Which Factors Are Associated With Positive Resection Margin in Pancreatic Ductal Adenocarcinoma for Different Surgical Procedures? Results Based on a Standard Pathological Evaluation System-retrospective Cohort Study

Background/Objectives: The present study identied the independent risk factors of R1 resection in pancreaticoduodenectomy (PD) and distal pancreatosplenectomy (DP) for patients with pancreatic ductal adenocarcinoma (PDAC). Methods (cid:0) Consecutive patients who were operated from December 2017 to 2018 with curative intent were analyzed retrospectively. A standardized pathological examination with digital whole-mount slide images (DWMSIs) was utilized for the resection margin status. R1 was dened as microscopic tumor inltration within 1 mm to the resection margin. The potential risk factors of R1 resection for PD and DP were analyzed separately by univariate and multivariate logistic regression analyses. Results (cid:0) For the 240 patients who underwent PD, and the 142 patients who underwent DP, the R1 resection rates were 30.8% and 35.6 %, respectively. Univariate analysis on risk factors of R1 resection for PD were tumor location, absence of tumor necrosis, N staging, TNM staging, and surgical type of PD; while those for DP were nerve invasion, T staging, and TNM staging. Multivariate logistic regression analysis showed that the risk factors of R1 resection for PD were tumor location (neck vs. head; uncinate process vs. head) and N staging, while those for DP were T staging . Conclusions (cid:0) The location of tumor in the neck and uncinate process, and N1/2 staging were independent risk factors of R1 resection for PD; while those for DP were T3/4 staging. on demographics and potential risk factors between the R0 and R1 resection groups of patients Patients with R1 resection had a high incidence of nerve invasion (96.2% vs. 85.1%, P = 0.041). The prevalence of T3 cancer (76.9% vs. 42.6%, P < 0.001) and TNM III cancer (34.6% vs. 19.1%, P = 0.038) were signicantly higher in the R1 resection group as compared to the R0 group. The prevalence of N1/2 cancer (69.2% vs. 56.4%, P = 0.095) was higher in the R1 resection group. Multivariable analyses showed that T3 staging (OR = 5.26, P < 0.001) was an independent risk factor of R1 resection.


Introduction
Pancreatic cancer is a lethal disease, being characterized by increasing incidences and mortality rates, and with a dismayed 5-year survival rate of < 9% [1,2]. Currently, radical surgery is the only potentially curative therapy for pancreatic ductal adenocarcinoma (PDAC) [3]. However, only about 15-20% of patients are eligible for upfront radical resection at the time of diagnosis [4,5]. Even when surgery is followed by chemotherapy or radiotherapy, the majority of patients develop tumor recurrence within 2 years after surgery [6]. Tumor relapse can occur either locally or as a metastatic disease [7]. Recurrence relating to microscopic positive resection margins (R1) is common, accounting for 46.7% of patients with pancreatic body/tail and 8.2% with pancreatic head PDAC [8]. Recent studies showed that the R1 margin status signi cantly lowered the disease-free survival (DFS) and the overall survival (OS) compared to the clear resection (R0) margin status after pancreaticoduodenectomy (PD) [9] or distal pancreatosplenectomy (DP) [10]. Thus, to achieve R0 resection is critical to prognosis of patients. The risk factors of R1 resection should be identi ed to improve patients' management.
The independent risk factors of R1 resection in PD and DP for PDAC have not been extensively investigated. Furthermore, the de nition of R1 and the protocol on pathological examination of resected specimens have not been standardized [11][12][13]. In our high volume pancreatic center, R1 is de ned as detection of any cancer cells within 1 mm of one or more surfaces or margins in the resected specimens. Meanwhile, the Leeds Pathology Protocol (LEEPP) was used to standardize our protocol of pathological examination since November 2016. The pathological reports were rigorously led following the latest National Comprehensive Cancer Network (NCCN) guidelines (Version 3, 2019) for pancreatic adenocarcinoma. The R status has since been assessed by digital whole-mount slide images (DWMSIs) which has the advantages of identifying the R status more precisely and comprehensively compared to the ordinary microscopic study. The present study, based on a standardized protocol of pathological examination and R1 de nition, aimed to identify independent risk factors of R1 resection for PD and DP in patients with PDAC. These ndings, hopefully, could provide useful information for clinical practice.

