What began out of sheer curiosity for Harry Agress, Jr, MD, eventually became an exercise in the importance of follow-up.
In 2000, Agress, director of the Division of Nuclear Medicine and PET Center, and senior attending radiologist in the Department of Radiology at Hackensack University Medical Center, Hackensack, NJ, began tracking unexpected findings uncovered during the routine reading of fluorine 18-labeled deoxyglucose (FDG) positron emission tomography (PET) scans of patients who either had or were suspected of having a malignancy.
Agress wanted to determine the clinical importance of the unexpected findings and whether they had any malignant potential. “We would see abnormalities that didn’t fit with the normal spread of disease for a particular cancer patient,” he recalls. “For example, if someone had lung cancer, we would expect them to potentially have metastases in the liver, the adrenal glands, and the bones. But we would see a focus somewhere in an unusual location, such as the breast or bowel, where it was very focal or very intense, and we weren’t quite sure of its significance.”
In reports to referring physicians, Agress began recommending further testing. He also started to keep track of unexpected findings on a makeshift PET follow-up registry—a piece of paper on which he listed the patient’s name, the referring physician’s name and telephone number, the primary disease, the unexpected finding, follow-up suggestions, and any other notes.
After about a year of following up on various cases, and uncovering more and more positive findings, Agress began to consider launching a study on PET follow-up. His decision was ultimately made on the results of one particular patient, who had been diagnosed with rectal cancer. A PET scan revealed a focus in the region of the gallbladder, and a CT showed very mild nonspecific focal thickening on the wall of the gallbladder. When the patient underwent surgery for the rectal cancer, the surgeon decided to scope the gallbladder and discovered that the patient, who was asymptomatic, had primary gallbladder cancer.
|Harry Agress, MD|
“I finally decided that I had to write this up,” Agress says. His findings were published in the February 2004 issue of Radiology .
Eventually, Agress transferred the information onto a computer spreadsheet, and digitized the images and clinical information necessary for patient follow-up and future presentations.
One month after first reviewing a case, Agress would contact the referring physician to follow up, assessing what, if anything, had been done, and asking to see the results if a follow-up procedure or test had been completed. Often, the follow-up required more than one telephone call. “There was initially some resistance [on the part of the referring physicians], especially when a patient had no symptoms relative to the unexpected finding,” he says. “But, as we identified more and more findings that ended up being significant, it became easier to get the follow-ups.
“Very early on, we started to realize that we were picking up either unexpected cancers or precancerous lesions, such as colonic adenomas, that had nothing to do with the cancer for which the patient was originally scanned,” Agress says. “It was done purely for curiosity. We really didn’t know if these findings were important; however, it became clear that they were significant, as time went on.”
Between November 2000 and July 2002, Agress and study coauthor Benjamin Cooper, MD, performed FDG PET scans on 1,750 patients. Of the 1,750 scans, 58 unexpected abnormalities were identified in 53 patients. Of the 58 instances, 45 abnormalities were further evaluated by CT, MRI, and/or mammography; and 42 underwent subsequent biopsies. Of those biopsies, 30—71%—in 25 patients were either malignant or premalignant tumors not related to the cancer for which the patient was originally scanned. It is important to note that the incidence of malignancies and premalignancies might be higher in this particular group than in the average population because all of those patients either had cancer or were suspected of having cancer.
One case involved a female patient initially evaluated for a lung abnormality. A PET scan ordered for staging purposes revealed activity with the lesion, which was subsequently removed and found to be a carcinoid tumor. Agress, however, noted an additional finding.
“When I was looking at that scan, I saw a small, subtle focus in her left breast, and I wasn’t sure if that was an adenoma or possibly a small cancer,” he recalls. Neither of those possibilities was the reason why the woman had initially received the PET scan.
Looking at the patient’s follow-up CT scan in the same area where the PET was positive, Agress saw a small soft tissue abnormality that he says would otherwise be extremely difficult to identify on a CT without knowing exactly where to look. He then recommended a detailed mammogram with special views, and ultrasound.
The patient underwent those procedures at another facility located closer to her, but did not undergo them exactly as Agress had ordered: The mammogram had been completed as a screening test without a radiologist present. The test was read as normal.
The patient was then told by a breast surgeon that nothing was palpable, and since the mammogram was read as normal, there was nothing to be concerned about, and to come back in a year.
Agress followed up with the primary physician and told him that the mammogram needed to be more detailed, and ultrasound might be necessary. Agress offered to perform the tests gratis at Hackensack University Medical Center—once again, out of curiosity.
“I wanted to learn if this was the kind of thing I needed to pay attention to on a PET scan,” he says. “Certainly, it could help the patient, if we found something early.”
