Journal of Gastrointestinal Surgery

, Volume 16, Issue 10, pp 1875–1882

Tumor Infiltration in the Medial Resection Margin Predicts Survival After Pancreaticoduodenectomy for Pancreatic Ductal Adenocarcinoma

Authors

  • Yaojun Zhang
    • HPB Surgical Unit, Department of Surgery and CancerImperial College
  • Adam E. Frampton
    • HPB Surgical Unit, Department of Surgery and CancerImperial College
  • Patrizia Cohen
    • Department of Histopathology, Imperial College Healthcare NHS TrustHammersmith Hospital
  • Charis Kyriakides
    • HPB Surgical Unit, Department of Surgery and CancerImperial College
  • Jan J. Bong
    • HPB Surgical Unit, Department of Surgery and CancerImperial College
  • Nagy A. Habib
    • HPB Surgical Unit, Department of Surgery and CancerImperial College
  • Duncan R. C. Spalding
    • HPB Surgical Unit, Department of Surgery and CancerImperial College
  • Raida Ahmad
    • Department of Histopathology, Imperial College Healthcare NHS TrustHammersmith Hospital
    • HPB Surgical Unit, Department of Surgery and CancerImperial College
Original Article

DOI: 10.1007/s11605-012-1985-4

Cite this article as:
Zhang, Y., Frampton, A.E., Cohen, P. et al. J Gastrointest Surg (2012) 16: 1875. doi:10.1007/s11605-012-1985-4

Abstract

Background

Microscopic tumor involvement (R1) in different surgical resection margins after pancreaticoduodenectomy (PD) for pancreatic ductal adenocarcinoma (PDAC) has been debated.

Methods

Clinico-pathological data for 258 patients who underwent PD between 2001 and 2010 were retrieved from a prospective database. The rates of R1 resection in the circumferential resection margin (pancreatic transection, medial, posterior, and anterior surfaces) and their prognostic influence on survival were assessed.

Results

For PDAC, the R1 rate was 57.1 % (48/84) for any margin, 31.0 % (26/84) for anterior surface, 42.9 % (36/84) for posterior surface, 29.8 % (25/84) for medial margin, and 7.1 % (3/84) for pancreatic transection margin. Overall and disease-free survival for R1 resections were significantly worse than those for R0 resection (17.2 vs. 28.7 months, P = 0.007 and 12.3 vs. 21.0 months, P = 0.019, respectively). For individual margins, only medial positivity had a significant impact on survival (13.8 vs. 28.0 months, P < 0.001), as opposed to involvement in the anterior (19.7 vs. 23.3 months, P = 0.187) or posterior margin (17.5 vs. 24.2 months, P = 0.104). Multivariate analysis demonstrated R0 medial margin was an independent prognostic factor (P = 0.002, HR = 0.381; 95 % CI 0.207–0.701).

Conclusion

The medial surgical resection margin is the most important after PD for PDAC, and an R1 resection here predicts poor survival.

Keywords

Pancreatic ductal adenocarcinomaPancreaticoduodenectomyMedial resection marginPrognostic factorsSurvival

Introduction

Pancreatic cancer is the sixth most common cancer and fourth cause of cancer-related death in the Western world.1,2 Surgery remains the only chance for cure, with a dismal 5-year survival ranging from 0 to 5 % for pancreatic ductal adenocarcinoma (PDAC).3 Few patients (10–15 %) are potentially resectable because of advanced disease at presentation.1,2,46

Resection margin status is used to define the extent of resection in patients with PDAC.2,46 An R0 resection has no tumor present at the resection margins of the specimen. R1 is defined as microscopic involvement, while R2 is macroscopic involvement of a margin and definitely has a negative impact on outcome after surgery.4,6 With improved surgical techniques and accurate preoperative assessment, R2 resection should be avoided.4,6 The impact of R1 on outcome is debatable.711 Both ESPAC-1 and ESPAC-312,13 data reported resection margin status is not an independent risk factor for survival. The main reason for this disparity lies in the lack of international consensus regarding fundamental issues such as the definition of microscopic margin involvement, the constituents of the circumferential resection margin (CRM) in pancreaticoduodenectomy (PD) specimens, and a standardized protocol (SP) for the pathological examination of these specimens.14 In the USA, many pathologists consider a margin positive if tumor cells are present at the surface of the margin (i.e., tumor extending to the inked margin).15,16 Whereas in Europe, the R1 definition varies, but most regard an involved margin if tumor is present within 1 mm of the resection margin.5,14,17,18 Furthermore, the definition of the CRM is also variable. For example, the anterior margin is usually included in the examination protocol in Europe14,17,18 and Japan,19 but in the USA the notion of the anterior surface or margin has been introduced only recently.20 Reports from Europe have demonstrated the use of SP driven reporting for the pathological examination of the specimen correlates more accurately with treatment outcomes.17,18,21

