1.0 GUIDELINES—SUMMARY DOCUMENT

The following recommendations are introduced by brief statements which summarise the evidence and discussion presented in the relevant section of the full text of the guidelines.

1.1 Incidence, mortality rates, and aetiology

Pancreatic cancer is an important health problem for which no simple screening test is available. The strongest aetiological association is with cigarette smoking, although at risk groups include patients with chronic pancreatitis, adult onset diabetes of less than two years’ duration, hereditary pancreatitis, familial pancreatic cancers, and certain familial cancer syndromes. Periampullary cancers are a feature of familial adenomatous polyposis.

1.2 Pathology

Most pancreatic cancers are of ductal origin and present at a stage when they are locally advanced, and exhibit vascular invasion and lymph node metastases. Variants of ductal carcinomas and other malignant tumours of the pancreas are rare.

1.3 Clinical features

In the majority of patients, the clinical diagnosis is fairly straightforward, although there are no positive clinical features which clearly identify a patient group with potentially curable disease. There are associated conditions, such as late onset diabetes mellitus or an unexplained attack of acute pancreatitis, which may point to an underlying cancer. A number of clinical features (persistent back pain, marked and rapid weight loss, abdominal mass, ascites, and supraclavicular lymphadenopathy) usually indicate an incurable situation

1.4 Investigations

The workup of patients with suspected pancreatic cancer should logically focus initially on establishment of the diagnosis and an assessment of the patient’s fitness to undergo potentially curative treatment. In selected patients, further investigation involves tumour staging and the assessment of local respectability.

1.5 Tissue diagnosis

1.6 Treatment

This largely centres around palliative surgery undertaken to relieve symptoms, resectional surgery with intent to cure, and endoscopic or percutaneous biliary stenting to relieve jaundice. There is an increasing use of chemotherapy and radiotherapy, both as palliative treatments as well as in an adjuvant setting in conjunction with surgery, although much of this practice is not evidence based. Appropriately designed multicentre clinical trials remain essential.

1.7 Organisation of services

The provision of effective services requires local cancer units as well as specialist centres.

1.7.1 Cancer units

These require sufficient diagnostic and therapeutic facilities to establish a likely diagnosis, assess the patient’s overall level of fitness to withstand potentially curative forms of treatment, and provide appropriate therapeutic facilities to ensure that adequate symptom palliation can be achieved.

Until services can be reorganised as specified by the NHS Executive, it is accepted that at some cancer units a specialist pancreatic surgeon may be available, and if the case load is sufficient, then resectional surgery may be justified on an interim basis This is only appropriate if the cancer unit has been approved to undertake resections by the Regional Upper Gastro-Intestinal and/or Hepato-Biliary-Pancreatic Cancer Network Group.

The minimum requirements for a cancer unit are:

• An integrated system of clinical care involving medical and surgical gastroenterology, clinical oncology, radiology, and pathology.

• Adequate radiological facilities to establish a diagnosis and the likely stage of disease. This should include abdominal ultrasound and a whole body imaging technique (CT or MRI). Guided biopsy techniques should be available for patients considered not suitable for surgical resection.

• Therapeutic facilities should include both endoscopic and radiological biliary stenting and, at least on an interim basis, facilities for surgical palliation.

• A variety of ancillary services are required, including palliative care, acute and chronic pain services, and clinical nutrition.

Local cancer units should provide guidance to primary health care physicians to ensure adequate patient referral. The following patient groups merit general practitioner referral to a local cancer unit:

• Obstructive jaundice.

• Unexplained weight loss.

• Unexplained gastrointestinal bleeding or iron deficiency anaemia thought to be of gastrointestinal origin in the absence of an upper gastrointestinal or colorectal cause.

• Unexplained upper abdominal or back pain.

• Unexplained steatorrhoea

• “Idiopathic” acute pancreatitis (no gallstones, no alcohol) in patients over 50 years of age.

• Unexplained diabetes in patients over 50 years of age (no family history, obesity, or steroids).

1.7.2 Specialist centres

These require all of the services provided by cancer units, with increased facilities for precise pretreatment staging of disease with particular emphasis on assessment of resectability, increased therapeutic resources, and adequate surgical expertise for pancreatic resections. They also require additional services in histopathology, intensive care, palliative care, and medical and clinical oncology, along with facilities for the organisation and conduct of local, national, and international trials. The Regional Cancer Network Group plan must ensure the timely establishment of the Regional Pancreas Tumour Centre based on a minimum of two million population that will undertake all pancreatic cancer resections in accordance with the National plans.

Specialist centres require all of the services provided at cancer units with further additions. These are:

• Facilities to include the majority of: spiral or multislice CT, MRI, endoscopic ultrasonography, and laparoscopic ultrasonography for precise pretreatment staging of disease with particular emphasis on assessment of resectability.

• Increased therapeutic resources, including expertise in radiological and endoscopic intervention and adequate surgical expertise for pancreatic resections.

• Additional services in histopathology (see pathology reporting), intensive care, palliative care, and oncology.

• Facilities for the organisation and conduct of local, national, and international trials, evaluating new modalities for diagnosis and treatment as well as involvement in basic science research in pancreatic cancer.

1.8 Audit and audit standards

Comprehensive clinical audit is essential. The minimum data set for the performance of an effective audit process is outlined below. The data set required in patients undergoing resection and the necessary information to complete this appropriately appear as an appendix.

1.8.1 Minimum data set for audit

• Accurate demographic information on all diagnosed cases.

• Duration of symptoms until first consultation.

• Duration from first consultation to referral to local cancer unit.

• Duration from date of referral to date of treatment.

• Accurate information on stage of disease involving the use of standardised histopathological assessments.

• Treatments received (the time from initial to definitive treatment should not exceed six weeks).

• Duration of hospital stays.

• Complications of treatment.

• Duration of survival.

• Quality of life assessments using validated instruments (for example, EORTC QLQ-C30) with a pancreatic cancer specific module (for example, QLQ PAN26), should be applied to all patients involved in prospective clinical trials.

The following standards are appropriate for clinical audit.

• Cancer units should respond to general practitioner requests within two weeks and specialist centres should respond to cancer unit referrals within a further two weeks.

