Pancreatic cancer: surgical management and outcomes after 6 years of follow-up

Pancreatic cancer is the fourth leading cause of cancer-related mortality in Australia.1 Median survival from diagnosis is about 5 months. The 5-year survival rate is less than 5%, but higher rates (between 6.8% and 17%) have been reported in those selected for surgery.2-4 Patients with unresectable disease have short survival times but benefit from palliative treatment.5,6 Results of resection and treatment in trials or published case series may not be achieved in clinical practice, as many patients are older and have more extensive disease than those included in randomised trials or referred to tertiary centres.

The Victorian Cooperative Oncology Group, in collaboration with the Victorian Cancer Registry (VCR), examined patterns of care for patients diagnosed with pancreatic cancer over a 2-year period. Here, we describe the surgical management of and outcomes for these patients. Chemotherapy and radiotherapy in this cohort have been described previously.7

We conducted a survey of all Victorian residents diagnosed with pancreatic carcinoma during 2002 or 2003 who were identified by the VCR. Ethics approval was obtained from the Cancer Council Victoria Human Research Ethics Committee.

Treating doctors were sent a questionnaire relating to treatment intent, management and outcomes. Data collection commenced in 2005. All patients were followed up for 6 years through mortality data from the Victorian Registry of Births, Deaths and Marriages and the National Death Index.

A total of 1044 patients with a diagnosis of pancreatic cancer in 2002 or 2003 were identified. Of these, 117 (11.2%) were deemed ineligible because the diagnosis was found to have occurred outside the study interval, the patient was not a usual resident of Victoria, or no treatment information was available because the notification to the VCR was by death certificate or autopsy only.

Completed questionnaires were received for 830 of the 927 eligible patients (response rate, 89.5%). For the remaining 97 patients, the patient record was unavailable (16), doctors declined to participate or did not respond (57), or there was insufficient information to include the patient in the survey (24). Patients with neuroendocrine tumours or tumours of the ampulla of Vater (67) were also excluded, leaving 763 patients with cancer of the pancreas in the study sample.

The median age of the included patients was 72 years, and 391 (51.2%) were male. Only 44 patients (5.8%) were younger than 50 years, and 196 (25.7%) were aged 80 years or older.

The findings of this study represent an unbiased population-based picture of the surgical management and outcomes of patients diagnosed with carcinoma of the pancreas in 2002 and 2003 in Victoria. Strengths of the study include a response rate of almost 90%, together with high case ascertainment by the VCR due to mandatory cancer reporting, and the near complete follow-up of survival through mortality data. A limitation is the retrospective collection of data.

The management of carcinoma of the pancreas is often benchmarked on the results of surgical resection. Resection rates should be greater than 10% in unselected patients, and early mortality, observed median survival and 5-year survival are important end points.8,9 In this study, Whipple resection was performed in 12% of all patients presenting with tumours of the head and neck of the pancreas, comparable to reported rates of 10%–15%.9,10 Thirty-day mortality in these patients (5.3%) was comparable to reports from individual hospitals.9 A large epidemiological study in the United States reported 30-day mortality of 7.5%, and a survey of 23 hospitals in England and Wales reported 12%.11,12 Although inhospital or 90-day mortality are better measures of perioperative mortality, they are not as widely reported. Median survival in the patients undergoing Whipple resection in this study (16.3 months) is comparable with other reports, and their 5-year survival of 8% is comparable to results from a review of eight similar series (range, 3.7%–16.4%).9,13

We found that a relatively large number of surgeons each performed a modest number of Whipple resections during the 2-year study period, but early mortality and survival were not related to surgeon workload. However, investigations in larger populations have found that early mortality rate and survival improve with increasing workloads for both surgeons and hospitals.11 The US National Comprehensive Cancer Network guidelines recommend pancreatic resection be done at hospitals that perform more than 20 resections annually.14 Chang and colleagues report excellent results from a high-volume centre in New South Wales and suggest most Whipple resections should be performed in high-volume specialist units.15 Implementation of this strategy has been difficult in other countries,16 and others argue that volume-based referrals undermine local expertise and may leave patients’ care far from supportive social networks, and therefore suggest that only high-risk patients be referred to high-volume centres.11

The proportion of patients in this study with positive resection margins (38.7%) was higher than in most other reports (14%–40%).17 Identification of involved resection margins depends on both the definition being used and the individual pathologist. In experienced hands, the rate of positive resection margins should be about 20%.18 As the rate of positive margins decreases with increasing workload,17 the higher rate in our study may be related to the low annual workload of most of the surgeons.