Currently, resection margins are considered uniformly, with tumors at or close to any margin rated of equal prognostic significance. However, Jamieson and colleagues21 reported that positive mobilization margins alone do not influence survival following PD for PDAC. Furthermore, Schmidt et al. demonstrated that survival after PD was improved by extending resections to achieve negative margins,22 but Hernandez et al. did not.23 It is therefore crucial to recognize the prognostic influence of different resection margins and identify which is the most important.

The present study assessed the frequency of R1 status in the different surgical resection margins after PD to determine their effect on survival in PDAC.

Materials and Methods

Ethics Statement

The study was approved by our institutional review board (Hammersmith Hospital, Imperial College Healthcare NHS Trust), and informed written patient consent was obtained before treatment.

Patients

A retrospective study, based on prospectively collected data, was performed in our hospital. Data on patients who underwent PD, including standard pancreaticoduodenectomy (Whipple’s procedure) and pylorus-preserving pancreaticoduodenectomy (PPPD), for periampullary malignancies between January 2001 and December 2010 were retrieved. This study was limited to patients who underwent PD for PDAC; those patients with ampullary carcinoma, distal cholangiocarcinoma, duodenal carcinoma, neuroendocrine tumors, cystadenocarcinoma, metastatic tumors, and benign disease were excluded. Patients who underwent pancreatic resections other than PD (e.g., distal pancreatectomy or total pancreatectomy) were also excluded. Since this study aimed to analyze the prognostic factors for long-term survival, those patients who died perioperatively <30 days following surgery or during the original hospital stay of >30 days were excluded (n = 3). After standard preoperative evaluation, decision to offer surgery was made by our pancreatic multidisciplinary team (MDT), which includes surgeons, radiologists, gastroenterologists, and oncologists. No patients were treated with preoperative chemoradiation therapy. The criteria for resectability were (1) absence of extrapancreatic disease, (2) no involvement of superior mesentery artery (SMA) or celiac axis, and (3) no encasement of superior mesenteric vein (SMV). For patients with localized lesions adjacent to either SMV or portal vein (PV), resection and reconstruction of the SMV, PV, or superior mesenteric–portal vein (SMPV) confluence were performed.

Our pancreaticoduodenectomy surgical techniques have been previously described and have not changed significantly over the last decade.24,25 In brief, standard PD with resection of peripancreatic tissues and lymph nodes (N1) is performed without extended lymphadenectomy (N2/N3). Retroportal resection extends to tissues in the groove between the uncinate process of pancreas and SMA. Segmental resection of the SMV, PV, or SMPV confluence was performed by all surgeons, when the pancreatic head and/or uncinate process could not be separated from the vein without leaving gross tumor on the vessel or risking a venotomy. Arterial involvement was considered to be a contraindication to surgery. Transection margin frozen sections are performed with resection of the pancreatic body until a negative tumor status is obtained.

Postoperatively, all patients were considered for adjuvant therapy by the MDT. The options given included 5-FU with folinic acid, gemcitabine alone or in combination, paclitaxel, capecitabine, radiotherapy, and 5-FU with radiotherapy. Nine patients were considered unsuitable for adjuvant chemo-/chemoradiotherapy on the basis of poor performance status, and ten patients refused any adjuvant therapies after surgery.

