The trio of biomarkers—tested in three separate cohorts, including two blinded validation studies—improved the detection of patients with early-stage disease compared to healthy or benign disease controls.
Addition to CA 19-9
“Adding these two biomarkers provided statistically significant improvement for all early-stage cancer vs healthy controls as well as other subcohorts when used with the current gold standard biomarker, CA 19-9,” said Ann Killary, PhD, Professor of Translational Molecular Pathology at MD Anderson.
While CA 19-9 is the only biomarker approved for use by the U.S. Food and Drug Administration, and only for monitoring treatment for the disease, according to first author Seetharaman Balasenthil, PhD, Instructor in Translational Molecular Pathology at MD Anderson. The marker of antigens produced by pancreatic cancer has a low positive predictive value for identifying early-stage disease.
At early stages, pancreatic cancer can be successfully removed with surgery, but 80% of patients are diagnosed with either locally advanced disease (stage III) or cancer that has spread to other organs (stage IV), when surgery is no longer a curative option.
“Our goal is to identify more patients at those earlier, resectable stages, when treatment could lead to a 5-year survival rate of 30% or more, depending on stage,” Dr. Killary said. Only about 7% of patients survive for 5 years following diagnosis of the disease.
Additional studies in larger cohorts will be needed to validate these findings, and more biomarkers will be needed to get the completely accurate set needed to screen the general population with the long-term goal of identifying precursor lesions before they become malignant.
Dr. Killary and colleagues earlier identified a cluster of genes involved in cancer migration and then analyzed the proteins produced by those genes. Two, known as tissue factor pathway inhibitor (TFPI) and tenascin C (TNC), emerged as the strongest biomarker candidates. The team first compared their predictive ability by comparing their presence in healthy volunteers and those with primarily stage IV pancreatic cancer.
In the current study, deploying TFPI and an isoform of TNC (TNC-FN IIIC) with CA 19-9 improved performance discriminating stage I and II disease from healthy or benign disease controls.
Any biomarker test for the general public would need to have an area under the receiver operating characteristic curve (AUC) for sensitivity and specificity close to 1.0, Dr. Killary noted, given the infrequency of pancreatic cancer in the population at large. (In this case, the AUC is a measure of how well a parameter can distinguish between two diagnostic groups.) In the group’s analysis of three cohorts of samples, AUC scores for the combination consistently outscored CA 19-9 alone.
For the first cohort of samples, comparing stage I/IIA and healthy controls yielded an AUC of 0.72 for CA 19-9 alone compared to 0.84 for the combination. For stage IIB vs controls, the combination raised CA 19-9’s performance from an AUC of 0.87 to 0.98. The second blinded cohort also showed improvements in AUC over CA 19-9.
A third, larger blinded validation study of 252 samples of cases and controls from the Early Detection Research Network of the National Cancer Institute further confirmed those findings and also permitted the team to measure the trio’s ability to detect disease in patients without a history of pancreatitis or diabetes. The AUC for those patients was 0.87 for stage I/IIA, 0.93 for stage IIB, and 0.89 for all early stages of the disease using the combination, pointing to a potential ability to detect disease in those who lack either of those known risk-raising conditions.
Possible Use for High-Risk Individuals
Dr. Killary said the team is working to use the biomarkers in MD Anderson’s high-risk clinic, which monitors people who are already at high risk of pancreatic cancer due to family history or having known risk-raising genetic mutations.
“In this population, our biomarker panel might prove very useful in early detection,” Dr. Killary said.
Coauthor Subrata Sen, PhD, also a Professor of Molecular Translational Pathology, noted, “These findings are a significant advance, considering that there’s nothing else available now to detect early-stage pancreatic cancer that has gone through blinded validations in multiple patient cohorts.”