• A full minimum data set should be available for all patients

• Resection rate in unselected patients should be more than 10%, and associated hospital mortality rate after pancreatic resection should be less than 10%.

2.0 PREPARATION OF THE GUIDELINES

This document covers a variety of areas which impact upon the production of clinical guidelines. The conclusions drawn at the end of each subsection have been used to generate a summary document. Due to considerable clinical similarities, pancreatic, periampullary, and ampullary cancers have been considered together.

These guidelines have been produced to help clinicians in the management of pancreatic and periampullary cancers. They were developed at the request of the Clinical Services Section of the British Society of Gastroenterology, with the support and endorsement of the Pancreatic Society of Great Britain and Ireland, the Association of Upper Gastrointestinal Surgeons of Great Britain and Ireland, the Royal College of Pathologists, and the Special Interest Group for Gastro-Intestinal Radiology. The guidelines were drawn up by a drafting committee under the Chairmanship of Professor Derek Alderson. The final document was prepared by a small writing committee and incorporates comments from members of the drafting committee and other interseted parties.

The evidence and recommendations have been assessed using a system designed by the Health Services Research Unit, University of Aberdeen. This system is summarised below.

2.1 Grading of evidence

• Grade Ia: meta-analysis of randomised controlled trials (RCT).

• Grade Ib: at least one RCT.

• Grade IIa: at least one well designed controlled study without randomisation.

• Grade IIb: at least one other type of well designed quasi-experimental study.

• Grade III: well designed non-experimental descriptive studies (for example, comparative, correlation, case studies).

• Grade IV: expert committee reports or opinions and/or clinical experiences of respected authorities.

2.2 Grading of recommendations

• Grade A: at least one RCT (Ia, Ib).

• Grade B: well conducted clinical studies (IIa, IIb, III).

• Grade C: respected opinions but absence of directly applicable good quality clinical studies (IV).

As the management of pancreatic cancer continues to evolve, new evidence will inevitably become available at regular intervals so that guidelines will need to be updated accordingly. The drafting committee considers that these guidelines will require revision within five years.

3.0 INCIDENCE AND MORTALITY RATES

The incidence of pancreatic cancer appears to have increased steadily in many countries for most of the 20th century. Mortality doubled in the UK between 1930 and 1970, but has risen much more slowly since then and it is the sixth most common cancer death in this country.^1,2 The incidence is higher in Western or industrialised countries in general.³ Pancreatic cancer is rare before the age of 45 years and 80% of cases occur in the 60–80 year age group.^4,5

Although there are considerable limitations to interpretation of epidemiological data,⁶ a study in the West Midlands indicated an age standardised incidence between 1960 and 1984 of approximately 10 cases per 100 000 population.⁷ There seems to have been some levelling of the annual incidence reported in this and other series.^8,9 Because the five year survival of this condition is so poor, incidence and mortality rates are virtually identical.

Pancreatic cancer has been more common in men than women but this is now beginning to change. In the USA, the Surveillance, Epidemiology, and End Results (SEER)¹⁰ programme has shown a fall in the total incidence of pancreatic cancer from 12.3 per 100 000 in 1973 to 10.7 per 100 000 in 1999.¹⁰ During the same period, the decline in rates for men was from 16.1 per 100 000 population to 12.1 per 100 000, and for women from 9.6 per 100 000 to 9.5 per 100 000, respectively.

Other periampullary tumours (of the ampulla, lower common bile duct, or duodenum) present with similar symptoms and signs to pancreatic cancer; without careful histological evaluation the differential diagnosis of tumour type may be impossible. The numbers of periampullary cancers are lower than pancreatic cancers, but they are more often resectable, so as many as half of pancreatic resections are for these periampullary tumours.

4.0 AETIOLOGY

The causes of pancreatic and periampullary cancer are not known. A variety of risk factors have been identified. The risk factor most consistently identified is cigarette smoking which may account for approximately 25–30% of cases.^11–20

Other factors including diet (high fat and protein, low fruit and vegetable intake), coffee consumption, alcohol, occupation, and the effects of other diseases such as diabetes mellitus, pernicious anaemia, chronic pancreatitis, cholelithiasis, and previous gastric surgery, have also been studied in detail. Of these, only in chronic pancreatitis and adult onset diabetes of less than two years’ duration does there seem to be clear evidence of an increased risk of pancreatic cancer.^19,21–23 Chronic pancreatitis is associated with an increased risk of cancer of the order of 5–15-fold.^19,21 Hereditary pancreatitis is associated with a 50–70-fold risk and a cumulative lifetime risk to the age of 75 years of 40%.^24,25

Pancreatic cancer may also occur in three other settings in which there is an inherited predisposition. Firstly, there appears to be an inherited component to pancreatic cancer in up to 10% of patients with pancreatic cancer in the absence of familial pancreatic cancer and other cancer syndromes.^26,27 Secondly, there is an increased incidence of pancreatic cancer in individuals from families with familial pancreatic cancer in which the disease appears to be transmitted in an autosomal dominant manner with impaired penetrance. Two recent studies have shown that approximately 17–19% of these families may have disease causing BRCA2 mutations in both Jewish and non-Jewish populations.^28,29 Thirdly, an increased risk of pancreatic cancer may occur as part of another cancer syndrome, including familial atypical multiple mole melanoma, Peutz-Jeghers syndrome, hereditary non-polyposis colorectal carcinoma (HNPCC), familial breast-ovarian cancer syndromes, and familial adenomatous polyposis (FAP) but probably not Li-Fraumeni syndrome.^30–36

The diagnosis and management of genetic predispositions to pancreatic cancer are developing rapidly. Consensus Guidelines of the International Association of Pancreatology advise that patients with an inherited predisposition to pancreatic cancer should be referred to specialist centres capable of providing expert clinical assessment of pancreatic diseases, genetic counselling, and advice on secondary screening.³⁷ In the UK, the national co-coordinating centre for secondary screening for pancreatic cancer is the European Registry of Hereditary Pancreatic Diseases (EUROPAC).³⁸