Pathological Assessment

Gross and microscopic examination of PD specimens was performed according to the guidelines of the UK Royal College of Pathologists (RCP).17,21,26 Excluding the pancreatic transection, bile duct, gastric, and jejunal resection margins, the CRM in PD specimens consists of the anterior surface, medial margin, and posterior surface. Multicolor inking of the specimen is used to clearly identify the margins (Fig. 1). The medial margin has a shallow groove-like shape, and although the width and depth of the groove may vary, it is readily identified as it runs from the inferior to the superior border of the pancreatic head in a slightly curved fashion, passing underneath and behind the transected pancreatic neck. It is on this surface that segments of SMV, PV, or SMA are to be found. The anterior margin comprises of the pancreatic surface lying in front of the medial margin. The posterior margin comprises the usually smooth pancreatic surface lying behind the medial margin.
https://static-content.springer.com/image/art%3A10.1007%2Fs11605-012-1985-4/MediaObjects/11605_2012_1985_Fig1_HTML.gif
Fig. 1

Identification of all four margins of a resected pancreatic tumor specimen. Multicolor inking of the specimen is used to clearly identify the medial resection margin (green), anterior surface (red), posterior/retroperitoneal surface (blue), and pancreatic transection margin (black). The medial margin (white arrow and green area), which is the transection margin of the uncinate process, runs from the inferior to the superior border of the pancreatic head in a slightly curved fashion, passing underneath and behind the transected pancreatic neck. Segments of the SMV, PV, or superior mesenteric artery are found on this margin, and if the tumor has infiltrated into these vessels, vascular resection and reconstruction of the SMV, PV, or SMPV confluence are performed

In recent years, a novel histopathological dissection technique has been increasingly used in European centers and is based on serial slicing of the entire pancreatic head in a single axial plane (i.e., perpendicular to the longitudinal axis of the duodenum). It does not prescribe opening of the pancreatic or bile duct, leaving the CRM intact. The pancreatic head is sliced at 5 mm intervals producing 10–13 slices. These slices are embedded whole in megablocks allowing extensive views of the lesion and its relation to the entire CRM and key anatomical structures.17,21,26 Detailed photographs and figures of the CRM can be seen in the articles by Verbeke et al.17,27 and Jamieson et al.21 Furthermore, the Journal of Gastrointestinal Surgery has an excellent teaching video available online describing the standardized protocol for the pathological examination of PD specimens.28

Histological findings were reviewed and specimens were re-examined where necessary by two experienced pathologists (PC and RA) to validate the diagnosis, tumor characteristics, and R1 status of the four margins. Staging was based on the American Joint Committee of Cancer (AJCC) system.29 R1 resection was defined as tumors within 1 mm of a CRM or transection margin. R0 resection was defined as no microscopic or macroscopic tumor, and R2 as macroscopic residual tumor.

Follow-up

Clinical follow-up of patients included computed tomography (CT) every 3 months in the 1st year, 6 months in the 2nd year, and then yearly. If necessary, further imaging was performed to diagnose any metastasis and/or recurrence. Follow-up for surviving patients was until August 2011.

Date and causes of death were determined from phone interviews with the patient’s family practitioner, medical records, and death certificates. Overall survival (OS) was defined as the interval between the surgery and death or the last follow-up. Disease recurrence (local and/or distant) was defined as the interval between the date of surgery and the first time of detectable disease, as determined by radiological imaging.

Statistical Analyses

Statistical analyses were performed using SPSS 13.0 software (SPSS, Chicago, IL, USA). Comparisons between two groups were made using Student’s t test for continuous data and chi-square test for categorical data. The OS and disease-free survival (DFS) were calculated by Kaplan–Meier method (compared by log-rank test). Prognostic factors for OS and DFS were assessed by multivariate Cox proportional hazards regression analysis (using factors found at univariate analysis). Logistic regression was performed to examine the impact of perioperative variables on margin status. Results are given as mean ± SD or median (range). All tests were two sided, and a P < 0.05 was considered significant.

Results

Patient Characteristics

PD was attempted in 258 patients with 45 patients undergoing a palliative bypass and/or biopsy during that period; the resection rate was 82.6 % (213/258). PD for PDAC was performed in 89 patients (41.8 %, 89/213); the remainder underwent PD for other histological diagnoses. Five patients were excluded; two patients (2.2 %, 2/89) had an R2 resection, and three patients (3.4 %, 3/89) died postoperatively from multi-organ failure (n = 1) or sepsis from a pancreatic leak with delayed hemorrhage (n = 2). Therefore, data for 84 patients were analyzed.

