Is Intraoperative Confirmation of Malignancy During ... - Springer Link

J Gastrointest Surg (2012) 16:370–375 DOI 10.1007/s11605-011-1728-y

ORIGINAL ARTICLE

Is Intraoperative Confirmation of Malignancy During Pancreaticoduodenectomy Mandatory? Giuseppe Garcea & Matthew S. Metcalfe & David P. Berry & Gavin S. Robertson & David M. Lloyd & Ashley R. Dennison

Received: 18 March 2011 / Accepted: 5 October 2011 / Published online: 28 October 2011 # 2011 The Society for Surgery of the Alimentary Tract

Abstract Introduction Differentiating between chronic pancreatitis and pancreatic adenocarcinoma can be difficult due to considerable overlap in disease presentation and radiological signs and the frequent co-existence of the two conditions. In this situation, surgeons may have to proceed to “blind” pancreaticoduodenectomy or attempt to confirm malignancy intraoperatively with frozen section (FS) histology. Methods This study attempted to ascertain the false-negative and false-positive rates of undertaking pancreaticoduodenectomies (PD) based on clinical suspicion (CS) or after intraoperative confirmation of malignancy using FS histology. Results Of patients, 13.6% (nine out of 66) underwent a benign PD in the CS group; 6.7% of patients had a missed malignancy in the FS group (n=62), but intraoperative histology prevented PD in 35% of patients with benign disease in the FS group. Specificity and sensitivity of intraoperative FS in detecting malignancy was 100% and 89.7%, respectively. Sensitivity of clinical assessment in detecting malignancy was 86.4%. Conclusions In experienced hands, intraoperative confirmation of malignancy is effective and will avoid resection in patients with benign disease. However, for many surgeons the chance of missing a small tumour with a false-negative biopsy will be unacceptable and they would prefer to undertake a “blind” resection and accept the mortality risk of pancreaticoduodenectomy for benign disease. Keywords Pancreatic ductal adenocarcinoma . Resection . Chronic pancreatitis . Biopsy, frozen section . Diagnosis

Introduction Differentiating between chronic pancreatitis and pancreatic adenocarcinoma can be difficult due to considerable overlap in disease presentation and radiological signs and the frequent co-existence of the two conditions. As a consequence, the incidence of benign disease in postoperative specimens from presumed cancer resection varies from 5% to 12%.1 Better-quality cross-sectional G. Garcea (*) : M. S. Metcalfe : D. P. Berry : G. S. Robertson : D. M. Lloyd : A. R. Dennison Department of Hepatobiliary and Pancreatic Surgery, University Hospitals of Leicester, Leicester LE5 4PW, UK e-mail: [email protected]

imaging,2 serum CA 19.9 levels,3 Doppler ultrasonagraphy,4 PET-CT,5 soluble interleukin-2 receptor assays6 and endoscopic ultrasound (EUS)7 with cytology have all been employed in an attempt to further characterise pancreatic head lesions. Despite these measures, a significant proportion of patients will present at laparotomy with pancreatic head masses of uncertain aetiology. The decision which must be addressed in this scenario is whether “blind” pancreaticoduodenectomy is justified or whether intraoperative confirmation of malignancy should be sought. In a series of 217 patients where 65 patients underwent a “blind” resection, 7% of those were found to have benign disease. In addition, frozen section (FS) of pancreatic biopsies revealed a 3% false-positive and 4% false-negative rate.8 In our study, we attempted to define the role of intraoperative FS from consecutive patients presenting with head of pancreas masses. The surgeons within the unit have different philosophies and hence approaches. Some will

J Gastrointest Surg (2012) 16:370–375

371

always attempt to obtain intraoperative confirmation of malignancy prior to resection while others have evolved their practice with time, from routine confirmation of malignancy, to a more targeted approach regarding intraoperative histology. Finally, others would always proceed to a “blind” resection rather than miss a potentially curable periampullary malignancy.

patients had a clinically suspected cancer; or that genuine equipoise existed as to whether the radiological changes were due to chronic inflammation or malignancy. In all cases a definite intention to resect any lesion (if confirmed on intraoperative biopsy or clinical assessment) existed. None of the patients included in the study had lesions which were radiologically inoperable in the event of malignant disease being discovered. A total of 128 patients were subsequently included into the study (Fig. 1).

