Published reports from screening programs for high-risk patients show diagnostic yields of 4% to 23%. At Johns Hopkins, screening of individuals with at least three first-degree relatives diagnosed with pancreatic cancer and of those with Peutz-Jeghers syndrome found lesions in 5.3% using EUS and 10.2% using EUS plus CT. Of the 302 individuals in these programs, screening identified six cancers (five subjects are still alive), 20 cases of single or multiple IPMNs, and three cases of pancreatic endocrine neoplasia.
"At Johns Hopkins University, over 380 patients have been screened to date in our very selective screening program, and 24 underwent 30 operations," Dr. Canto noted. "Many had PanIN multifocality and 26% had high-grade neoplasia. All patients are alive and undergoing surveillance."
She noted that screening and surveillance of high-risk patients has a diagnostic yield that averages 10% for prevalent neoplasms and 96% for moderate-grade or high-grade dysplasia. In the Hopkins series, surgical resection has proven to be safe, with no deaths or severe morbidity among patients undergoing partial resection, partial followed by completion resection, or total pancreatectomy.
"We believe you can argue for this approach in some individuals," Dr. Canto said.
Closing in on biomarkers
Recent genomic, transcriptomic, and proteomic studies have provided leads for molecular diagnostics in pancreatic cancer. Dr. Hanash, head of molecular diagnostics at Fred Hutchinson, described the ideal molecular diagnostic exam: A blood-based test that would measure the novel circulating proteins associated with cancer development and their related signaling pathways. Also, the test should involve the novel tumor antigens that induce an immune response such as autoantibodies that indicate an early-stage neoplasm, he explained.
For identifying high-risk patients, the insulin growth factor (IGF) pathway has already been singled out. Low levels of IGFBP-1 are associated with a twofold to fourfold increased cancer risk, although levels of IGF-1, IGF-2, and IGFBP-3 have not been linked to the same increase in risk, he said.
"For establishing risk, we are still relying on patient characteristics, including age, smoking [history], diabetes, chronic pancreatitis, and family history. There is suggestive evidence that obesity, diet, and alcohol(Drug information on alcohol) may have a minor role," Dr. Hanash said. With biomarkers, the emphasis is more on very early detection. This is a daunting proposition because the characteristics of the marker must be outstanding, given the low overall incidence of pancreatic cancer, Dr. Hanash pointed out.
Such biomarkers would be applied to persons deemed at risk for the tumor, but there are issues with that, as well. "Some preneoplastic lesions never progress to cancer. It is helpful to detect them, but you must be cautious in intervening unnecessarily. It is a challenge," he said.
The circulating protein CA 19-9 is now being used to monitor disease; rising levels of CA 19-9 after resection are a sign of relapse, and high levels correlate with unresectable tumors. But CA 19-9 tests suffer from low specificity because high levels of the circulating protein are seen in chronic pancreatitis as well as cancer. So by itself, CA 19-9 is not reliable for screening, Dr. Hanash said.
However, several other proteins appear promising as biomarkers, including Muc1, which can discriminate between newly diagnosed cancer and case controls, and cytokeratin 18, which is associated with advancedstage disease. Combinations of proteins, or panels, will probably be the most useful, he predicted.
Aside from proteins, nucleic acids in the blood are interesting candidates as screening tools. MicroRNAs, especially miR-155, have potential in early detection efforts. In a recent study, a panel of four mRNAs (including miR-155 and others) demonstrated 64% sensitivity and 89% specificity in discriminating cancer cases from controls (Cancer Prev Res 2:807-813, 2009).
