|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 22, No 20 (October 15), 2004: pp. 4147-4156 © 2004 American Society of Clinical Oncology. DOI: 10.1200/JCO.2004.12.115 Clinical Decisions Associated With Positron Emission Tomography in a Prospective Cohort of Patients With Suspected or Known Cancer at One United States CenterFrom the Department of Internal Medicine and Radiology, Virginia Commonwealth University, and Massey Cancer Center, Richmond, VA Address reprint requests to Bruce E. Hillner, Division of General Medicine, Virginia Commonwealth University, 1101 E. Marshall St, Sanger Hall, Room 7-083, Richmond, VA, 23298; e-mail: Hillner{at}mail2.vcu.edu
PURPOSE: In 2001, Medicare approved reimbursement of F-18 fluorodeoxyglucose positron emission tomography (PET) for a variety of cancers. PET has been observed to be more accurate than other imaging in cancer patients, but the impact of PET on management in routine practice is uncertain. PATIENTS AND METHODS: We studied a prospective cohort having noninvestigational PET at one university center. Before and after PET, a questionnaire was administered to solicit information regarding each physician's preceding actions, intended management, and probability estimates. RESULTS: Seventy-one physicians provided data on 248 patients, of whom 40% had new or suspected cancer and 60% were undergoing restaging or had suspected recurrence. Lung, lymphoma, and head/neck cancers accounted for two thirds of cases. Sixteen physicians made 64% of requests. Physicians changed their intended management in 61% of patients (95% CI, 54% to 66%). For individual physicians ordering at least 10 scans, the average kappa was 0.16 (range, –0.04 to 0.36), reflecting only slight level of agreement between their before and after PET plan. PET was associated with a change in 90 (79%) of 114 patients if the pre-PET intended plan involved more testing or biopsy. In 32% of cases, physicians changed to a treatment from a nontreatment strategy. The therapeutic goal and mode changed in 22 (7%) and 21 cases (8%), respectively. CONCLUSION: This study confirms that physicians often change their decision making based on PET. This impact is likely due to combined effects of PET's improved accuracy and reduced physician uncertainty. Physicians may also be overconfident in interpreting PET and use it as the final arbiter after an extensive evaluation in lieu of tissue biopsy.
Medical imaging technologies have a recurring history of diffusing into routine marketplace use before scientifically sound technology assessment. Over the last 25 years, if and how to use computed tomography (CT) and, subsequently, magnetic resonance imaging of various anatomic sites and patient symptoms occurred in an expensive, chaotic manner. Positron emission tomography (PET) using 18-F fluorodeoxyglucose (FDG) has recently been approved in the United States for patients with suspected cancer or the staging of most cancers. This technology is a prototypic example of the challenges facing medicine: the tool is likely to be superior to current techniques, more expensive, and used as supplement rather than a replacement for the current modality. An extensive literature search has found that PET has an improved sensitivity and specificity for staging patients compared with other imaging methods, predominantly CT scans.1-4 In addition, there is a growing body of literature suggesting that PET has a role in predicting response to treatment and prognosis and can detect residual tumor in lymphomas.5,6 However, improved test performance does not necessarily translate into meaningful changes in treatment decisions nor does altered treatment necessarily translate into improved patient outcomes. There is limited literature, primarily in lung cancer, showing PET changed or influenced decision making in 40% to 60% of cases.7-10 The only known United States reports in breast cancer, lymphoma, and colorectal cancer assert that PET is associated with a 30% to 40% change in management.11-13 However, these studies did not collect information about the intended management plan before PET and had low (approximately 40%) response rates. In this report, we describe a prospective cohort from one university center addressing the clinical impact of PET in all cancer types that assessed physicians' intended management before and after PET. These data were supplemented by pre- and post-PET probability estimates of either the presence of cancer, its resectability, or recurrence. This population is likely to be representative of most referral centers in the United States.