Study population
This retrospective study included consecutive patients who were diagnosed to have PDAC by pathological examination after pancreatic resectional surgery between December 2017 and 2018. The inclusion criteria were: (1) PD or DP with curative intention, and (2) underwent a standardize pathological protocol for the resected specimen and a standard reporting record for resection margin (R) status according to the NCCN guidelines for pancreatic cancer. The exclusion criteria were: (1) total pancreatectomy (TP); (2) macroscopic evidence of margin involvement (R2); (3) patients received neoadjuvant chemotherapy; (4) resection for metastatic disease. The Changhai Hospital Review Board approved the study, and no additional informed consent was required to review the patients' medical records.

Pathological Examination
The Leeds Pathology Protocol (LEEPP) was routinely used [14,15] for the pathological examination.
For PD specimens: Brie y, after multicolor inking of the posterior, medial margin, and anterior surfaces of the pancreatic head, the specimen of PD was serially sliced in an axial plane, perpendicular or paralleled to the longitudinal duodenal axis.
For DP specimens: After multicolor inking of the posterior and anterior surfaces of the pancreatic body and tail, the specimen of DP was serially sliced in a plane parallel to the main pancreatic duct or the pancreatic transection margin when the tumor invades to the surface of the pancreas according to the NCCN guidelines.
The entire specimen was sliced in 5-mm-thick sections, which resulted in 10-35 (average 24.5 ± 6.7) formalin-xed para n-embedded (FFPE) block for each specimen. Then, each FFPE block was cut in 4-µm-thick sections on whole-tissue glass slides measuring 7.8 × 5.4 cm 2 . These slides were scanned by the Hamamatsu S60 Whole Slide Scanner (Hamamatsu Photonics, Hamamatsu city, Japan) to obtain the DWMSIs with an average of 6.47 GB in le size [16]. The DWMSIs can also be observed by the NanoZoomer Digital Pathology (NDP) viewer2 version 2.7.25 ( Fig. 1), and the distance between the resection margins and the tumor cells can be measured precisely down to 0.01 mm on the screen ( Supplementary Fig. 1).

Classi cation and De ning of the R Status
The pathological reports on a detailed form were reassessed retrospectively. The slides were rechecked if the reports were not clearly depicted. According to the NCCN guidelines, the R status records for PD were comprised of the transection margin, the anterior surface, the posterior margin, the bile duct margin, the enteric margin, and the medial margin, which included the SMV (superior mesenteric vein)/PV (portal vein) resection margin and the SMA (superior mesenteric artery) resection margin. There were just three margins for DP. R1 was de ned as one or more cancer cells within 1 mm of one or more surfaces or margins (R1 < 1 mm) or at the surface or margin (R1-direct). R0 was de ned as > 1 mm from any surface or margin (R0 > 1 mm).

Variables Analyzed
For all the patients, the following demographic, clinical, and pathological variables, including sex, age, preoperative carbohydrate antigen (CA) 19 − 9, tumor location (head, neck, or uncinate process for PD; neck and body, or body and tail for DP), necrosis, nerve invasion, microvascular invasion, duodenum invasion, duodenal papilla invasion, common bile duct invasion, fat invasion around pancreas, tumor differentiation grade (G1/2 or G3/4), vascular resection. The TNM staging was also recorded using the 8th edition of the AJCC Cancer Staging Manual for Pancreatic Cancer. Resectability status was recorded according to the NCCN guideline for pancreatic adenocarcinoma (version 3. 2019).

Statistical Analysis
Categorical data were presented as percentage proportion, and continuous data were transformed into categorical data by the median value. For incomplete data, a dummy variable classi ed as "missing" was included in the analysis. To compare the differences in the categorical data between the two groups, the Fisher's exact test or the χ 2 test was used as appropriate. All variables found to be statistically signi cant (P < 0.1) on univariate analysis were subsequently included in a multivariate binary logistic regression model to determine the independent risk factors associated with R1 resection. For all the analyses, a two-tailed P < 0.05 was considered as statistically signi cant. The analyses were performed using the SPSS version 22.0 (SPSS, Chicago, IL, USA).

Patient Characteristics
From December 2017 to 2018, 457 patients underwent pancreatic surgery and were diagnosed to have PDAC on pathological examination. A total of 71 (15.5%) patients were excluded: 6 patients because of TP, 7 patients due to a lack of standard records for the R status, 9 patients because of R2, 41 patients because of neoadjuvant chemotherapy before surgery, and 8 patients because of resection for metastatic disease, which were discovered during the operations. The remaining 386 patients were included in this study (M:F = 1.47:1; age range 28-83, mean age 61.9±9.5 years), with 240 patients underwent PD and 146 patients underwent DP. The R status of the two groups of patients was assessed using DWMSIs. (Flowchart in Fig. 2) No signi cant differences were detected in the baseline data between R0 and R1 resections in the patients who underwent PD and DP, respectively.