4.1 Periampullary cancers

Periampullary cancers can be broadly considered as those tumours arising out of or within 1 cm of the papilla of Vater and include ampullary, pancreatic, bile duct, and duodenal cancer. There is a high incidence of these tumours in patients with FAP.^35,39,40 The median interval between colectomy for FAP and the development of upper gastrointestinal cancer is 22 years³⁹ and cancer is often preceded by ampullary or duodenal adenomas^39,40 or arises in an adenoma.⁴¹ The frequency of periampullary neoplasms in FAP patients is sufficient to warrant a policy of regular duodenoscopy and biopsy of suspicious lesions. Duodenoscopy should be started when colorectal polyps have been diagnosed, and repeated at intervals of five years (stage 0/1 polyposis), three years (stage 2 polyposis), and one or two years for patients with stage 3 duodenal polyposis.⁴² Patients with stage 4 polyposis should be advised to have surgical resection by pylorus preserving pancreaticoduodenectomy.⁴²

Conclusions

• Pancreatic cancer is an important health problem.

• No simple screening test is available for the general population.

• With an increasingly elderly population, there can be no expectation of a marked reduction in incidence.

• The strongest aetiological association is with cigarette smoking.

• At risk groups include:

  • – Patients with chronic pancreatitis.

  • – Adult onset diabetes of less than two years’ duration.

  • – Patients with hereditary pancreatitis, familial pancreatic cancer, and certain other cancer family syndromes, notably ovarian and breast cancer syndrome and the familial multiple mole melanoma syndrome.

• Periampullary cancer is a feature of familial adenomatous polyposis

5.0 PATHOLOGY

Although a variety of exocrine pancreatic tumours exist, by far the most common is ductal adenocarcinoma which accounts for well over 90% of all tumours. In surgical resection series, 80–90% occur in the head of the gland.⁴³ Lymph node metastases are common and are present at the time of surgery in 40–75% of primary tumours less than 2 cm in diameter.⁴³ Perineural infiltration and vascular invasion are both frequently seen in resection specimens.

A variety of other exocrine tumours arise from the pancreas (see appendix for details) and because of their rarity they often require specialist pathological interpretation. Some, such as serous and mucinous tumours, intraductal-mucinous tumour, and solid-pseudopapillary tumour, have a very much better prognosis than pancreatic adenocarcinoma.^44,45 Endocrine tumours and lymphomas can be confused clinically and radiologically with pancreatic carcinoma. Some endocrine tumours have characteristic presentations such as insulinoma, glucagonoma, and gastrinoma. Management of these hormonally active neoplasms lies outside the scope of this document but the possibility of a clinically silent endocrine tumour should be considered when a mass is identified in the absence of other clinical features characteristic of pancreatic cancer. A tissue diagnosis is thus important in the management of a patient with a mass in the pancreas.

Conclusions

• Most pancreatic carcinomas are of ductal origin. They are usually locally advanced, exhibit vascular invasion, and lymph node metastases.

• Variants of ductal carcinomas and other malignant tumours of the pancreas are rare.

• Perineural and vascular invasion is extremely common in ductal adenocarcinoma

6.0 CLINICAL FEATURES AND DIAGNOSIS

The three main symptoms of pancreatic cancer are pain, loss of weight, and jaundice. Nausea, anorexia, malaise, and vomiting are also common. Persistent back pain is associated with retroperitoneal infiltration and usually incurability.⁴⁶ Severe and rapid weight loss are features that are usually also associated with unresectability.^47,48 Jaundice draws attention to ampullary tumours at a relatively early stage, which accounts for their higher resectability and may account for the better cure rates than for tumours further from the papilla. Conversely, jaundice in patients with carcinoma of the body or tail of the pancreas is usually caused by hepatic or hilar metastases and therefore indicates inoperability. Some 5% of patients with pancreatic cancer will have developed diabetes mellitus within the previous two years⁴⁹ and recent onset diabetes in older patients may therefore serve as a warning sign. As noted above, recent onset of diabetes mellitus without predisposing features is associated with an increased risk of diagnosis of pancreatic cancer. Acute and chronic pancreatitis are also possible presentations of pancreatic cancer as 5% of cancer patients will present with an atypical attack of acute or subacute pancreatitis.⁵⁰ In the absence of another recognised aetiology for an attack of pancreatitis, the possibility of an underlying carcinoma should be considered.

Migratory thrombophlebitis is rarely the first symptom of the disease. The same applies to the physical signs, apart from jaundice and a palpable gall bladder (Courvoisier’s sign). Other findings are conspicuous by their absence. A palpable and fixed epigastric mass, ascites, or an enlarged supraclavicular lymph node (Virchow’s node) are all signs of inoperability.

Conclusions

• In the majority of patients, the clinical diagnosis is fairly straightforward.

• There are no positive clinical features which clearly identify a patient group with potentially curable pancreatic or periampullary carcinoma.

• There are associated conditions, notably late onset diabetes mellitus and an unexplained attack of acute pancreatitis, which may point to an underlying pancreatic carcinoma.

• There are a number of clinical features (persistent back pain, marked, and rapid weight loss, abdominal mass, ascites, and supraclavicular lymphadenopathy) that usually indicate an incurable situation.

7.0 INVESTIGATIONS

There are no specific blood tests for the diagnosis of pancreatic carcinoma. Abnormal liver function tests cannot reliably distinguish biliary obstruction (of any cause) from hepatic metastases. The most useful initial investigation seems to be abdominal ultrasonography which can identify the pancreatic tumour, as well as dilated bile ducts, and will save considerable time and inconvenience if liver metastases are identified. The reported sensitivity of ultrasonography in the detection of pancreatic carcinoma is as high at 80–95%.^51–53 The technique however, becomes less sensitive in evaluating the body and tail and provides less accurate staging information than other modalities, such as CT.^54,55 Technical difficulties with bowel gas compromise interpretation in 20–25% of subjects,⁵⁶ and interobserver variation continues to be a problem.⁵³ Improvements in ultrasound technology, with inclusion of colour Doppler, may improve staging accuracy, particularly with respect to vascular invasion.⁵⁷

