Pathology (April 2014) 46(3), pp. 188–192
ANATOMICAL PATHOLOGY
Frozen section of the pancreatic neck margin in pancreatoduodenectomy for pancreatic adenocarcinoma is of limited utility TONY C. Y. PANG1,2, OLIVER WILSON1, MANUEL A. ARGUETA1, THOMAS J. HUGH1,2, ANGELA CHOU3,4, JASWINDER S. SAMRA1,2 AND ANTHONY J. GILL1,3,5 1Sydney Medical School, University of Sydney, 2Department of Upper GIT Surgery, Royal North Shore Hospital, St Leonards, 3Cancer Diagnosis and Pathology Group, Kolling Institute of Medical Research, St Leonards, 4Anatomical Pathology, Sydpath, St Vincent’s Hospital, Darlinghurst, and 5Department of Anatomical Pathology,
Royal North Shore Hospital, St Leonards, NSW, Australia
Summary The use of frozen section to assess resection margins intraoperatively during pancreaticoduodenectomy facilitates further resection. However, it is unclear whether this actually improves patient survival. We reviewed the overall survival and resection margin status in consecutive pancreaticoduodenectomies performed for carcinoma. An R1 resection was defined as an incomplete excision (1 mm margin); R0(p) resection as complete excision without re-resection and R0(s) resection as an initially positive neck margin which was converted to R0 resection after re-resection. Between 2007 and 2012, 116 pancreatoduodenectomies were performed for adenocarcinoma; 101 (87%) underwent frozen section of the neck margin which was positive in 19 (19%). Sixteen of these patients had negative neck margins after re-excision but only seven patients had no other involved margins [true R0(s) resections]. Median survival for the R0(p), R0(s) and R1 groups were 29, 16, 23 months, respectively ( p ¼ 0.049; R0(p) versus R0(s) p ¼ 0.040). Intra-operative frozen section increased the overall R0 rate by 7% but this did not improve survival. Our findings question the clinical benefit of intraoperative margin assessment, particularly if re-excision cannot be performed easily and safely. Key words: Frozen section, pancreatic adenocarcinoma, pancreatoduodenectomy, surgical margins. Received 27 August, revised 15 October, accepted 17 October 2013
INTRODUCTION The use of frozen section to assess resection margins intraoperatively during pancreaticoduodenectomy (PD) is based on two premises. The first is that frozen section allows for the further resection of involved margins and therefore increases the rate of R0 resection.1 The second is the assumption that a complete R0 resection improves outcome.2–4 Whilst frozen section has been shown to increase the rate of R0 excision, two large retrospective studies have suggested that it does not improve overall survival.1,5 A major area of contention in the recent literature has been over the prognostic significance of positive resection margins in pancreatic cancer. Whilst some groups have reported that excision margin status does not hold prognostic significance, others have reported that it does. A fair reading of the literature Print ISSN 0031-3025/Online ISSN 1465-3931 DOI: 10.1097/PAT.0000000000000072
#
suggests that when sub-specialist pathologists rigorously and systematically assess resection margin status after PD, there is usually a high rate of R1 excision (from 57% to 85%) and in these studies the excision status seems to be a strong predictor of outcome.6–10 In essence it appears that a low rate of incomplete excision may be a reflection of less rigorous (or at least less predictive) pathological assessment rather than being a true reflection of the excision margin status. Indeed without any other intervention, three separate groups have reported that the introduction of a standardised and rigorous approach to the pathological reporting of pancreatic cancers increased the number of PD reported as being incompletely excised from 40 to 57%, from 45 to 59% and from 14 to 76%.6,7,9 Studies which have previously failed to show a survival advantage for intraoperative frozen section examination have been criticised for demonstrating a low rate of R1 excision (23–30%) with the implication that pathological assessment of the final resection margin status may have been incomplete and that therefore some of the apparently R0 resections may actually represent false negatives of the final pathological assessment.10 Since mid-2006 all our PD specimens have been reported by experienced sub-specialist pathologists using a standardised reporting protocol with particular emphasis placed on rigorous assessment of resection margin status, particularly the periuncinate retroperitoneal margin.7 Therefore, we sought to examine the impact of frozen section examination on this cohort of patients specifically to determine whether it improved the R0 excision rate or overall survival.
PATIENTS AND METHODS All patients who underwent PD between 2007 and 2012 for pancreatic carcinoma at Royal North Shore Hospital, a tertiary referral centre in Sydney, were identified using the Department of Anatomical Pathology database. Specifically, ampullary cancers, neuroendocrine tumours, cholangiocarcinomas, intraductal papillary mucinous neoplasms and benign diseases were excluded from this study whilst pancreatic ductal adenocarcinomas, acinar cell carcinomas (n ¼ 1), adenosquamous carcinomas (n ¼ 2) and undifferentiated carcinomas (n ¼ 2) were included. Basic demographic, clinical and pathological data were extracted from patient records. An R1 resection margin was defined as tumour within 1 mm of any of the resection margins. R0 resection was defined as tumour beyond 1 mm of the
2014 Royal College of Pathologists of Australasia
Copyright © Royal College of pathologists of Australasia. Unauthorized reproduction of this article is prohibited.
PANCREATIC NECK MARGIN IN PD
resection margin. An R0 resection which was achieved without the need for re-resection was defined as ‘primary R0 resection’ [R0(p)]. Those patients with initially positive pancreatic neck margins who were converted to R0 resection after re-resection were classified as ‘secondary R0 resections’ [R0(s)]. None of the patients underwent an R2 resection. Overall survival was defined by survival censored at last follow-up or at death by any cause.
189
margins (single section each), anterior surface of the pancreas (single section minimum), the true posterior margin (which in vivo would abut the aorta and vena cava, single section minimum) and the superior mesenteric vein bed (minimum of one longitudinal section). Most importantly, the periuncinate retroperitoneal soft tissue margin (also known as the true retroperitoneal margin or the ‘mesopancreatic margin’ which is defined as the soft tissue on the posteromedial aspect of the uncinate process) was carefully identified and either all embedded or thoroughly sampled (3–4 blocks minimum).
Clinical protocol Patients were excluded from resection if there was metastatic disease, tumour encasement or abutment of more than 180 degrees of the superior mesenteric artery or coeliac axis, or if there was an occluded mesentericoportal venous system.11 Our surgical technique involves an aggressive approach to resection and reconstruction of the portal vein if it is safe to do so provided R0/R1 resection is thought to be possible.12,13 Frozen section assessment of the neck and common bile duct (CBD) margin was requested at the surgeons’ discretion after resection and prior to reconstruction. If the pancreatic neck margin was found to be positive (1 mm margin), re-resection was performed with the aim of achieving a clear pancreatic neck margin.