Methods Patient Cohort Patient Identification All patients discussed at a multidisciplinary team meeting (MDT) with a possible periampullary malignancy were identified from 2003 to the present day. Patients undergoing exploration/resection for pre-malignant mucinous lesions, duodenal adenomas/adenocarcinomas and neuroendocrine tumours were excluded. Patients with a preoperative histological confirmation of malignancy were also excluded (Fig. 1). Criteria for inclusion into the study were patients undergoing exploration for a presumed distal cholangiocarcinoma or pancreatic adenocarcinoma. Case notes and MDT summaries were carefully examined to ensure that

The remaining patients (after exclusion) were those undergoing exploration±pancreaticoduodenectomy for a possible periampullary malignancy. Of this group, 62 had intraoperative FS of the head of pancreas or bile duct sent (patients undergoing FS for extra-pancreatic disease, either lymphatic tissue, peritoneal, or hepatic tissue were not included (frozen section (FS) group)). Sixty-six patients proceeded to a resection on clinical suspicion (CS) of malignancy alone (CS group), nine of which had benign disease (Fig. 1). Of the patients undergoing FS, 35 underwent a resection for adenocarcinoma on FS (subsequently re-confirmed on formal histology) (Fig. 1). Twenty-

25 Exclusions

153 Explorations

128 Inclusions Clinical suspicion of malignancy at operation, no intraoperative histology

66 without intraoperative frozen section

62 with intra-operative frozen section

C 35 Resections For Confirmed Adenocarcinoma

57 Resections For Confirmed Adenocarcinoma 27 Biliary Bypasses

Intraoperative frozen section histology of pancreas suggestive of benign disease

2 Exclusions

A

B

21 Biliary Bypasses for True Benign Disease

4 Biliary Bypasses for Missed Adenocarcinoma

9 Resections For Subsequent Benign Disease

Bypass within 2 years of study date, therefore inadequate follow-up

Patient re-presented within 2 years of bypass with histologically confirmed adenocarcinoma

Fig. 1 Flow diagram summarising numbers of patients and subsequent outcome undergoing exploratory laparotomy for suspected periampullary cancers. a Groups 1, b 2 and c 3

372


seven patients in this group underwent a biliary bypass because intraoperative frozen-section histology did not reveal carcinoma. All bypasses for biopsy-proven benign disease within 2 years of data analysis were excluded (n=2) leaving 25 biliary bypasses with putative benign disease. Four re-presented with new symptoms and radiological signs and all four patients had histologically confirmed adenocarcinoma of periampullary origin on subsequent follow-up. Data Analysis Patients were divided into three groups for the purposes of data analysis. Group 1 (labelled A in Fig. 1) underwent biliary bypass on the strength of intraoperative histology from the pancreas which confirmed benign disease only. Group 2 underwent biliary bypass under the same criteria, but re-presented within 2 years of their surgery with histologically proven adenocarcinoma of periampullary

origin (labelled B in Fig. 1). Group 3 underwent pancreaticoduodenectomy (without intraoperative histology) because of clinical intraoperative assessment suspicious of malignancy, but on subsequent specimen, processing were found to have benign disease (labelled C in Fig. 1). Radiological features, preoperative imaging and other demographic data were collected and compared between the groups. Median survival was calculated using Kaplan– Meier survival curves using Medcalc™ version 9.3. Mortality rates, median survival and 1-, 3- and 5-year survivals were derived from a cross-section of published literature regarding both curative pancreaticoduodenectomy and palliative biliary bypasses.

Results Figure 1 summarises the patient population included in the study. Table 1 displays demographic and radiological data

Table 1 Table displaying demographics and mode of presentation for groups 1, 2 and 3

Age (years) History of alcohol abuse Jaundice Dilated pancreatic duct on CT imaging Mass in head of pancreas on CT or intraoperatively Ca 19.9 (units/mL) Size of mass (mm) Intraoperative histology Chronic pancreatitis AIPa Alive at follow-up Yes No Follow-up (months) Interval (months) from bypass to confirmation of malignancy

ns not significant a

Autoimmune pancreatitis

b

Death from other myocardial infarction

Group 1

Group 2

Group 3

Significance

Median Range Number Percentage Number Percentage Number Percentage Number Percentage Median Range Median Range

