Cost-Effectiveness of Staging Computed Tomography of the Chest in Patients with T2 Soft Tissue Sarcomas

197 Cost-Effectiveness of Staging Computed Tomography of the Chest in Patients with T2 Soft Tissue Sarcomas Geoffrey A. Porter, M.D. 1 Scott B. Cantor, Ph.D. 1 Syed A. Ahmad, M.D. 1 Jeffrey T. Lenert, M.D. 1 Matthew T. Ballo, M.D. 1 Kelly K. Hunt, M.D. 1 Barry W. Feig, M.D. 1 Shreyaskumar R. Patel, M.D. 1 Robert S. Benjamin, M.D. 1 Raphael E. Pollock, M.D., Ph.D. 1 Peter W. T. Pisters, M.D. 1 1 Multidisciplinary Sarcoma Center, The University of Texas M. D. Anderson Cancer Center, Houston, Texas. 2 Section of Health Services Research, Department of Biostatistics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas. Presented at the 36th annual meeting of the American Society of Clinical Oncology, New Orleans, Louisiana, May 20 23, 2000. The authors gratefully acknowledge the assistance of Vivian Z. Garcia and Melissa Burkett in the preparation of this article. The current address for Geoffrey A. Porter, M.D., is 7-007 Victoria Building, Dalhousie University, QE II Health Sciences Center, 1278 Tower Road, Halifax, Nova Scotia B3H 2Y9, Canada. Address for reprints: Peter W. T. Pisters, M.D., Department of Surgical Oncology, Box 444, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030-4009; Fax: (713) 792-7829; E-mail: ppisters@mdanderson.org Received May 8, 2001; revision received August 30, 2001; accepted September 4, 2001. BACKGROUND. Published practice guidelines recommend routine chest computed tomography (CT) scanning as part of the staging evaluation for patients with T2 soft tissue sarcomas (STS), although there is no direct evidence to support this practice. The objective of this study was to determine the yield and cost-effectiveness of routine versus selective chest CT scanning for the staging of patients with T2 STS and to identify any subgroups for whom a more selective approach to chest CT scanning could be considered. METHODS. Six hundred consecutive patients with primary, nonthoracic, T2 ( 5 cm) STS underwent both chest X-ray (CXR) and chest CT scanning to evaluate the presence of pulmonary metastatic disease (M1). The authors constructed a decision tree that modeled the outcomes of diagnostic testing for two hypothetical diagnostic strategies: 1) routine chest CT (rct) or 2) CXR and selective chest CT (sct). The yield and cost of each strategy were determined; the incremental cost-effectiveness ratio (ICER) was calculated as the cost per additional patient with pulmonary metastases identified by rct versus sct. RESULTS. The yield of rct was higher than that of sct (M1 disease identified in 19.2% vs. 16.0% of patients, respectively), but rct was more costly ($1301 vs. $418 per patient, respectively). The ICER of rct compared with sct was $27,594 per patient identified with pulmonary metastasis. The expected yields, costs, and ICERs of the diagnostic strategies varied across patient subgroups based on grade, anatomic site, and tumor size. CONCLUSIONS. For patients with T2 STS, rct was most cost-effective in patients with high-grade lesions or extremity lesions. The findings of this study do not support the routine use of chest CT scanning in all patients with T2 STS. Cancer 2002;94:197 204. 2002 American Cancer Society. KEYWORDS: soft tissue sarcoma, chest computed tomography, cost-effectiveness, staging. The outcomes research movement has emphasized the importance of maximizing health care for a finite budget. 1,2 Applied to patients with malignant disease, there is increasing evidence that several radiologic and laboratory investigations that previously were considered important for staging and follow-up, in fact, are of little benefit. 3 9 Consequently, the cost-effectiveness of certain investigations the cost in relation to outcome has been questioned in the assessment of patients with several common malignancies. 7,10 However, the cost-effectiveness of staging procedures has not been examined in patients with soft tissue sarcoma (STS). Approximately 8100 new cases of STS are diagnosed annually in the United States, and fewer than 50% of patients with this diagnosis will be cured with current treatment modalities. 11 The majority of patients who die of STS will have pulmonary metastases, and it is 2002 American Cancer Society DOI 10.1002/cncr.10184

