test, which assesses the equality of survival distributions between groups, revealed a statistically significant difference (Chi-square=13.020, df=1, p<0.001). This finding indicates that the GRIm score creates a statistically significant difference in survival duration, demonstrating that patients with a low GRIm score have significantly longer survival times compared to those with a high GRIm score (Table 3, Fig. 1).

Table 3 Relationship of quantitative 18F-FDG PET/CT parameters, histopathological features, and GRIm score with overall survival

Fig. 1. Overall survival (OS) analysis by Kaplan-Meier method, confirming the significant prognostic value of the gustave roussy immune score (GRIm score) and lymphovascular invasion (LVI) status. Both variables were further validated as independent poor prognostic factors in multivariate Cox regression analysis (Log-rank p<0.001 for GRIm score, p=0.017 for LVI).

There were 8 patients in the group without LVI; 6 of these died during the follow-up period, while 2 remained alive. Conversely, the LVI-positive group consisted of 33 patients, with 29 dying during the follow- up and 4 surviving. The estimated mean overall survival for the LVI-negative group was 22.6 months (95% CI: 13.5-31.7 months), and the median survival was 17.6 months (95% CI: 13.5-21.6 months). In contrast, the LVI-positive group had an estimated mean overall survival of 10.6 months (95% CI: 7.4-13.8 months), and a median survival of 9.400 months (95% CI: 6.0-12.7 months). The Log-Rank (Mantel-Cox) test, used to compare survival distributions between groups, revealed a statistically significant difference (Chi-square=5.659, df=1, p=0.017). This finding indicates that the presence of LVI significantly impacts survival duration, showing that patients without lymphovascular invasion have considerably longer survival times than those with it (Table 3, Fig. 1).

No significant relationship was observed between other parameters and survival, and these findings are summarized in Table 3.

Multivariate Survival Analysis
Multivariate Cox Proportional Hazard Regression analysis was performed including variables with prognostic significance in the univariate analysis (GRIm Score and LVI) and other clinically relevant factors (Grade, pT/pN stages, SUV_max categories).

The analysis demonstrated that a High GRIm Score (HR=10.258, 95% CI: 2.935?35.852, p<0.001) and the presence of LVI (HR=14.899, 95% CI: 1.555-142.720, p=0.019) were independent poor prognostic factors for overall survival. Other parameters included in the multivariate model were not found to have significant independent prognostic value (Table 4).

Table 4 Multivariate cox proportional hazard regression analysis for overall survival

Discussion

In our study, a significant relationship was observed between the GRIm score and the overall survival durations of operable PDAC patients. Patients with a high GRIm score were found to have significantly shorter mean survival times compared to those with a low GRIm score (6.2 months vs. 16.7 months). A study by Ma et al.,[12] which compared different immune scoring systems in advanced pancreatic cancer patients, also demonstrated that the GRIm score had higher predictive value and that patients with a high GRIm score had shorter overall survival times. In a study conducted by Basoglu et al.[11] on operable PDAC patients, it was also shown that patients with a high GRIm score had shorter survival. Our study's results align with existing literature, indicating that an elevated GRIm score is a poor prognostic factor for patient outcomes.

Albumin, a key component of the GRIm score, reflects a patient"s nutritional status and the inflammatory process. Decreased nutrition and increased inflammation, often associated with disease severity, progression, and prognosis, reduce albumin synthesis. LDH, another component, is a marker of tumour hypoxia. Elevated LDH levels indicate increased tumour hypoxia and heightened macrophage-mediated angiogenesis and invasion. NLR (neutrophillymphocyte ratio), the third marker, signifies a rise in neutrophil count alongside a decrease in lymphocyte count. Neutrophils, through the mediators they secrete, can suppress lymphocyte proliferation and lead to a reduction in lymphocyte numbers, particularly CD8+ tumour-infiltrating lymphocytes. This imbalance disrupts immune system homeostasis and weakens the anti-tumoural effect on tumour tissue. [11,12,15-17] The GRIm score has the potential to offer a more holistic approach to prognosis by integrating these individual biomarkers.