Twenty patients in our study (2.6%) survived at least 5 years. However, half of these patients were managed palliatively without histological confirmation of carcinoma and presumably represent false-positive diagnoses. A review of 23 3-year survivors found only 11 had histological confirmation of carcinoma, and cautioned that including patients without histological confirmation would falsely inflate the survival rate.19 Therefore, the actual 5-year survival in our study is best recorded as 10 patients (1.3%). A previous report of outcomes in Victoria found a more optimistic 5-year survival rate of 5%.2 The difference may be explained by lack of detailed clinical follow-up in that study, resulting in the inclusion of false-positive diagnoses reported to the VCR.

Unfortunately, 5-year survival does not equal cure. We found that disease recurred after 5 years in three of 10 patients treated with resection. During long-term follow-up after Whipple resection, one study found recurrent carcinoma in five of 12 patients between 5 and 6 years of follow-up, and another reported that four of 11 5-year survivors later died of recurrent disease.13,20 These findings suggest that 6-year survival is a better proxy for cure than 5-year survival.

Older age, more advanced disease, and poor performance status are significant predictors of poor survival.21-23 Comparisons of outcomes from different management strategies should be adjusted for the patient characteristics in each group. The influence of patient selection can be seen in the outcomes of bypass surgery in this study. Bypass patients who had been referred for Whipple resection but found to have unresectable disease at laparotomy were younger and had better outcomes than those referred initially for palliative surgery.

Guidelines recommend that biliary obstruction in frail elderly patients with limited life expectancy is best palliated with endoscopic stenting.8 Compared with bypass surgery, endoscopic stenting has been found to have lower early mortality and morbidity, although recurrent jaundice is more common due to stent blockages.24 In our study, bypass surgery was used as initial palliation for unresectable tumours in similar proportions of younger and older patient groups, including 6.6% of patients aged 80 years or older. The selection of elderly patients for bypass surgery may be due to patient preference for local management in regional hospitals or to difficulty accessing endoscopic stenting.

Chemotherapy with gemcitabine reduces symptoms, with a modest improvement in survival, and should be offered to all patients with advanced disease and good performance status.25 In this study, gemcitabine was used more often in younger patients with a good performance status, as judged by 30-day mortality. Other studies have found that a decreasing proportion of patients receive chemotherapy with advancing age.21 Our results have been reported in detail previously.7

An important question when discussing major surgery is: What is the chance of cure? A survey of oncologists and their patients in NSW found that 98% of patients would like realistic information about their prognosis.26 The rate of 5-year survival after Whipple resection is often quoted as about 20%.14,15 However, this outcome refers to those with a clear resection margin — a subgroup that can only be identified after surgery. The outcomes for all those undergoing a laparotomy for Whipple resection are more relevant for informing the process of consent before surgery. We found that these patients have a 5-year survival of 6%, but the 6-year survival (1.5%) is a more realistic estimate of the probability of cure. On the other hand, patients with tumours in the head or neck of the pancreas managed with best supportive care had median survival of only 2.3 months — much less than the overall median survival of 4.5 months — and benefit from early referral to palliative care.

The population-based nature of this study avoids the biases inherent in descriptions of outcomes from a single institution. We found that outcomes for carcinoma of the pancreas in Victoria in 2002–2003 were comparable to those reported in other multicentre studies and from single institutions. Carcinoma of the pancreas has a grim prognosis, with few long-term survivors. We found that half of the long-term survivors among patients reported to the VCR had false-positive diagnoses, and previous estimates of long-term survival may therefore have been optimistic. Disease recurrence in 5-year survivors is common, and 6-year survival is a better proxy for cure.

1 Surgical management of patients with pancreatic cancer,* by site of tumour

* Patients not ultimately accounted for in the flow charts did not have any of the specified treatment options.

† Patients may have received more than one type of treatment.

4 Proportions of patients in each age group receiving each treatment, for all tumour types

* Includes both Whipple resection and distal pancreatectomy. † Head and neck of pancreas tumours — stent for initial palliation. ‡ Head and neck of pancreas tumours — biliary bypass for initial palliation. § No resection, stent or biliary bypass. ¶ With or without any other treatment.