198 CANCER January 1, 2002 / Volume 94 / Number 1 estimated that 20 38% of patients will develop isolated pulmonary metastases at some point in the course of their disease. 12 14 The most commonly employed studies to evaluate patients for pulmonary metastases are chest X-ray (CXR) and chest computed tomography (CT) scanning. Of these, chest CT scanning is the most sensitive and is used commonly when patients who present with ostensibly localized STS undergo staging evaluation. This approach, however, is largely empiric. In critically evaluating the approaches to the identification of synchronous pulmonary metastases in patients with STS who present with ostensibly localized disease, it is important to define the potential benefits of such information. Knowledge of the presence of lung metastases may alter the surgical treatment of the primary lesion and/or may alter the use of adjuvant therapies. Most important, however, the identification of lung metastases changes the prognosis of a patient otherwise believed to have clinically localized disease and, depending on the extent of metastatic disease, may define the situation as incurable. Several studies have demonstrated that a subgroup of patients with primary STS and isolated, resectable pulmonary metastases may be treated by pulmonary metastasectomy with curative intent. 13 18 Those studies showed that pulmonary resection was associated with minimal perioperative mortality rates, actuarial 3-year survival rates of 20 54%, 14,15 and 5-year survival rates of 21 26%. 17,18 Unfortunately, many of those studies were limited by median follow-up of 3 years 16,17 and reported actuarial overall and/or disease specific survival rates; thus, they did not reflect the actual percentage of patients who survive without recurrence for at least 5 years. In fact, the proportion of STS patients with lung metastases who were cured by pulmonary metastasectomy, defined by absolute disease free survival 5 years, is only 9%. 14 Nevertheless, pulmonary resection, when possible, has been advocated as the only potentially curative treatment for patients with STS who have lung metastases. The practice guidelines published by the Society of Surgical Oncology and the National Comprehensive Cancer Network advocate a liberal approach to the use of chest CT scanning as a method of assessing for the presence of synchronous pulmonary metastases in patients who present with T2 extremity, visceral, or retroperitoneal STS. 19,20 However, these guidelines are consensus-based, not evidence-based. Thus, the objectives of this study were to assess the yield and cost-effectiveness of routine versus selective chest CT scanning in patients with T2 ( 5 cm) STS and to identify any subgroups for whom a more selective approach to chest CT scanning could be considered. FIGURE 1. Decision tree. Summary of the analytic model comparing two hypothetical diagnostic strategies. The algorithm models possible outcomes of 1) routine chest computed tomography (CT) scanning and 2) chest X-ray (CXR) with selective chest CT scanning. T2 STS: T2 soft tissue sarcoma; 3/12: 3 months later. MATERIALS AND METHODS Decision Analysis Using a clinical decision-analysis approach, 21,22 we modeled the outcomes of two diagnostic strategies for the pulmonary staging of patients with T2 primary STS, namely, routine chest CT scanning and selective chest CT scanning based on CXR results. The decisionanalysis model is presented in Figure 1. In the routine chest CT scanning approach, all patients would undergo chest CT scans, with possible positive, negative, or indeterminate results. For positive or negative chest CT findings, no further imaging would be required to complete the pulmonary staging assessment. However, for patients with indeterminate initial chest CT findings (e.g., 5 mm noncalcified pulmonary nodules), repeat chest CT scans would be performed at 3 months and could be either positive or negative for metastatic disease. In the patient cohort examined, five patients with indeterminate initial CT findings did not undergo repeat chest CT scans until 6 12 months after initial staging; all such repeat scans were negative. Thus, for the purposes of the analysis, it was assumed that a 3-month repeat chest CT study would have been negative in these five patients. For the selective chest CT approach, it was assumed that patients who had CXR results showing metastatic disease would undergo chest CT scanning to delineate further the extent and anatomic distribution of the pulmonary metastases. Patients with negative CXR results would undergo no further imaging. Patients with indeterminate CXR results would un-