Discussion
Previous reports have shown a close relationship between R1 resection and OS (or DFS) in PDAC patients [10,[17][18][19]. Even for patients who underwent PD for PDAC with positive resection margins shown on frozen section, further surgical resection to achieve R0 did not have any signi cant positive impact on OS [20]. These studies suggested that it is vital to achieve R0 resection on the rst attempt of surgical resection. Knowledge on independent risk factors of R1 resection can help surgeons to make better decisions on surgical treatment strategies.
Studies on independent risk factors of R1 resection have been made di cult because of the differences in the de nition between the Union for International Cancer Control (UICC) criteria (R1: 0 mm de nition) and the UK Royal College of Pathologists (RCPath) criteria (R1: 1 mm de nition) [21]. Furthermore, most studies on risk factors of R1 resection did not apply a standard pathological examination protocol, thus making any comparisons among these studies impossible. Also, most studies focused on pancreatectomy did not discriminate between PD and DP. The risk factors of R1 resection in these two different surgical operations for PDAC differ signi cantly.
With accurate assessment of resection margins by a standardized pathological examination protocol using LEEPP and DWMSIs, and with a single de nition of R1 using the 1 mm de nition, our study showed marked differences in the R1 resection rates for PD and DP when compared with the published rates of R1 resection [22][23][24][25][26]. In the current study, the rates of R1 resection were similar to the recently reported data for patients who underwent upfront surgery [27]. Previous studies reported R1 resection was frequently present in PV/SMV margins [28] and SMA margins [29], which was closely related to prognosis [30]. Our data also showed similar results in PD. Previous studies who demonstrated lesions in the neck (OR = 5.48) or uncinate process (OR = 2.996) [11], tumor size > 30 mm (OR = 1.13) [12] and grade 3 tumors (OR = 4.05) [28] were independent risk factors of R1 resection. However, there were no studies to clarify independent risk factors separately for R1 in PD and DP. The NCCN guidelines for pancreatic adenocarcinoma recommend operation and resection margins should be assessed separately in PD and DP. Our study supported neck/uncinate tumor, and the N1/2 stage were independent risk factors of R1 resection for PD, and T3 and surgery type were independent risk factors for DP.
The pancreatic neck is a narrow anatomical region between the pancreatic head and body, lying just anterior to the PV, gastroduodenal artery, and common hepatic artery (CHA) [31]. The uncinate process is closely related to the SMA and the SMV [32]. As a consequence, pancreatic neck or uncinate process cancer are prone to invade these major vessels to become either a borderline or unresectable tumor. R0 resection is technically di cult with involvement of theses vessels. In our study, neck tumor had the highest R1 resection rate, followed by uncinate process tumors. In addition, the most frequently R1 invaded margin in PD (Fig. 3A-F)  Pancreatic body and tail tumors present late as they do not have symptoms in the early stages [33]. Thus, the pancreatic body and tail tumors are likely to invade beyond the pancreas. Our study suggested that T3 was an independent risk factor of R1 resection.
The present study has several limitations. First, this retrospective study has the intrinsic defects of any retrospective study. Second, the de nition of R1 resection margin and the standard procedure for pathological examination of resection margin used in this study made it di cult to compare our results with other published studies. Third, long-term surgical outcomes could not be obtained due to the short study period.

Conclusions
A standardized pathological examination using the LEEPP, DWMSIs and the de nition of R1 resection margin within 1 mm to the tumor were used in this study to analyze independent risk factors for PD and DP separately.

Declarations Ackowledgments
We would like to thank Prof. Wan Yee Lau for English language polishing.
Authors' contributions LB,GSW and SXH contributed to analysis and interpretation of data, drafting the article and nal approval of the version to be published. NCM, GSZ, LG and NCR contributed to conception and design of data, revising the article critically for important intellectual content and nal approval of the version to be published. JH, LWY and JG contributed to analysis and interpretation of data, drafting the article and nal approval of the version to be published. All authors read and approved the nal manuscript.

Availability of data and materials
Please contact the corresponding authors for data requests.

Ethics approval and consent to participate
The study was approved by the Institutional Review Board of Changhai Hospital (Shanghai Changhai Hospital Ethics Committee, reference number: CHEC2020-043).

Consent for publication
All patients enrolled in the study signed the consent for publication. The written consent is available for review by the Editor-in-Chief of this journal.
Competing interests