Furthermore, our study found no correlation between the GRIm score and quantitative 18F-FDG PET/CT parameters or histopathological features. This suggests that the GRIm score provides a distinct prognostic insight for operable PDAC patients, separate from the tumour"s metabolic and morphological characteristics. This finding was strongly validated by our Multivariate Cox Regression Analysis, which demonstrated that the GRIm score is an independent poor prognostic factor for overall survival (HR=10.258, p<0.001). This result reinforces the potential of the GRIm score as a simple and cost-effective tool that complements complex imaging and pathology data in predicting prognosis.

In our study, we concluded that patients with histopathological LVI had a shorter mean overall survival duration compared to those without LVI (22.6 months vs. 10.6 months). Furthermore, the Multivariate Cox Regression Analysis confirmed LVI as a second independent poor prognostic factor (HR=14.899, p=0.019). Many studies in the literature have shown that the presence of LVI is associated with shorter overall survival in resected PDAC patients, and our study reached similar conclusions. A study by Takahashi et al.[18] reported that the median overall survival for patients with LVI was 17 months, while it was 22.5 months for those without LVI. Similarly, a study by Chatterjee et al.[19] demonstrated that patients with muscular venous invasion, a specific histological subtype of LVI, had shorter overall survival compared to those without LVI or those with invasion confined to non-muscular lymphovascular spaces. A recently published meta-analysis by Javed et al.[20] also identified LVI as being associated with long-term survival in resected PDAC patients, demonstrating shorter long-term survival in those with LVI.

In our study, no significant relationship was found between quantitative 18F-FDG PET/CT parameters and overall survival duration. However, literature reviews indicate that tumours with higher SUVmax, MTV, and TLG values are associated with shorter overall survival.[6,21,22] Our results, however, contradict findings in the existing literature. Possible reasons for this discrepancy might include the relatively smaller patient cohort in our study. Furthermore, our study specifically included patients with a clear indication for surgical resection who had not received any prior treatment, while excluding those with borderline surgical indications or those who underwent neoadjuvant therapy before surgery. This selection criterion likely resulted in a more homogeneous patient cohort. Variations in the 18F-FDG PET/CT imaging protocols used across studies could also account for these differences.

In our study, we found no significant association between overall survival and other histopathological features, aside from LVI. However, existing literature consistently reports that advanced pT, pN, and TNM stages, higher tumour grade, and the presence of PNI are all poor prognostic indicators, with numerous studies demonstrating significantly shorter overall survival in tumours exhibiting these characteristics. [23-28] We believe the primary reasons we couldn't replicate these findings in our study are the relatively small patient cohort and the homogeneous distribution of most patient characteristics. For instance, 87.2% of patients had Grade 2 tumours, and PNI was present in 95% of patients. While a large proportion of patients (65.9%) had T2 tumours, T1 (4.9%) and T4 (7.3%) tumours were less common. This distribution of histopathological features was relatively homogeneous, which we believe was insufficient to yield statistical significance. Despite a more heterogeneous distribution for the pN stage, we hypothesize that the small patient number still prevented us from achieving statistically significant results.

Limitations of Our Study
Our study has several limitations, primarily due to its retrospective nature and the relatively small patient cohort. Regarding the clinical management, we only included patients eligible for upfront surgery who did not receive neoadjuvant therapy. While this selection created a homogeneous cohort, the exact details of received adjuvant chemotherapy (e.g., specific regimen, duration, tolerance) were not standardized or prospectively collected. However, we state in the Methods that all patients were routinely referred for standard adjuvant protocols. The most critical missing clinical data is the information regarding patient comorbidities. This absence is a limitation because underlying systemic conditions, independent of the tumour, can significantly influence the GRIm score components, particularly albumin and NLR, potentially confounding the survival analysis. Future prospective studies must be conducted to validate the GRIm score's prognostic value while prospectively collecting and controlling for these vital clinical variables.

Conclusion

Ethics Committee Approval: The study was approved by the Sivas Cumhuriyet University Faculty of Medicine Ethics Committee (no: 2025-04/88, date: 24/04/2025).