Chest CT for T2 Soft Tissue Sarcoma/Porter et al. 199 dergo chest CT scanning, which could be positive, negative, or indeterminate. Patients with positive or negative chest CT results would undergo no further imaging, whereas patients with indeterminate CT results would undergo follow-up chest CT scanning in 3 months in the manner described above. In no case were initial CXR results interpreted with knowledge of the chest CT results. A uniform technique for chest CT scanning was not used, because many patients had the investigation performed at their referring institutions. Results were classified as positive, negative, or indeterminate based on the initial radiologist s report, with selective review of outside studies done at the discretion of the treating physician. Patients, Yield, and Cost-Effectiveness of Chest CT Scanning This study was based on data derived from a prospective STS data base that included demographic, clinical, and pathologic characteristics for all adult patients (age 16 years) with a diagnosis of primary STS who presented at The University of Texas M. D. Anderson Cancer Center s Multidisciplinary Sarcoma Center since June 24, 1996. The size of the primary tumor was determined by macroscopic pathologic examination or, for patients who underwent initial nonsurgical treatment, by CT scan or magnetic resonance imaging. From June 24, 1996 to September 30, 1999, 699 patients presented with primary, nonrecurrent STS measuring 5 cm in greatest dimension (T2). We excluded patients with primary tumors of the thorax or chest wall (n 36 patients), because chest CT scans usually would be required to evaluate the primary lesion adequately. In addition, we excluded patients with histologic subtypes known to have a low risk for metastatic spread, specifically, desmoid tumors (n 13 patients) and dermatofibrosarcoma protuberans (n 7 patients). Of the remaining 643 patients, 43 patients (7%) did not undergo both CXR and chest CT scan and, thus, were excluded from the study. Six hundred patients (93%) had undergone both CXR (posterior-anterior and lateral projections) and chest CT scans at or before initial evaluation at The M. D. Anderson Cancer Center and are the subject of this report. Data from this cohort of 600 patients with primary T2 STS staged by both CXR and chest CT scan were used to determine the expected yield and costs for each of the two strategies in the decision-analysis model. A variety of CT scanning techniques were used in these patients: Four hundred twenty-one patients (70%) had chest CT scans performed on spiral CT scanners. Chest CT scans were performed at a slice thickness of 5 mm, 6 9 mm, and 10 mm in 215 patients (36%), 96 patients (16%), and 289 patients (48%), respectively. Costs, not charges, were used in all economic analyses. Total costs were calculated as the sum of fixed and variable direct costs. Variable costs, which represent costs that are saved if a given procedure is not performed (e.g., the cost of a radiologist interpreting a chest CT scan), were determined with the aid of specialized computer software (Hospital Cost Consultants, Chicago, IL) and relied on information from a general ledger. Included in these variable costs were professional costs incurred by the hospital. Fixed costs, which represent costs that are incurred even if a procedure is not performed (e.g., the overhead costs associated with a CT scanner), were obtained by means of a statistical allocation from the cost accounting software. Total costs did not include any measurement or estimation of patient time costs (formerly called indirect costs), such as lost wages. In addition, costs associated with subsequent treatment were not estimated it was assumed that such costs would be similar for the two strategies and thus would not significantly impact the cost-effectiveness analysis. All cost data were indexed to 1999 U.S. dollars by multiplying recorded costs by the percentage inflationary increase, based on the medical component of the consumer price index, for each fiscal year, in a compound manner. Using this methodology, the cost of a CXR was calculated as $162, and the cost of a chest CT study was calculated as $1172. The primary outcome was the presence of pulmonary metastases, as identified by chest CT scan (either initial positive chest CT scan or initial indeterminate chest CT scan with a positive repeat CT scan at 3 months); histologic confirmation was not required for this diagnosis. The yield of each diagnostic strategy was defined as the number of patients with pulmonary metastases divided by the number of patients going through the diagnostic strategy. The measure of cost-effectiveness was the incremental cost-effectiveness ratio (ICER), which is the difference in the costs of the two strategies divided by the difference in the yields of the two strategies. Otherwise stated, the ICER expressed the additional cost per additional patient with pulmonary metastases detected by routine versus selective use of chest CT scanning. The ICER is a widely used, standard measure of cost-effectiveness. 23,24 In this study, the ICER was calculated as follows: ICER C r C s Y r Y s, where C r is the cost per patient for routine chest CT