CT and, more recently, MR imaging, both reliably demonstrate the primary tumour and evidence of extrapancreatic spread, particularly in the presence of liver metastases.^58–61 Contrast enhanced CT, particularly using helical scanners with arterial and portal phases of contrast enhancement, accurately predicts resectability in 80–90% of cases.^62–67 Assessment of local tumour extension with contiguous organ invasion, vascular involvement, hepatic metastases, and lymph node metastases correlate well with surgical findings in large tumours. CT is, however, much less accurate in identifying potentially resectable small tumours and where alternative diagnoses may need to be considered.⁶⁸ Some centres believe that fine needle aspiration cytology under CT guidance is appropriate in these circumstances but this may be inadvisable if peritoneal seeding of cancer cells occurs, which might then eliminate the possibility of cure in otherwise potentially curable cases⁶⁹(see section on tissue diagnosis). Early results suggest that spiral CT allied to multislice technology and three dimensional reconstruction may prove advantageous in the identification of small tumours and resectability.^70–76 MR imaging detects and predicts resectability with accuracies similar to CT.^59,76,77 MRCP provides detailed ductal images without the risk of ERCP induced pancreatitis and may clarify diagnostic uncertainty (chronic pancreatitis versus cancer) as well as being informative regarding intraductal tumours^78–80). MRA can demonstrate vascular anatomy, and some have proposed a “one stop” investigation with MR, MRCP, and MRA. However, the value of this approach remains to be proven, and current practice is to obtain appropriate images with various techniques according to individual diagnostic questions and local expertise.

ERCP is important in the diagnosis of ampullary tumours by direct visualisation and biopsy. All other pancreatic tumours are detectable only if they impinge on the pancreatic duct so that small early cancers and those situated in the uncinate process can be missed by this technique. ERCP has the advantage of providing an opportunity to sample for cytology or histology and an important therapeutic modality via biliary stenting, to provide relief of jaundice and the associated symptom of pruritus.

Recent progress includes the use of endosonography (EUS) and the selective use of laparoscopy. EUS is highly sensitive in the detection of small tumours and invasion of major vascular structures^81,82 and can be used to avoid unnecessary surgery. EUS is superior to spiral CT, MR, or positron emission tomography in the detection of small tumours.^83–87 Laparoscopy, including laparoscopic ultrasound, can detect occult metastatic lesions in the liver and peritoneal cavity not identified by other imaging modalities.^88–90

Selective angiography has no place in establishing the diagnosis of pancreatic cancer but its use has been advocated by some authors as a means of detecting arterial anomalies and defining resectability. Most centres can now obtain this information non-invasively with CT or MR. While arterial anomalies are present in about a third of all patients undergoing pancreatic resection, this is nearly always an aberrant right hepatic artery, supplied from the superior mesenteric artery, and is detected at operation as pulsation posterior to the bile duct. This is easily recognisable and can be confirmed by intraoperative ultrasonography. Similarly, angiography is an unreliable method of predicting unresectability, with an overall predictive value in one recent series of only 61%.⁹¹

The workup of patients with suspected pancreatic cancer should logically focus initially on establishment of the diagnosis and an assessment of the patient’s fitness to undergo potentially curative treatment. In selected patients, further investigation involves tumour staging and the assessment of local resectability.

Conclusions

• Neither endosonography nor laparoscopic ultrasonography is widely available in the UK. Expertise and further evaluation of these techniques requires development.

8.0 TISSUE DIAGNOSIS

Tissue can be obtained by a variety of methods. Aspiration or brushing of the duct systems at ERCP have high specificity but low sensitivity.⁹² Guided biopsy or fine needle aspiration cytology can also be performed under EUS guidance.^93,94

The alternative approach involves a transperitoneal approach. This can be undertaken transcutaneously under ultrasound or CT guidance, or at the time of laparoscopy with either visual or ultrasound guidance. These techniques have high specificity with a low risk of procedure related complications.^95–97

There are however two concerns regarding transperitoneal techniques, particularly relevant to patients with small and potentially resectable tumours. Firstly, there is a risk of a false negative result. Failure to obtain histological confirmation of a suspected diagnosis of malignancy does not exclude the presence of a tumour, and should not delay appropriate surgical treatment. Secondly, there are concerns regarding tumour cell seeding along the needle track or within the peritoneum.^98–100 Although the study by Warshaw⁶⁹ showed that previous percutaneous biopsy significantly increased the incidence of positive peritoneal cytology in pancreatic tumours, most of the patients in this series who had positive cytology had advanced disease. In subsequent studies, fine needle aspiration did not increase the risk of positive peritoneal cytology.^101,102

The consequence of attempted resection without efforts to obtain a preoperative tissue diagnosis is that some patients will undergo resection for benign disease. This is probably the case in approximately 5% of all pancreaticoduodenal resections¹⁰³ and provided that pancreaticoduodenectomy can be undertaken with low morbidity and mortality, this represents an acceptable risk.

Given the above concerns, there seems little justification for transperitoneal biopsy in patients thought to have potentially resectable malignant lesions and those likely to benefit from surgery, even if benign disease is present. Conversely, reasonable efforts to obtain a tissue diagnosis should be made in patients selected to undergo palliative forms of therapy, to exclude variant tumour types which might have a better prognosis, and ensure patient eligibility for participation in trials evaluating new therapies.

9.0 TREATMENT

The treatment of pancreatic cancer has centred largely around palliative surgery undertaken to relieve symptoms, resectional surgery undertaken with intent to cure, and endoscopic or percutaneous biliary stenting to relieve jaundice. Chemotherapy and radiotherapy may also be used as palliative treatments, as well as in an adjuvant setting in conjunction with surgery.

9.1 Palliation by stent or surgery

There have been three controlled trials of palliation of obstructive jaundice by stenting or surgical bypass but the results do not favour one method for use in all cases.^104–106 The advantages of stenting include fewer immediate complications and shorter initial treatment time whereas surgery has better long term patency. Mortality rates at 30 days and median survival times are similar with the two techniques. It seems reasonable to reserve surgery for patients with good performance status and small tumours who are likely to survive longer than average, and to place a stent in patients with advanced tumours who are unlikely to survive longer than the usual patency time of the stent. The decision should also take account of the greater risk of early complications with the surgical approach.

