Recent advances in the surgical treatment of pancreatic cancer

Abstract

Key Words: pancreatic neoplasms/surgery, pancreatic neoplasms/diagnosis

INTRODUCTION

Pancreatic cancer remains the fourth commonest cause of cancer related death in the western world[1]. The prognosis remains dismal due partly to late presentation, with associated low resectability rates, and the aggressive biological nature of these tumors. The median survival time from diagnosis in unresectable tumors remains only 4-6 mo.

For those patients amenable to surgical resection over the last 20 years have seen marked improvements in postoperative mortality and morbidity, especially in specialist pancreatic centres[2,3]. Despite these changes long-term survival remains low, with a total 5-year survival rate remaining less than 5%. Patients with ampullary cancer have a better 5-year survival of 40%-60%.

Resection, however, remains the only chance of long term survival with adjuvant therapies providing disappointing results. Operability remains low due to the local and distant extent of the disease. Assessment of this extent has been greatly advanced by modern radiological techniques.

PREOPERATIVE ASSESSMENT

Once distant disease has been excluded, selection of patients for resection is crucial if the rate of irresectability discovered at operation is to be kept to a minimum.

Angiography

Once considered crucial in the assessment of operability, angiography is now virtually unnecessary. It was argued that the venous phase of the arteriogram was fundamental if invasion of the superior mesenteric and portal vein were to be excluded. This has now been superseded by improvements in helical CT scanning, which is able to accurately determine venous involvement. The issue of preoperative detection of vascular anomalies should not be an indication for routine angiography in the hands of experienced pancreatic surgeons.

Computerised tomography (CT)

CT is still the traditional imaging modality for staging pancreatic cancer, although it lacks specificity and sensitivity. Spiral CT with intravenous contrast offers higher resolution than conventional CT and improves diagnostic and staging accuracy[4]. Unfortunately CT is still limited by its ability to differentiate between benign and malignant lesions and also may miss subcentimetre hepatic deposits and peritoneal seedlings. The impact of helical CT scanning using defined pancreatic protocols with multiplanar reconstruction has yet to be assessed, but the extent of local disease, involvement of arteries and veins and lymphatic spread can be assessed with improved accuracy (as illustrated in the following CT images).

Magnetic resonance imaging (MRI/MRCP)

Although a relatively new technique for assessing pancreatic lesions, MRI is particularly useful at differentiating inflammatory from neoplastic pancreatic lesions. Pancreatic adenocarcinomas are usually low signal on T1 and T2 weighted images[5], although it has as yet not been shown to be superior to CT in assessing operability. The use of magnetic resonance endoscopy may in the future improve the accuracy of MRI. At present, both MRI and CT scanning give additional information such that both techniques are of value in assessment.

Laparoscopy and laparoscopic ultrasound

Laparoscopy has the advantage of being able to detect small (＜ 10 mm) hepatic deposits (with and without laparoscopic ultrasound), assess degree of lymph node involvement and identify peritoneal disease over that detected by conventional imaging. The position of the pancreas in the retroperitoneal area does, however, limit its ease of application.

When combined with laparoscopic ultrasound, a sensitivity up to 90% for predicting operability have been suggested[6,7] which may be further improved with the addition of peritoneal lavage cytology[8]. Positive lavage cytology for tumor cells is associated with lower resectability and lower survival rates[9].

Determination of irresectability by laparoscopy in patients deemed operable by CT is usually due to the detection of occult liver metastases and peritoneal seedlings. Laparoscopic ultrasound allows direct placement of the probe onto the pancreas providing views of the tumor in relation to the superior mesenteric and portal veins. Such reports have led to the routine use of staging laparoscopy in some units prior to resection, although laparoscopic ultrasound is still relatively uncommon.

The issue of port site metastases is a risk in this technique, although the mechanism is not completely understood.