200 CANCER January 1, 2002 / Volume 94 / Number 1 TABLE 1 Distribution of Clinicopathologic Prognostic Factors in 600 Patients with T2 Soft Tissue Sarcoma Clinicopathologic factor No. of patients % Gender Male 310 52 Female 290 48 Age (yrs) Median 53 Range 16 88 Tumor size (cm) 5.1 10.0 350 58 10.0 250 42 Tumor grade Low 99 17 Intermediate 87 15 High 414 69 Tumor location Extremity 228 38 Retroperitoneum 181 30 Viscera 123 21 Trunk 49 8 Head and neck 19 3 Histologic subtype MFH 112 19 Leiomyosarcoma 101 17 Unclassified sarcoma 99 17 Liposarcoma 95 16 GIST 30 5 Synovial sarcoma 25 4 Fibrosarcoma 15 3 Angiosarcoma 11 2 Other 112 19 MFH: malignant fibrous histiocytoma; GIST: gastrointestinal stromal tumor. scanning, C s is the cost per patient for selective chest CT scanning, Y r is the yield of routine chest CT, and Y s is the yield of selective chest CT. We performed subgroup analyses based on various clinicopathologic characteristics to estimate comparative ICERs for clinically relevant subgroups of patients. In addition, one-way sensitivity analyses were performed by altering the cost of chest CT scanning and recalculating the ICER. Two-way sensitivity analyses also were performed by recalculating the ICER with altered chest CT scanning costs across different tumor grade subgroups. RESULTS Patients, Costs, and Yield of Routine and Selective CT Scanning The distribution of clinicopathologic factors in the study cohort staged by both CXR and chest CT scanning is shown in Table 1. Forty-two percent of patients had large ( 10 cm) T2 tumors, and the majority of tumors were of high grade. The most common tumor location was an extremity (38%). The most common histologic subtypes were malignant fibrous histiocytoma (19%), leiomyosarcoma (17%), liposarcoma (16%), and unclassified sarcomas (17%). The routine chest CT strategy cost $1301 per patient and identified 115 patients (19.2%) with lung metastases (M1). The selective CT approach to pulmonary staging based on initial CXR results was significantly less expensive at $418 per patient and identified 96 patients (16.0%) with lung metastases. Otherwise stated, the selective chest CT approach identified 83.5% of patients who had lung metastases. This resulted in an ICER of $27,594, which represents the additional cost of routine chest CT scanning over selective chest CT scanning to identify one additional patient with of pulmonary metastases (Table 2). In examining the patients in whom the selective chest CT scanning approach did not identify pulmonary metastases that would have been identified using the routine chest CT scanning approach, no significant associations with patient demographics, tumor location, tumor size, or tumor grade were identified. Sensitivity Analyses Subgroup analyses were performed as a type of sensitivity analysis to determine the impact of histologic grade, anatomic site, and tumor size on cost-effectiveness (Table 2). Histologic grade of the primary lesion had a marked effect on the resulting ICER: The ICER for low-grade lesions was $99,800, whereas the ICER for high-grade lesions was only $21,538. Similarly, routine chest CT scanning for retroperitoneal lesions was associated with a relatively high ICER of $52,588, whereas it was found that this approach for extremity lesions was more cost-effective, particularly for large lesions (ICER, $12,524). A subgroup analysis examining the combined effect of grade and anatomic site suggested that grade was the critical factor driving the ICER (Table 3). The effect of the cost of chest CT scanning on the cost-effectiveness of the routine chest CT scanning approach was examined in a one-way sensitivity analysis (Fig. 2A). This was performed by recalculating the ICER for various costs of chest CT scans, assuming that the yield and the cost of CXRs remained constant. There was a linear correlation between the cost of chest CT scanning and the ICER of the routine CT chest scan approach compared with the CXR plus selective chest CT scan approach; as the cost of chest CT scans increased, so did the ICER. When the cost of chest CT scanning was reduced from the $1172 determined and used in this study, the ICER decreased such that, once the cost of chest CT scanning was $224 per