Statistical methods Descriptive statistics used mean (standard deviation) and median (interquartile range) as appropriate. Inferential statistical comparisons between groups used Fisher’s exact test, Student’s t-test (or analysis of variance) and Kruskal–Wallis test for categorical, parametric and non-parametric data, respectively. Each pair of groups was compared as a post-test, with p value correction with Sidak’s method unless otherwise specified. Survival was described with Kaplan–Meier curves and comparison between groups was performed using log-rank test. Multivariate analysis was not performed due to the small sample size. A p value of less than 0.05 was considered statistically significant. All statistical analyses were performed with Stata SE version 11.2 for Windows (Statacorp, USA).
Pathological protocol A standardised synoptic protocol was used for both macroscopic and microscopic pathological assessment as outlined previously.7 Briefly, both the frozen section dissections and the final dissection of the formalin fixed, paraffin embedded tissue were performed under the direct supervision of an experienced pathologist with particular expertise in pancreatic cancer. At frozen section the entire pancreatic neck margin was embedded as serial sections perpendicular to the transected neck margin, usually three to four sections per case. Once the specimen was fixed, the pancreas was bivalved and as a minimum the following additional margins were sampled for histological analysis: the common bile duct margin (single transverse section, all embedded), duodenal and intestinal
Patient and tumour characteristics The patient and tumour characteristics are summarised in Table 1 and Fig. 1. Briefly, there were 116 patients with a mean age of 68 years. Sixty-four (55%) were males. There was one acinar cell carcinoma, two adenosquamous carcinomas and two undifferentiated carcinomas in the series, with the remainder being conventional adenocarcinomas.
Summary of patient and tumour characteristics
Table 1
All patients
R0(p)
Mean (SD, range) or n (%) Patient characteristics n Age (years) Sex Tumour characteristics Size (mm) Location Grade{
T stage
RESULTS
{
LN Involved Perineural invasion Vascular invasion Survival Median OS (months)
R0(s)
R1
Mean (95% CI) or n (%)
P value
Overall
F M
116 68 52 64
(100) (10, 34–85) (45) (55)
42 69 23 19
(36) (66–72) (55) (45)
7 61 1 6
(6) (46–76) (14) (86)
67 68 28 39
(58) (66–70) (42) (58)
– 0.17 0.10
Head Unc G1 G2 G3 G4 T1 T2 T3 T4 No Yes No Yes
37 108 8 8 71 31 5 3 6 104 3 30 86 33 83
(14, 8–100) (93) (7) (7) (62) (27) (4) (3) (5) (90) (3) (26) (74) (28) (72)
34 41 1 27
(30–39) (98) (2) (66)
45 7 0 5
(28–61) (100) (0) (71)
38 60 7 47
(35–41) (90) (10) (70)
0.17 0.24
No Yes
14 (34)
2 (29)
20 (30)
7 (17)
1 (14)
1 (1)
35 (83)
6 (86)
66 (99)
15 27 20 22
2 5 3 4
13 54 10 57
(36) (64) (48) (52)
40 (34) 76 (66)
18 (43) 24 (57)
25 (18–31)
29 (18–NR)
(29) (71) (43) (57)
2 (29) 5 (71) 16 (5–28)
(19) (81) (15) (85)
Multiple comparison*
0.89
0.013
R0(p) vs R1 0.009; R0(s) vs R1 0.13; R0(p) vs R0(s) 1.00
0.16 0.001
20 (30) 47 (70)
0.36
23 (16–31)
0.049
R0(p) vs R1 0.0003; R0(s) vs R1 0.18; R0(p) vs R0(s) 1.00
R0(p) vs R0(s) 0.040; R0(s) vs R1 0.31; R0(p) vs R1 0.42
The neck margin described is that which was present prior to re-resection *
Multiple comparison post-test with Sidak’s correction. Group comparisons only shown if global test were significant. Grade and T stage divided into two groups for statistical analysis (G1–2, G3–4 and T1–2, T3–4). CI, confidence interval; F, female; LN, lymph node; M, male; NR, not reached; OS, overall survival; SD, standard deviation; Unc, uncinate process.
{
Copyright © Royal College of pathologists of Australasia. Unauthorized reproduction of this article is prohibited.
190
PANG et al.
Pathology (2014), 46(3), April PD for adenocarcinoma n = 116
Frozen section (neck margin) performed n = 101 (87%)
Frozen section (neck margin) not performed n = 15 (13%)
Positive margin n = 19 (19%)
Negative margin n = 82 (81%)
Positive margin 3 (16%)
Negative neck margin n = 13 (87%)
Negative margin 16 (84%)
Neck and other margins involved 2 (13%)
Legend Primary R0 (R0(p)) Other margins involved 3 (100%)
Other margins involved 9 (56%)
Other margins clear 7 (44%)
False negative 2 (2.4%) Other margins involved 38 (46%)
Other margins clear 42 (51%)
Other margins involved 13 (100%)
Secondary R0 (R0(s)) R1 resection
Fig. 1 Flow chart illustrating the proportion of patients undergoing frozen section, pancreatic neck margin and other margin status. Area of circles is proportional to the number of patients.
