The PET and the Pendulum FREE

We live in an era in which new medical procedures, diagnostic tests, and treatments are being introduced at a rapid pace. Some important new technologies were widely adopted without rigorous initial outcomes testing, with subsequent reconsideration and reduction in use, as exemplified by the pulmonary artery catheter in critically ill patients (1). Other procedures, such as lung volume reduction surgery for severe chronic obstructive pulmonary disease, underwent careful evaluation relatively early on (after a modern reintroduction in the case of lung volume reduction surgery), which strongly influenced their role in current practice (2). These well-studied procedures must be reexamined in light of newer refinements, such as bronchoscopic lung volume reduction (3).

Somewhere between these extremes is the ongoing evolution of positron emission tomography (PET), which has engendered enormous interest since its introduction into clinical oncology in the late 1980s. This technique has a very wide range of potential applications in diagnosis, staging, guiding biopsies, assessment of response to therapy, and identification of recurrent disease, as reflected in a burgeoning literature. In lung cancer, PET considerations are particularly complex—and promising—because the technique provides information about the primary lesion, mediastinum, and distant metastases.

The stakes in this very common form of cancer are enormous for individual patients, physicians, and a health care system struggling mightily to control costs, because lung cancer staging is the foundation for most treatment decisions. The staging system, which grows more complex with each iteration, serves to triage individual patients via the TNM system (tumor, node, metastasis) to a wide range of treatments based on the extent of disease and prognosis. So a single test—especially a noninvasive one—that could accurately stage lung cancer at a reasonable cost would be a major step in avoiding futile surgery and inappropriate undertreatment.

For many, the proverbial pendulum has already swung to PET, especially when integrated with computed tomography (CT), as an indispensable test for noninvasive detection of distant metastatic disease in lung cancer. Fortunately, acceptance by clinicians has been increasingly validated by many detailed studies that have served to clarify its clinical role; an example is the admirable study by Maziak and colleagues (4) in this issue. In this study, derived from 5 Canadian academic institutions, Maziak and colleagues randomly assigned 337 patients with non–small cell lung cancer—thought to be stage I, II, or IIIA on the basis of initial chest radiography and CT—to staging with PET-CT or a combination of abdominal CT and radionuclide bone scanning (conventional staging). Three statistically and clinically significant findings emerged. First, PET-CT correctly upstaged disease to unresectable stage IIIB and IV in 13.8% of patients; conventional staging correctly upstaged disease in only 6.8%. Therefore, PET-CT staging resulted in proportionately fewer inappropriate attempts at surgical cure. Second, false-positive PET-CT findings incorrectly upstaged disease in 4.8% of patients, versus 0.6% in the conventionally staged group. In the latter patients, disease was eventually correctly staged by using additional diagnostic procedures (including biopsy when necessary), then treated appropriately. Maziak and colleagues note that subsequent clarifying tests were performed more frequently in the conventionally staged group. Finally, PET-CT incorrectly downstaged disease in fewer patients (15%) than did conventional staging (29.6%), which also helped avoid futile thoracotomies and allowed more stage-appropriate chemotherapy plus radiation.

Maziak and colleagues' study (4) certainly has some notable limitations, many of which the authors mention. The sample sizes were small, especially for some of the subgroup analyses; for example, only 1 patient in the conventional staging group had incorrect upstaging. Small sample sizes mean wide confidence intervals and considerable uncertainty about the true rates. Generalizability is unknown because PET-CT was performed on only 5 machines at top referral centers with strict quality control guidelines. The patients' true state—the presence or absence of metastatic disease—was ascertained through clinical judgment, with concurrence of an adjudication committee. This diagnostic reference standard is imperfect but necessary for a study of this type because it avoids unnecessarily subjecting patients to biopsy of every suspicious abnormality on scans. The authors did not report the costs of care in the 2 groups, so they do not advance our limited understanding of the economic implications of PET-CT. Finally, because the authors did not provide information on long-term survival, we do not know whether ordering PET-CT for lung cancer staging ultimately helps patients live longer. Nevertheless, Maziak and colleagues are among the first to specifically compare PET-CT with conventional staging, and the results certainly suggest that PET-CT has important advantages. The head-to-head comparative study design also distinguishes the current study from other studies that focused on the addition of PET to conventional staging and evaluated PET rather than PET-CT (5 - 7).

Many issues remain, and it is particularly difficult to formulate recommendations for optimal preoperative assessment when the number of possible tests is increasing, along with important technical refinements in individual tests. For example, is integrated PET-CT so superior that CT or PET alone is now obsolete for staging lung cancer (8 - 13)? Are the putative advantages worth the increased cost (current charges at my university hospital are $3060 for chest CT, $3680 for PET, and $4180 for PET-CT)? How shall we incorporate new refinements in PET technology, such as dual–time-point imaging (14), into our judgments about when to test and how to interpret the results? Should we routinely do endoscopic ultrasonic biopsy—in addition to or instead of imaging—even if mediastinal nodes are small and PET results are negative, in an attempt to stage the mediastinum histologically (15)? What level of evidence is necessary to establish a clear preference for one of several tests, especially given the many available tests that we could compare in various combinations?

In interpreting PET-CT results, one must be particularly mindful of false-negative and false-positive results (16 - 17), identifying them with additional imaging studies and even biopsy if necessary for key management decisions. Clearly, decisions to do additional testing must be individualized. A PET-CT finding that is discordant with the pretest probability of neoplastic involvement is a key clue to a false-negative or a false-positive PET-CT result. In other words, a surprising PET-CT result (for example, a negative PET-CT result when the probability of metastases is high) is a good indication for further testing.

At this point in an evolving saga, it is reasonable to order PET-CT for lung cancer staging, especially for patients who seem to be candidates for curative therapy. A recommendation to use PET-CT does not imply that other means of preoperative assessment are invalid or unacceptable; preferences among imperfect tests always depend on local expertise and test availability. We will need additional comparisons between PET-CT and other staging tests—preferably studies that measure clinical outcomes and include cost analyses—as lung cancer staging continues to transform and improve.

Mitchell L. Margolis, MD

Philadelphia Veterans Affairs Medical Center and University of Pennsylvania

Philadelphia, PA 19104