We are not aware of any randomised comparison of expanding metal stents and bypass surgery for the relief of obstructive jaundice. There are reports of the use of expanding metal stents in duodenal obstruction but there is no convincing evidence that this approach offers a better outcome than surgical bypass.

9.2 Stent insertion

Endoscopic stent insertion into the biliary tree at the time of ERCP has been established for many years.¹⁰⁷ A number of studies have shown that the endoscopic approach is associated with lower morbidity and procedure related mortality rates than the transhepatic approach, by minimising the risk of bile leaks and bleeding.¹⁰⁸ Brush cytology and/or biopsies can be taken from within the bile duct at the time of ERCP, prior to stenting. If the stricture cannot be negotiated with a catheter and guidewire system, a combined approach involving insertion of a transhepatic catheter and guidewire, which can be retrieved by the endoscopist, will allow successful stent placement in a group of patients where endoscopic stenting alone is unsuccessful.^109,110 However, in most centres such patients are now treated by percutaneous stent placement.

Modern techniques and equipment for percutaneous stenting with a self expanding metal stent are associated with fewer complications than percutaneous plastic stent placement and may be appropriate for patients who have better than average life expectancy but who are unsuitable for surgical palliation, after occlusion of a plastic stent, or when endoscopic stent placement has failed.

Insertion of biliary stents is associated with complications such as cholangitis and perforation. After stent insertion, the most important clinical problem is stent occlusion due to deposition of a bacterial biofilm and precipitation of biliary sludge within stents made of plastic.¹¹¹ Recurrent jaundice usually indicates stent occlusion, rather than progressive disease. Such patients may need re-evaluation with a view to further stent placement. Occlusion is less problematic with self expanding metal stents, which open to a diameter of approximately 10 mm. As the lumen of this type of stent is so large, biliary drainage is superior to that seen with plastic prostheses, so that blockage due to debris hardly ever occurs. Conversely, tumour ingrowth through the mesh can occur. The use of thin membranes to cover self expanding stents may minimise this problem. It would appear however that the average patency of metal stents in the distal bile duct is about twice that of polyethylene stents, the latter usually lasting for about four months.^112,113 Some selection of patients thought likely to survive for greater than this length of time might be used to identify those patients who should receive a self expanding metal stent. Because at least two thirds of patients with pancreatic cancer will be successfully palliated with a single stent¹⁰⁶ and because the cost of a plastic prosthesis is approximately 3% or 4% of the cost of a self expanding metal prosthesis, both stent types should still be used appropriately.

Stenting is clearly best suited to patients with significant comorbid disease who are deemed unsuitable for surgery and those with proven widespread disease. While many clinicians view symptomatic gastrointestinal obstruction as a relative contraindication to biliary stenting, gastric outlet obstruction can be effectively palliated in some patients by a self expanding metal stent.¹¹⁴

9.3 Endoscopic stenting before resection

The role of endoscopic stenting as a preliminary to attempted resection, in an attempt to reduce surgical morbidity and mortality related to jaundice, remains controversial. Retrospective data indicating that this reduces surgical morbidity¹¹⁵ have not been supported by a prospective randomised controlled trial although the numbers of patients studied were small.¹¹⁶ Several other non-randomised studies confirm that similar results can be obtained in jaundiced patients without relief of biliary obstruction, as in those who are operated on after relief of jaundice by endoscopic stenting.^117–121 It is well established that preliminary external biliary drainage does not favourably influence hospital morbidity or mortality prior to pancreas resection in jaundiced patients.^122–125

There is agreement based on anecdotal experience that surgical resection is made more difficult by the preoperative insertion of self expanding metal stents. This is attributed to the tissue reaction provoked by these stents, and the potential difficulty that may arise if the stent crosses the preferred line of bile duct division.

9.4 Resectional surgery

There is wide variation in resection rates and operative mortality rates in pancreatic cancer surgery. There is considerable evidence that operative mortality rates can be kept to low single figure values when undertaken in specialist centres.^126–128 These results contrast markedly with those obtained in the West Midlands where the resection rate in the two decades to 1976 and 1986 was only 2.6%, with an operative mortality of 45% and 28% in the two periods.⁷ A similar study conducted by the New York State Department of Public Health demonstrated a clear correlation between caseload and surgical mortality. When surgeons performed less than nine resections annually, mortality was 16% compared with less than 5% for surgeons performing more than forty cases per year.¹²⁷ Similar relationships between hospital volume and mortality have been reported by other authors.^130,131 A survey of 2.5 million complex surgical procedures showed a large inverse relationship between hospital volume and case mortality rates for pancreatic resection.¹³²

In specialist centres, resectability rates are high at approximately 20%, reflecting referral practices and case selection.^{126,133–135} The most widely employed procedure is the Whipple pancreaticoduodenectomy, with a five year survival following resection of approximately 10%.^136,137 More radical approaches have been adopted, such as total pancreatectomy or portal vein excision,^138–140 as well as more conservative approaches to include pylorus preservation in order to improve the quality of survival.^141,142 There are four acceptable types of operation: proximal pancreaticoduodenectomy with pylorus preservation; proximal pancreaticoduodenectomy with antrectomy (Kausch-Whipple); total pancreaticoduodenectomy; and left (distal) pancreatectomy.

9.4.1 Proximal pancreaticoduodenectomy

Large series have indicated that the pylorus preserving operation does not compromise long term survival figures compared with the standard Whipple’s operation for carcinoma for the head of the pancreas.¹⁴³ The potential drawbacks of the pylorus preserving operation are tumour involvement of the duodenal resection line and incomplete removal of regional lymph nodes.^144,145 These risks can be obviated by patient selection so that the pylorus preserving operation is avoided in patients where there is proximal duodenal involvement or the tumour is close to the pylorus.^146–148

The advantages of pylorus preservation have not been conclusively established but may include a reduction in post gastrectomy complications, a reduction in enterogastric reflux, and improved postoperative nutritional status and weight gain compared with the standard Whipple operation.^{145,149–152}

9.4.2 Total pancreaticoduodenectomy

This has no advantage in long term survival compared with Whipple’s resection^153,154 and has its own troublesome nutritional and metabolic sequelae.^140,155 The procedure may be justified where there is diffuse involvement of the whole pancreas without evidence of spread.