Consent and Study Design The project was reviewed and approved by the university institutional review board (approval No. 2091). Patient informed consent was explicitly considered and determined to not be necessary because the project did not restrict patient access to PET. Because the research subject was the ordering physician and not the patient, informed consent was obtained from all participating physicians. Before starting the project, one investigator (B.E.H.) met and sought consent from Virginia Commonwealth University Health System (VCUHS) physicians who were expected to be frequent requestors of PET. If an individual agreed to participate, consent was required only with the first patient. For other VCUHS physicians and non-university referring physicians, a consent form was sent concurrent with the pre-PET questionnaire. For each complete question set, the physician received a $15 voucher at a video store. All but three VCUHS and five outside physicians agreed to participate (71 of 79 physicians). All patient requests for a PET scan at the VCUHS Department of Radiology from December 2002 to September 2003 were eligible. The only study exclusion was if the PET was done as part of a clinical trial. When a PET study was ordered, the physician was sent a two-page form that combined demographic, insurance, and clinical information needed to establish that PET was clinically indicated and reimbursable. The physician was required to give the indication for the study (suspected cancer, surgical resectability, initial staging, restaging, or suspected recurrence of previously treated cancer), the cancer type, the extent of prior imaging evaluation, usually CT and magnetic resonance imaging, the availability of those images, and a free text summary of the clinical question. A second page of targeted questions (Fig 1) included probability estimates and intended management if PET was not available. Approximately 1 week after the completion of PET, the ordering physician was sent a copy of PET interpretation and a one-page survey (Fig 2). The survey asked about their intended management plan, probabilities of cancer and whether PET results suggested new lesions or were consistent with metastatic disease and allowed them to avoid additional tests or procedures. Data were requested regarding whether a tissue biopsy had be done or was planned. Repeat faxes or telephone calls were made weekly until the follow-up form was returned.
The FDG PET procedure was performed using a standardized protocol on a GE Advance scanner (General Electric Medical Systems, Milwaukee, WI). Patients fasted at least 8 hours before imaging. FDG (Eastern Isotopes, Sterling, VA) 15 mCi + 20% was administered intravenously 120 minutes before imaging. Two-dimensional emission and corresponding transmission acquisitions extending from the midhead to the midthighs were done. Attenuation correction was used and images were reconstructed using filtered back projections. Tomographic views from trans-axial, coronal, and sagittal planes and a maximum-intensity projection were reviewed for study interpretation. The PET scan was interpreted with all available clinical information and imaging studies. The PET report format was not altered according to whether the physician was or was not participating.
Sample Size and Statistics Statistical methods were chosen that specifically control for dependencies induced between PET scans ordered by the same physician. The relationship between pre- and post-PET intended management, categoric variables, was assessed via the Cochran-Mantel-Haenzel method. Kappa values, which quantify the extent of agreement, were assessed for physicians with at least 10 PET scans in the study. A useful ad hoc benchmark for the strength of the agreement is poor (0), slight (0.00 to 0.20), fair (0.20 to 0.40), moderate (0.40 to 0.60), substantial (0.60 to 0.80), and almost perfect (0.80 to 1.00).14 Mixed model analysis of variance was used to assess overall change in estimated probability of cancer and metastatic disease plus the effect of these physician probability estimates on intended management.15 Least squares means and their SEs were calculated. The effectiveness of the PET scan in identifying new lesions, detecting lesions suggestive of metastatic disease, or avoiding more tests on intended management was assessed via the Cochran-Mantel-Haenzel method. SAS (version 9.0; SAS Institute, Cary, NC) and SPSS (version 11.0; SPSS Inc, Chicago, IL) were used for statistical analysis.
Classifying Management
Patient Profile A total of 248 patients made up the study cohort. Table 2 lists patient characteristics. Median age was 55.0 years. Fifty-seven percent of patients were men, and 8% were diabetic. Two hundred eight patients (84%) had prior radiology imaging before PET, with the vast majority being CT scans. Lung cancer and lymphomas accounted for approximately half of the PET studies.
Physician Profile In the 10-month study period, 71 physicians (54 VCUHS, 17 private) ordered studies. Table 3 lists the 16 physicians who ordered five or more studies and accounted for 64% of the cohort. Of these high-volume users, the number participating and the current total VCUHS faculty were six medical oncologists (11 on faculty), two radiation oncologists (seven on faculty), two thoracic surgeons (two on faculty), one gynecologic oncologist (one on faculty), two ear, nose, and throat (three on faculty), two surgical oncologists (four on faculty), and one pulmonologist (five on faculty). Medical oncologists and surgeons requested a disproportion share of the studies.
Clinical Indication Table 4 stratifies the five clinical indication categories for PET by the type of cancer and the pre-PET intended plan. Approximately 20% of scans were done for diagnosis or surgical resectability questions, 20% as part of initial staging, 29% for restaging, and 31% for suspected recurrences. Cases where diagnosis or surgical resectability was the question were predominantly in patients with lung or head and neck cancers.