59 25–72 14 66.7 19 90.5 7 33.3 16 76.2 121 3–3,232 36 20–70

59 48–64 2 50 4 100 1 25 3 75 167 78–206 20 15–20

62 43–65 5 55.6 8 88.9 2 22.2 7 77.8 156 75–747 32 25–59

ns

Number Percentage Number Percentage

20 95.2 1 4.8

4 100 0 0

– – – –

ns

Number Percentage Number Percentage Median Range Median Range

20 95.2 1b 4.8 95.5 30–125 – –

1 25 3 75 18 8–22 12.8 8–16

9 100 0 0 44.2 10–100 – –

ns

ns ns ns ns ns ns

ns

ns ns ns


373

between groups 1, 2 and 3. No significant differences existed between the two groups. In group 2 with missed adenocarcinoma, the median interval from histological confirmation of malignancy following initial surgery was 12.8 months. There was no post-operative mortality in any of the three groups. Resection for suspected malignancy without intraoperative confirmation resulted in 13.6% (nine out of 66 explorations) of patients undergoing a pancreaticoduodenectomy for benign disease. Intraoperative histological confirmation of malignancy, excluding the two biliary bypasses within 2 years of the study, resulted in 6.7% (four out of 60 explorations) of patients with a missed diagnosis of cancer (false negative) but prevented a resection for benign disease in 35% of patients (true negative). Specificity and sensitivity of intraoperative FS in detecting malignancy was 100% and 89.7%, respectively. Specificity of clinical assessment in detecting malignancy could not be determined since cross-over between the two groups (and hence the true negative rate) was unknown. Sensitivity of clinical assessment in detecting malignancy was 86.4%. Figure 2 demonstrates survival curves for groups 1, 2 and 3, and those patients undergoing curative resection for adenocarcinoma. As expected group 2, undergoing biliary bypass for missed adenocarcinoma, had the worst survival, groups 3 and 4 had the best survival. Median survival for patients undergoing biliary bypass for missed malignancy was 13 months.

100 Group 3 (n= 9)

Survival Probability (%)

90

Group 1 (n= 21)

80 70 60 50 Curative Resection of Periampullary Malignancy (n= 89)

40 Group 2 (n= 4)

30 20 0

20

40

60 80 Time (months)

100

120

140

Fig. 2 Survival curves for patients undergoing resection of periampullary malignancies, patients undergoing bypasses for benign disease (group 1), patients undergoing bypasses with occult malignant disease (group 2) and patients undergoing pancreaticoduodenectomies for benign disease (group 3)

Discussion The aim of this study was to compare two contrasting philosophies in the intraoperative management of suspected pancreatic malignancies. When there is no pre-operative histological confirmation of the nature of a pancreatic lesion, surgeons adopt different approaches to enable them to decide how to proceed. Their approaches are often very different and although there will inevitably be some modification (and consequent overlap) depending on surgical experience and the exact intraoperative findings. Results of this study confirm, as suspected, that intraoperative confirmation of histology by the use of FS reduces the rate of resection for benign lesions compared with a resection based on clinical grounds alone (0% versus 13.6%). In addition, FS biopsy is 100% specific and has a sensitivity rate of 89.7%, slightly higher than the sensitivity of clinical assessment alone (86.7%). From the data in this study, it is not possible to determine the degree of cross-over between the two cohorts (FS and CS). It is inevitable that even operating surgeons who did not routinely employ histological confirmation prior to resection, would have encountered a number of patients whose pre-operative assessment strongly suggested a malignant pathology but who intra-operatively were not found to have overtly malignant lesions. The higher percentage of patients with benign disease in the FS group compared with the CS group, does suggest the groups are not balanced in this regard. In these patients, the vast majority of surgeons will perform a FS in an attempt to confirm the presence of malignancy. If the FS does not confirm malignancy then the decision frequently depends on the presence or absence of an alternate pathology principally chronic pancreatitis and the patients’ fitness. If chronic pancreatitis can be demonstrated the majority of surgeons will perform a biliary bypass but if FS fails to demonstrate any pathology then the decision is more difficult and although it will generally default to the surgeons’ philosophy, the lack of a pathological explanation for the mass makes the majority of surgeons uncomfortable. Under these circumstances even operating surgeons who would normally routinely require histology prior to resection may proceed to a resection based on a strong CS. Crossover between the FS and “nonfrozen” section (CS) groups explains the lower rate of benign disease in the CS group. The data do show, however, that intraoperative histology is an accurate, reliable and feasible method of confirming the presence of malignancy prior to proceeding to resection. This study does highlight the difficulty in comparing the different approaches and particularly attempts to make an objective assessment based on the data available from the literature. Figure 3 outlines some hypothetical mathematical concepts using the data from this study with an assumed