The vast majority (90%) of tumours were T3 tumours and nearly three-quarters had nodal involvement. Perineural and vascular invasion were found in the majority. The most commonly involved margins were periuncinate retroperitoneal and superior mesenteric vein bed margins, each involved in one-third of all patients. Thirty-four patients (29%) had resection of portions of the superior mesenteric vein performed as part of the procedure. A total of 101 patients (87%) underwent intra-operative frozen section of the neck margin. Nineteen (19%) of these were positive for malignancy. Further resection was undertaken in all of these patients. Sixteen (84%) of the patients with initially positive neck margins were negative at this margin after further excision(s). However, nine (56%) of the cases converted from R1 to R0 excision at the neck margin were positive at other margins. Therefore, intraoperative frozen section examination resulted in only seven patients (7% of
Table 2
all those undergoing frozen section) being converted from R1 to true R0(s) resection status (Fig. 1). Patients were divided into three groups [R0(p), R0(s), R1] according to margin status and compared in Table 1. Adverse pathological features (such as advanced stage, lymph node status, and perineural spread) became progressively more frequent from R0(p) to R0(s) to R1, suggesting that the resection margin status reflected worsening tumour biology or at least more anatomically advanced and infiltrative disease. R0(s) tumours were larger [45 mm, 95% confidence interval (CI) 28–61 mm] than R0(p) tumours (34 mm, 95% CI 30–39 mm) or R1 tumours (38 mm, 95% CI 35–41 mm), although this failed to achieve statistical significance. Pancreatic neck margin involvement was not found to be associated with involvement of any of the other margins and was not predictive of the number of other margins involved (Table 2). Also, none of the tumours located in the uncinate
Pattern of margin involvement (prior to re-resection) All patients (n ¼ 116)
Margin involved Neck Periuncinate retroperitoneal Posterior Superior mesenteric vein bed Common bile duct No. of margins involved No other margins 1 other margin >1 other margin 0 margin involved 1 margin involved 2 margins involved 3 margins involved
23 40 25 42 0
Pancreatic neck not involved
(20) (34) (22) (36) (0)
– – – 42 (36) 33 (28) 28 (24) 13 (11)
0 31 21 30 0
(0) (33) (23) (32) (0)
42 (45) 25 (27) 26 (28) – – – –
Pancreatic neck initially involved
23 9 4 12 0
(100) (39) (17) (52) (0)
8 (35) 7 (30) 8 (35) – – – –
p value
– 0.63 0.78 0.09 – 0.74 –
Involved margin defined as tumour present 1 mm of resection margin. p value compares patients with and without pancreatic neck involvement.
Copyright © Royal College of pathologists of Australasia. Unauthorized reproduction of this article is prohibited.
PANCREATIC NECK MARGIN IN PD
process involved the neck margin. However, all but one (n ¼ 7, 87%) of the uncinate process tumours involved margins. An involved periuncinate retroperitoneal margin was associated with an increased chance of involvement of the posterior (35% versus 14%, p ¼ 0.05) and superior mesenteric vein bed margins (55% versus 26%, p ¼ 0.012) (Table 3). There were two false negative frozen section margins. In one case, the margin was reported to be 1.3 mm on frozen section (therefore not involved) whilst the final margin was 0.1 mm. In the other instance, carcinoma was not present in the frozen section slides but was subsequently found at the margin in the final paraffin sections. Two cases with involved neck margins did not undergo frozen section. Survival The median period to last follow-up or death was 16 months. The three groups were statistically significantly different on the global test ( p ¼ 0.0485). Comparing each group individually revealed a statistically significant difference between the R0(p) and R0(s) groups only ( p ¼ 0.040, corrected). The median survival times for the R0(p), R0(s), and R1 groups were 29, 16 and 23 months, respectively (Fig. 2). Failure to reach the median survival upper confidence limits for R0(p) is likely a reflection of the short follow-up period.
DISCUSSION The justification for frozen section in PD is that it facilitates reresection of an involved margin and therefore increases the likelihood of achieving a complete R0 resection. Of the 101 patients who underwent frozen section assessment in our study, seven (7%) patients were converted from R1 to R0(s) margin status on re-resection. This finding is in line with the previous studies indicating that frozen section analysis increases the rate of complete excision by 6.0–8.4%.1,5,14 However, the question of whether this definite increase in the R0 excision rate actually results in an improved outcome is less straightforward. In our study there was worse survival in the R0(s) group (16 months) versus the R0(p) group (31 months); p ¼ 0.04. This worse survival may have been expected given that the R0(s) tumours must necessarily be larger, more locally advanced and more infiltrative than R0(p) tumours and therefore represent more unfavourable disease. However, we were surprised by the worse survival of R0(s) tumours compared to R1 tumours (24 months). Interestingly, this unexpected result was also found by Hernandez et al.5 Taken at face value our findings suggest that frozen section may not improve overall Table 3
191
survival and this would be in keeping with three studies which also found no statistically significant survival difference despite their improved R0 rates.1,5,14 An older study of patients from 1992–2006 demonstrated an improved survival in patients who underwent total pancreatectomy to achieve R0 resection compared to those who did not (and therefore had an isolated positive pancreatic neck margin). It should be noted however that both groups had poor survival (median 17.9 versus 9.7 months) and that the patients included in this study were a highly selected subgroup of all those who underwent surgery (33 and 28 out of 1579 patients who underwent PD during the same period).15 The question of whether a complete R0 resection improves prognosis has been controversial and much may depend on both the pathological sampling employed and the definition of an involved margin.6,7,9,16–18 Chang et al. found that a positive resection margin remained prognostically significant even when the definition of R1 extended to a margin 1.5 mm.4 Were we to apply this definition to our data, we would find the number of R0(s) resections would reduce to five, the median overall survival of groups would be unchanged and no statistically significant difference would be found in survival between the three resection margin groups (data not shown). We followed a strict protocol for histological examination of the resection margins, resulting in a rate of R1 resection of 52% with re-resection and 59% without. This R1 excision rate was much greater than the two previously reported series which suggested no survival benefit of frozen section (23–30% R1 rate before re-excision).1,5 However it was not as high as that reported by groups which serially sectioned the entire specimen (76–85%).6,8,9 Therefore, it is possible that some of our R0(p) and R0(s) cases actually represent false negatives of the margin assessment. However, we also perform an aggressive surgical clearance of the retroperitoneal soft tissue adjacent to the uncinate process of the pancreas so that all the periuncinate soft tissue is stripped from the adventitia of the right side of the superior mesenteric artery to ensure that the entire mesopancreas is removed en bloc.19 Therefore, given our comprehensive approach to pathological sampling of resection margins, we suspect that this aggressive approach to retroperitoneal dissection may also be a major cause of our increased complete excision rate. Our number of R0(s) tumours (n ¼ 7) was small, our followup limited (median 16 months) and we had access only to all cause mortality rather than disease specific mortality data; therefore, it is possible that our study was subject to a selection bias and was underpowered to show statistical difference.
Illustration of the correlation between the involvement of each individual margin Posterior
Margin Neck Periuncinate Uninvolved (n) Involved (n) Posterior
Periuncinate retroperioneal
Uninvolved
0.63 –
Involved
Uninvolved
0.78 0.017 (0.050) 65 (86%) 26 (65%)
–
Superior mesenteric vein bed
0.092 0.004 (0.012) 11 (14%) 14 (35%)
–
Involved
56 (74%) 18 (45%)
20 (26%) 22 (55%) 0.24
Each cell describes the p value of the Fisher’s exact test (2-sided) result without multiple comparison correction (to maximise sensitivity of detecting an association). The p values in bold indicate a statistically significant result, in which case the corrected p value (Sidak method) is shown in parenthesis. The contingency table for these significant tests is also shown. Common bile duct margin is not shown as the incidence of involvement was zero.
