| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Genitourinary Imaging |
1 From the Departments of Radiology (G.J.J., S.H.J.R., J.O.B.) and Urology (J.J.M.C.H.d.l.R.), University Hospital Nijmegen, Geert Grooteplein zuid 18, 6500 HB, Nijmegen, the Netherlands; the Department of Medical Technology Assessment, University of Nijmegen, the Netherlands (J.L.S.); and the Department of Radiology, University of Wisconsin-Madison (J.R.T.). From the 1998 RSNA scientific assembly. Received February 1, 1999; revision requested March 31; final revision received August 16; accepted August 26. Address reprint requests to G.J.J. (e-mail: G.Jager@rdiag.azn.nl).
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIAL AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Literature review was performed by using the principles of evidence-based medicine and medical technology assessment. A decision analytic model was used to compare (a) the strategy that radical prostatectomy is performed on the basis of clinical staging with (b) the strategy that extracapsular disease detected at MR imaging contraindicates radical prostatectomy in patients who were considered surgical candidates on the basis of clinical staging.
RESULTS: After review of the literature, expert panel opinion did not recommend MR staging. No studies in which therapeutic efficacy was addressed were found. The decision analytic model indicated that the strategy including MR staging decreased costs (MR imaging, $10,568; radical prostatectomy, $11,669) and resulted in almost equal life expectancy (MR imaging, 12.59 years; radical prostatectomy, 12.60 years) and quality-adjusted life-years ([QALYs] MR imaging, 12.53; radical prostatectomy, 12.52). Results of sensitivity analyses demonstrated that the MR strategy was both more effective and less costly if the prior probability of extracapsular disease was at least 39% when considering QALY and 50% when considering unadjusted life expectancy.
CONCLUSION: It is not yet conclusively determined whether preoperative MR staging is appropriate, but results of decision analysis suggest that MR staging is cost-effective for men with moderate or high prior probability of extracapsular disease.
Index terms: Cost-effectiveness, 844.1214 • Diagnostic radiology, observer performance, 844.1214 • Economics, medical, 844.1214 • Prostate neoplasms, 844.32 • Prostate neoplasms, MR, 844.1214, 844.32 • Prostate neoplasms, surgery, 844.1267 • Radiology and radiologists, outcome studies, 844.1214 • Radiology and radiologists, socioeconomic issues, 844.1214
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIAL AND METHODSRESULTSDISCUSSIONReferences |
|---|
Literature evidence indicates that localized prostate cancer can be treated successfully by means of radical prostatectomy in the patient group having a life expectancy of 10–15 years or more. Nevertheless, benefits of surgical treatment over watchful waiting in terms of quality-adjusted life-years (QALYs) are often small, which leads to controversies about the preferred treatment (2–6).
Among men with prostate cancer, at presentation 60% have tumors apparently confined to the prostate (7,8). However, clinical staging appears to be unreliable. Understaging occurs in approximately 40% of cases (7,9). Since 1984, magnetic resonance (MR) imaging has been available for use as a staging modality for prostate cancer. The results are diverse and conflicting (10–12). Urologists, in general, advise against the routine use of MR imaging in staging prostate cancer (13,14).
The purpose of our study was to determine the appropriate use of MR imaging for preoperative staging of prostate cancer. In the radiology literature, the role of MR imaging in local staging is still debated, although currently many radiologists consider MR imaging the best staging modality (15). However, the use of MR imaging to select patients with prostate cancer for curative therapy is justified only if implementation of MR imaging will improve patient outcome and if there is a real justification on the basis of cost, risk, and benefit. To evaluate these questions, we performed a thorough review and evaluation of the literature. We then performed a formal decision analysis with data, assumptions, and conditions obtained by means of literature review.
|
MATERIAL AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIAL AND METHODSRESULTSDISCUSSIONReferences |
|---|
Abstracts were checked for eligibility. Articles containing data on staging strategies; MR staging; and costs and effects of screening, staging, and treatment were retrieved.
