SAN FRANCISCO – A pancreatic cancer’s molecular markers might make it possible to predict whether a patient will benefit from surgery, suggests a study from the New South Wales Pancreatic Cancer Network in Australia.
In the cohort of 372 patients who underwent surgical resection for pancreatic cancer, the markers S100A2 and S100A4, which are associated with more aggressive tumor behavior, predicted worse survival after surgery.
Median survival was almost three times longer when a tumor was immunohistochemically negative for both markers, compared with positive for S100A2 (34.3 vs. 11.9 months). A third group of patients having tumors positive only for S100A4 had intermediate survival (15.6 months).
"Perhaps we can use these two markers to stratify resectable pancreatic cancer into three different phenotypes of prognosis and response to therapy," lead investigator Dr. David K. Chang said at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.
"So we can make the argument that for those patients in the good prognostic group, we can subject them to a potentially more morbid operation and more aggressive locoregional control, such as synchronous arterial-venous resection," he elaborated. "And for the group of patients [that] we know are going to do very poorly after surgery, perhaps it’s best to place them on the neoadjuvant pathways."
Surgery remains the only potentially curative therapy for pancreatic cancer, a disease whose survival has improved little over 45 years, according to Dr. Chang, a doctoral student at the Garvan Institute of Medical Research in Sydney.
"We know that surgery works," he commented. "However ... we do not have the ability to predict who is going to benefit from surgery preoperatively. Currently, we select patients for surgery based on CT criteria. ... And we predict prognosis based on clincopathologic variables that are only available after surgery, and sometimes they are fairly inconsistent."
Dr. Chang and his coinvestigators assessed the two markers in a cohort of 372 patients who underwent resection of their pancreatic cancer. Immunohistochemistry was performed on the tumors to ascertain whether their epithelial cells expressed S100A2 in the cytoplasm and S100A4 in the cytoplasm and/or nucleus.
Both markers are calcium-binding proteins. S100A2 has been associated with metastasis, invasiveness, and larger tumor size. S100A4 (also known as MTS1, metastasin, FSP1, and p9ka) has been associated with metastasis, chemoresistance, and a mesenchymal phenotype.
Multivariate analysis showed that each marker independently predicted an increased risk of death, with a hazard ratio of 1.76 (P = .005) for patients having S100A2+ tumors and 1.69 (P = .002) for patients having S100A4+ tumors.
Combination of the two markers stratified patients into three distinct groups having different survival after surgery, according to Dr. Chang. Median survival was 34.3 months for patients with S100A2–/A4– tumors, 15.6 months for patients with S100A2–/A4+ tumors, and 11.9 months for patients with S100A2+ tumors (P less than .0001).
In the same cohort, the investigators compared a new nomogram integrating the two markers with preoperative factors such as tumor size against a previously validated prognostic nomogram for patients undergoing resection that relies on some variables available only after surgery (Ann. Surg. 2004;240:293-8).
The new nomogram "performed just as good if not slightly better than the postoperative nomogram using clinicopathologic variables that are only available postoperatively," Dr. Chang reported.
A subsequent pilot experiment has shown good correlation between quantitative RT-PCR levels of the markers in biopsy tissue obtained by endoscopic ultrasound with fine-needle aspiration and the immunohistochemical results on resected surgical specimens. The team is now assessing the use of the assays on tissue obtained by fine-needle core biopsy.
"Aberrant expression of S100A4 and S100A2 stratified pancreatic cancer into three distinct prognostic phenotypes," Dr. Chang said, summing up the study’s findings. "This may be used to improve the accuracy of prognostic nomograms and to help us better prognosticate and predict prognosis preoperatively."
And from the larger perspective, "It may help us to better select patients for more appropriate and personalized therapies so that we only operate on the right patients and, as a consequence, it may improve the overall outcomes," he concluded.
Dr. Chang reported having no conflicts of interest related to the research.