Copyright © Royal College of pathologists of Australasia. Unauthorized reproduction of this article is prohibited.
192
Pathology (2014), 46(3), April
Address for correspondence: Dr A. Gill, Department of Anatomical Pathology, Royal North Shore Hospital, Pacific Highway, St Leonards, NSW 2065, Australia. E-mail: [email protected]
0.75
P = 0.049 R0(p) vs R0(s), P = 0.04 R0(s) vs R1, P = 0.31 R0(p) vs R1, P = 0.42
0.50
References
0.00
0.25
Overall survival
1.00
PANG et al.
0
12
24
36
48
60
72
Months from surgery R0(p) R1
R0(s)
Fig. 2 Kaplan–Meier curves for overall survival for each of the resection groups. The p value for the log rank test comparing all three curves was 0.049. Group comparisons p values are also shown; these have been corrected for multiple comparison.
However, considering that all R0(s) patients died within the follow-up period in this study, that the R0(s) group had a poorer survival than even the R1 group, and that our results are in line with the literature, they strongly suggest that there is no survival advantage to re-resection. Furthermore, the finding that R0(s) tumours were of more advanced stage and had a higher incidence of perineural invasion than R0(p) tumours, supports the idea that R0(s) tumours may be inherently more infiltrative and aggressive than R0(p) tumours. That is, R0(s) tumours may demonstrate a worse biology than R0(p) tumours, and this worse biology (which is reflected by prominent perineural invasion, larger size and infiltrative growth pattern) may be only partly overcome by complete excision.
CONCLUSION In conclusion, our study is not definitive but raises serious doubts about the utility of intraoperative margin assessment in PD for pancreatic adenocarcinoma. Because of its limitations and in the belief that complete local excision offers a better basis for long-term survival than incomplete excision, we will continue to perform intra-operative frozen sections. However, we will do so with the knowledge that re-resection of involved margins may be of marginal, if any, benefit and therefore should only be undertaken if it can be performed with minimal additional morbidity. Conflicts of interest and sources of funding: The authors state that there are no conflicts of interest to disclose.
1. Dillhoff M, Yates R, Wall K, et al. Intraoperative assessment of pancreatic neck margin at the time of pancreaticoduodenectomy increases likelihood of margin-negative resection in patients with pancreatic cancer. J Gastrointest Surg 2009; 13: 825–30. 2. Yeo CJ, Cameron JL, Sohn TA, et al. Six hundred fifty consecutive pancreaticoduodenectomies in the 1990s: pathology, complications, and outcomes. Ann Surg 1997; 226: 248–57. 3. Winter JM, Cameron JL, Campbell KA, et al. 1423 pancreaticoduodenectomies for pancreatic cancer: A single-institution experience. J Gastrointest Surg 2006; 10: 1199–210. 4. Chang DK, Johns AL, Merrett ND, et al. Margin clearance and outcome in resected pancreatic cancer. J Clin Oncol 2009; 27: 2855–62. 5. Hernandez J, Mullinax J, Clark W, et al. Survival after pancreaticoduodenectomy is not improved by extending resections to achieve negative margins. Ann Surg 2009; 250: 76–80. 6. Esposito I, Kleeff J, Bergmann F, et al. Most pancreatic cancer resections are R1 resections. Ann Surg Oncol 2008; 15: 1651–60. 7. Gill AJ, Johns AL, Eckstein R, et al. Synoptic reporting improves histopathological assessment of pancreatic resection specimens. Pathology 2009; 41: 161–7. 8. Menon KV, Gomez D, Smith AM, et al. Impact of margin status on survival following pancreatoduodenectomy for cancer: the Leeds Pathology Protocol (LEEPP). HPB (Oxford) 2009; 11: 18–24. 9. Verbeke CS, Leitch D, Menon KV, et al. Redefining the R1 resection in pancreatic cancer. Br J Surg 2006; 93: 1232–7. 10. Verbeke CS, Smith AM. Survival after pancreaticoduodenectomy is not improved by extending resections to achieve negative margins. Ann Surg 2010; 251: 776–7; author reply 777–778. 11. Katz MH, Pisters PW, Evans DB, et al. Borderline resectable pancreatic cancer: the importance of this emerging stage of disease. J Am Coll Surg 2008; 206: 833–46. 12. Wang F, Arianayagam R, Gill A, et al. Grafts for mesenterico-portal vein resections can be avoided during pancreatoduodenectomy. J Am Coll Surg 2012; 215: 569–79. 13. Christians KK, Tsai S, Tolat PP, et al. Critical steps for pancreaticoduodenectomy in the setting of pancreatic adenocarcinoma. J Surg Oncol 2013; 107: 33–8. 14. Lad NL, Squires MH, Maithel SK, et al. Is it time to stop checking frozen section neck margins during pancreaticoduodenectomy? Ann Surg Oncol 2013; 20: 3626–33. 15. Schmidt CM, Glant J, Winter JM, et al. Total pancreatectomy (R0 resection) improves survival over subtotal pancreatectomy in isolated neck margin positive pancreatic adenocarcinoma. Surgery 2007; 142: 572–8. 16. Raut CP, Tseng JF, Sun CC, et al. Impact of resection status on pattern of failure and survival after pancreaticoduodenectomy for pancreatic adenocarcinoma. Ann Surg 2007; 246: 52–60. 17. Schnelldorfer T, Ware AL, Sarr MG, et al. Long-term survival after pancreatoduodenectomy for pancreatic adenocarcinoma: is cure possible? Ann Surg 2008; 247: 456–62. 18. Kimbrough CW, St Hill CR, Martin RC, et al. Tumor-positive resection margins reflect an aggressive tumor biology in pancreatic cancer. J Surg Oncol 2013; 107: 602–7. 19. Samra JS, Gananadha S, Gill A, et al. Modified extended pancreatoduodenectomy: en block resection of the peripancreatic retroperitoneal tissue and the head of the pancreas. ANZ J Surg 2006; 76: 1017–20.
Copyright © Royal College of pathologists of Australasia. Unauthorized reproduction of this article is prohibited.