Informed Consent: Informed consent was obtained from all participants.

Conflict of Interest Statement: The authors have no conflicts of interest to declare that are relevant to the content of this article.

Funding: The authors did not receive support from any organization for the submitted work.

Use of AI for Writing Assistance: No AI technologies utilized. Author Contributions: Concept - K.Ş., S.K., H.Ö., Z.H.; Design - K.Ş., S.K., H.Ö., Z.H.; Supervision - K.Ş., S.K., H.Ö., Z.H.; Data collection and/or processing - K.Ş., S.K.; Data analysis and/or interpretation - K.Ş., S.K.; Literature search - K.Ş., S.K., H.Ö., Z.H.; Writing - K.Ş., Z.H.; Critical review - Z.H., H.Ö.

Peer-review: Externally peer-reviewed.

References

1) Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74(3):229-63.

2) Arnone A, Laudicella R, Caobelli F, Guglielmo P, Spallino M, Abenavoli E, et al. Clinical impact of 18FFDG PET/CT in the diagnostic workup of pancreatic ductal adenocarcinoma: A systematic review. Diagnostics 2020;10(12):1042.

3) Siegel RL, Miller KD, Wagle NS, Jemal A. Cancer statistics, 2023. CA Cancer J Clin 2023;73(1):17-48.

4) Wartski M, Sauvanet A. 18F-FDG PET/CT in pancreatic adenocarcinoma: A role at initial imaging staging? Diagn Interv Imaging 2019;100(12):735-41.

5) Ahn SJ, Park MS, Lee JD, Kang WJ. Correlation between 18F-fluorodeoxyglucose positron emission tomography and pathologic differentiation in pancreatic cancer. Ann Nucl Med 2014;28(5):430-5.

6) Kitasato Y, Yasunaga M, Okuda K, Kinoshita H, Tanaka H, Okabe Y, et al. Maximum standardized uptake value on 18F-fluoro-2-deoxy-glucose positron emission tomography/computed tomography and glucose transporter-1 expression correlates with survival in invasive ductal carcinoma of the pancreas. Pancreas 2014;43(7):1060-5.

7) Zhu D, Wang L, Zhang H, Chen J, Wang Y, Byanju S, et al. Prognostic value of 18F-FDG-PET/CT parameters in patients with pancreatic carcinoma: A systematic review and meta-analysis. Medicine 2017;96(33):e7813.

8) Lewis R, Drebin JA, Callery MP, Fraker D, Kent TS, Gates J, et al. A contemporary analysis of survival for resected pancreatic ductal adenocarcinoma. HPB 2013;15(1):49-60.

9) Shi H, Wei Y, Cheng S, Lu Z, Zhang K, Jiang K, et al. Survival prediction after upfront surgery in patients with pancreatic ductal adenocarcinoma: Radiomic, clinic-pathologic and body composition analysis. Pancreatology 2021;21(4):731-7.

10) Bigot F, Castanon E, Baldini C, Hollebecque A, Carmona A, Postel-Vinay S, et al. Prospective validation of a prognostic score for patients in immunotherapy phase I trials: The Gustave Roussy Immune Score (GRIm-score). Eur J Cancer 2017;84:212-8.

11) Basoglu T, Babacan NA, Ozturk FE, Arikan R, Demircan NC, Telli TA, et al. Prognostic value of Gustave Roussy immune score in operable pancreatic adenocarcinoma. Indian J Cancer 2023;60(2):179-84.

12) Ma LX, Wang Y, Espin-Garcia O, Allen MJ, Jang GH, Zhang A, et al. Systemic inflammatory prognostic scores in advanced pancreatic adenocarcinoma. Br J Cancer 2023;128(10):1916-21.

13) Im HJ, Bradshaw T, Solaiyappan M, Cho SY. Current methods to define metabolic tumor volume in positron emission tomography: Which one is better? Nucl Med Mol Imaging 2018;52(1):5-15.