A new mammogram, this time with spot compression views, was completed at the center, and a subtle abnormality was found. An ultrasound probe at the spot uncovered a nodule that was too small to feel on palpation, and a needle biopsy under the ultrasound guidance revealed cancer cells.
The patient eventually underwent surgery, which demonstrated no evidence of spread. “All this was because of the PET scan. She was completely asymptomatic,” Agress says. “Part of it was the fact that the PET was able to identify this sort of thing in an asymptomatic patient, and the other part was follow-up. Had we not done the follow-up, the lesion would have the potential for growth and spread.”
In another case, a female patient who had an abnormality on a chest x-ray underwent a CT scan, which showed a large mass, and had mediastinoscopy, which was benign. A PET scan was conducted to rule out any abnormal activity in the mass, and showed that the mediastinum was normal. (See Figure 1.)
Reviewing that same PET scan, however, Agress saw an abnormal focus behind the bladder, and recommended an MRI of the pelvis to find out whether it was in the rectum or the uterus. The patient underwent the MRI at another facility that was used more often by the referring physician, and more convenient for the patient. At the time of the scan, the facility had the patient’s PET report, but not the scan itself. The facility reported no abnormality on the MRI behind the bladder, in the cervix, or in the rectum.
A month later, Agress called the referring doctor, who informed him of the results.”This physician basically thought that I had overcalled the PET scan, and I had caused the patient unnecessary anxiety and additional testing,” he says. “We then asked to have the MR sent to us. Viewing the MR, directly in conjunction with the PET scan, it became apparent that there was an abnormality in the rectum, which was most likely, and not surprisingly, interpreted as stool. When correlating with the PET, it became clear that this was a rectal mass that needed biopsy.” A subsequent endoscopy uncovered a full-blown adenocarcinoma of the colon.
Another case involved a female patient who had received a CT scan that showed a lung abnormality. Follow-up PET showed no abnormal focus in the lungs; however, a hypermetabolic focus was noted, again posterior to the bladder. A subsequent MRI showed a mass in the uterus, which was biopsied and found to be cancer.
Even in retrospect, the patient was completely asymptomatic, with no pain or bleeding. The patient eventually underwent a hysterectomy, and there was no spread of tumor and no adenopathy.
Yet another female patient with a lung abnormality underwent a PET scan, which found gastric cancer. The patient was also completely asymptomatic, even in retrospect. “The more we got these, the more interesting it became,” Agress says.
Overall, unexpected findings were discovered in 3.3% of the scans (58 out of 1,750) in the study, with 1.7% (30 out of 1,750) incidence of malignant or premalignant tumors.
Not all of the cases turned out to be malignancies; others were benign, although sometimes also significant, including findings of cholecystitis, thyroiditis, and pigmented villonodular synovitis. “These findings were not the primary focus for the physician who ordered the PET scan,” he says. “They didn’t order the scan looking for other cancers. They were looking for the presence of a primary cancer or extent of involvement or how their therapy had worked. Here, we bring in a whole other finding that has potentially nothing to do with why they ordered the scan, and it just creates more of a problem for them. But it turns out that although we were identifying new abnormalities, they were at an early stage, almost all asymptomatic, and the vast majority could be removed or treated before they had a chance to spread.”
In the study’s findings, Agress reported that personal communication with referring physicians was a major component for follow-up, and for emphasizing the importance of the incidental findings as potential indicators of malignancies. “I really wanted to know the answers myself, and sometimes it required a physician talking to a physician to get these things pushed a little bit. But you could certainly do follow-up if you have good nurses or physician assistants,” Agress says.
While he received some resistance when he first started following up, “it’s a routine part of our practice now, and I don’t have to make so many follow-up phone calls because frequently the referring physicians call me first,” he says.
Among the other discoveries uncovered through the follow-ups were that 23 of the 25 patients with malignant or premalignant findings were asymptomatic, even in retrospect; that follow-up imaging was most helpful when it was correlated directly with the PET images, and not just the report; and that the use of combination PET/CT units could prove useful in correlating findings immediately.
“With PET/CT, you would most likely not have to do another imaging study, as you would be able to localize the finding on the CT portion of the exam. You would still have to biopsy to determine the pathology; however, PET/CT might identify the optimal procedure for doing so, for example whether to do a CT-guided or endoscopic biopsy,” Agress says.
“Follow-up is really ideal for developing your own confidence and competence in PET interpretation, and for building stronger relationships and credibility with referring physicians because you are showing a genuine interest in the patient,” Agress continues. “You also want the results so you can better understand the potential and limitations of PET. I think it makes it more intellectually stimulating for the person reading PET, and it clearly improves patient care.”
Danielle Cohen is associate editor of Decisions in Axis Imaging News.