Endoscopic ultrasound (EUS)

This technique involves real time ultrasound scanning from within the lumen of the stomach and duodenum. Initial reports suggested highly reproducible results in expert hands in relation to staging of pancreatic cancer, especially in regard to vascular invasion, providing more accurate results than conventional CT[10]. It is less useful at demonstrating malignant lymph node involvement due to confounding factors such as nodal reactivity and inability to differentiate nodal micrometastases[11]. EUS is also used to obtain fine needle aspiration cytology of pancreatic masses and has yielded almost 80% diagnostic sensitivity[12]. One of the main drawbacks to this technique is that it is highly user dependent.

Positron emission tomography (PET)

PET involves the injection of radiolabelled glucose which is preferentially picked up by malignant cells and metabolised. PET scanning has found increasing applications throughout oncological assessment and when combined with CT scanning provides both anatomical and functional information. The most commonly used agent is 2-Fluoro-2-deoxy-D-glucose (FDG), which when taken up by cells and metabolised emits positrons that collide with electrons creating gamma rays, which are then picked up by a ring detector.

PET aids diagnosis of adenocarcinoma of the pancreas and also improves the sensitivity of conventional imaging in the detection of peritoneal and hepatic disease, enabling correlation between CT detected abnormalities and PET findings. Initial studies suggest that when PET is combined with EUS it may be more accurate than CT in detecting metastatic disease[13].

In addition to aiding diagnosis and staging, the extent of uptake of the radiolabelled substrate may correlate with the aggressiveness of the tumor, with a higher uptake predicting shorter survival[14,15]. This extent of expression may also help differentiate benign from malignant lesions[16].

SURGICAL APPROACH TO RESECTABLE PANCREATIC CANCER

Ninety percent of all malignant pancreatic exocrine tumors are accounted for by pancreatic duct cell cancer, with about two thirds occurring in the head of the gland. Convention states that tumors arising from the left side of the gland have a worse prognosis due to later presentation. Whilst it is true that tumors arising in the head may be detected at an earlier stage due to the associated jaundice, more recent data suggests that resectable tumors in the head, uncinate process and neck have a similar survival to those in the body and tail[3].

A number of issues arise in relation to the surgical approach to pancreatic cancer and include: preoperative biliary drainage, extent of lymphadenectomy, vascular resection and pylorus-preserving versus standard Whipple.

Preoperative biliary drainage

The issue of preoperative biliary drainage either endoscopically or transhepatically remains controversial. Trede and Schwall, in a large retrospective study, demonstrated that patients who underwent preoperative endoscopic stenting had a reduced incidence of postoperative infective complications[17]. Other studies failed to show any benefit and suggested an increased postoperative risk associated with stenting in relation to sepsis introduced with instrumentation. Povoski et al[18] demonstrated that the primary predictor of postoperative complications was the preoperative placement of biliary stents.

For patients in whom the date for resection is to be delayed, or in those undergoing palliation of jaundice, biliary stenting is usually recommended.

Extent of lymphadenectomy

A wide variety of opinion exists regarding the extent of lymphatic dissection for both tumors on the right and left side of the pancreas. Results vary depending upon the institution/country of origin and is partly due to the disparity that exists in the definitions used. Recent standardisation of terminology in this regard may clarify future studies, with precise definitions of ‘standard’, ‘radical’ and ‘extended radical’ lymphadenectomy.

Although extended surgical procedures popularised in Japan may increase the resectability rate up to 50%, this does not necessarily translate into improved survival. The retrospective nature of much of this data renders interpretation difficult and requires validation. Node positivity in many series is a strong negative predictor for survival with a 5-year survival stated to be less than 5%[19,20] whilst others using radical lymphadenectomy demonstrate a 5-year survival of up to 70% for node negative patients[21], results not reproduced elsewhere. These results are supported by the finding of metastatic disease in lymph nodes removed during extended lymphadenectomies for early disease[22].