Analysis: Literature Review
The literature was reviewed for evidence concerning the use of MR imaging of prostate cancer. The level of evidence was graded in three categories: (a) randomized controlled trials, (b) other robust experimental or observational studies, (c) more limited evidence with advice that is based on expert opinion and that has the endorsement of respected authorities (16). We found two articles reporting level c evidence (17,18). In 1992, results from the Consensus Workshop on Screening and Global Strategy for Prostate Cancer did not recommend MR imaging as a staging tool (17). The National Comprehensive Cancer Network recently did not include MR staging in the staging guidelines (18).
Published research on the clinical efficacy of MR staging subsequently was classified into levels of assessment according to a six-tiered hierarchic model (19) (Table 1). Our literature review resulted in seven articles containing data about level 1, 14 about level 2, and five about level 3 of the hierarchic model. The feasibility of MR imaging for providing high-quality images of the prostate was reported extensively in the period before 1991. There is a wide range of reported accuracy rates that vary from 50% to 90% (20).
|
10–20 ng/mL, Gleason score of 5–7); endorectal coil MR images provided a more accurate prediction of extracapsular disease or seminal vesicle invasion than did clinical staging (21,22). (b) MR staging results should be interpreted in combination with clinical findings (23). (c) To be more cost-effective, MR staging should be performed with a high specificity (24).
Decision Analytic Model
Model.—Decision analysis, a modeling instrument for specific complex choices in situations of uncertainty, involves choosing a strategy after weighing the risks and benefits of the alternatives. We constructed a decision analysis model (Figure) to compare (a) the strategy that the decision to perform radical prostatectomy is based on clinical staging with (b) the strategy that extracapsular disease detected at MR imaging contraindicates radical prostatectomy in patients who are considered surgical candidates on the basis of clinical staging. If the patient did not undergo surgery, the treatment strategy was expectant management or "watchful waiting," including several approaches that ranged from reserving palliative interventions for patients who developed symptomatic local progression or metastatic disease to withholding potentially curative therapy until signs of disease progression were seen (25). Patients who did not benefit from surgery were considered to have the same course of disease as that of patients from the expectant management group.
|
Second, because the diagnostic and therapeutic efficacies of MR staging have not yet been defined, we had to make an assumption about how MR imaging affects treatment. We assumed that extracapsular disease on MR images contraindicates surgery. To be sure that only patients with gross extracapsular disease will not undergo surgery, MR imaging should be performed with high specificity, preferably almost 100%, and relatively low sensitivity. Langlotz et al (24) demonstrated that high-specificity MR imaging (specificity, 97%; sensitivity, 30%) was possible and more cost-effective than endorectal MR imaging, with a sensitivity of 84% and a specificity of 80%. Results of other studies demonstrated that high-specificity MR imaging is possible. Yu et al (28) achieved a specificity of 95% and a sensitivity of 38%. With a radiologist-computer system, Seltzer et al (15) achieved a receiver operating characteristic curve with a very high accuracy. The receiver operating characteristic curve was constructed with sensitivity-specificity pairs of 30%-100%, 39%-98%, 47.5%-96.5%, and 56.5%-96%.
Data for the decision analysis.—Data were obtained about utilities (health care status measured from 0, death, to 1, perfect health), probabilities, and costs (with respect to diagnosis, treatment, and complications), and likely treatment outcomes were derived from the reported number of tumors understaged (29); meta-analysis of morbidity and mortality (30); comprehensive decision models comparing radical prostatectomy with other treatment options (2,4,31,32); and cost-effectiveness models of estimating the risks, benefits, and costs of screening (25,33,34). These are summarized in Table 2.
|
To examine the robustness of the decision analysis findings concerning the necessary assumptions, a sensitivity analysis was performed. The most common form of sensitivity analysis is one-way sensitivity analysis in which one parameter at a time (for instance, cost estimates) is varied across a plausible range. Two-way sensitivity analysis varies two parameters at the same time. Sensitivity analyses were performed on the variables specificity, sensitivity (Table 3), number of tumors understaged (Table 4), surgical costs (Table 5), and surgical mortality (Table 6).
|
|
|
|
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIAL AND METHODSRESULTSDISCUSSIONReferences |
|---|
By using baseline estimates, however, the analysis of expected costs for each strategy resulted in favor of using MR staging ($10,568) versus immediate surgery ($11,669) (Table 3). Estimates of life expectancy predicted 0.01 year longer survival for the surgical strategy based on clinical staging. However, the incremental or marginal cost per life-year gained was high ([$11,669 - $10,568]/0.01 = $110,100). Compared with the MR strategy, the surgical strategy was not cost-effective (36). Expressed in QALYs, compared with surgery, the MR strategy was both slightly better (12.53 vs 12.52) and less costly, which indicates dominance.