374


Fig. 3 Mortality and “missed cancer” rates per 1,000 resections incorporating the diagnostic and error rates form FS and CS resection policies

mortality rate of 2.9% based on a meta-analysis of 154 studies detailing outcomes in over 25,000 patients.9 Within the CS group, there would be 3.9 deaths every 1,000 resections for patients undergoing resection for benign disease. If the figures from those patients who avoided resection surgery by virtue of FS, this mortality rises to 10.2 deaths per 1,000 patients. With a FS approach, there would be 8.7 preventable deaths (assuming 13% cure rate from pancreaticoduodenectomy),9 per 1,000 deaths. The numbers would appear to be broadly equivalent. These outcomes are very dependent on centre-based mortality rates and present long-term outcomes from pancreatic resection. Most of the published data for survival after palliative biliary bypass is not applicable in this situation as it refers to a planned bypass procedure (undertaken purely for palliation with disease known to be unresectable from pre-operative imaging) as well as biliary bypasses undertaken when intraoperatively; disease proves to be more extensive than believed when a resection was planned. However, the median survival of 13 months for the biliary bypasses in this study is commensurate with the previously reported survival from our group of 12.9 months for patients with locally advanced pancreatic adenocarcinoma.10 This is higher than the usual 4to 6-month survival reported following palliative biliary bypass11–14 but is probably applicable in this clinical scenario. Hence, in the remainder of the bypass group, the

87% of patients where pancreaticoduodenectomy would not have been curative, the reduction in survival would be minimal (from 15.5 months, the median survival reported following pancreaticoduodenectomy,9 to 13 months). In most centres, however, a formal bypass is the usual procedure undertaken in this situation and the risk associated with this operation is the most relevant. The majority of studies reporting results following biliary bypass procedures discuss procedures (a significant number of technically different procedures) employed for the palliation of malignant disease and the mortality is disproportionately high with a median value of 10% and a range of 1 to 10%.11–17 This high mortality rate is probably due to patient selection and the presence of advanced pancreatic malignancy and it is likely to be significantly lower when the decision is taken intraoperatively in patients who had been considered suitable for a pancreaticoduodenectomy. With the low mortality rates and morbidity which can be achieved following pancreaticoduodenectomy in specialised centres, the merit of avoiding such a major resection have reduced and results are likely to continue to improve. In addition, resection of the head of pancreas is a recognised option in the management of patients with chronic pancreatitis (although this will increase the likelihood of the subsequent onset of diabetes) and is consequently considered by many pancreatic surgeons to make the procedure more acceptable in


patients with this pathology (the most likely pathology to produce an intraoperative diagnostic dilemma).18, 19 Longterm results following pancreaticoduodenectomy are also good with patients having near-normal quality-of-life scores.20 In a retrospective such as this, quality-of-life assessment is not possible. However, this could represent a further confounding factor with regards to which operative approach is undertaken. These points do support the belief that a “blind” resection is justified when the nature of a circumscribed lesion in the head of pancreas is in doubt. If this approach is to be adopted however careful patient counselling is essential and the risk of death from an operation for benign disease should be openly discussed. Equally, if intraoperative confirmation of malignancy is considered mandatory then the risk of a missed diagnosis should also be carefully explained. The discussion will (in addition to explaining the risks of surgery for benign disease and the possibility of a missed malignancy) also need to consider results of the individual units and the fitness of the patient. A significant departure from 3% mortality from pancreaticoduodenectomy (better or worse) will significantly shift the equation. A 1.5% rate would reduce the post-operative mortality by 100% and patients would potentially derive significantly more benefit from a “blind” approach whereas 4.5% mortality would increase the expected mortality by 50% and support the use of intraoperative FS assessment.

Conclusions For surgeons and patients, a missed diagnosis of cancer is an emotive event and renders a completely pragmatic approach to intraoperative assessment difficult. In experienced hands, intraoperative confirmation of malignancy is effective and will avoid resection in patients with benign disease. The two approaches of undertaking resection only with definitive histology or undertaking a blind resection would appear to be, on the data presented here, equivalent. However, for many surgeons (and patients) the chance of missing a small tumour with a false-negative biopsy will be unacceptable and they would prefer to undertake a “blind” resection and accept the risk of mortality associated with a pancreaticoduodenectomy for benign disease. References 1. Kennedy T, van Gulik TM, Mojen TM, van Geenen R, Rauws EA, Obertop H, Gouma DJ. Differential diagnosis of focal pancreatitis and pancreatic cancer. Annals of Oncology 1999; 10: 85–88. 2. Yamada Y, Mori H, Matsumoto S, Kiyosue H, Hori Y, Hongo N. Pancreatic adenocarcinoma versus chronic pancreatitis: differentiation with triple-phase helical CT. Abdominal Imaging 2010; 35: 163–71.