Extended lymphatic resection when performed by exponents of this technique report equivalent rates for morbidity and mortality to those found in standard resections[23]. Hirata et al[24] reported a retrospective series of 1001 resected pancreatic cancers from 77 centres in Japan. The majority of these were stage IV according to the standard of Japanese Pancreatic Society (JPS). One hundred and thirty-one patients had a D0 resection (no lymph nodes), 365 a D1, and a D2 in 505 (422 of whom also underwent a further resectional procedure such as IVC reconstruction). The mortality rate was 2.5%. Improved survival was only found in those patients with N1 nodes positive who had a D1 resection.

In a retrospective series, Naganuma et al[25] demonstrated a significant improment in survival for patients who underwent extended resections compared to earlier more ‘standard’ procedures. This study has several failings, one of which is its retrospective nature, and another is the highly selective process of determining which patients underwent a extended resection in the second group.

A relatively small (n = 81) randomised, multicentre, prospective study of standard versus extended lymphadenectomy was reported by Pedrazzoli et al[23]. There was a significant improvement in survival for the node positive patients in the extended lymphadenectomy group, although when the whole group was analysed there was no difference. The survival curve of node positive patients undergoing extended lymphadenectomy could be superimposed on that for node negative patients. There was no difference in postoperative morbidity and mortality between the two groups. Multivariate analysis of all the patients demonstrated that long-term survival was significantly affected by tumor differentiation, tumor size, nodal involvement and more than a four unit blood transfusion.

Vascular resection

Whilst technically feasible in some patients en bloc resection of infiltrated peripancreatic vessels does not appear to improve survival, with vessel involvement a strong negative predictor of outcome[26,27]. There are few groups reporting worthwhile survival figures for patients undergoing vascular resection and again in a retrospective format[28]. Morbidity is high due to the extensive autonomic neurectomy causing diarrhoea.

Some groups have advocated a more widespread use of venous resection in view of the high rate of vessel involvement by tumor in some patients thought to be macroscopically clear, although the relevance of such findings to survival is hard to quantify[29].

Pylorus-preserving versus standard Whipple

Traditional teaching had been that proximal pancreatic resection should be combined with distal gastrectomy to reduce gastric acid production and hence minimise the risk of postoperative stress ulceration and also to improve oncological clearance. This approach has now been modified with preservation of the en tire stomach and proximal duodenum. This reduces the complications associated with gastrectomy and does not appear to increase the risk of marginal ulceration. Apart from dorsally placed tumors abutting the duodenum oncological clearance does not appear to be affected although there is a higher incidence of postoperative delayed gastric emptying[30].

COMPLICATIONS OF PANCREATIC RESECTION

In specialist centres perioperative mortality should be less than 5% and in some are less than 1%, although the incidence of postoperative complications is still 30%-40%[2,3,34]. The complications most commonly found are leakage of the pancreatic anastomosis, haemorrhage, abdominal abscess and delayed gastric emptying.

Pancreatic fistulae occur with a reported frequency of 4%-24%[3,35] and may progress to a full leak with associated sepsis and haemorrhage. Such serious leaks occur in less than 5% in specialist units and may carry a mortality of up to 40%. Clearly, access to experienced interventional radiology is crucial in these cases to deal with associated intra-abdominal collections. There is a correlation between the frequency of leaks and the pancreatic consistency, duct diameter and extent of residual exocrine function.

Although some groups routinely stent the pancreatic duct there is currently no randomised data to substantiate this practice, which also applies to the practice of creating separate Roux loops for high risk cases. The method of reconstruction is also debatable, i.e. end-to-side or end-to-end, or invagination versus mucosa to mucosa, but again lacks randomised data. The experience of the surgeon is probably the most crucial factor in determining leak rates [36]. Inhibition of pancreatic exocrine function using the somatostatin analogue octreotide prophylactically in the perioperative period has been shown to reduce postoperative morbidity associated with pancreatic fistulae in a number of randomised studies in Europe[37,38].

