Christine Gourin, MD, is still haunted by a clinical case involving a woman diagnosed with HPV-related cancer of the tongue.
Explore This IssueSeptember 2013
Dr. Gourin, associate professor of otolaryngology and director of the clinical research program in head and neck cancer at Johns Hopkins University in Baltimore, said the woman was asymptomatic after being treated with chemoradiation for her cancer. Nevertheless, Dr. Gourin and other physicians ordered PET/CT surveillance scans at regular intervals to assure themselves that the tumor was really gone.
Coughing fits caused the patient to aspirate fluid into her lungs, which showed up on the scans as abnormalities that prompted her doctors to order biopsies. Those biopsies, in turn, led to significant complications. On one occasion, she was admitted to the hospital with a chest tube. On another, a non-diagnostic biopsy led doctors to open her chest for diagnosis.
When the woman’s cancer eventually returned, she detected it herself by noticing a new pain in her mouth. “So all the surveillance imaging didn’t do any good,” Dr. Gourin said. “It was the patient who picked up her recurrence.”
A few years ago, clinicians routinely repeated PET/CT imaging studies at regular intervals after treating patients for head and neck cancers. Physicians like Dr. Gourin, however, are now challenging protocols that they see as wasteful, unnecessary and even potentially harmful. “Frankly, when you do pick things up that the patient wasn’t aware of, like lung metastases, you don’t have very much to offer,” she said. “We don’t prolong life— and don’t even necessarily improve the quality of life—by looking for trouble early.” Although prices vary widely by institution, PET/CT scans can easily cost three times more than a CT scan. “The cost question needs to be raised, because we need to be better stewards of our resources,” Dr. Gourin said.
Other physicians, such as Barton F. Branstetter IV, MD, chief of neuroradiology at the University of Pittsburgh Medical Center, maintain that a retreat from PET/CT use may be just as imprudent if not backed by sufficient evidence. The “constant flux” in how institutions are now approaching imaging surveillance—whether increasing or decreasing its use—is based more on hunches than on scientific data, he said. “We’d like to move away from guesses and preferences to scientifically backed effectiveness, because that’s the way that all of medicine is moving,” Dr. Branstetter said.
When monitoring for local or lymph node-based recurrence of head and neck cancer, nearly everyone agrees that the National Comprehensive Cancer Network (NCCN) guidelines seem reasonable, if vague: a post-treatment baseline imaging of the primary site within six months of treatment. Many clinicians also agree that a PET/CT scan is likely the best imaging method for patients with advanced stage cancer.
A lack of agreement beyond those few points, however, has resulted in widely variable institution-specific protocols and spirited debates. Last year, for example, Dr. Gourin’s group practice agreed to restrict its surveillance to a single PET/CT scan at three months after non-surgical treatment, unless the patient displays a new symptom, a physical exam reveals an abnormality or the scan itself is abnormal or indeterminate. One exception to that rule is made for patients with a history of heavy smoking, who may benefit from NCCN guidelines for post-treatment lung cancer surveillance.
—Barton F. Branstetter IV, MD
A forthcoming study of 134 patients in the Journal of Nuclear Medicine suggests that the university’s single PET/CT scan identified unsuspected recurrence or metastasis in about 5 percent of asymptomatic patients. Conversely, the scan yielded a negative predictive value of 98 percent.
At the University of Iowa, surveillance scans are similarly reserved for symptomatic patients or those treated for advanced stage cancer. Doctors conduct two restaging exams, however: the first at three months and the second after a year. “If somebody has a squeaky clean, one-year post-treatment PET/CT scan, then I think you’re treading on pretty thin ice if you want to start ordering more in the absence of any symptoms or exam findings,” said Gerry Funk, MD, professor of otolaryngology-head and neck surgery and director of the division of head and neck oncology. Nevertheless, he cautioned that doctors have very little good data about the utility of long-term serial PET/CT scans in that patient population.
Among the medical center’s patients, Dr. Funk said, researchers found that a restaging PET/CT scan at 12 months came back positive about 10 percent of the time in asymptomatic patients. About 3 percent of all tested patients had treatable disease, a result he described as “a pretty reasonable yield.”
Meanwhile, at the University of Pittsburgh, clinicians perform the first PET/CT surveillance scan two months after treatment and repeat the process every three months through 14 months, meaning that many patients receive five scans in all. This protocol has drawn criticism from other researchers, and Dr. Branstetter readily acknowledged that the medical center is likely overutilizing PET/CT scans, but he argued that it’s part of an intentional big-picture research strategy. “We want to know, what if we hadn’t done that one? Would it still have worked out well for the patient?” he said. “By putting in place a protocol that overutilizes, it’s easier to measure how much you’ve overutilized. If you put in a protocol that underutilizes, you’ll never know what you could have accomplished.”
So far, insurance reimbursement hasn’t been an issue at the University of Pittsburgh; however, Dr. Branstetter sees trouble brewing on the horizon and has heard from plenty of colleagues in other states where insurers would never permit similar protocols. Even so, he contends that such restrictions have routinely taken effect, with limited or no science to support or refute them. “I recognize that our scheme is very much reliant on where we are practicing, and that our success is reliant on that as well,” he said. “We’re taking advantage of that while we can, so we can put science in place to justify either more or less utilization than what people are using.”
The university’s surveillance also comes with important caveats. A recent study of 512 patients by Dr. Branstetter and colleagues, for example, suggests that a single negative PET/CT scan carries a negative predictive value of 91 percent for recurrence of head and neck squamous cell carcinoma, an inadequate cure rate for deferring further surveillance (Am J Neuroradiol. 2013;34:1632-1636). Two consecutive negative results, however, yielded a negative predictive value of 98 percent. After the second negative result, Dr. Branstetter said, doctors can stop all imaging surveillance unless the patient develops symptoms. “It’s powerful information, and it genuinely changes the surveillance protocols in a subset of patients,” he said.
At Johns Hopkins, Dr. Gourin said her practice group has decided not to use subgroup analysis to determine which asymptomatic patients might benefit from post-treatment surveillance imaging. Although a patient treated for a more aggressive disease such as piriform sinus cancer may be at much higher risk for recurrence and distant disease, she said, cancer persistence after treatment would likely show up on the three-month scan. And if the cancer recurs after chemotherapy or radiation, she added, finding the tumor before signs or symptoms materialize offers no advantage. “Just to routinely scan the asymptomatic and otherwise healthy patient does not make sense,” she said. “We follow them clinically.”
—Gerry Funk, MD
As in the United States, protocols vary throughout Europe, though physicians abroad generally face a higher level of scrutiny with repeated use. In France, a clinical trial of 91 patients suggested that a single scan at 12 months after treatment is highly effective in detecting recurrence, despite a false positive rate of 23 percent (J Nucl Med. 2009;50:24-29).
A more recent retrospective analysis co-authored by Mike Yao, MD, clinical associate professor of otolaryngology at New York Medical College in Valhalla, instead suggests that patients with negative three-month scans derive limited benefit from subsequent PET/CT surveillance (Cancer. 2013;119:1349-1356). “What we say in the paper is that you probably should only use it if you have some sort of suspicion: The patient might have pain or there might be swelling; there might be something there that you can’t account for,” he said. “If you can’t find a problem, PET/CT probably isn’t going to change that patient’s outcome at all.”
Nevertheless, Dr. Yao said his group’s results are unlikely to yield much change in physician behavior, due to lingering perceptions that a PET/CT scan remains a sensitive method for detecting recurrences. “Unless someone comes in and says, ‘You need to save some money and this is not a good use of resources,’” he said, “I think doctors are going to continue to order them.”