There were 50 male and 34 female patients, with median age of 63.7 years (range, 40.7–79.5 years). The majority (60.7 %, 51/84) were referred with a biliary stent in situ. Pancreatic resection included Whipple’s procedure (n = 39, 46.4 %, 39/84) and PPPD (n = 45, 53.6 %, 45/84). Six patients (7.1 %, 6/84) had stage I, 17 (20.2 %, 17/84) had stage IIA, and 61 (72.6 %, 61/84) had stage IIB disease, respectively. Vascular resection was performed in 24 patients (28.6 %, 24/84; of which 11 cases had PV transection with venovenous anastomosis and 13 cases had PV side wall resection without anastomosis). For these 24 patients, 20 cases had true vascular wall invasion confirmed on histology and four cases had only inflammatory adherence. Adjuvant therapy after PD included chemotherapy (5-FU with folinic acid, n = 16; gemcitabine alone, n = 22; gemcitabine in combination, n = 20,) and chemoradiotherapy (5-FU with radiotherapy, n = 7). Adjuvant therapy was not given in 19 patients due to poor performance status (n = 9) and patient refusal (n = 10).

Resection Margin Status

R1 resection was reported in 48 patients (57.1 %, 48/84). The clinicopathological characteristics of patients are detailed in Table 1. Patients with R1 resection were more likely to have more advanced T stage (P = 0.003) and larger tumor size (P = 0.021) than patients with R0.
Table 1

Clinic-pathologic characteristics of patients with R1 and R0 resection margins

Variables

R0 (n = 36)

R1 (n = 48)

P

Age

64.0 ± 8.8

63.4 ± 10.2

0.788

Sex (M/F)

19/17

31/17

0.368

Preoperative drainage (no/yes)

13/23

20/28

0.772

Preoperative bilirubin (mmol/L)

50.0 ± 87.1

49.2 ± 63.5

0.961

Surgical procedure (Whipple/PPPD)

15/21

24/24

0.591

Vascular resection (no/yes)

26/10

34/14

1.000

Differentiation (well or moderate/poor)

7/29

8/40

0.967

Lymph node status (N0/N1)

10/26

13/35

1.000

T stage (1/2/3)

3/5/28

0/0/48

0.003

Tumor size (cm)

2.4 ± 0.8

2.8 ± 0.9

0.021

Perineural invasion (no/yes)

12/24

8/40

0.130

Lymphovascular invasion (no/yes)

13/23

9/39

0.124

Adjuvant therapy (yes/no)

27/9

38/10

0.851

PPPD pylorus-preserving pancreaticoduodenectomy

R1 resections consisted of anterior margin in 26 patients (31.0 %, 26/84), posterior margin in 36 (42.9 %, 36/84), medial margin in 25 (29.8 %, 25/84), pancreatic transection margin in 3 (7.1 %, 3/84), with no positive bile duct, gastric, or jejunal margins. Of the 48 patients with R1 resections, 17 patients (35.4 %, 17/48) had only a single margin involved, 21 (43.8 %, 21/48) had two margins involved, and 10 (20.8 %, 10/48) had three or more margins involved.

Overall Survival and Disease-Free Survival

Mean follow-up was 26.4 ± 2.0 months (range, 2.5–73.5 months). At final follow-up, there were 72 deaths, 11 surviving patients, and 1 lost to follow-up (overseas patient). The 1-, 3-, and 5-year OS were 80.6, 26.2, and 7.2 %, respectively, and the median OS 21.0 months. In subgroup analysis, the 1-, 3-, and 5-year OS for patients with R0 resection were 91.3, 37.6, and 11.3 %, respectively, and 72.6, 17.9, and 0 %, respectively, for patients with R1 resection. The median OS were 28.7 and 17.2 months, respectively. Patients with R1 resection had a significantly poorer OS than those with R0 resection (P = 0.007, Fig. 1).

When the OS was analyzed against the involved margins, the median OS for patients with R0 and R1 anterior margin were 23.3 and 19.7 months, respectively (P = 0.187); posterior margin 24.2 and 17.5 months, respectively (P = 0.104); and medial margin 28.0 and 13.8 months, respectively (P < 0.001). Analysis was not performed for pancreatic transection margin because only three cases were R1. The OS was significantly influenced by positivity in the medial resection margin but not in the anterior or posterior margin (Fig. 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs11605-012-1985-4/MediaObjects/11605_2012_1985_Fig2_HTML.gif
Fig. 2

Overall survival following pancreaticoduodenectomy for pancreatic ductal adenocarcinoma. a For all resection margins, the median overall survival for R0 resections (n = 36) was 28.7 months compared with 17.2 months for R1 resections (n = 48; P = 0.007). b For the medial margin, the median overall survival for R0 resections (n = 59) was 28.0 months compared with 13.8 months for R1 resections (n = 25; P < 0.001)