The physician's intended management if PET was unavailable was observation with close follow-up in 38 cases (15%), additional noninvasive tests or imaging in 66 cases (27%), tissue biopsy in 48 cases (19%), or treatment (96 cases; 39%) with curative intent in 60 cases (24%) or palliative intent 36 cases (15%). In cases where diagnosis or surgical resectability was the question, the intended management in lieu of PET was most commonly to do a tissue biopsy (27 of 51 cases; 53%). If PET was unavailable and the indication was initial staging, the physician would have initiated therapy in the majority of cases (28 of 47 cases; 60%). In cases of restaging, the most common strategy if PET was unavailable would be additional tests or other imaging in 26 (30%) of cases. In cases of suspected recurrence, there was no dominant intended plan.
Impact of PET on Management
The impact of PET was greatest in patients where the management if PET were unavailable was additional testing or tissue biopsy. In 27 patients (24%), the management plan was revised to watch, and in 63 patients (55%), the management plan was revised to a treatment strategy of the initial 114 patients. In patients for whom treatment would have been initiated if PET were unavailable, the PET result changed intended management in 22 (23%) of 96 patients. Figure 3 summarizes the concordance and changes in intended management associated with PET after including changes in treatment goal. In 97 (39%) of patients, the management plan remained unchanged. In about one quarter, patients who would have had additional testing or biopsy switched to treatment. Conversely, approximately one in 10 patients went from additional testing or biopsy to watching. Slightly more patients (23%) went from a planned treatment strategy to a watch strategy than its opposite of a watch to a treatment plan (16%). Sixteen patients (6%) had a change in the goal of treatment. In only seven patients (3%) did the post-PET plan include additional testing or biopsy that was not part of the initial management plan.
Changes in Intended Management Including Specific Treatments Table 6 shows the impact of PET when including three nontreatment actions (watch, test or image, biopsy) and six specific treatments. Using this more rigorous comparison, 75 (32%) of 235 patients had no change in their management plan. Seventy-five patients (32%) had a change from a nontreatment plan to a treatment one. A variety of therapies were listed in revised post-PET strategies, with chemotherapy alone or in combination planned in 42 patients. Twenty-six patients (10%) switched from any form of intended treatment to a post-PET plan of nontreatment.
Physician Probability Estimates and Perceived Contribution of PET Surgical resection was at issue in 18 initial-diagnosis and eight suspected-recurrence patients. Because only 22 patients had both pre- and post-PET estimates, the average estimate change from 50% pre-PET to 34% post-PET was not statistically significant. Physician probability estimates for metastatic disease in any given patient covered the entire range from 0% to 100%. The pre-PET probability estimate of metastases was less than 20% in 69 patients (30%) and was greater than 80% in 48 patients (20%). After the PET scan, using these same thresholds, 72 (31%) of the estimates were less than 20%, and 95 estimates (40%) were greater than 80%. The mean estimated probability of metastases increased from 45% to 58%, which was statistically significant (P = .005). After PET, the physician summary impression was that 108 (47%) had more advanced disease, 96 (41%) had no change, and only 29 (12%) had lower or less extent of disease. Physicians recorded that based on the PET findings, 57% of patients would avoid future test or procedures. However, the value of PET in avoiding additional actions was not associated with changes in intended management (P = .5). Table 7 shows factors associated with a change in intended management from nontreatment to treatment or vice versa. These exploratory results adjusted for the physician clustering of PET request per physician. There was no difference in the pre-PET probability of metastatic disease between groups. Patients switching from a nontreatment to a treatment strategy had a greater increase in the physician probability estimate of metastases (44% to 71%; P = .0004). Similar but smaller differences were seen in the physician's impression that PET identified new sites of disease in patients switching to a treatment strategy (P = .032).