Performing sensitivity analysis of the prior probability of having extracapsular disease, we found that the calculated costs for both strategies were equal (threshold value) at a prior probability of 12% (Table 4) and that the threshold value for life expectancy was 50%; however, the effect on life expectancy was negligible (Table 4). The threshold value for QALYs was at a prior probability of 39% (Table 4). At performance of sensitivity analysis of surgical mortality, the threshold value was at 1.4% (Table 6). Sensitivity analysis of surgical cost is shown in Table 5.
A two-way sensitivity analysis of the variables sensitivity and specificity, with the use of data from a recently published receiver operating characteristic curve (15), is demonstrated in Table 3. If one moves from left to right on the receiver operating characteristic curve, the 30%-100% sensitivity-specificity pair increased both cost and QALYs compared with the baseline MR strategy. The ratio of incremental cost to QALY is $17,400 to QALY, which means that this sensitivity-specificity pair should be recommended in terms of cost-effectiveness (36). Moving one or two steps to the right on the receiver operating characteristic curve increases QALY and saves costs, which indicates dominance of the MR strategy with these sensitivity-specificity pairs compared with the baseline MR strategy. Lower specificity rates were not calculated because these were not according to our assumptions.
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIAL AND METHODSRESULTSDISCUSSIONReferences |
|---|
Literature Review
Decision making in health care is a complex process that is often strongly influenced by intuition and unsystematic clinical experience. It has been emphasized that evidence-based decisions should be made (16). Guidelines and recommendations should be developed based on strong evidence derived from the literature.
With respect to MR staging in prostate cancer, only limited evidence is available. The published level c evidence (more limited evidence with advice that is based on expert opinion and that has the endorsement of respected authorities) may be subject to debate because there are no estimates of what data the recommendations were based on and because no radiologists were involved in the development of these guidelines.
A diagnostic test ideally should be evaluated adequately before widespread introduction. Medical technology assessment is used to define the clinical efficacy of a new diagnostic test. The basic definition of efficacy is "the probability of benefit to individuals in a defined population from a medical technology applied for a given medical problem under ideal conditions of use" (37). Efficacy overlaps partly with effectiveness, which reflects the use of a medical technology in ordinary clinical practice rather than in ideal conditions. Thornbury and Fryback (38) advocated a hierarchic model to study the efficacy of a diagnostic imaging method (Table 1).
Most MR studies have been focused on the differentiation between T2 and T3 tumors, or diagnostic accuracy efficacy. There is a wide range of reported accuracy rates, which vary from 50% to 90% (20). To our knowledge, the effect of MR imaging on diagnostic thinking is still unknown, although the diagnostic thinking efficacy of MR staging has been evaluated to some extent (15,21,39,40).
To our knowledge, the efficacy on a higher level (therapeutic efficacy, patient outcome efficacy, and society outcome level) has not yet been reported. Thus, by reviewing the literature, we have no arguments to include routine MR imaging in the staging protocol.
Decision Analysis
One of the main reasons to use models in the economic evaluation of health care interventions is that relevant clinical trials have not been conducted. In these cases, modeling is considered "an unavoidable fact of life" (41). Because differences in outcomes between different treatment strategies for prostate cancer have not yet been assessed with randomized controlled trials, partly owing to the indolent course of the disease (42), it is obvious that the effect of MR staging on patient outcome could not be studied with randomized controlled trials.
Our decision analysis model is a valuable instrument in the evaluation of the merits of MR imaging in prostate cancer staging. The decision analysis was meant to evaluate the optimum use of MR imaging in staging prostate cancer in patients who were considered candidates for radical prostatectomy. Our baseline results, including the reported understaging rate of 40% (7,9), indicated that MR staging was cost-effective for routine use.