The 1-, 3-, and 5-year DFS were 63.6, 19.5, and 4.3 %, respectively, and the median DFS was 15.8 months. In subgroup analysis, the 1-, 3-, and 5-year DFS for patients with R0 resection were 71.3 , 24.2, and 8.1 %, respectively, and 53.7, 13.7, and 0 %, respectively, for patients with R1 resection. The median DFS were 21.0 and 12.3 months, respectively. Patients with R1 resection had a significantly shorter DFS than those with R0 resection (P = 0.019, Fig. 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs11605-012-1985-4/MediaObjects/11605_2012_1985_Fig3_HTML.gif
Fig. 3

Disease-free survival following pancreaticoduodenectomy for pancreatic ductal adenocarcinoma. (a) For all resection margins, the median disease-free survival for R0 resections (n = 36) was 21.0 months compared with 12.3 months for R1 resections (n = 48; P = 0.019). b For the medial margin, the median disease-free survival for R0 resections (n = 59) was 21.0 months compared with 8.4 months for R1 resections (n = 25; P < 0.001)

The median DFS for patients with R0 and R1 anterior margin were 16.6 and 12.2 months, respectively (P = 0.388); posterior margin 17.3 and 12.7 months, respectively (P = 0.225); and medial margin 21.0 and 8.4 months, respectively (P < 0.001). The DFS was significantly influenced by positivity in the medial resection margin but not in the anterior and posterior margin (Fig. 3).

Medial Resection Margin

The clinicopathological characteristics of patients with R0 and R1 resection in the medial margin are detailed in Table 2. Patients with a R1 medial margin were more likely to have a larger tumor (3.1 vs. 2.4 cm, P = 0.021) and more posterior margin involvement (P = 0.021) than patients with R0 medial margin. Interestingly, 44 % (11/25) with an R1 medial margin had additional vascular resection (Table 2), indicating that tumor has almost always invaded beyond the SMV/PV to involve the groove and tissue between the pancreas and SMA (i.e., medial RM).
Table 2

Clinic-pathologic characteristics of patients with R1 and R0 medial resection margin

Variables

R0 (n = 59)

R1 (n = 25)

P value

Age

62.7 ± 9.6

66.0 ± 9.1

0.146

Sex (M/F)

30/24

15/10

1.000

Preoperative drainage (no/yes)

22/37

11/14

0.740

Preoperative bilirubin (mmol/L)

46.9 ± 79.2

55.8 ± 61.4

0.619

Surgical procedure (Whipple/PPPD)

25/34

14/11

0.365

Vascular resection (no/yes)

46/13

14/11

0.076

Differentiation (well or moderate/poor)

10/49

5/20

0.739

Lymph node status (N0/N1)

18/41

5/20

0.472

T stage (1/2/3)

3/5/51

0/0/25

0.154

Tumor size (cm)

2.4 ± 0.8

3.1 ± 1.0

0.021

Perineural invasion (no/yes)

15/44

5/20

0.800

Lymphovascular invasion (no/yes)

19/40

3/22

0.098

Anterior margin (R0/R1)

44/15

14/11

0.154

Posterior margin (R0/R1)

39/20

9/16

0.021

Adjuvant therapy (yes/no)

48/11

17/8

0.293

Analysis of perioperative covariates revealed that the tumor size was the only determinant factor directly associated with R1 medial resection margin (P = 0.007, HR = 2.679; 95 % CI 1.304–5.504, Table S1).

Prognostic Factors for Survival

Univariate analysis showed OS was directly influenced by differentiation of tumor (P = 0.036), lymph node status (P = 0.015), tumor size (P = 0.046), and medial resection margin (P = 0.000; Table 3). Of note, for the 24 patients with large vessel involvement, who had vascular resection (true vascular wall invasion, n = 20; inflammatory adherence, n = 4), the median OS were 10.4 and 17.5 months, respectively (P = 0.339). When we compared true vascular wall invasion (n = 20) to no vascular wall invasion (n = 64), we found that OS was significantly worse (median OS 10.7 vs. 24.0 months, respectively; P < 0.001).
Table 3