9.4.3 Left pancreatectomy

This resection is indicated for lesions in the body and tail of the pancreas. Ductal carcinoma is seldom resectable in this location¹⁵⁶ but this procedure may be appropriate for a variety of the other slow growing malignant tumours (see histopathology appendices).

9.4.4 Radical and extended resections

Modifications of these standard operations to include the portal vein and a block of lymphatic tissue around the origins of the coeliac and superior mesenteric arteries was proposed by Fortner and colleagues.¹⁵⁷ In most centres, postoperative morbidity and mortality were higher than that encountered in the standard Whipple resection, although more recently a number of centres have reported mortality rates in the range 3–7%.^158–161 There are no data to indicate that this more radical approach is associated with increased survival.^139,162,163 A randomised controlled trial of extended versus standard lymphadenectomy also failed to demonstrate survival benefit.¹⁶⁴

9.4.5 Venous involvement

Most surgeons agree that resection should not be undertaken with intent to excise tumours where there is clear preoperative evidence of venous encasement. It is believed that this situation is more hazardous for the patient, as a result of preoperative segmental portal hypertension, and some evidence exists that survival is not greatly different to that seen in patients who are not resected.¹⁶⁵ Resection of the portal or superior mesenteric vein as a means of ensuring that resection with tumour free margins becomes feasible is appropriate if vein involvement is discovered during pancreaticoduodenectomy. This extension of the procedure does not increase operative morbidity or mortality¹⁶⁶ and long term outcome is not affected by the need for vein resection.¹⁶⁷

9.5 Palliative surgery

A number of prospective randomised studies have been undertaken to compare palliative biliary drainage surgery with stenting, performed either endoscopically or by a transhepatic approach. In a direct comparison of plastic stent placement, procedure related morbidity and mortality rates were lower when the endoscopic route was used compared with the transhepatic route.¹⁰⁸ Similarly, endoscopic stenting has a lower procedure related complication rate and mortality than surgical bypass, although this is at the expense of a higher risk of recurrent jaundice and a greater risk of gastric outlet obstruction.¹⁰⁶ There is no recent published comparison of surgery and other methods of palliation: it is appropriate to consider surgery in low risk patients with potential for longer than average survival. Operative risk can be assessed using scoring systems,^168–170 and the absence of an acute phase protein response has been shown in one study to be associated with longer survival.¹⁷¹ These features may help select patients for surgical palliation. While these procedures can be carried out by laparoscopic, as well as by open, means there are no data at present to indicate superiority of either approach.

A variety of bypasses have been employed. Relief of jaundice is more reliably attained when the bile duct is used rather than the gall bladder.^172,173 Addition of a duodenal bypass when there is gastric outflow obstruction does not increase operative risk.^174,175 Approximately 17% of patients treated by biliary bypass alone subsequently require a gastroenterostomy.¹⁷³ Prophylactic gastrojejunostomy decreases the incidence of late gastric outlet obstruction.¹⁷⁶

9.6 Non-surgical therapies

The objectives of radiotherapy and chemotherapy in pancreatic cancer may be considered under three headings:(1) neoadjuvant or adjuvant therapy—therapy given prior to, during, or after surgery where the aim is to improve survival;(2) in the management of locally advanced disease, not amenable to surgical therapy; and (3) metastatic disease where the primary objective is palliation and prolongation, where possible, of a symptom free life.

9.6.1 Adjuvant and neoadjuvant treatments

9.6.1.1 Adjuvant therapy

A prospective randomised controlled study of adjuvant chemoradiation (5-fluorouracil (5-FU) for six days and 40 Gy of radiation followed by maintenance chemotherapy with 5-FU) after pancreaticoduodenectomy conducted by the Gastrointestinal Tumour Study Group¹⁷⁷ demonstrated a survival advantage for multimodal therapy compared with resection alone. However, the total number of patients in this trial was only 43, and because of slow postoperative recovery, 24% of the patients in the adjuvant chemoradiation arm did not begin chemoradiation until more than 10 weeks after surgery.

Two other randomised controlled trials have examined the role of postoperative chemoradiation therapy. An EORTC study of pancreatic and ampullary cancers found no benefit on survival for patients treated with radiation and 5-FU in a chemoradiation protocol similar to the GITSG study but without maintenance chemotherapy.¹⁷⁸ The European Study Group for Pancreatic Cancer (ESPAC) reported a large trial (ESPAC-1) of 546 patients which compared adjuvant chemoradiotherapy with or without maintenance chemotherapy (5-FU with folinic acid) against no treatment.¹⁷⁹ This showed no benefit for chemoradiotherapy and a probable survival advantage for prolonged chemotherapy after resection. Specific analysis according to resection margin status also failed to show any benefit for chemoradiotherapy but with the same proportional benefit for chemotherapy.¹⁸⁰ A further study is in progress to compare adjuvant 5-FU with folinic acid, gemcitabine, and no adjuvant therapy (ESPAC-3 trial).

A survival advantage was also demonstrated for adjuvant chemotherapy (5-FU, doxirubicin, mitomycin C) in another randomised controlled trial. Median survival was 23 months in 30 patients randomised to receive adjuvant therapy compared with 11 months in 31 patients treated with surgery alone.¹⁸¹ However, 46 additional patients were ineligible for the study following surgery and the toxicity of chemotherapy was significant. Only one third of the patients allocated actually received all six planned cycles of chemotherapy. This study is open to criticism of selection bias for protocol entry, selecting such therapy for patients who recover rapidly from surgery and have good performance status. Other studies have shown broadly similar effects without clear evidence of survival benefit.^137,182,183 At present, adjuvant therapy is not considered standard therapy. Further studies are planned or in progress, which should provide additional data regarding the potential benefits of adjuvant therapies.

9.6.1.2 Neoadjuvant therapy

An alternative strategy is to give non-surgical therapies before or during surgery. At present, reported studies rely on external beam radiotherapy or chemoradiation and are non-randomised. These studies suggest that there may be an improvement in locoregional control but no significant improvement in survival.^184–191 Neoadjuvant therapy remains investigational in pancreatic cancer.