Chest CT for T2 Soft Tissue Sarcoma/Porter et al. 201 TABLE 2 Cost-Effectiveness of Routine versus Selective Approach to Chest Computed Tomography Staging for Patients with Primary T2 Soft Tissue Sarcoma Routine CT Selective CT Group Cost per patient Yield Cost per patient Yield ICER All patients $1301 0.192 $418 0.160 $27,594 Histologic grade Low $1314 0.060 $316 0.050 $99,800 Intermediate $1293 0.115 $337 0.092 $41,565 High $1299 0.239 $459 0.200 $21,538 Anatomic site Extremity $1347 0.219 $404 0.167 $18,135 Retroperitoneum $1263 0.133 $369 0.116 $52,588 Extremity tumor size (cm) 5.1 10.0 $1311 0.189 $359 0.154 $27,200 10.0 $1553 0.277 $501 0.193 $12,524 CT: computed tomography; ICER: incremental cost-effectiveness ratio (measured in dollars per additional patient identified with pulmonary metastases). TABLE 3 Cost-Effectiveness of Routine versus Selective Approach to Chest Computed Tomography Staging According to Soft Tissue Sarcoma Site and Grade Routine CT Selective CT Site and grade Cost per patient Yield Cost per patient Yield ICER Extremity, low grade $1310 0 $320 0 a Extremity, high grade $1328 0.270 $431 0.204 $13,591 Retroperitoneum, low grade $1284 0.047 $340 0.047 a Retroperitoneum, high grade $1280 0.180 $410 0.156 $36,250 CT: computed tomography; ICER: incremental cost-effectiveness ratio (measured in dollars per additional patient identified with pulmonary metastases). a No difference was seen in the yield for increased costs with routine CT; thus, ICER approaches infinity. CT study, routine CT scanning was no longer a more expensive option. The effect of chest CT sensitivity on the decisionanalysis model also was examined in a one-way sensitivity analysis. Although no published sensitivities of chest CT scanning for the detection of sarcomatous pulmonary metastases exist, we assumed a conservative 85% sensitivity based on the literature examining the sensitivity of spiral chest CT scanning (8-mm slice thickness) for the detection of pulmonary nodules. 25 The correlation between the sensitivity of chest CT scanning and the ICER, holding the sensitivity of CXR constant, is depicted in Figure 3. In line with expectations, as the sensitivity of chest CT scanning increased, the ICER decreased. Finally, a two-way sensitivity analysis of the effect of tumor grade and cost of chest CT scanning on the ICER was performed (Fig. 2B) In that analysis, as the hypothetical cost of chest CT scanning was reduced for each tumor grade, the ICER declined progressively. For hypothetical chest CT scanning costs of $201, $221, and $242 in patients with low-grade, mediumgrade, and high-grade tumors, respectively, routine chest CT scanning was no longer the more expensive option. DISCUSSION The management of patients with localized STS has improved over the past 25 years. For patients with extremity STS, there has been a transition from amputation to conservative surgery and radiotherapy as the primary form of local therapy. This has resulted in a decline in amputation rates 26 but no readily discernible decrease in distant metastases or sarcoma-related mortality. 27 29 Unfortunately, distant metastases remain a common pattern of treatment failure, 30,31 and, although adjuvant chemotherapy has been used to reduce this risk, its efficacy is modest at best. 32 Thus, the optimal staging and management of pulmonary

202 CANCER January 1, 2002 / Volume 94 / Number 1 FIGURE 3. Effect of the sensitivity of chest computed tomography (CT) scanning on the incremental cost-effectiveness ratio (ICER). The solid circle represents the literature-based assumed chest CT scan sensitivity used in this study. FIGURE 2. (A) Effect of the cost of chest computed tomography (CT) scanning on the incremental cost-effectiveness ratio (ICER). There is a linear correlation between the cost of chest CT scanning and the ICER; i.e., as the hypothetical cost of chest CT scanning decreases, so does the ICER of the routine chest CT strategy. The solid circle indicates the actual cost of chest CT scanning used in this study. (B) Effect of the tumor grade and the cost of chest CT scanning on the ICER. There is a linear correlation between the cost of chest CT scanning and the ICER for each tumor grade. The effect of an increase in the cost of chest CT scanning on the ICER is greatest for low-grade lesions. The solid circles indicate the actual cost of chest CT scanning used in this study. metastases remain major issues for physicians treating patients with STS. A previous study of patients with T1 extremity STS demonstrated that the yield of chest CT scanning in identifying patients with synchronous pulmonary metastases is low ( 1%) and suggested that chest CT scanning should not be used for patients with T1 lesions, regardless of tumor grade. 