Univariate and multivariate analyses of prognostic factors of overall survival following pancreaticoduodenectomy for pancreatic ductal adenocarcinoma

Variables

Median survival (months)

Univariate (log rank)

Multivariate (Cox regression)

Hazard ratio (95 % CI)

Age (≤60/>60 years)

26.1/18.3

0.143

  

Sex (M/F)

19.2/23.2

0.554

  

Preoperative drainage (no/yes)

23.3/19.0

0.526

  

Bilirubin (≤200/>200 mmol/L)

23.4/19.0

0.562

  

Surgical procedure (Whipple/PPPD)

23.3/19.7

0.622

  

Vascular resection (no/yes)

23.4/18.2

0.189

  

Differentiation (well or moderate/poor)

24.2/16.1

0.036

0.020

0.553(0.336–0.909)

Lymph node status (N0 /N1)

28.7/19.0

0.015

0.012

0.448(0.239–0.841)

T stage (1–2/3)

30.6/19.7

0.057

0.590

0.803(0.361–1.786)

Tumor size (cm)

28.7/19.1

0.046

0.018

1.842(1.112–3.057)

Perineural invasion (no/yes)

23.4/19.1

0.494

  

Lymphovascular invasion (no/yes)

24.7/14.0

0.065

0.501

0.817(0.453–1.473)

Anterior margin (R0/R1)

23.4/19.8

0.187

  

Posterior margin (R0/R1)

24.2/17.5

0.102

  

Medial margin (R0/R1)

28.0/13.9

0.000

0.002

0.381(0.207–0.701)

Adjuvant therapy (yes/no)

21.0/19.4

0.203

  

Multivariate analysis showed that histological differentiation (P = 0.020, HR = 0.553; 95 % CI 0.336–0.909), negative lymph node status (P = 0.012, HR = 0.448; 95 % CI 0.239–0.841), large tumor size (P = 0.018, HR = 1.842; 95 % CI 1.112–3.057) and R0 medial margin (P = 0.002, HR = 0.381; 95 % CI 0.207–0.701) significantly affected OS (Table 3).

Discussion

The variability of R1 resection rate in PDAC after PD has been extensively debated.10,14,17,18 Published R1 resection rates vary greatly ranging from 10 to 84 %718 and are lower than the loco-recurrence rates (67 to 87 %).10,14,17,18 Some authors believe that these studies were underreporting the involved resection margin, mostly due to a non-standardized protocol for the pathological examination and definition of microscopic margin involvement.14 Recently, Verbeke et al.17 reported their standardized protocol (SP) according to the guidelines of the UK RCP, which is similar to our protocol, showing an R1 rate of 84.6 % in resected T3 and T4 PDAC. Esposite et al.18 reported their similar protocol developed by the European Study Group for Pancreatic Cancer had an R1 rate of 76 % in resected T3 PDAC. We identified a comparable R1 rate of 67.1 % (48/76) for T3 disease indicating that, with an SP, the R1 rate is more reproducible and higher than in previous reports without SP.

Further debated is the impact of R1 resection on long-term outcomes.10,14,17,18 Many studies have failed to show R1 resection as an independent prognostic factor for long-term survival.1013 The majority of studies with low rates of R1 resection have failed to demonstrate that R1 status influenced outcome. A multivariate analysis by the University of Indiana did however demonstrate that achieving an R0 margin status influences actual long-term survival.30 The ESPAC-1 trial12 reported an R1 rate of 18 % which was a borderline but not an independent risk factor for survival. The studies with SPs, reporting a higher R1 rate, demonstrated R1 to be an independent prognostic factor, as shown in our study.17,18,21 Jamieson et al.21 and Verbeke et al.17 reported the R1 rate to be as high as 74 and 84.6 %, respectively, and both studies showed R1 resection impacted the survival significantly. More importantly, in these two studies, as well as in ours, the OS rates were similar to those with lower R1 rates, identifying more significant survival benefit of patients in the R0 subgroup. These data suggested that the confusion of the impact of R1 resection on long-term survival might have been due to the underestimated resection margin involvement and the incorrect reporting of R0 resection blurring the difference between the R0 and R1 subgroups.