9.6.1.3 Intraoperative radiotherapy

At present, no centre in the UK is using intraoperative radiotherapy. Despite some reports from centres with access to the appropriate equipment, there is at present no evidence of benefit with this technique to support its development in the UK.

9.6.2 Combined therapy for locally advanced disease

Patients with locally advanced non-metastatic disease have a median survival of 6–10 months. Reports of treatment without a control group provide no useful evidence to judge efficacy. However, improved median survival in a study of 64 such patients was demonstrated with a combination of external beam radiotherapy plus 5-FU compared with radiotherapy alone (10.4 versus 6.3 months, respectively)¹⁹²; 5-FU has remained the mainstay of chemoradiotherapy since then.¹⁹³

To control metastases outside the radiation field, chemoradiotherapy has been combined with maintenance chemotherapy. Two GITSG studies^194,195 and an Eastern Cooperative Oncology Group (ECOG) trial¹⁹⁶ showed no survival benefit for chemoradiotherapy and maintenance chemotherapy with a variety of agents. Overall, the results are not convincingly better than for chemotherapy alone.

9.6.3 Chemotherapy for non-resectable localised, metastatic, or recurrent disease

Patients with metastatic disease have a limited survival of 3–6 months, dependent on the extent of disease and performance status. Many patients will not wish or be suitable for anticancer therapy. Well motivated patients with good performance status may gain psychological benefit from palliative chemotherapy; increased duration of survival has been shown in a few trials.^197–199 The best objective response rates historically were achieved with 5-FU and mitomycin C.²⁰⁰ Chemotherapy regimens that use 5-FU based doublet or triplet therapies have tended to be associated with greater toxicity without any survival advantage.²⁰¹ However, since the introduction of gemcitabine the scene appears to be changing.

Gemcitabine is a deoxycitidine analogue that has been extensively evaluated, including in a randomised trial against bolus 5-FU.²⁰² Patients treated with gemcitabine achieved modest but significant improvements in response rate and survival. There was also evidence of improvement in disease related symptoms, including a clinical benefit response (based on pain control, performance status, and weight gain) in 24% of gemcitabine treated patients, as opposed to 5% with 5-FU. This was the pivotal trial used to obtain licensing of gemcitabine. A recent NICE evaluation concluded that gemcitabine may be considered as a treatment option for patients with advanced or metastatic adenocarcinoma of the pancreas and a Karnofsky performance score of 50 or more, where firstline chemotherapy is to be used. If chemotherapy is to be used in patients with pancreatic adenocarcinoma, gemcitabine appears to be the agent of choice.²⁰³ There are now numerous phase II and phase III studies of doublet and triplet regimens that include gemcitabine as one of the active agents.²⁰⁴

There remains continued interest in fluoropyrimidines, as seen in several studies of protracted venous infusion 5-FU and the development of orally active agents (including capecitabine (Xeloda), ZD9331, and Tegafur) as well as other antimetabolites (including raltitrexed (Tomudex) and pemetrexed (Alimta, LY231514)). Maisey and colleagues²⁰⁵ randomised patients to protracted venous infusion of 5-FU with or without mitomycin C. The response rate was significantly higher in the combination arm (17.6% and 8.4%, respectively) and toxicities in both arms were mild but the difference in response rates did not translate into a significant difference in median survival (6.5 v 5.1 months, respectively). Cancer Research UK has recently launched the Gem-Cap Trial which will compare gemcitabine with or without capecitabine in a large phase III study in advanced pancreatic cancer.

9.6.4 Other treatment approaches

Pancreatic tumours contain sex hormone receptors. Suggestion of a survival benefit for tamoxifen has however not been confirmed in a randomised study.²⁰⁶ Metalloproteinase inhibitors such as marimastat have shown considerable promise, both as a single agent and in combination with gemcitabine,^207,208 but their clinical utility has not been supported by a larger study²⁰⁹ benefit with any of these at present.

10.0 OTHER ASPECTS OF MEDICAL MANAGEMENT

10.1 Relief of pancreatic pain

Pain is a common presenting feature and in patients with advanced disease can be intolerable, providing a major therapeutic challenge. Various factors are thought to produce pancreatic pain, including increased parenchymal pressure secondary to ductal obstruction, neural infiltration, superimposed pancreatic inflammation, and associated biliary stenosis.²¹⁰ The World Health Organisation analgesic ladder recommends three steps, from non-opioids, to opioids for mild to moderate pain, and then to opioids for moderate to severe pain. A variety of measures have been proposed to alleviate pancreatic pain in addition to oral and parenteral analgesics. Adjunctive approaches include pancreatic ductal decompression by endoscopic and surgical means.^210,211 Percutaneous, laparoscopic, or open ablation of the coeliac ganglia using 5% phenol or 50% ethanol produces effective palliation of pain in approximately 70% of patients.^{173,212–214} The technique is most effective when used early rather than late in the course of disease and does reduce the consumption of other analgesics.^212,214 Thoracoscopic division of the splanchnic nerves has also been described as an effective method.^215,216

Pancreatic pain may be palliated by external beam radiotherapy, particularly when this recurs after coeliac plexus blockade.¹⁷³ While the survival benefit of chemoradiation compared with chemotherapy alone is questionable, phase II studies typically report temporary pain relief in as many as 40–80% of patients.^217,218

10.2 Nutritional aspects of care

10.2.1 Pancreatic enzyme supplements

Compared with untreated patients, patients with advanced pancreatic cancer who are given pancreatic enzyme supplements enjoy a better quality of life and improved symptom score.²¹⁹

10.2.2 Lipid supplements

There is some evidence that lipid supplements with unsaturated fats, such as fish oil, may reduce weight loss and cachexia, and may prolong survival.^220,221

10.3 Palliative and supportive care

There is good evidence that dying patients and their families benefit from the specialist attention which can be provided by palliative care units²²² and hospices.²²³ In addition to pain, depression is a common problem in pancreatic cancer patients^224,225 which may require treatment in its own right.