ANATOMICAL PATHOLOGY
Frozen section of the pancreatic neck margin in pancreatoduodenectomy for pancreatic adenocarcinoma is of limited utility TONY C. Y. PANG1,2, OLIVER WILSON1, MANUEL A. ARGUETA1, THOMAS J. HUGH1,2, ANGELA CHOU3,4, JASWINDER S. SAMRA1,2 AND ANTHONY J. GILL1,3,5 1Sydney Medical School, University of Sydney, 2Department of Upper GIT Surgery, Royal North Shore Hospital, St Leonards, 3Cancer Diagnosis and Pathology Group, Kolling Institute of Medical Research, St Leonards, 4Anatomical Pathology, Sydpath, St Vincent’s Hospital, Darlinghurst, and 5Department of Anatomical Pathology,
Royal North Shore Hospital, St Leonards, NSW, Australia
Summary The use of frozen section to assess resection margins intraoperatively during pancreaticoduodenectomy facilitates further resection. However, it is unclear whether this actually improves patient survival. We reviewed the overall survival and resection margin status in consecutive pancreaticoduodenectomies performed for carcinoma. An R1 resection was defined as an incomplete excision (1 mm margin); R0(p) resection as complete excision without re-resection and R0(s) resection as an initially positive neck margin which was converted to R0 resection after re-resection. Between 2007 and 2012, 116 pancreatoduodenectomies were performed for adenocarcinoma; 101 (87%) underwent frozen section of the neck margin which was positive in 19 (19%). Sixteen of these patients had negative neck margins after re-excision but only seven patients had no other involved margins [true R0(s) resections]. Median survival for the R0(p), R0(s) and R1 groups were 29, 16, 23 months, respectively ( p ¼ 0.049; R0(p) versus R0(s) p ¼ 0.040). Intra-operative frozen section increased the overall R0 rate by 7% but this did not improve survival. Our findings question the clinical benefit of intraoperative margin assessment, particularly if re-excision cannot be performed easily and safely. Key words: Frozen section, pancreatic adenocarcinoma, pancreatoduodenectomy, surgical margins. Received 27 August, revised 15 October, accepted 17 October 2013
INTRODUCTION The use of frozen section to assess resection margins intraoperatively during pancreaticoduodenectomy (PD) is based on two premises. The first is that frozen section allows for the further resection of involved margins and therefore increases the rate of R0 resection.1 The second is the assumption that a complete R0 resection improves outcome.2–4 Whilst frozen section has been shown to increase the rate of R0 excision, two large retrospective studies have suggested that it does not improve overall survival.1,5 A major area of contention in the recent literature has been over the prognostic significance of positive resection margins in pancreatic cancer. Whilst some groups have reported that excision margin status does not hold prognostic significance, others have reported that it does. A fair reading of the literature Print ISSN 0031-3025/Online ISSN 1465-3931 DOI: 10.1097/PAT.0000000000000072
#
suggests that when sub-specialist pathologists rigorously and systematically assess resection margin status after PD, there is usually a high rate of R1 excision (from 57% to 85%) and in these studies the excision status seems to be a strong predictor of outcome.6–10 In essence it appears that a low rate of incomplete excision may be a reflection of less rigorous (or at least less predictive) pathological assessment rather than being a true reflection of the excision margin status. Indeed without any other intervention, three separate groups have reported that the introduction of a standardised and rigorous approach to the pathological reporting of pancreatic cancers increased the number of PD reported as being incompletely excised from 40 to 57%, from 45 to 59% and from 14 to 76%.6,7,9 Studies which have previously failed to show a survival advantage for intraoperative frozen section examination have been criticised for demonstrating a low rate of R1 excision (23–30%) with the implication that pathological assessment of the final resection margin status may have been incomplete and that therefore some of the apparently R0 resections may actually represent false negatives of the final pathological assessment.10 Since mid-2006 all our PD specimens have been reported by experienced sub-specialist pathologists using a standardised reporting protocol with particular emphasis placed on rigorous assessment of resection margin status, particularly the periuncinate retroperitoneal margin.7 Therefore, we sought to examine the impact of frozen section examination on this cohort of patients specifically to determine whether it improved the R0 excision rate or overall survival.
PATIENTS AND METHODS All patients who underwent PD between 2007 and 2012 for pancreatic carcinoma at Royal North Shore Hospital, a tertiary referral centre in Sydney, were identified using the Department of Anatomical Pathology database. Specifically, ampullary cancers, neuroendocrine tumours, cholangiocarcinomas, intraductal papillary mucinous neoplasms and benign diseases were excluded from this study whilst pancreatic ductal adenocarcinomas, acinar cell carcinomas (n ¼ 1), adenosquamous carcinomas (n ¼ 2) and undifferentiated carcinomas (n ¼ 2) were included. Basic demographic, clinical and pathological data were extracted from patient records. An R1 resection margin was defined as tumour within 1 mm of any of the resection margins. R0 resection was defined as tumour beyond 1 mm of the
2014 Royal College of Pathologists of Australasia
Copyright © Royal College of pathologists of Australasia. Unauthorized reproduction of this article is prohibited.
PANCREATIC NECK MARGIN IN PD
resection margin. An R0 resection which was achieved without the need for re-resection was defined as ‘primary R0 resection’ [R0(p)]. Those patients with initially positive pancreatic neck margins who were converted to R0 resection after re-resection were classified as ‘secondary R0 resections’ [R0(s)]. None of the patients underwent an R2 resection. Overall survival was defined by survival censored at last follow-up or at death by any cause.
189
margins (single section each), anterior surface of the pancreas (single section minimum), the true posterior margin (which in vivo would abut the aorta and vena cava, single section minimum) and the superior mesenteric vein bed (minimum of one longitudinal section). Most importantly, the periuncinate retroperitoneal soft tissue margin (also known as the true retroperitoneal margin or the ‘mesopancreatic margin’ which is defined as the soft tissue on the posteromedial aspect of the uncinate process) was carefully identified and either all embedded or thoroughly sampled (3–4 blocks minimum).
Clinical protocol Patients were excluded from resection if there was metastatic disease, tumour encasement or abutment of more than 180 degrees of the superior mesenteric artery or coeliac axis, or if there was an occluded mesentericoportal venous system.11 Our surgical technique involves an aggressive approach to resection and reconstruction of the portal vein if it is safe to do so provided R0/R1 resection is thought to be possible.12,13 Frozen section assessment of the neck and common bile duct (CBD) margin was requested at the surgeons’ discretion after resection and prior to reconstruction. If the pancreatic neck margin was found to be positive (1 mm margin), re-resection was performed with the aim of achieving a clear pancreatic neck margin.