33 The current report provides data for an evidence-based approach to staging of patients with T2 STS. On the basis of the yield and cost-effectiveness analyses reported in this study, routine chest CT scanning appears to be indicated for patients with primary T2 STS of high grade (at any site) or arising in an extremity. In contrast, the routine use of staging chest CT scans does not appear to be warranted in patients with low-grade or retroperitoneal STS. These findings do not substantiate current consensus-based Society of Surgical Oncology staging guidelines, 20 which recommend chest CT scans for all patients with T2 STS, or National Comprehensive Cancer Network staging guidelines, 19 which advocate staging chest CT scans for all patients with retroperitoneal STS. In interpreting the aggregate data reported in this study, it is important to recognize the relative strengths and limitations of our methodology. This study used cost data (the amount of money required to actually provide a medical service) and not charge data (the amount of money a patient or third-party payer is asked to pay for a given medical service). Although charge data generally are easier to obtain, the use of cost data is less prone to the potential biases of different payer profit margins. Moreover, costs are less prone to the large variations in charges seen between different institutions. Finally, because costs theoretically do not include profits and deficits incurred by hospitals and third-party payers to provide a medical service, it is believed that they are more reflective of the patient, physician, and societal perspectives. The lack of a uniform CT scanning protocol and an independent review of all chest CT scans may be viewed as potential limitations in this study. However, the methodology used in this study was intended to be generalizable to the real world clinical situation, in which differing CT scanning techniques and reporting criteria are used. Moreover, the use of sensitivity analysis in the cost-effectiveness analysis enabled us to analyze our results with hypothetical alterations to clinicopathologic data, cost, sensitivity of chest CT scanning, and other underlying assumptions. For example, a one-way sensitivity analysis of the effect of

Chest CT for T2 Soft Tissue Sarcoma/Porter et al. 203 the cost of chest CT scans (simulating other institutions or technology in which chest CT scanning is less costly) showed that the ICER decreased such that, once the cost of chest CT scanning was $224 per CT study, routine CT scanning was no longer a more expensive option. We did not incorporate potential cost differences associated with subsequent treatment between the routine and selective chest CT scanning approaches; we assumed that such costs were similar for the two strategies. A proportion of patients with STS who had CT scan-identified, CXR-occult pulmonary metastases will not undergo surgery for resection of the primary lesion, resulting in a potential cost savings with the routine chest CT scanning approach. However, some such patients may receive additional chemotherapy, resulting in higher treatment-related costs in the routine chest CT scan group. It is noteworthy that, at the M. D. Anderson Cancer Center, the median cost associated with six cycles of doxorubicin-based chemotherapy in a cohort of 26 patients was $91,650 (1999 U.S. dollars; unpublished data) several times greater than the costs of major abdominal oncologic surgery (for example, the median cost of pancreaticoduodenectomy in a cohort of 80 patients was $36,627 in 1999 U.S. dollars). 34 Another difficulty with the type of cost-effectiveness analysis that we performed is the outcome measure used, which, in this study, was the cost to identify an additional patient with pulmonary metastasis. The cost per patient detected has been used for cost-effectiveness analyses in patients with other diseases. 35 37 However, it may be difficult to interpret and, specifically, to judge in terms of value with regard to exactly what represents an acceptable threshold cost for the detection of one patient. If long-term outcome data from a large cohort of prospectively followed patients with STS are used, then an estimated cost per life saved can be determined. Based on data from 719 patients who were followed prospectively at the Memorial Sloan-Kettering Cancer Center, only 22% of patients with pulmonary metastases are able to undergo complete resection of all pulmonary metastases, and only 9% of such patients who undergo complete resection will be alive and free of disease at 5 years (cured). 