Fifty Years of Pancreas Cancer Care

Abstract

Resulting from 50 years of innovation, operations for pancreatic neoplasms can now be performed safely, albeit with significant but manageable morbidity. Molecular diagnosis has allowed for the identification of multiple distinct histopathologies with variable natural histories. Observation is now a strategy for selected indolent cysts and some neuroendocrine neoplasms. For ductal pancreatic adenocarcinoma, a long-term cure remains elusive and will require more than surgical resection for meaningful progress.

1. Introduction

Over the last five decades Pancreas cancer has evolved from a single generic term with a uniformly lethal outcome to a multitude of varying histopathologies with unique diagnostic signatures, variable natural histories and histology specific approaches to treatment. Memorial Sloan Kettering (MSK) has had a continuous involvement in understanding of such diseases, and has been a pioneer in improving outcome initially based on improved safety and outcome of surgical procedures through to definitive imaging, molecular diagnosis and histology specific treatments. .

2. The 1970s

During the 1970s, there was an increasing level of understanding of the metabolic consequences of pancreas cancer. This era ushered in the rise of total parenteral nutrition (TPN) as a support modality for surgical recovery and pediatric life support. TPN called for the definition of requirements in man of trace metals, essential fatty acids, vitamins, iron supplementation, and an awareness of phosphate depletion syndromes with non-supplemented TPN.¹

It was a decade of change in the ability to evaluate patients preoperatively, highlighted by the development of the computed tomography (CT) scan, which resulted in the Nobel Prize for Allan Cormack and Godfrey Hounsfield in 1979. The standard approach to ductal pancreas cancer in the 1970s was total pancreatectomy. Since there was only a limited awareness of the consequences of brittle diabetes after total pancreatectomy at this time, the surgeons who managed such patients erred on the side of avoiding hypoglycemia; this meant that patients often remained hyperglycemic after surgery.

While there was still little focus on the use of underlying histopathology for targeted surgery, familial syndromes commonly associated with endocrine tumors of the pancreas began to be recognized during this time. Localization of functional pancreatic endocrine tumors, initially with angiography and portal vein selective catheter hormonal sampling, was developed. With the advent of H2 blockers, there was also a change from total gastrectomy to the search for and removal of pancreatic or duodenal gastrinomas.²

The first studies of intraperitoneal chemotherapy for any cancer³ were conducted in the 1970s as well, but there were limited opportunities for evaluation since there was only occasional use of CT scanning, ultrasound, and physical examination.

3. The 1980s

The recognition of the value of prospective databases was an important development in the 1980s. Institutions created their own databases, with administrative databases at the state and national level following soon after. These data led to the detection of prolonged hospital stays after pancreatic resection. For example, in this decade, for pancreaticoduodenectomy, the average hospital stay at Memorial Sloan Kettering Cancer Center (MSK) was 28 days. Initial diagnosis was difficult, and most patients received a barium swallow test to look for duodenal encroachment or obstruction. The only effective preoperative evaluation to assess resectability consisted of conducting an in-hospital angiogram, looking for arterial and portal venous narrowing, and determining aberrant anatomy. At the time, the great concern was not to miss median arcuate syndrome, which turned out to be less important once it was discovered that 25% of such patients would have some coeliac artery stenosis often unrelated to the median arcuate ligament. A simple intraoperative temporary occlusion of the superior pancreaticoduodenal artery established that there was flow in the hepatic arteries prior to the division of the artery. All such concerns for arterial occlusions or aberrant vasculature were addressed when the use of CT scans increased.

For nearly all cases after pancreatectomy, nutritional support by TPN was slowly recognized to be unnecessary, and the metabolic consequences of TPN in cancer patients were progressively defined.⁴ Total pancreatectomy was also gradually abandoned as the treatment of choice once the toxicity of long-term diabetes was recognized, especially fatty liver failure from prolonged hyperglycemia (although some attempts were made to improve liver failure after pancreatectomy by using a glucagon infusion).⁵ The introduction of H2 blockers was evaluated in Zollinger-Ellison syndrome patients along with the focused identification of the site of the primary tumor as the mainstay for curative treatment.²

In the 1980s, early studies of radioactive implants for unresectable pancreas cancer were described and investigated for pancreatic adenocarcinoma⁶, with a similar approach used for intraluminal radiotherapy for malignant biliary obstruction. However, the technical challenges of intraoperative radioactive implants in the pancreas, with the potential for accompanying complications, limited their applicability. During this decade, the introduction of endoscopic ultrasound⁷ for the evaluation of upper gastrointestinal tract malignancy was advanced to evaluate the resectability and provide a histological diagnosis of pancreatic adenocarcinoma.