Statistical methods Descriptive statistics used mean (standard deviation) and median (interquartile range) as appropriate. Inferential statistical comparisons between groups used Fisher’s exact test, Student’s t-test (or analysis of variance) and Kruskal–Wallis test for categorical, parametric and non-parametric data, respectively. Each pair of groups was compared as a post-test, with p value correction with Sidak’s method unless otherwise specified. Survival was described with Kaplan–Meier curves and comparison between groups was performed using log-rank test. Multivariate analysis was not performed due to the small sample size. A p value of less than 0.05 was considered statistically significant. All statistical analyses were performed with Stata SE version 11.2 for Windows (Statacorp, USA).
Pathological protocol A standardised synoptic protocol was used for both macroscopic and microscopic pathological assessment as outlined previously.7 Briefly, both the frozen section dissections and the final dissection of the formalin fixed, paraffin embedded tissue were performed under the direct supervision of an experienced pathologist with particular expertise in pancreatic cancer. At frozen section the entire pancreatic neck margin was embedded as serial sections perpendicular to the transected neck margin, usually three to four sections per case. Once the specimen was fixed, the pancreas was bivalved and as a minimum the following additional margins were sampled for histological analysis: the common bile duct margin (single transverse section, all embedded), duodenal and intestinal
Patient and tumour characteristics The patient and tumour characteristics are summarised in Table 1 and Fig. 1. Briefly, there were 116 patients with a mean age of 68 years. Sixty-four (55%) were males. There was one acinar cell carcinoma, two adenosquamous carcinomas and two undifferentiated carcinomas in the series, with the remainder being conventional adenocarcinomas.
Summary of patient and tumour characteristics
Table 1
All patients
R0(p)
Mean (SD, range) or n (%) Patient characteristics n Age (years) Sex Tumour characteristics Size (mm) Location Grade{
T stage
RESULTS
{
LN Involved Perineural invasion Vascular invasion Survival Median OS (months)
R0(s)
R1
Mean (95% CI) or n (%)
P value
Overall
F M
116 68 52 64
(100) (10, 34–85) (45) (55)
42 69 23 19
(36) (66–72) (55) (45)
7 61 1 6
(6) (46–76) (14) (86)
67 68 28 39
(58) (66–70) (42) (58)
– 0.17 0.10
Head Unc G1 G2 G3 G4 T1 T2 T3 T4 No Yes No Yes
37 108 8 8 71 31 5 3 6 104 3 30 86 33 83
(14, 8–100) (93) (7) (7) (62) (27) (4) (3) (5) (90) (3) (26) (74) (28) (72)
34 41 1 27
(30–39) (98) (2) (66)
45 7 0 5
(28–61) (100) (0) (71)
38 60 7 47
(35–41) (90) (10) (70)
0.17 0.24
No Yes
14 (34)
2 (29)
20 (30)
7 (17)
1 (14)
1 (1)
35 (83)
6 (86)
66 (99)
15 27 20 22
2 5 3 4
13 54 10 57
(36) (64) (48) (52)
40 (34) 76 (66)
18 (43) 24 (57)
25 (18–31)
29 (18–NR)
(29) (71) (43) (57)
2 (29) 5 (71) 16 (5–28)
(19) (81) (15) (85)
Multiple comparison*
0.89
0.013
R0(p) vs R1 0.009; R0(s) vs R1 0.13; R0(p) vs R0(s) 1.00
0.16 0.001
20 (30) 47 (70)
0.36
23 (16–31)
0.049
R0(p) vs R1 0.0003; R0(s) vs R1 0.18; R0(p) vs R0(s) 1.00
R0(p) vs R0(s) 0.040; R0(s) vs R1 0.31; R0(p) vs R1 0.42
The neck margin described is that which was present prior to re-resection *
Multiple comparison post-test with Sidak’s correction. Group comparisons only shown if global test were significant. Grade and T stage divided into two groups for statistical analysis (G1–2, G3–4 and T1–2, T3–4). CI, confidence interval; F, female; LN, lymph node; M, male; NR, not reached; OS, overall survival; SD, standard deviation; Unc, uncinate process.
{
Copyright © Royal College of pathologists of Australasia. Unauthorized reproduction of this article is prohibited.
190
PANG et al.
Pathology (2014), 46(3), April PD for adenocarcinoma n = 116
Frozen section (neck margin) performed n = 101 (87%)
Frozen section (neck margin) not performed n = 15 (13%)
Positive margin n = 19 (19%)
Negative margin n = 82 (81%)
Positive margin 3 (16%)
Negative neck margin n = 13 (87%)
Negative margin 16 (84%)
Neck and other margins involved 2 (13%)
Legend Primary R0 (R0(p)) Other margins involved 3 (100%)
Other margins involved 9 (56%)
Other margins clear 7 (44%)
False negative 2 (2.4%) Other margins involved 38 (46%)
Other margins clear 42 (51%)
Other margins involved 13 (100%)
Secondary R0 (R0(s)) R1 resection
Fig. 1 Flow chart illustrating the proportion of patients undergoing frozen section, pancreatic neck margin and other margin status. Area of circles is proportional to the number of patients.
