Diagnosis of pancreatic ductal adenocarcinoma and chronic pancreatitis by measurement of microRNA abundance in blood and tissue

Abstract

A solid process for diagnosis could have a substantial impact on the successful treatment of pancreatic cancer, for which currently mortality is nearly identical to incidence. Variations in the abundance of all microRNA molecules from peripheral blood cells and pancreas tissues were analyzed on microarrays and in part validated by real-time PCR assays. In total, 245 samples from two clinical centers were studied that were obtained from patients with pancreatic ductal adenocarcinoma or chronic pancreatitis and from healthy donors. Utilizing the minimally invasive blood test, receiver operating characteristic (ROC) curves and the corresponding area under the curve (AUC) analysis demonstrated very high sensitivity and specificity of a distinction between healthy people and patients with either cancer or chronic pancreatitis; respective AUC values of 0.973 and 0.950 were obtained. Confirmative and partly even more discriminative diagnosis could be performed on tissue samples with AUC values of 1.0 and 0.937, respectively. In addition, discrimination between cancer and chronic pancreatitis was achieved (AUC = 0.875). Also, several miRNAs were identified that exhibited abundance variations in both tissue and blood samples. The results could have an immediate diagnostic value for the evaluation of tumor reoccurrence in patients, who have undergone curative surgical resection, and for people with a familial risk of pancreatic cancer.

Figure 1. Barplots detailing intensity values recorded for particular miRNAs that exhibited significant expression variations.

Only one typical result each is shown for blood and tissue. In the two panels on the left, the intensity values of miR-320a in blood samples are presented. The two panels to the right show the values of miRNA-103 as recorded in tissue samples. The horizontal lines in each panel represent the respective median.

Figure 2. Number of significantly informative miRNAs in blood.

The distribution of the 863 miRNAs across the entire p-value range is shown. The blue line denotes a significance value of 0.05. The bar to the left therefore indicates the number of significantly differentially expressed miRNAs. For classification purposes, however, also molecules of less significance could be informative.

Figure 3. ROC curves calculated on the basis of the miRNA measurements.

Receiver operating characteristic (ROC) curves are a widely accepted indicator of diagnostic utility. Measure of accuracy is the corresponding area under the ROC curve, denoted as AUC. It ranges in value from 0.5 (chance) to 1.0 (perfect discrimination).

Figure 4. Venn diagram of the overlap of miRNA biomarkers.

The miRNAs that are significant and specific for a blood-based diagnosis in patients with cancer versus healthy donors are shown in the yellow circle. The blue circle represents the markers that discriminate cancer and normal tissue, while the green circle stands for the markers that permit a distinction of cancer and inflammation in tissue, which is not possible in blood. The overlap in miRNAs between the tissue marker sets is significant; the relevant molecules are named.