New magnetic resonance imaging technique is useful in early detection of pancreatic cancer

September 01, 2005
CHAPEL HILL — Research conducted at the University of North Carolina Hospitals shows that a new magnetic resonance imaging (MRI) technique is very successful for early detection of pancreatic cancer.”This work is an important, major advance. The essence of this work is that very early detection of pancreatic cancer is possible with MRI that is not possible with other modalities,” said Dr. Richard C. Semelka, the study’s principal investigator.

Semelka is a radiologist and vice chairman for clinical research in the University of North Carolina at Chapel Hill School of Medicine’s department of radiology.

Results of the study by Semelka and colleagues at UNC Hospitals are published in September’s issue of the American Journal of Roentgenology. Their article is believed to be the first published in the medical literature that describes the evaluation of suspected pancreatic cancer by dynamic gadolinium-enhanced, 3-D gradient-echo MRI images, Semelka said.

This particular application of MRI technology, called 3D GRE for short, is less than three years old.

About 32,180 people nationwide will be diagnosed with pancreatic cancer in 2005 and another 31,800 will die from the disease, the American Cancer Society estimates. Pancreatic cancer is the fourth leading cause of cancer death.

Pancreatic cancer is not detected in most patients until the disease has reached an advanced state. Unfortunately, successful treatment of pancreatic cancer occurs only between three and four months after a tumor begins developing, Semelka said.

Until recently, the ability of radiologists to detect pancreatic cancer was limited by problems associated with existing MRI and computed tomography (CT) techniques. For example, distortions in MRI images caused by the pulsing of blood through the aorta or by the patient’s breathing made it difficult to obtain images of the pancreas that were good enough for early cancer detection. And while dynamic enhanced CT also is useful in evaluating pancreatic cancer, that technique has difficulty detecting tumors smaller than two centimeters in size because of limited soft-tissue resolution, Semelka said.

The development of 3D GRE has enabled radiologists to obtain higher quality images of the pancreas. As a result, Semelka and colleagues at UNC Hospitals believed 3D GRE might be an excellent tool for evaluating the pancreas and detecting pancreatic cancer.

To evaluate the usefulness of 3D GRE for this purpose, they reviewed all 3D GRE images from 57 patients referred to UNC Hospitals for examination because of clinical suspicion of pancreatic cancer between June 2002 and August 2003. The reviewers were two experienced radiologists with training in body MRI techniques. The patients’ clinical histories and original MRI interpretations were withheld from the reviewers, to prevent observer bias.

The reviewers found signs of pancreatic cancer in 27 of the 57 patients, with a level of confidence they rated as high or very high. A review of these patients’ records showed that 21 of the 27 did indeed have pancreatic cancer. In addition, the reviewers found tumors of less than two centimeters in size in eight patients, Semelka said.

“Most of these patients had recently had high-quality CT scans that failed to find these tumors,” he said. “So, without 3D GRE, their tumors would not have been found at such an early stage in their development.”

Semelka and colleagues concluded that the identification of pancreatic cancer using dynamic gadolinium-enhanced 3D GRE images “can be performed with a high degree of confidence and accuracy, making them very useful in the detection of pancreatic cancer.”

In addition to Semelka, the study’s UNC authors from the department of radiology are Dr. Katherine R. Birchard, a resident physician at UNC Hospitals; Dr. W. Brian Hyslop, an attending physician; and Dr. Diane Armao, a pathologist. Three authors who took part in the study have since left UNC: Dr. Alfonso Brown, an internal medicine specialist; Zeynap Firat, a medical technologist; and Dr. Georgeta Vaidean, formerly of UNC’s School of Public Health, who performed the study’s statistical analysis.

– 30 –

This release was researched and written by Tom Hughes of the UNC School of Medicine.

UNC School of Medicine contact: Stephanie Crayton, (919) 966-2860 or

For further information please contact Ramona DuBose by email at