The vast majority (90%) of tumours were T3 tumours and nearly three-quarters had nodal involvement. Perineural and vascular invasion were found in the majority. The most commonly involved margins were periuncinate retroperitoneal and superior mesenteric vein bed margins, each involved in one-third of all patients. Thirty-four patients (29%) had resection of portions of the superior mesenteric vein performed as part of the procedure. A total of 101 patients (87%) underwent intra-operative frozen section of the neck margin. Nineteen (19%) of these were positive for malignancy. Further resection was undertaken in all of these patients. Sixteen (84%) of the patients with initially positive neck margins were negative at this margin after further excision(s). However, nine (56%) of the cases converted from R1 to R0 excision at the neck margin were positive at other margins. Therefore, intraoperative frozen section examination resulted in only seven patients (7% of
Table 2
all those undergoing frozen section) being converted from R1 to true R0(s) resection status (Fig. 1). Patients were divided into three groups [R0(p), R0(s), R1] according to margin status and compared in Table 1. Adverse pathological features (such as advanced stage, lymph node status, and perineural spread) became progressively more frequent from R0(p) to R0(s) to R1, suggesting that the resection margin status reflected worsening tumour biology or at least more anatomically advanced and infiltrative disease. R0(s) tumours were larger [45 mm, 95% confidence interval (CI) 28–61 mm] than R0(p) tumours (34 mm, 95% CI 30–39 mm) or R1 tumours (38 mm, 95% CI 35–41 mm), although this failed to achieve statistical significance. Pancreatic neck margin involvement was not found to be associated with involvement of any of the other margins and was not predictive of the number of other margins involved (Table 2). Also, none of the tumours located in the uncinate
Pattern of margin involvement (prior to re-resection) All patients (n ¼ 116)
Margin involved Neck Periuncinate retroperitoneal Posterior Superior mesenteric vein bed Common bile duct No. of margins involved No other margins 1 other margin >1 other margin 0 margin involved 1 margin involved 2 margins involved 3 margins involved
23 40 25 42 0
Pancreatic neck not involved
(20) (34) (22) (36) (0)
– – – 42 (36) 33 (28) 28 (24) 13 (11)
0 31 21 30 0
(0) (33) (23) (32) (0)
42 (45) 25 (27) 26 (28) – – – –
Pancreatic neck initially involved
23 9 4 12 0
(100) (39) (17) (52) (0)
8 (35) 7 (30) 8 (35) – – – –
p value
– 0.63 0.78 0.09 – 0.74 –
Involved margin defined as tumour present 1 mm of resection margin. p value compares patients with and without pancreatic neck involvement.
Copyright © Royal College of pathologists of Australasia. Unauthorized reproduction of this article is prohibited.
PANCREATIC NECK MARGIN IN PD
process involved the neck margin. However, all but one (n ¼ 7, 87%) of the uncinate process tumours involved margins. An involved periuncinate retroperitoneal margin was associated with an increased chance of involvement of the posterior (35% versus 14%, p ¼ 0.05) and superior mesenteric vein bed margins (55% versus 26%, p ¼ 0.012) (Table 3). There were two false negative frozen section margins. In one case, the margin was reported to be 1.3 mm on frozen section (therefore not involved) whilst the final margin was 0.1 mm. In the other instance, carcinoma was not present in the frozen section slides but was subsequently found at the margin in the final paraffin sections. Two cases with involved neck margins did not undergo frozen section. Survival The median period to last follow-up or death was 16 months. The three groups were statistically significantly different on the global test ( p ¼ 0.0485). Comparing each group individually revealed a statistically significant difference between the R0(p) and R0(s) groups only ( p ¼ 0.040, corrected). The median survival times for the R0(p), R0(s), and R1 groups were 29, 16 and 23 months, respectively (Fig. 2). Failure to reach the median survival upper confidence limits for R0(p) is likely a reflection of the short follow-up period.
DISCUSSION The justification for frozen section in PD is that it facilitates reresection of an involved margin and therefore increases the likelihood of achieving a complete R0 resection. Of the 101 patients who underwent frozen section assessment in our study, seven (7%) patients were converted from R1 to R0(s) margin status on re-resection. This finding is in line with the previous studies indicating that frozen section analysis increases the rate of complete excision by 6.0–8.4%.1,5,14 However, the question of whether this definite increase in the R0 excision rate actually results in an improved outcome is less straightforward. In our study there was worse survival in the R0(s) group (16 months) versus the R0(p) group (31 months); p ¼ 0.04. This worse survival may have been expected given that the R0(s) tumours must necessarily be larger, more locally advanced and more infiltrative than R0(p) tumours and therefore represent more unfavourable disease. However, we were surprised by the worse survival of R0(s) tumours compared to R1 tumours (24 months). Interestingly, this unexpected result was also found by Hernandez et al.5 Taken at face value our findings suggest that frozen section may not improve overall Table 3
191
survival and this would be in keeping with three studies which also found no statistically significant survival difference despite their improved R0 rates.1,5,14 An older study of patients from 1992–2006 demonstrated an improved survival in patients who underwent total pancreatectomy to achieve R0 resection compared to those who did not (and therefore had an isolated positive pancreatic neck margin). It should be noted however that both groups had poor survival (median 17.9 versus 9.7 months) and that the patients included in this study were a highly selected subgroup of all those who underwent surgery (33 and 28 out of 1579 patients who underwent PD during the same period).15 The question of whether a complete R0 resection improves prognosis has been controversial and much may depend on both the pathological sampling employed and the definition of an involved margin.6,7,9,16–18 Chang et al. found that a positive resection margin remained prognostically significant even when the definition of R1 extended to a margin 1.5 mm.4 Were we to apply this definition to our data, we would find the number of R0(s) resections would reduce to five, the median overall survival of groups would be unchanged and no statistically significant difference would be found in survival between the three resection margin groups (data not shown). We followed a strict protocol for histological examination of the resection margins, resulting in a rate of R1 resection of 52% with re-resection and 59% without. This R1 excision rate was much greater than the two previously reported series which suggested no survival benefit of frozen section (23–30% R1 rate before re-excision).1,5 However it was not as high as that reported by groups which serially sectioned the entire specimen (76–85%).6,8,9 Therefore, it is possible that some of our R0(p) and R0(s) cases actually represent false negatives of the margin assessment. However, we also perform an aggressive surgical clearance of the retroperitoneal soft tissue adjacent to the uncinate process of the pancreas so that all the periuncinate soft tissue is stripped from the adventitia of the right side of the superior mesenteric artery to ensure that the entire mesopancreas is removed en bloc.19 Therefore, given our comprehensive approach to pathological sampling of resection margins, we suspect that this aggressive approach to retroperitoneal dissection may also be a major cause of our increased complete excision rate. Our number of R0(s) tumours (n ¼ 7) was small, our followup limited (median 16 months) and we had access only to all cause mortality rather than disease specific mortality data; therefore, it is possible that our study was subject to a selection bias and was underpowered to show statistical difference.
Illustration of the correlation between the involvement of each individual margin Posterior
Margin Neck Periuncinate Uninvolved (n) Involved (n) Posterior
Periuncinate retroperioneal
Uninvolved
0.63 –
Involved
Uninvolved
0.78 0.017 (0.050) 65 (86%) 26 (65%)
–
Superior mesenteric vein bed
0.092 0.004 (0.012) 11 (14%) 14 (35%)
–
Involved
56 (74%) 18 (45%)
20 (26%) 22 (55%) 0.24
Each cell describes the p value of the Fisher’s exact test (2-sided) result without multiple comparison correction (to maximise sensitivity of detecting an association). The p values in bold indicate a statistically significant result, in which case the corrected p value (Sidak method) is shown in parenthesis. The contingency table for these significant tests is also shown. Common bile duct margin is not shown as the incidence of involvement was zero.
