College of American Pathologists
Printable Version

  GIST prognosis based on histology


cap today

September 2004
Cover Story

Related Article:
GI stromal tumors a model for what’s coming’
Antibodies and kits for KIT

William Check, PhD

While most anatomic pathologists use tumor size and mitotic count to estimate the likelihood that a GIST is malignant, Henry Appelman, MD, professor of surgical pathology at the University of Michigan Medical Center, believes this quantitative approach is basically flawed. "I don’t buy the quantitative system," he says. "Neither mitotic counts nor gross measurements are reproducible."

Dr. Appelman bases his GIST prognoses on histology. "I have been studying these tumors for close to 40 years now, and I think I have looked at enough to be able to distinguish virtually all the time between those that are benign and those that are malignant," he says. "This is what we in pathology do almost all the time. We don’t take epithelial neoplasms of the colon and count mitoses. We look at them."

Dr. Appelman acknowledges that stromal tumors are so uncommon that few pathologists have seen enough of them to use his system, so they are more comfortable with the "counting" approach. "I can train people to use the morphological approach," he says. "But it is easier to sit down and look at a series of slides together than it is to define the morphological criteria."

Dr. Appelman and two anatomic pathologists that he trained evaluated 77 gastric stromal tumors looking for morphological features that he believes define benign GISTs. ("I have found over the years that the most difficult thing to figure out is what is benign, rather than defining malignant," he explains.) Characteristics of benign epithelioid GISTs include a rounded or polygonal shape, abundant cytoplasm which often looks peripherally clear, and cells with large or multiple nuclei. Benign spindle cells tend to be "annoyingly uniform," Dr. Appelman says, and are arranged in long sweeping palisades. They tend to have small vacuoles indenting the nucleus at one edge. "We thought we did a great job," Dr. Appelman says. Sensitivity was 100 percent and specificity was 92 percent (Trupiano JK, et al. Am J Surg Pathol. 2002;26:705-714).

Mainly because of this disagreement, Dr. Appelman refused to sign on to the 2002 report of the consensus conference on diagnosis of GISTs.

"Dr. Appelman says he can look at GISTs under the microscope and see which ones will not metastasize," says GI pathologist Elizabeth Montgomery, MD, of Johns Hopkins Medical Institutions. "I have tried to apply his criteria to some cases that I published. It worked really well—except for one case" (Am J Surg Pathol. 2004;28:168-177). Dr. Montgomery notes that there was one missed call in Dr. Appelman’s experience as well, which he has mentioned. "If you asked him about the outlier, he would probably say he didn’t get a good sampling and see the right area," she says. "But that doesn’t matter to that patient. Which is why other people like to use quantitative criteria that simply attempt a forecast."

"What is true about these tumors," Dr. Appelman acknowledges, "is that they are hard to diagnose and it is hard to determine whether they are benign or malignant unless you have experience with them. Maybe we will eventually find specific genetic or chromosomal changes that will be better than histology at these tasks."

William Check is a medical writer in Wilmette, Ill.