For clinical purposes, however, it is current practice to assess the risk of having extracapsular disease. This assessment is based on PSA level and Gleason score, which are derived from the nomograms of Partin et al (29). It is feasible to categorize patients into three subgroups with low (PSA level < 10 ng/mL and biopsy Gleason score of 2–7; PSA level of 10–20 ng/mL and biopsy Gleason score of 2–5), moderate (PSA level of 10–20 ng/mL and biopsy Gleason score of 5–7), and high (PSA level > 20 ng/mL or biopsy Gleason score of 8–10) risk of having extracapsular disease.
For some patients at intermediate and high risk, we consider MR staging a valuable procedure to avert unnecessary surgery. Even for the low-risk group, MR imaging is cost-effective. Laupacis et al (36) addressed the question of how attractive a technology has to be to warrant adoption and use. With the use of five "grades of recommendation," the magnitude of the incremental net benefit of a technology can be classified. Our analysis demonstrated that our alternative MR strategy clearly falls within the B grade, which indicates a technology that costs less than $20,000 per QALY, which is clearly not dominant but seems to be acceptable. MR staging lowers costs because it lowers the surgical rate. This is different from the urologic point of view that the accuracy of MR imaging is too low to justify its cost (13).
The assumption, however, that extracapsular disease on MR images will obviate surgery is untenable in patients with a very low risk of having extracapsular disease. For example, if the prior probability of extracapsular disease is 15%, the sensitivity is 33%, and the specificity is 97%, then true-positive extracapsular disease will be diagnosed in only 3% of patients, which is almost equal to the number of false-positive diagnoses (2,5). Therefore, one may assume that the surgeon will not consider the MR imaging results in making the treatment decision. For that reason, we recommend MR staging in the patient group with moderate or high risk of having extracapsular disease on the basis of the tables of Partin et al (29). In our hospital, about 50% of the patients are categorized into these groups; however, this rate is decreasing.
Our approach is slightly different from the approach of D'Amico et al (21,39). D'Amico et al performed staging to determine whether a patient is at high risk for extracapsular disease and should undergo radiation therapy or at low risk and should undergo surgery. They recommend performing MR imaging in the group of patients at intermediate risk (ie, patients with a 33%–66% chance of extracapsular disease). This is in about 14%–25% of patients. MR staging in this group considerably increases staging accuracy on the basis of clinical staging, PSA level, and Gleason score.
For two reasons, our analysis was from the third-party payer (Medicare) perspective. First, available data came from studies that evaluated the cost-effectiveness of PSA screening and the cost-effectiveness of different treatment options for localized prostate cancer and that also had the same perspective. Second, the purpose of our analysis was to justify the use of MR staging from the provider point of view. One should be aware that the outcome may differ if one chooses another economic perspective, such as that of the individual, hospital, health care system, or society. For example, from the societal perspective, we should also consider the cost of temporary inability to work, and the cost of relatives traveling for hospital visits may be considered.
Modeling always involves a compromise between simplicity and reality. A decision analytic model should synthesize the best available data. However, because of the indolent course of the disease and consequently long time horizon of the model, it is hard to determine the validity of the data used. For example, it is likely that in the near future other palliative treatment options will become available, which affects the outcomes of the model.
Our decision analysis study contains some specific limitations. Regarding our local situation, we have data only about the costs of MR imaging and radical prostatectomy—approximately $500 and $7,500, respectively. Therefore, we used data available from the literature in the United States. Results of recent studies suggested that both the cost and the mortality rate of radical prostatectomy have decreased (30,35). Therefore, the sensitivity analysis of surgical costs did not include the range $10,000–$20,000.
We performed the evaluation in only one age cohort, 65-year-olds, and we did not compare other treatment alternatives such as radiation or implantation of radioactive seeds. The main reasons were that the expectant management strategy was the most favored strategy outside the United States and that available data were not sufficient to perform a decision analysis with radiation treatment (25). Furthermore, we did not perform discounting; that is, taking into consideration that future costs and benefits have less value than costs today. In general, discounting is performed at an arbitrary rate of 5% (43).
Results from the model were in favor of surgery. We assumed that all men with confined disease were cured by means of prostatectomy; however, approximately 25% of men with apparently localized disease received additional treatment within 5 years of cancer diagnosis. This may be balanced by the observation that patients with only focal capsular penetration may be cured by means of radical prostatectomy. In our model, however, if we assume very strict criteria, almost all patients described as having extracapsular disease would have overt extracapsular disease. Furthermore, the two-way sensitivity analysis showed more favorable outcome for the sensitivity-specificity pairs derived from the receiver operating characteristic curve of Seltzer et al (15) (Table 3).