375 3. Bedi MM, Gandi MD, Jacob G, Lekha V, Venogopal A, Ramesh H. CA 19.9 to differentiate benign and malignant masses in chronic pancreatitis: is there any benefit? Indian Journal of Gastroenterology 2009; 28: 24–27. 4. Scialipi M, Midir M, Bartolotta TV, Cazzolla MP, Rotondo A, Resta MC, Lagall R, Cardianle AE. Pancreatic carcinoma versus chronic focal pancreatitis: contrast-enhances power Doppler ultrasonagraphy findings. Abdominal Imaging 2005; 30: 222–227. 5. Stollfuss JC, Glatting G, Friess H, Kocher F, Berger HG, Reskse SN. 2-(Flurine-18)-fluror-2-D-glucose PET in detection of pancreatic cancer: value of quantitative image interpretation. Radiology 1995; 195: 339–344. 6. Rabitti PG, Pacelli L, Uomo G, Laccetti M, Spada OA, Esposito G, Visconti M. Soluble interleukin-2 receptor: anew marker in pancreatic adenocarcinoma. Minerva Gastroenterol Dietol 1994; 40: 101–103. 7. Nakamura R, Machado R, Amikura K, Ruebner B, Frey CF. Role of fine needle aspiration cytology and endoscopic biopsy in the preoperative assessment of pancreatic and peripancreatic malignancies. International Journal of Pancreatology 1994; 16:17–21. 8. Lee YT. Tissue Diagnosis for carcinoma of the pancreas and periampullary structures. Cancer 1982; 49: 1035–1039. 9. Garcea G, Dennison AR, Pattenden CJ, Neal CP, Sutton CD, Berry DP, Survival following curative resection for pancreatic ductal adenocarcinoma; a systematic review of the literature. Journal of the Pancreas 2008; 9:99–132. 10. Scott EN, Garcea G, Doucas H, Steward WP, Dennison AR, Berry DP. Surgical bypass vs. endoscopic stenting for pancreatic ductal adenocarcinoma. HPB (Oxford) 2009; 11:118–24. 11. Mann CD, Thomasset SC, Johnson NA, Garcea G, Neal CP, Dennison AR, Berry DP. Combined biliary and gastric bypass procedures as effective palliation for unresectable malignant disease. Australian New Zealand Journal of Surgery 2009; 79: 471–512. 12. Watanapa P, Williamson RC. Surgical palliation for pancreatic cancer: developments during the past two decades. British Journal of Surgery 1992; 79: 8–20; 13. Sarr MG, Cameron JL. Surgical palliation of unresectable carcinoma of the pancreas. World J Surg 1984; 8: 906–918 14. Sarr MG, Cameron JL. Surgical management of unresectable carcinoma of the pancreas. Surgery 1982; 91: 123–133. 15. Mukherjee S, Kocher HM, Hutchins RR, Bhattacharya S, Abraham AT.Palliative surgical bypass for pancreatic and peri-ampullary cancers. Journal of Gastrointestinal Cancer 2007;38:102–107 16. Singh S, Sachdev AK, Chaudhary A, Agarwal AK.Palliative surgical bypass for unresectable periampullary carcinoma. Hepatobiliary and Pancreatic Diseases International 2008; 7: 308–312. 17. Kuhlmann KF, van Poll D, de Castro SM, van Heek NT, Busch OR, van Gulik TM, Obertop H, Gouma DJ. Initial and long-term outcome after palliative surgical drainage of 269 patients with malignant biliary obstruction. European Journal of Surgical Oncology 2007;33:757–762. 18. Kennedy T, Preczweksi L, Stocker SJ, Rao SM, Parsons WG, Wayne JD, Bell RH, Talamonti MS. Incidence of benign inflammatory disease in patients undergoing Whipple’s procedure for clinically suspected carcinoma: a single-institution experience. American Journal of Surgery 2006; 191: 437–441. 19. Barens SA, Lillemoe KD, Kaufman HS, Sauter PK, Yeo CJ, Talamini MA, Pitt HA, Cameron JL. Pancreaticoduodenectomy for benign disease. American Journal of Surgery 1996; 171: 131–134. 20. Huang JJ, Yeo CJ, Sohn TA, Lillemoe KD, Sauter PK, Coleman J, Hruban RH, Cameron JL. Quality of life and outcomes after pancreaticoduodenectomy. Annals of Surgery 2000; 231: 890– 898.