Copyright © Royal College of pathologists of Australasia. Unauthorized reproduction of this article is prohibited.
192
Pathology (2014), 46(3), April
Address for correspondence: Dr A. Gill, Department of Anatomical Pathology, Royal North Shore Hospital, Pacific Highway, St Leonards, NSW 2065, Australia. E-mail: [email protected]
0.75
P = 0.049 R0(p) vs R0(s), P = 0.04 R0(s) vs R1, P = 0.31 R0(p) vs R1, P = 0.42
0.50
References
0.00
0.25
Overall survival
1.00
PANG et al.
0
12
24
36
48
60
72
Months from surgery R0(p) R1
R0(s)
Fig. 2 Kaplan–Meier curves for overall survival for each of the resection groups. The p value for the log rank test comparing all three curves was 0.049. Group comparisons p values are also shown; these have been corrected for multiple comparison.
However, considering that all R0(s) patients died within the follow-up period in this study, that the R0(s) group had a poorer survival than even the R1 group, and that our results are in line with the literature, they strongly suggest that there is no survival advantage to re-resection. Furthermore, the finding that R0(s) tumours were of more advanced stage and had a higher incidence of perineural invasion than R0(p) tumours, supports the idea that R0(s) tumours may be inherently more infiltrative and aggressive than R0(p) tumours. That is, R0(s) tumours may demonstrate a worse biology than R0(p) tumours, and this worse biology (which is reflected by prominent perineural invasion, larger size and infiltrative growth pattern) may be only partly overcome by complete excision.
CONCLUSION In conclusion, our study is not definitive but raises serious doubts about the utility of intraoperative margin assessment in PD for pancreatic adenocarcinoma. Because of its limitations and in the belief that complete local excision offers a better basis for long-term survival than incomplete excision, we will continue to perform intra-operative frozen sections. However, we will do so with the knowledge that re-resection of involved margins may be of marginal, if any, benefit and therefore should only be undertaken if it can be performed with minimal additional morbidity. Conflicts of interest and sources of funding: The authors state that there are no conflicts of interest to disclose.
1. Dillhoff M, Yates R, Wall K, et al. Intraoperative assessment of pancreatic neck margin at the time of pancreaticoduodenectomy increases likelihood of margin-negative resection in patients with pancreatic cancer. J Gastrointest Surg 2009; 13: 825–30. 2. Yeo CJ, Cameron JL, Sohn TA, et al. Six hundred fifty consecutive pancreaticoduodenectomies in the 1990s: pathology, complications, and outcomes. Ann Surg 1997; 226: 248–57. 3. Winter JM, Cameron JL, Campbell KA, et al. 1423 pancreaticoduodenectomies for pancreatic cancer: A single-institution experience. J Gastrointest Surg 2006; 10: 1199–210. 4. Chang DK, Johns AL, Merrett ND, et al. Margin clearance and outcome in resected pancreatic cancer. J Clin Oncol 2009; 27: 2855–62. 5. Hernandez J, Mullinax J, Clark W, et al. Survival after pancreaticoduodenectomy is not improved by extending resections to achieve negative margins. Ann Surg 2009; 250: 76–80. 6. Esposito I, Kleeff J, Bergmann F, et al. Most pancreatic cancer resections are R1 resections. Ann Surg Oncol 2008; 15: 1651–60. 7. Gill AJ, Johns AL, Eckstein R, et al. Synoptic reporting improves histopathological assessment of pancreatic resection specimens. Pathology 2009; 41: 161–7. 8. Menon KV, Gomez D, Smith AM, et al. Impact of margin status on survival following pancreatoduodenectomy for cancer: the Leeds Pathology Protocol (LEEPP). HPB (Oxford) 2009; 11: 18–24. 9. Verbeke CS, Leitch D, Menon KV, et al. Redefining the R1 resection in pancreatic cancer. Br J Surg 2006; 93: 1232–7. 10. Verbeke CS, Smith AM. Survival after pancreaticoduodenectomy is not improved by extending resections to achieve negative margins. Ann Surg 2010; 251: 776–7; author reply 777–778. 11. Katz MH, Pisters PW, Evans DB, et al. Borderline resectable pancreatic cancer: the importance of this emerging stage of disease. J Am Coll Surg 2008; 206: 833–46. 12. Wang F, Arianayagam R, Gill A, et al. Grafts for mesenterico-portal vein resections can be avoided during pancreatoduodenectomy. J Am Coll Surg 2012; 215: 569–79. 13. Christians KK, Tsai S, Tolat PP, et al. Critical steps for pancreaticoduodenectomy in the setting of pancreatic adenocarcinoma. J Surg Oncol 2013; 107: 33–8. 14. Lad NL, Squires MH, Maithel SK, et al. Is it time to stop checking frozen section neck margins during pancreaticoduodenectomy? Ann Surg Oncol 2013; 20: 3626–33. 15. Schmidt CM, Glant J, Winter JM, et al. Total pancreatectomy (R0 resection) improves survival over subtotal pancreatectomy in isolated neck margin positive pancreatic adenocarcinoma. Surgery 2007; 142: 572–8. 16. Raut CP, Tseng JF, Sun CC, et al. Impact of resection status on pattern of failure and survival after pancreaticoduodenectomy for pancreatic adenocarcinoma. Ann Surg 2007; 246: 52–60. 17. Schnelldorfer T, Ware AL, Sarr MG, et al. Long-term survival after pancreatoduodenectomy for pancreatic adenocarcinoma: is cure possible? Ann Surg 2008; 247: 456–62. 18. Kimbrough CW, St Hill CR, Martin RC, et al. Tumor-positive resection margins reflect an aggressive tumor biology in pancreatic cancer. J Surg Oncol 2013; 107: 602–7. 19. Samra JS, Gananadha S, Gill A, et al. Modified extended pancreatoduodenectomy: en block resection of the peripancreatic retroperitoneal tissue and the head of the pancreas. ANZ J Surg 2006; 76: 1017–20.
Copyright © Royal College of pathologists of Australasia. Unauthorized reproduction of this article is prohibited.
