Stanford Report Online

Stanford Report, February 28, 2001
Study assesses accuracy of new technology in diagnosing lung cancer


A relatively new, noninvasive imaging test can improve the diagnosis of patients who have lung tumors. That's the finding from the first large-scale review of cases involving positron emission tomography, or PET, as a diagnostic tool for lung cancer.

PET promises an alternative to invasive biopsies and surgeries. It also has an advantage over computed tomography, or CT, scanning -- a standard noninvasive technique used to diagnose lung cancer, said the leader of the study, Michael Gould, MD, MS, assistant professor of medicine at the Veterans Affairs Palo Alto Health Care System and, by courtesy, of health research and policy.

CT scans simply measure tissue size but PET can better reveal whether the tissue is malignant. That's because PET detects and measures metabolic activity, a characteristic that sets apart malignancies from many other tissues. Malignant tissues tend to be extremely active as a result of taking up large amounts of the chemical glucose.

The PET scan procedure involves injecting a patient with a radio-labeled glucose molecule known as FDG (for 18-fluorodeoxyglucose) that is preferentially taken up by the tumor cells. Because the molecule is a glucose analog, it's not well-metabolized by the cell and accumulates there, causing the malignant tissue to "light up" on a PET scan. While the technology is limited largely to academic medical centers, it's gaining acceptance because of its many oncology applications. Such cancers as breast, colon, lung and melanoma light up on a PET scan.

"People are also using PET imaging to identify tumor recurrence after treatment," said Gould, lead author of the meta-study published in the Feb. 21 issue of the Journal of the American Medical Association. "We're also evaluating it for lung cancer staging to determine if the tumor has spread outside the lung."

Gould's team found that a PET scan was 95 percent accurate at identifying lung cancer when it was present and 80 percent accurate at identifying when the disease was not present.

PET is one of five conventional methods used for diagnosing lung cancer. The other methods are CT scan, biopsy, surgery and repeated X-ray observation. A newer method, in which a conventional gamma camera is modified to measure FDG levels, is sometimes used as a less-expensive alternative to PET.

Lesions of potentially malignant tissue are typically first spotted by chance on chest X-rays conducted as part of an unrelated procedure, Gould said. Usually a CT scan is carried out to confirm the abnormal tissue's location and measure it more accurately. Abnormalities less than 3-4 centimeters in diameter are described as nodules, while larger potential malignancies are called masses. "Masses are much more likely to be cancer -- 75 percent or higher," said Gould. "If you have a nodule that's between 1 and 4 centimeters, there's about a 50-50 chance it's cancer."

PET equipment isn't able to reliably detect tumors smaller than 0.7 centimeters. For the smallest lesions, detectable at 2-3 millimeters through CT technology, diagnosis usually relies on observation.

"CT also gives a better sense of the edge characteristics of the nodule, whether smooth or irregular --more irregularity being more suspicious for cancer," said Gould. "Finally, CT can sometimes detect otherwise invisible specks of calcium that don't show up on regular X-ray and which suggest a nodule more likely to be benign. But because CT doesn't measure the metabolism of the tissue, most often it's indeterminate."

An alternative test involves needle biopsy, which is considered 99 percent accurate when indicating cancer but has several disadvantages. Even under X-ray guidance the needle might not penetrate the nodule; also the tests are potentially risky and inconvenient for patients. Currently, the only certain measure of malignancy is to remove a tumor surgically.

The large meta-study provides insight into how best to use PET. "I think PET is most useful for looking at nodules between 1 and 3 centimeters ... particularly for people at lower risk for cancer. If a patient is young, not a smoker, has a nodule with smooth borders that isn't very big, and has a normal PET scan with nothing lighting up, it might be okay to say, 'We don't have to pursue this further.'"

Limitations to the meta-study are worth noting. Although 1,474 lesions of all sizes were reviewed from cases dating back to 1990, relatively few involved no cancer. "In these studies there's an over-representation of people with cancer," Gould explained. Perhaps due to the small subject pool, the PET scan's ability to accurately exclude cancer ranged widely.

Cost is ultimately a factor in how PET will be used. Medicare reimbursement for FDG-PET imaging is approximately $1,912. In contrast, reimbursement for CT of the thorax is $276, while CT-guided needle biopsy is reimbursed at $560. Because of the relative cost and scarcity of PET, interest has grown regarding the use of a modified gamma camera to measure FDG levels.

"The gamma camera is more or less a poor man's PET," said Gould. "It's widely available, and there's a way to rig these things so you can do metabolic imaging with FDG." As yet there is insufficient data to compare their accuracy with PET, he noted.

The meta-analysis provides important data for a large-scale cooperative Veterans Affairs hospital study comparing the diagnoses of dedicated PET scanners against other methods. The study is near the middle of its anticipated five- to six-year run, although Gould and his colleagues will complete a preliminary cost-effectiveness analysis in April.

"The cost-effectiveness analysis considers not only accuracy and cost and clinical outcomes, but the larger question of how best does PET fit with these other diagnostic modalities," said Gould. "It turns out there are about 40 different clinically reasonable ways to combine the six diagnostic tests."

Gould's Stanford colleagues in the review include Ware Kushner, MD, assistant professor of medicine at the VA-Palo Alto; and Douglas Owens, MD, associate professor of medicine and, by courtesy, of health research policy. Funding and support came from the VA Health Services Research and Development Service and the VA Cooperative Study in Health Services.