Medical imaging with PET is the most recent example of another so-called break-through technology. An extensive literature search, while noting methodology flaws or limitations, has suggested that PET is superior to other approaches in the diagnosis of suspected or recurrent cancer. Distinct from its imaging role, other preliminary studies have suggested that FDG activity has a role in predicting prognosis and response to chemotherapy. After almost two decades of development, the United States Center for Medicare and Medicaid Services recently began to approve FDG PET for reimbursement on indication-by-indication bias. In 2003, the approved list includes a wide variety of cancers and cancer indications.16 Although PET may be more accurate than other approaches, the value or impact of PET depends on if it is used as an add-on, not replacement, technique, and subsequently if and how it changes clinician behavior.17 This potential impact must be balanced against its financial cost of currently approximately $2,000 per study. In this prospective, single-center, before-after study, the impact of PET on physician decision making in a wide unselected cross-section of cancers was considerable. Our finding that PET altered management in approximately 60% of cases. Although this finding is consistent with other studies, primarily in lung cancer, it should not be assumed that the changes in management were necessarily correct or optimal.18 The strengths of this study were its diversity of cancers and requesting physicians, its prospective design, the collection of intended management before reporting of the PET findings, and its high (> 90%) response rate. Previously uninvestigated variables of interest related to the impact value of PET were our inclusion of probability estimates before and after of metastatic disease and the physician's impression of whether the PET findings would reduce subsequent tests or procedures. The dominant and inherent limitation of the project was the use of intended management or decision making as the primary preplanned end point. This was chosen because the period of interest was shortly after PET completion and our anticipation that many PET studies would be ordered by VCUHS physicians serving as consultants to outside physicians. Nevertheless, the project could not determine whether the changes in management were appropriate or the subsequent consequences in patient outcomes. Other studies have strived to understand the value of PET in improving diagnostic understanding, how it affected treatment choice, and physician confidence.8-10 The current study did not attempt to dissect PET's value. Changes in subsequent action were used to indicate value and included measuring a change in therapeutic intent from palliative or curative. Our study found that approximately 40% of PET studies were requested in patients with new or suspected cancer and the other 60% for reassessing known cancer. In either setting, in patients where the intended management in the absence of PET would have been additional imaging, other testing, or tissue biopsy, PET had the greatest impact. Such patients accounted for 46% of the cohort. In 79% of these patients, PET led to a revised plan of treatment or observation. Among patients whose intended management was treatment, the treatment goal changed after PET in 16% and the specific mode in 17%. As might have been expected, physicians changed three times as often from a nontreatment strategy to a treatment one than vice versa. Increases in physician probability estimate of metastases were associated with switching to a treatment plan. Although physicians' impression in more than half the patients was that PET would allow them to avoid further tests or procedure, this was not associated with changes in their intended management. A major concern raised from this study is the potential over-confidence of physician interpretation of the PET findings. There is an extensive body of literature showing that physicians are inaccurate in probability estimation, revision, and prognosis.18,19 These interpretative errors are distinct from a test's sensitivity and specificity. Because of the concern of false-positive findings, many experts recommend histologic confirmation of unexpected PET abnormalities.20,21 A limitation of this study was its failure to ask clinicians if the PET findings were consistent with other clinical data or were unexpected findings. This is most warranted in the 56 patients with suspected recurrence or restaging of their cancer whose management plan switched from nontreatment to chemotherapy or radiation. Only 24 patients (43%) had a tissue biopsy done or planned before initiating these treatments. In conclusion, our study confirms similar studies from outside the United States that PET study findings change clinical management in more than half of patients who have had extensive preceding imaging. Although PET has previously been most extensively evaluated related to surgical resection of primary lesions or suspected liver metastases, our study found current United States use was most often for suspected cancer recurrences. Clinical management changed from watching or testing to active treatment in one of three patients and from treatment to observation in one in 10 patients. No specific probability thresholds were identified for changes in actions. As others have pointed out, this substantial impact is likely due to a combination of PET's improved accuracy that is intertwined with changes in physician confidence or reduced uncertainty based on PET scan results. Systematic assessments in using PET earlier in the diagnostic process are warranted.
The authors indicated no potential conflicts of interest.
We thank the participating physicians and Donna McClish for assistance in the statistical evaluation.
The primary support was provided by a grant from the Agency for Healthcare Research and Quality to B.E.H. (grant No. R03 HS13244-01). Additional support to B.E.H. provided by a Research Scholar Grant for Health Services, Health Policy and Outcomes Research grant No. RSGHP-04-003-01-CPHPS from the American Cancer Society. Authors' disclosures of potential conflicts of interest are found at the end of this article.