14) Shin DW, Lee JC, Kim J, Woo SM, Lee WJ, Han SS, et al. Validation of the American Joint Committee on Cancer 8th edition staging system for the pancreatic ductal adenocarcinoma. Eur J Surg Oncol 2019;45(11):2159-65.

15) Alagappan M, Pollom EL, von Eyben R, Kozak MM, Aggarwal S, Poultsides GA, et al. Albumin and neutrophil- lymphocyte ratio predict survival in patients with pancreatic adenocarcinoma treated with SBRT. Am J Clin Oncol 2018;41(3):242-7.

16) Albain KS, Crowley JJ, LeBlanc M, Livingston RB. Survival determinants in extensive-stage non-small-cell lung cancer: The Southwest Oncology Group experience. J Clin Oncol 1991;9(9):1618-26.

17) Coussens LM, Werb Z. Inflammation and cancer. Nature 2002;420(6917):860-7.

18) Takahashi H, Katsuta E, Yan L, Tokumaru Y, Katz MHG, Takabe K. Transcriptomic profile of lymphovascular invasion, a known risk factor of pancreatic ductal adenocarcinoma metastasis. Cancers 2020;12(8):2033.

19) Chatterjee D, Rashid A, Wang H, Katz MH, Wolff RA, Varadhachary GR, et al. Tumor invasion of muscular vessels predicts poor prognosis in patients with pancreatic ductal adenocarcinoma who have received neoadjuvant therapy and pancreaticoduodenectomy. Am J Surg Pathol 2012;36(4):552-9.

20) Javed AA, Mahmud O, Fatimi AS, Habib A, Grewal M, He J, et al. Predictors for long-term survival after resection of pancreatic ductal adenocarcinoma: A systematic review and meta-analysis. Ann Surg Oncol 2024;31(7):4673-87.

21) Lee JW, Kang CM, Choi HJ, Lee WJ, Song SY, Lee JH, et al. Prognostic value of metabolic tumor volume and total lesion glycolysis on preoperative 18F-FDG PET/ CT in patients with pancreatic cancer. J Nucl Med 2014;55(6):898-904.

22) Shimomura A, Oshima M, Suto H, Matsukawa H, Kondo A, Ando Y, et al. Prognostic significance of 18F-FDG PET/CT and tumor metabolic changes in patients with pancreatic ductal adenocarcinoma. Anticancer Res 2024;44(8):3321-30.

23) Felsenstein M, Lindhammer F, Feist M, Hillebrandt KH, Timmermann L, Benzing C, et al. Perineural invasion in pancreatic ductal adenocarcinoma: A saboteur of curative intended therapies? J Clin Med 2022;11(9):2367.

24) Garcea G, Dennison AR, Pattenden CJ, Neal CP, Sutton CD, Berry DP. Survival following curative resection for pancreatic ductal adenocarcinoma: A systematic review of the literature. JOP 2008;9(2):99-132.

25) Javed AA, Habib A, Mahmud O, Fatimi AS, Grewal M, Mughal N, et al. Prognostic factors in localized pancreatic ductal adenocarcinoma after neoadjuvant therapy and resection: A systematic review and metaanalysis. J Natl Cancer Inst 2025;117(5):840-67.

26) Javed AA, Rompen IF, van Goor I, Stoop TF, Andel P, Mahmud O, et al; Dutch Pancreatic Cancer Group and the PANC-PALS Consortium. Poor prognostic factors in long-term survivors of resected pancreatic ductal adenocarcinoma: An international, multicenter cohort study. Ann Surg 2024;2024: SLA.0000000000006539.

27) Nozzoli F, Catalano M, Messerini L, Cianchi F, Nassini R, De Logu F, et al. Perineural invasion score system and clinical outcomes in resected pancreatic cancer patients. Pancreatology 2024;24(4):553-61.

28) Piper M, Ross RB, Hu J, Watanabe S, Knitz M, Mehrotra S, et al. Vasculitis, CA19-9, and perineural invasion differentially predict response and surgical outcome in pancreatic ductal adenocarcinoma. Int J Radiat Oncol Biol Phys 2023;116(3):627-39.