Different resection margins should theoretically have different impacts on treatment outcome. As gastric, duodenal, jejunal, and bile duct resection margins are rarely reported positive,14,1719,21,23 in this study the four margins of the CRM were analyzed. With improvements in surgical techniques, standardization of PD resection and general utilization of intraoperative frozen section, the rate of R1 in the pancreatic transection margin has been reduced to <10 % (7.1 % in our cohort, 3/84 patients).14,1719,21,23 The other resection margins cannot be as easily assessed intra-operatively in this way, and this is perhaps reflected by the positive margins in those 25 patients with an R1 medial margin (i.e., 16 had a R1 posterior margin and 11 also had a R1 anterior margin), indicating that these tumors had spread microscopically as the resections were not R2 by visual assessment.

The anterior margin is not a true resection margin as the surgeon does not transect any tissues in this area. It has been an integral part of the Japanese Pancreas Society classification for many years, and the presence of tumor cells at this surface is likely to increase the risk of local recurrence.19,31,32 However, an R1 anterior surface has not been shown to be a predictor for survival.21 In our study, there was no difference in either OS or DFS between R0 and R1 anterior surface patients. The posterior surface is generally reported to be the most frequently involved margin with an R1 rate of 32.4 to 53.8 % (42.9 % in our cohort).14,17,18,21 The posterior margin is the mobilization margin between pancreas and retroperitoneal fat and is separated by developing embryological planes during PD. Our results demonstrated that R1 posterior margin does not negatively influence survival after PD for PDAC. Similar results were also reported by Jamieson et al.,21 although they grouped anterior, posterior, and duodenal serosal margin together as mobilization margins.

The medial margin is frequently confused, being defined as the “mesenteric,” “uncinate,” “SMV (SMA) groove,” or even “retroperitoneal” margin.10,14,33 In our study, medial margin refers to a true transection margin produced when surgically dividing the uncinate process as close to the mesenteric vasculature as possible. It is on this surface that segments of the SMV, PV, or SMA are found. If the tumor is adhered to major vessels, requiring vascular resection, it is generally regarded as the most difficult and challenging part during PD and may not result in a R0 margin. Our results show that the larger the tumor size, the more frequent is R1 resection in the medial margin. This is potentially due to more advanced disease rather than surgical technique.

Our study confirmed that the medial margin is the most important surgical resection margin predicting long-term OS and DFS for PDAC after PD. Poor survival was seen in patients with R1 compared to R0 medial margins (13.8 vs. 28.0 months), which was more significant than that for R1 in any resection margin (17.2 vs. 28.7 months), even though an R1 medial margin was present in fewer cases. Medial R1 resection was an independent prognostic factor in multivariate analysis, as R1 resection in anterior and posterior surfaces failed to predict long-term survival. Jamieson et al.21 also demonstrated that transection margins (medial, pancreatic, bile duct, gastric, and jejunal margins), but not mobilization margins (anterior, posterior, and duodenal serosal margins), influence survival following PD. Raut et al.10, however, failed to demonstrate significant impact of SMA margin status on survival, but in their study, the SMA resection margin did not include the SMV groove and the R1 rate was relatively low, only 16.7 % (60/360) for all margins and 14.7 % (53/360) for SMA margin.

Pancreatic resection with en bloc vascular resection is generally performed in experienced centers to ensure a complete resection, although systematic review reported that it has no impact on survival.34 Nakagohri et al. verified that survival rates were similar after PD with or without vascular resection, but microscopic invasion of the PV was significantly associated with poorer survival.35 Their findings support our results that R1 tumor in the medial margin predicts poorer survival.

In conclusion, we show that with an SP for pathological reporting, the medial surgical margin is the most important resection margin in PD for PDAC and R1 resection here predicts poor survival.

Further Considerations

First, the hazard of an R1 medial margin is greater (reducing survival time more than half) than any of the clinico-pathological variables previously established to impact survival in PDAC (e.g., T stage, lymph-node status, etc). While we have attempted to include as many of these relevant factors as possible, there may be unknown variables affecting the analysis. Second, an R1 medial margin is associated with poor outcomes; however, positive anterior or posterior surfaces were not. This may indicate that an R1 medial margin is a surrogate of aggressive biology and further molecular analysis of these primary tumors may help to explain these phenomena.

Supplementary material

11605_2012_1985_MOESM1_ESM.doc (31 kb)
Table S1Logistic regression for covariates associated with R1 medial resection margin. PPPD pylorus-preserving pancreaticoduodenectomy. (DOC 31 kb)

Copyright information

© The Society for Surgery of the Alimentary Tract 2012