SURVIVAL POST RESECTION

According to the review on the resection of pancreatic cancer published in 1996 by Sperti et al[39], the rate of resection was approximately 20% with a mean operative mortality of 9% and a 5-year survival rate of 12%.

Tumor biology appears to be the most important predictor of survival, with multivariate analysis demonstrating that aneuploidy, tumor size and nodal status are all independent predictors of outcome[40].

Clear resection margins is a crucial measure of long-term survival, especially in relation to the posterior aspect of the tumor[41]. This mode of spread occurs predominantly along the perineural sheath and lymphatics.

Extended resections, whilst improving resectability rates, do not appear to improve survival but await results of randomised studies.

The largest single institution series to date reported by Sohn et al[42] has an overall 5-year survival rate of 17% for the 208 patients followed for at least five years (of a total of 616 patients included in the series). Improved survival was found in those with less than 750 mL intra-operative blood loss, no transfusions, negative nodes, clear margins and tumors less than 3 cm. In addition, there was a demonstrable survival advantage for those patients receiving adjuvant chemoradiotherapy. For those patients with tumors less than 3 cm, node negative with clear margins, the 5-year survival rate was 31% in this series. The issue of blood loss has been noted previously to be a predictor of survival[40] and should be minimised in all cases.

Two other large retrospective series from France[43] and USA[44] of 555 and 327 resections demonstrated 5-year survival rates of 15% and 12% respectively for the groups as a whole.

The use of postoperative chemoradiotherapy has been the subject of much debate. After an apparently curative resection a large number of patients will subsequently develop liver and/or peritoneal metastases. The North American Gastrointestinal Tumor Study Group (GITSG) randomised patients to receive radiotherapy and 5-fluorouracil or supportive therapy only, with treatment continuing for two years. Patients receiving adjuvant therapy had a significant survival advantage[45].

These results were contradicted by the results of two European studies-ESPAC-1 [46] and The EORTC gastrointestinal study group[47]. The EORTC study randomised 218 patients with T1-2 N0-1a M0 pancreatic head or T1-3 N0-1a M0 periampullary cancer to receive post resection either 5FU and external beam radiotherapy or observation alone between 1987 and 1995. There was no significant improvement in either survival or local recurrence rates in all groups studied. The only difference in treatment protocol between this study and the GITSG trial[45] was a reduced dose of 5FU, 5FU was only given in the first week of chemotherapy. Currently adjuvant therapy cannot be recommended in the context of a randomized control trial.

Some success has been reported in the use of neoadjuvant therapy to down-stage tumors to an operable state, although these are all small, non-randomised studies. Snady et al[48] looked at two groups of patients over an 8-year period. Those with initially diagnosed locally invasive but regional tumors were given a combination of 5FU, streptozocin, cisplatin and radiotherapy. The others were patients who underwent resection without preoperative treatment (the majority of whom underwent postoperative adjuvant chemotherapy with or without radiotherapy). Those patients who became down-staged and underwent subsequent resection had significantly better survival than those in the non-pretreated group.

QUALITY OF LIFE (QOL) AFTER PANCREATICODUODENECTOMY

Few studies have addressed the issues about quality of life after pancreaticoduo denectomy, concentrating rather on morbidity, mortality and survival. McLeod et al[31] compared 25 age- and sex-matched controls undergoing cholecystectomy with an equivalent number undergoing pancreaticoduodenectomy, measuring 6 quality of life assessments. This study demonstrated no difference between the two groups in terms of the quality of life measures and also showed a return to normal preoperative body weight and nutritional status in the pancreaticoduodenectomy patients. Melvin et al[32] compared the quality of life in patients undergoing a standard Whipple with those undergoing a pylorus-preserving procedure. Mental health was the same but physical health quality of life was lower in the Whipple group, with no association noted with the pathology found, i.e. benign or malignant. Huang et al[33] assessed 192 patients undergoing pancreaticoduodenectomy using the three domains of quality of life, physical, psychological and social and compared these to healthy patients and those undergoing laparoscopic cholecystectomy. The pancreatectomy patients manifested similar quality of life profiles to the control groups, but did experience problems related to weight loss, abdominal pain, steatorrhoea and diabetes. When a subgroup analysis was performed, the patients undergoing surgery for chronic pancreatitis scored significantly lower in all three parameters when compared to the control group, presumably secondary to the chronicity of their disease.

