UF Researcher: New Classification System Helps Assess Prognosis For Patients With Pancreatic Tumors

June 20, 2002

GAINESVILLE, Fla. — Patients with certain pancreatic tumors could benefit from a new way of assessing which growths constitute potentially aggressive cancers that are likely to recur or kill, report scientists at the University of Florida Shands Cancer Center and the Memorial Sloan-Kettering Cancer Center.

The method, which involves sampling tumors to see whether they contain large numbers of dividing cells or areas of dead cells-characteristics closely associated with more virulent tumors and therefore worse prognoses-should help doctors better predict a patient’s outcome and identify who to track more closely after surgery, said Dr. Steven N. Hochwald, an assistant professor of surgery, molecular genetics and microbiology at UF’s College of Medicine. Details are reported in the June 1 issue of the Journal of Clinical Oncology.

These types of pancreatic tumors, also known as pancreatic endocrine neoplasms, are relatively rare, representing about 5 percent of all pancreatic cancers. Yet surgeons say the number of patients diagnosed with the growths is rising, thanks to advances in abdominal imaging and the increasingly elderly age of the population.

“It’s frequently very difficult to determine if these tumors are a more benign type or if they’re going to behave in a malignant fashion,” Hochwald said. “Up to now, pathologists have used fairly rough criteria to determine if tumors were more aggressive or malignant based on whether they detected spread to the lymph nodes or other organs.”

Pancreatic endocrine neoplasms can arise in cells that produce a variety of hormones, including insulin, that the body uses to store and process sugars from food. These cells often begin pumping out high levels of hormones, sending blood sugar levels plummeting or, alternatively, skyrocketing. Other forms of pancreatic endocrine neoplasms, meanwhile, do not ramp up hormone production and are detected when patients complain of abdominal pain and other symptoms. To complicate matters, some tumors that arise in these hormone-producing cells actually turn out to be relatively harmless.

Existing classification systems don’t consistently identify which tumors are not very dangerous, and which are especially dangerous and likely to recur in other parts of the body after surgery, Hochwald said. This ambiguity often makes it difficult for physicians to estimate a patient’s prognosis.

About half of all patients with pancreatic endocrine neoplasms are eventually cured, he said.

Researchers studied tumor samples taken from 136 patients with pancreatic endocrine neoplasms who had been treated at New York’s Memorial Sloan-Kettering Cancer Center between 1979 and 1998. Cases were first grouped according to the standard Armed Forces Institute of Pathology classification system, which looks at a variety of factors including metastatic spread, tumor size, invasion of blood vessels and rate of cell division. Researchers then regrouped the cases into low- and intermediate-grade categories, according to the rate at which cells were dividing within the tumor and whether zones of dead cells were present or absent.

The new classification system more accurately predicted the likelihood of disease recurrence or death from the tumors, said Hochwald, whose collaborators at Memorial Sloan-Kettering included pathologist Dr. David S. Klimstra and surgeons Drs. Kevin C. Conlon and Murray F. Brennan.

“Certain tumors did have a higher incidence of dividing cells and areas of dead cells (necrosis),” Hochwald added. “For instance, we looked at insulin-secreting tumors compared to other hormone-secreting tumors and found that the tumors that did not secrete insulin had a higher chance of having an elevated number of dividing cells and dead cells being present. That might help explain why patients with insulin-secreting tumors generally do better and have a better prognosis.”

The scientists also assessed other approaches that factored in size, the types of hormones tumors secreted and other criteria, but in the long run these were not as reliable, Hochwald said. In one comparison, three patients whose tumors were previously considered benign and four who had been classified as borderline were found to have had a recurrence or died. Under the new system, their tumors would have been categorized more accurately, he said.

A similar classification system has proved effective in evaluating endocrine tumors found in other organs, such as the lung.

“The advantage of our classification system is that it’s fairly easily done by pathologists in multiple places,” Hochwald said. “Areas of dead cells are fairly easy to identify, and determining the number of dividing cells is not that difficult-pathologists do that on a fairly regular basis for other tumors. If standardized in a strict fashion it can be accomplished without too much difficulty. This is a fairly simple classification system that should be able to be adapted on a widespread scale.”

Meanwhile, UF scientists plan to continue studying these tumors.

“The next step is to determine whether the presence of certain genes or molecular markers in cells can help predict how the tumors will behave,” Hochwald said.

These pancreatic endocrine tumors are variable in their behavior, said Dr. Gerard M. Doherty, the Thompson professor and chief of general surgery at the University of Michigan-Ann Arbor.

“Some patients’ tumors grow slowly, and they may co-exist with them for years,” Doherty said. “Other people have tumors that grow quickly, spread widely and kill them rapidly. If we can distinguish one from the other, then we may be able to tailor the therapy to fit the disease.

“Accurate and reproducible tumor classification systems are important for being able to give people an accurate prognosis and for making reasonable plans for tumor therapy,” he said. “Our near future will certainly include the frequent molecular classification of tumors in order to distinguish the inner workings of tumors cells; Dr. Hochwald’s findings should reflect the molecular function of the tumor. This can allow specific designs for therapy to treat the disease.”