4. The 1990s

By the 1990s, CT scans replaced angiography for the diagnosis and evaluation of the resectability of pancreatic adenocarcinoma. Progressively, the recognition of characteristic CT findings for lesion description allowed for the more accurate prediction of underlying pathology. The early application of laparoscopy to define undetected metastatic disease during this time resulted in the rapid change from open evaluation and 25% resection rate, to open operation post-laparoscopy with 76% resection rate.⁸

During this decade, portal vein involvement stopped being considered a barrier to resection after results including the portal vein were found to be parallel to the long-term outcomes of patients who did not require portal vein resection.⁹ Around the same time, early trials of routine perioperative drainage suggested poor limited benefit and increased associated complications.¹⁰ There was a recognition of the benefits and lack thereof of biliary stenting, and the importance of intraoperative bile cultures in those with prior stenting was defined.¹¹ Additionally, the recognition that the bile organisms were the source of intra-abdominal and wound infections resulted in a change of the prescribed prophylactic antibiotic.¹²

The increased use of prospective databases in 1990s was demonstrable at MSK when its surgical pancreatic database exceeded 5,000 patients and became a rich resource for perioperative outcomes and natural history for individual histopathologies. The use of prospective institutional databases and of administrated databases drew attention to the importance of institutional and surgical volume in decreasing operative mortality. In New York State in 1995, for example, operative mortality for pancreaticoduodenectomy was as high as 28%, whereas in high-volume institutions it was 4%.¹³ Perhaps more important was the recognition that of those New York State surgeons who performed a major pancreatic resection in the late 1980s and early 1990s, 96% performed less than 1 per year, linking volume of cases by surgeon and hospital volume as a major factor in operative outcome.^13,14

This observation led to the concept of regionalization of complex pancreatic cancer cases, as the support infrastructure of high-volume hospitals resulted in improved perioperative outcomes even for the lower-volume surgeons in the same high-volume hospital. The length of stay for pancreaticoduodenectomy subsequently declined from 20+ days to 10 days.

The increased use of laparoscopy to help determine resectability identified some patients with positive cytology but without visible metastasis, which confirmed unresectability as an outcome was equivalent to stage IV disease.⁸ Concurrently, with the advances made in technique and surgical management during this decade, chronological age was no longer seen as a barrier to resection, and similar results were obtainable in elderly patients with careful preoperative selection.¹⁵ A number of studies also confirmed that routine preoperative biliary drainage by stenting was not necessary and often harmful.¹⁶

The mandatory use of TPN was questioned and prospective randomized trials of routine TPN and enteral nutrition for pancreatic and other upper gastrointestinal malignancies showed no improvement in perioperative complications or in long-term outcomes. In fact, routine applications of artificial nutritional support were more likely to be harmful. Such findings led to the recognition that only patients with nutritional debilitation should be considered for perioperative nutritional support.^17,18 Similar concerns about the use of intraoperative drainage resulted in randomized controlled trials of drain versus no drain following liver and pancreas resections.^19,20 These trials did not show a benefit of routine intraabdominal drainage.

In the 1990s, there was increased awareness that adult-onset diabetes in the non-obese patient was an early indicator of pancreas cancer. In addition, early examination of molecular markers resulted in an awareness of the importance of KRAS mutations in pancreatic adenocarcinoma.²¹

5. The 2000s

By the 2000s, improvements in preoperative evaluation maximized awareness of comorbidities and recognition of early postoperative complications resulted in a decrease in hospital stay for pancreaticoduodenectomy to 7 days at MSK. Minimally invasive surgery or laparoscopy was progressively evaluated during this time, especially for those patients with distal lesions; similar oncological outcomes and perioperative morbidity, and a decrease in wound complications, were reported when compared with open operations.^22,23

In the same period, palliative surgery became increasingly defined while the need for prophylactic open bypass for potential stomach and bile duct obstruction was abandoned. There was also increased awareness and acceptance of less invasive palliative procedures, and non-operative biliary and gastric outlet obstruction by endoscopy became the standard first approach.²⁴ Subsequently, the efficacy of newer systemic regimens for advanced disease was examined as a neoadjuvant for resectable pancreatic cancer, with encouraging results, some of which had not been seen with earlier regimens.²⁵

The increased use of CT for cancer screening or other reasons resulted in many incidental or non-suspected pancreatic cystic lesions. With time and prospective data, it became recognized that not all cystic neoplasms need to be resected,²⁶ which was reinforced when clinical and biological risk factors for malignancy in cystic neoplasm were identified.²⁷

During this decade, a progressive definition of cystic neoplasms, in particular defining the risk of developing adenocarcinoma, in patients with intraductal papillary mucinous neoplasm (IPMN), based on the presence of main duct dilation, size greater than 3 cm, and the presence of any internal solid component in the dominant cysts, became accepted.²⁶ The definition of two subtypes of pancreatic adenocarcinoma associated with IPMN was associated with very different outcomes—improved survival for the colloid subtype while survival for the tubular subtype was similar to other classical pancreatic adenocarcinomas.^28,29