1. Gould M, Maclean C, Kuschner W, et al: Accuracy of positron emission tomography for diagnosis of pulmonary nodules and mass lesions: A meta-analysis. JAMA 285: 914-924, 2001
2. Huebner R, Park K, Shepherd J, et al: A meta-analysis of the literature for whole-body FDG PET detection of recurrent colorectal cancer. J Nucl Med 41: 1177-1189, 2000 3. Toloza EM, Harpole L, McCrory DC: Noninvasive staging of non-small cell lung cancer: A review of the current evidence. Chest 123: 137S-146S, 2003
4. Kinkel K, Lu Y, Both M, et al: Detection of hepatic metastases from cancers of the gastrointestinal tract by using noninvasive imaging methods (US, CT, MR imaging, PET): A meta-analysis. Radiology 224: 748-756, 2002 5. Naumann R, Vaic A, Beuthien-Baumann B, et al: Prognostic value of positron emission tomography in the evaluation of post-treatment residual mass in patients with Hodgkin's disease and non-Hodgkin's lymphoma. Br J Haematol 115: 793-800, 2001[CrossRef][Medline]
6. Spaepen K, Stroobants S, Dupont P, et al: Early restaging positron emission tomography with (18)F-fluorodeoxyglucose predicts outcome in patients with aggressive non-Hodgkin's lymphoma. Ann Oncol 13: 1356-1363, 2002
7. Pieterman R, van Putten J, Meuzelaar J, et al: Preoperative staging of non-small-cell lung cancer with positron-emission tomography. N Engl J Med 343: 254-261, 2000 8. van Tinteren H, Hoekstra O, Smit E, et al: Effectiveness of positron emission tomography in the preoperative assessment of patients with suspected non-small-cell lung cancer: The PLUS multicentre randomised trial. Lancet 359: 1388-1393, 2002[CrossRef][Medline]
9. Kalff V, Hicks R, MacManus M, et al: Clinical impact of (18)F fluorodeoxyglucose positron emission tomography in patients with non-small-cell lung cancer: A prospective study. J Clin Oncol 19: 111-118, 2001
10. Herder G, Van Tinteren H, Comans E, et al: Prospective use of serial questionnaires to evaluate the therapeutic efficacy of 18F-fluorodeoxyglucose (FDG) positron emission tomography (PET) in suspected lung cancer. Thorax 58: 47-51, 2003
11. Yap C, Seltzer M, Schiepers C, et al: Impact of whole-body 18F-FDG PET on staging and managing patients with breast cancer: The referring physician's perspective. J Nucl Med 42: 1334-1337, 2001
12. Meta J, Seltzer M, Schiepers C, et al: Impact of 18F-FDG PET on managing patients with colorectal cancer: The referring physician's perspective. J Nucl Med 42: 586-590, 2001
13. Schoder H, Meta J, Yap C, et al: Effect of whole-body 18F-FDG PET imaging on clinical staging and management of patients with malignant lymphoma. J Nucl Med 42: 1139-1143, 2001 14. Landis J, Koch G: The measurement of observer agreement for categorical data. Biometrics 33: 159-174, 1977[CrossRef][Medline] 15. Laird NM, Ware JH: Random-effects models for longitudinal data. Biometrics 38: 963-974, 1982[CrossRef][Medline] 16. FDG Positron Emission Tomography (CAG-00065) Decision Memorandum: National Medicare Coverage Decision on FDG PET. Federal Register December 15, 2000
17. Deyo R, Jarvik J: New diagnostic tests: Breakthrough approaches or expensive add-ons? Ann Intern Med 139: 950-951, 2003 18. Black ER, Bordley DR, Tape TG, et al: Diagnostic Strategies for Common Medical Problems (ed 2). Philadelphia, PA, American College of Physicians, 1999
19. Lamont EB, Christakis NA: Prognostic disclosure to patients with cancer near the end of life. Ann Intern Med 134: 1096-1105, 2001
20. Detterbeck F, Falen S, Rivera M, et al: Seeking a home for a PET, Part 2: Defining the appropriate place for positron emission tomography imaging in the staging of patients with suspected lung cancer. Chest 125: 2300-2308, 2004
21. Detterbeck F, Falen S, Rivera M, et al: Seeking a home for a PET, Part 1: Defining the appropriate place for positron emission tomography imaging in the diagnosis of pulmonary nodules or masses. Chest 125: 2294-2299, 2004 Submitted December 16, 2003; accepted July 30, 2004.
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2004 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