SURGICAL PALLIATION OF ADVANCED PANCREATIC CANCER

Given the low rate of resectability for patients with pancreatic cancer, the issue of palliation of symptoms becomes of paramount importance. Surgery may play a part in the management of jaundice, duodenal obstruction and pain. Relief of jaundice plays a key role in improving quality of life and survival in patients with advanced pancreatic and periampullary cancer. The two options are endoscopic/percutaneous stenting (either metal or plastic) and open/laparoscopic surgical drainage. Endoscopic drainage has a success rate of 85%-90% with a relatively low complication rate. Stent occlusion may be a problem and in an effort to overcome this metal stents are now commonly utilized for proven inoperable malignant tumors. Their greater patency rates compared to plastic stents have been demonstrated in randomized controlled trials. Even if occlusion does occur, tumor ingrowth can be ablated using laser/thermal techniques, otherwise further stents can be placed through the middle, beyond the site of occlusion. For those patients in whom endoscopic access has failed, percutaneous transhepatic stents can be used, although with a higher risk of complications[49].

For those patients who at the time of surgery are found to be inoperable and who have biliary obstruction, surgical biliary drainage is necessary. A variety of techniques are available, although the original cholecystjejunostomy has now been superseded due to the low long-term patency rates. The preferred method of bypass in this institution is choledochoduodenostomy, as this is relatively simple to perform with minimal morbidity and long-term patency. Other groups used choledochojejunostomy, which is suggested by some to have longer patency rates[50]. Operative drainage has a higher morbidity and mortality than stenting[51] and should be reserved for those patients found to be inoperable at surgery and the rare instance in which stenting by either route has failed, bearing in mind the mortality rates of open drainage of 19%-24%.

The issue of duodenal obstruction is controversial, with debate over whether or not routine gastric bypass should be performed at the same time as surgical biliary bypass. Approximately 5% of patients with pancreatic cancer will have actual mechanical obstruction and 17% of patients undergoing biliary bypass alone will develop subsequent duodenal obstruction. A proportion of patients undergoing gastric bypass procedures will develop delayed gastric emptying, with significant morbidity attached to this procedure. Therefore, it seems sensible to reserve gastric bypass for those with definite obstruction or evidence of impending problems, especially given the availability of endoscopically placed metal duodenal stents.

Pain is a major problem in the management of patients with pancreatic cancer and aside from standard analgesic regimens a variety of nerve ablative techniques are available. Coeliac plexus blockade, either at the time of surgery or percutaneously, has been reported as having significant success rates[52]. More recently thoracoscopic division of the splanchnic nerves has reported variable success rates and the results of large randomised studies are underway. Initial enthusiasm was tempered by high complication rates which have now drastically reduced. Unilateral versus bilateral approaches have been assessed and the left side appears to be the most important for its analgesic effect in pancreatic cancer.

CONCLUSION

Pancreatic cancer remains a disease with a dismal prognosis despite improvements in surgical technique. Accurate staging of disease is crucial if inappropriate surgery is avoided, with the choice of technique depending upon available facilities and expertise.

Modifications in surgical approach are unlikely to modify the outcome, and only improvements in adjuvant treatments, coupled with a greater understanding of the biological nature of the disease process, will the survival be improved. Meanwhile, patients with pancreatic cancer should be managed by multidisciplinary teams in specialist centres, if morbidity and mortality associated with the management of this disease are to be minimized.