11.0 ORGANISATION OF SERVICES

The NHS Executive Evidence “Improving outcomes in upper gastrointestinal cancers” was published early in 2001.²²⁶ A key recommendation is the establishment of cancer centres and units, the former providing surgery for pancreatic cancers and dealing with population bases of between two and four million.

For a variety of reasons not all pancreatic centres are currently capable of offering a complete range of services to deal with all patients. Provision of effective services will require cancer units as well as specialist centres. It is acknowledged that appropriate service reconfiguration will require time and interim local arrangements will remain necessary, reflecting existing resource allocations.

11.1 Cancer units

Such units require sufficient diagnostic and therapeutic facilities to establish a likely diagnosis, assess the patient’s overall level of fitness to withstand potentially curative forms of treatment, and provide appropriate therapeutic facilities to ensure that adequate symptom palliation can be achieved.

It is accepted that in some cancer units a specialist pancreatic surgeon with appropriate training and experience to justify resectional surgery may be available and may be required to continue to provide this service until service reconfiguration can be achieved. Provision of pancreatic resection in such cancer units should continue only with the approval of the Regional Upper Gastro-Intestinal and/or Hepato-Biliary-Pancreatic Cancer Network Group.

The minimum requirements for a cancer unit are:

• An integrated system of clinical care involving medical and surgical gastroenterology, clinical oncology, radiology, and pathology.

• Adequate radiological facilities to establish a diagnosis and the likely stage of disease. This should include abdominal ultrasound and a whole body imaging technique (CT or MRI). Guided biopsy techniques should be available for patients considered not suitable for surgical resection.

• Therapeutic facilities should include both endoscopic and radiological biliary stenting and, at least on an interim basis, facilities for surgical palliation.

• A variety of ancillary services are required, including palliative care, acute and chronic pain services, and clinical nutrition.

Local cancer units should provide guidance to primary health care physicians to ensure adequate patient referral. The following patient groups merit general practitioner referral to a local cancer unit:

• Obstructive jaundice.

• Unexplained weight loss.

• Unexplained gastrointestinal bleeding or iron deficiency anaemia thought to be of gastrointestinal origin in the absence of an upper gastrointestinal or colorectal cause.

• Unexplained upper abdominal or back pain.

• Unexplained steatorrhoea.

• “Idiopathic” acute pancreatitis (no gall stones, no alcohol) in patients over 50 years of age.

• Unexplained diabetes in patients over 50 years of age (no family history, obesity, or steroids).

It can be anticipated that such a unit should be capable of providing effective palliation for 70–80% of patients in whom the diagnosis of pancreatic cancer is made. The implication is that 20–30% of patients will require referral to specialist centres.

11.2 Specialist centres

Specialist centres are justified for three main reasons. Existing data indicate that hospital mortality related to surgical resection is related to operative experience and volume. If it is accepted that approximately 20% of patients will benefit from resection, then from current epidemiological information, each year a centre would carry out 20 resections per million of population. Concentration of cases to achieve these numbers will be vital in the UK in future to provide adequate training for surgeons in upper gastrointestinal surgery. The Regional Cancer Network Group plan must ensure the timely establishment of the Regional Pancreas Tumour Centre based on a minimum of two million population that will undertake all pancreatic cancer resections in accordance with the National plans.

Specialist centres require all of the services provided at cancer units with further additions. These are:

• Facilities for precise pretreatment staging of disease with particular emphasis on assessment of resectability. These should include the majority of the following: spiral or multislice CT, MRI, endoscopic ultrasonography, laparoscopic ultrasonography.

• Increased therapeutic resources, including expertise in radiological and endoscopic intervention and adequate surgical expertise for pancreatic resections.

• Additional services in histopathology (see pathology reporting), intensive care, palliative care, and oncology.

• Facilities for the organisation and conduct of local, national, and international trials, evaluating new modalities for diagnosis and treatment as well as involvement in basic science research in pancreatic cancer.

12.0 AUDIT AND AUDIT STANDARDS

Comprehensive clinical audit is essential. The performance of an effective audit process includes the following:

• Accurate demographic information on all diagnosed cases.

• Duration of symptoms till first consultation.

• Duration from first consultation to referral to local cancer unit.

• Duration from date of referral to date of treatment.

• Accurate information on stage of disease involving the use of standardised histopathological assessments.

• Treatments received (the time from initial to definitive treatment should not exceed six weeks).

• Duration of hospital stays.

• Complications of treatment.

• Duration of survival.

• Quality of life assessments using validated instruments (for example, EORTC QLQ-C30) with a pancreatic cancer specific module (for example, QLQ PAN26) should be applied to all patients involved in prospective clinical trials.

The following standards are appropriate for clinical audit:

• Cancer units should respond to general practitioner requests within two weeks and specialist centres should respond to cancer unit referrals within a further two weeks.

• A full minimum data set should be available for all patients.

• Resection rate in unselected patients should be more than 10%, and associated hospital mortality rate after pancreatic resection should be less than 10%.

13.0 OTHER ORGANISATIONAL ISSUES

Because of wide variations in the extent of services between hospitals in the UK it remains difficult, in some aspects of practice, to provide firm guidelines which are immediately applicable. The following, however, represent elements which both cancer units and specialist centres should be capable of achieving for patients with pancreatic adenocarcinoma.

• Joint assessments involving appropriate physicians, surgeons, oncologists, radiologists, histopathologists, specialist nurses, research personnel, and representatives from intensive care, palliative care, and nutritional services.

• Appropriate high dependency, intensive care, and anaesthetic facilities for pancreatic surgery. An adequately equipped and staffed system of graduated care is important. Anaesthetists and intensivists at consultant level should be familiar with the specialised surgery involved, in particular the nature and duration of surgery, which can be prolonged.

14.0 HISTOPATHOLOGICAL REPORTING

This is of greatest importance in patients who have undergone surgical resection. Accurate and reproducible information demands an understanding of histological typing, grading, staging, and clinical residual tumour classification.

The appendix includes the minimum data set required for histopathological reporting for carcinomas arising from pancreas, bile duct, and ampulla of Vater.

16.0 APPENDIX