The use of a prospective cystic database revealed an increase in the diagnosis of smaller and smaller cystic neoplasms, defining a role of observation alone in >30%, and for the majority of lesions <2 cm without significant ductal dilatation or solid components.²⁶ Abandonment of nasogastric tubes, commenced in 1990s at MSK and removal of “Nil per Os,” received limited acceptance elsewhere marking a return in 2010s, to unnecessary and unindicated gastric decompression.³⁰

The adoption of synoptic reporting in the 2010s allowed precise and uniform data collection of all known intraoperative variables that influenced postoperative complications and long-term outcomes. The focus on the identification of risk groups for pancreas cancer emphasized the role of BRCA and familial pancreatitis with the associated development of adenocarcinoma; however, the screening of such high-risk groups for early-stage disease remained challenging.^31,32

The promise of tumor markers as a screening test was not fulfilled during this decade. There was limited specificity in the CA-19–9 and carcinoembryonic antigen blood tests other than the association of high levels of these markers with detectable or non-detectable liver metastasis. Such markers did have narrow value in monitoring persistence or recurrence in those patients who had elevated levels preoperatively that fell following resection.

Nomograms for postresection outcome prediction in adenocarcinoma, and subsequently other subtypes—especially IPMN, became readily available.^33,34 Extensive studies of aspirated cystic fluid for markers of underlying cancer risk provided some limited predictors for cyst progression to carcinoma.²⁸

6. The 2010s

In the 2010s, the preliminary evaluation of tumor vaccines and an attempt at the direct injection of intra-lesion heat shock protein were not successful. However, there were increasing reports of downstaging and pathological responses with the use of more effective regimens of preoperative chemotherapy that had not been seen with earlier regimens.

Following a report on the improved survival of metastatic pancreas cancer with a regimen of FOLFIRINOX (fluorouracil, leucovorin, irinotecan, and oxaliplatin) compared with gemcitabine³⁵, the FOLFIRINOX regimen was applied as a preoperative approach, which subsequently showed measurable pathological response rates.³⁶ A randomized controlled trial of limited intraoperative and postoperative fluid resuscitation suggested support for restricted rather than excessive fluid replacement in postoperative recovery.³⁷ A predicted nomogram for IPMN was also developed and validated during this time.³⁴

The association between inflammation and malignant progression in IPMN was reported.³⁸ In 2017, investigators at MSK published a report on more than 3000 cystic tumors evaluated at the institution, of which one-third were followed without operation,³⁹ which led to the definition of underlying risk factors for progression from premalignant to invasive disease.

The evaluation of the lymphocyte-to-neutrophil ratio as an outcome predictor was also validated during this decade.⁴⁰ In a study of 93 patients with pancreatic ductal adenocarcinoma (PDAC) treated with neoadjuvant therapy, 17% had evidence of a favorable response histologically. Patients treated with neoadjuvant therapy had higher neutrophil-to-lymphocyte ratios, but there was no significant correlation with response or survival.⁴¹ Other important trials from MSK during this time period included a randomized controlled trial of pasireotide to limit postoperative pancreatic complications, especially excess pancreatic drainage and fistula, which showed promising results but there was still controversy surrounding the uniform acceptance of this practice.⁴²

The increasing use of positron emission tomography (PET) scan was identified but shown to be of limited value in diagnosis or outcome prediction. The use of radiation preoperatively in patients with potentially resectable tumors was limited; however, preoperative chemotherapy was increasingly used in the marginally resectable with use of radiation reserved for the locally nonresectable. The recognition that FOLFIRINOX gave improved survival over gemcitabine in metastatic cancer allowed consideration of such a regimen in the neoadjuvant setting.⁴³

The 2010s also ushered in the increasing use of robotic resections, for both proximal and distal lesions, and this approach was progressively adopted in high-volume centers. The revival of extended operations including vascular resection, which had been performed without benefit in the 1980s, was reexamined in the light of chemotherapeutic response with newer combinations but shown to be of limited value.

A study of 921 patients submitted to resection of PDAC showed that initial hepatic recurrence was common (nearly one-third of patients) and was associated with a very poor prognosis compared with recurrence at extrahepatic sites (34% vs 61%). This study also showed the use of radiomics, also known as quantitative imaging analysis, of the liver remnant identified features associated with early liver recurrence and improved the predictive ability of a model based on high-risk clinical variables.⁴⁴

7. Summary

In summary, considerable progress in the identification and subsequent management of pancreatic lesions have been made over the past 50 years. This has allowed for the identification of numerous different pathological entities, each with variable outcomes. Operative procedures have become increasingly safer in recent decades, and minimally invasive procedures are now more applicable. More aggressive chemotherapeutic regimens, with responses in advanced disease are progressively being applied in a neoadjuvant setting, with conversion of the marginally resectable to resectable lesion. However, overall progress in advancing outcomes beyond earlier diagnosis in classical PDAC has been limited and requires further study.