The model should be validated with data obtained prospectively. A thorough evaluation of such data about costs, benefits, values, and risks of different treatment options, when these become available, should be used to test our conclusions.
In the future, computer modeling may be helpful in selecting the patients in whom MR staging is appropriate. MR imaging may add valuable information that can be used in counseling the patient about the best treatment.
Future studies should also investigate the effect of MR staging on disease management. Well-designed cohort studies should be performed to compare the results and outcomes of radical prostatectomy in sites in which MR staging is used with sites in which it is not used. Because test indexes may be dependent on the prevalence of extracapsular disease, patients should be stratified in clinical subgroups on the basis of PSA level and Gleason score (ie, low, intermediate, and high risk for extracapsular disease). Data should be gathered from multiple institutions to obtain sound state-of-the-art evidence for standard clinical protocols. Imaging strategies and guidelines should be developed by expert panels that include radiologists using and citing evidence of such high scientific quality.
Results with our decision analysis model support the opinion that MR staging in preoperative work-up of prostate cancer is cost-effective, specifically in the group of patients with high or intermediate risk of having extracapsular disease after clinical staging. MR imaging should be performed with a high specificity, and if images are positive for extracapsular disease, the patient should be spared surgery.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Author contributions: Guarantor of integrity of entire study, G.J.J.; study concepts, G.J.J., J.L.S., J.R.T.; study design, G.J.J.; definition of intellectual content, all authors; literature research, G.J.J.; data acquisition and analysis, G.J.J., J.L.S.; manuscript preparation, G.J.J., J.L.S., J.R.T.; manuscript editing, G.J.J., J.L.S., J.R.T., S.H.J.R., J.O.B.; manuscript review, all authors
|
References |
|---|
|
TOPAbstractIntroductionMATERIAL AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  V. Kundra, P. M. Silverman, S. F. Matin, and H. Choi Imaging in Oncology from The University of Texas M. D. Anderson Cancer Center: Diagnosis, Staging, and Surveillance of Prostate Cancer Am. J. Roentgenol., October 1, 2007; 189(4): 830 - 844. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. W. T. P. J. Heijmink, J. J. Futterer, T. Hambrock, S. Takahashi, T. W. J. Scheenen, H. J. Huisman, C. A. Hulsbergen-Van de Kaa, B. C. Knipscheer, L. A. L. M. Kiemeney, J. A. Witjes, et al. Prostate Cancer: Body-Array versus Endorectal Coil MR Imaging at 3 T--Comparison of Image Quality, Localization, and Staging Performance Radiology, July 1, 2007; 244(1): 184 - 195. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. J. Futterer, S. W. T. P. J. Heijmink, T. W. J. Scheenen, G. J. Jager, C. A. Hulsbergen-Van de Kaa, J. A. Witjes, and J. O. Barentsz Prostate Cancer: Local Staging at 3-T Endorectal MR Imaging--Early Experience Radiology, December 1, 2005; 238(1): 184 - 191. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. J. Futterer, M. R. Engelbrecht, H. J. Huisman, G. J. Jager, C. A. Hulsbergen-van De Kaa, J. A. Witjes, and J. O. Barentsz Staging Prostate Cancer with Dynamic Contrast-enhanced Endorectal MR Imaging prior to Radical Prostatectomy: Experienced versus Less Experienced Readers Radiology, November 1, 2005; 237(2): 541 - 549. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. McNaughton-Collins, E. Walker-Corkery, and M. J. Barry Health-Related Quality of Life, Satisfaction, and Economic Outcome Measures in Studies of Prostate Cancer Screening and Treatment, 1990-2000 J Natl Cancer Inst Monographs, October 1, 2004; 2004(33): 78 - 101. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Cornud, T. Flam, L. Chauveinc, K. Hamida, Y. Chretien, A. Vieillefond, O. Helenon, and J. F. Moreau Extraprostatic Spread of Clinically Localized Prostate Cancer: Factors Predictive of pT3 Tumor and of Positive Endorectal MR Imaging Examination Results Radiology, July 1, 2002; 224(1): 203 - 210. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