14 Using these data and the assumption that the behavior of synchronous and metachronous pulmonary metastases is similar, the cost per life saved in our study would have been $1.8 million (1999 U.S. dollars). It is noteworthy that this calculation ignores alterations in treatment strategies precipitated by the knowledge of metastatic disease, the significant costs associated with the surgical and nonsurgical treatment of pulmonary metastases, and the approximately 20% risk of late recurrence in these patients. 30 Thus, $1.8 million likely represents an underestimate of the true cost per life saved. A more accurate cost per life or life-year saved could be determined only by using objective, long-term outcome data as well as incorporating the costs of treating the pulmonary metastases. The former data do not exist currently in the literature on STS, and the treatment costs are extremely difficult to quantitate, because most patients with pulmonary metastases are treated with an individualized therapeutic plan based on disease status, performance status, and comorbidities. Notwithstanding the issues of yield and cost-effectiveness, some clinicians may believe that there is a secondary benefit to a policy of routine chest CT scanning. If post-treatment follow-up with serial chest CT scanning is planned, then normal baseline chest CT scans are useful when subsequent follow-up chest CT scans show new, small lesions that may be interpreted as indeterminate without a negative baseline chest CT scan. With the normal baseline study, such a scenario would be viewed by most clinicians as highly suspicious for metastatic disease. However, there is no evidence to suggest that earlier detection of small, CXR-occult, asymptomatic pulmonary metastases by post-treatment serial chest CT scan alters outcome. Moreover, few centers use serial chest CT scanning for routine post-treatment follow-up, minimizing the relative benefit of routine baseline chest CT scans. 12 In summary, this study found that, for patients who presented with primary T2 STS, routine chest CT scanning was most cost-effective for patients with high-grade lesions and patients with extremity lesions, especially large lesions. Routine chest CT scanning was less cost-effective if the lesion was retroperitoneal and/or low grade. This was due largely to the negligible absolute increase in yield of a routine approach versus a selective approach to chest CT scanning (yield rct yield sct) in these groups, although the relative difference in yields (yield sct/yield rct) was fairly constant across all patient subgroups. Taking into account previous work in conjunction with the results of this study, the routine use of chest CT scanning appears to be indicated only in patients with high-grade T2 lesions at any site patients with classic high-risk sarcoma. These observations do not support current consensus-based staging guidelines for patients with STS. Further studies of this type that specifically examine subsequent treatment and outcome in patients with synchronous lung metastases will be important to facilitate a more evidence-based approach to STS staging.

204 CANCER January 1, 2002 / Volume 94 / Number 1 REFERENCES 1. Gerszten PC. Outcomes research: a review. Neurosurgery 1998;43:1146 56. 2. Lee SJ, Earle CC, Weeks JC. Outcomes research in oncology: history, conceptual framework, and trends in the literature. J Natl Cancer Inst 2000;92:195 204. 3. Ravaiola A, Tassinari D, Pasini G, Polselli A, Papi M, Fattori PP, et al. Staging of breast cancer: what standards should be used in research and clinical practice? Ann Oncol 1998;9: 1173 7. 4. Ciatto SW, Pacini P, Andreoli C, Cecchini S, Iossa A, Grazzini G, et al. Chest X-ray survey in the follow-up of breast cancer patients. J Cancer 1989;60:102 3. 5. Isbister WH, al-sanea O. The utility of preoperative abdominal computerized tomography scanning in colorectal surgery. J R Coll Surg Edinburgh 1996;41:233 4. 6. McAndrew MR, Saba AK. Efficacy of routine preoperative computed tomography scans in colon cancer. Am Surg 1999; 65:205 8. 7. Berman JM, Cheung RJ, Weinberg DS. Surveillance after colorectal cancer resection. Lancet 2000;355:395 9. 8. Mooney MM, Mettlin C, Michalek AM, Petrelli NJ, Kraybill WG. Life-long screening of patients with intermediate thickness cutaneous melanoma for asymptomatic pulmonary recurrences. Cancer 1997;80:1052 64. 9. The GIVO Investigators. Impact of follow-up testing on survival and health-related quality of life in breast cancer patients. JAMA 1994;271:1587 92. 10. Virgo KS, Vernava AM, Longo WE, McKirgan LW, Johnson FE. Cost of patient follow-up after potentially curative colorectal cancer treatment. JAMA 1995;273:1837 41. 11. Greenlee RT, Murray T, Bolden S, Wingo PA. Cancer statistics 2000. CA Cancer J Clin 2000;50:6 32. 12. Whooley BP, Gibbs JF, Mooney MM, McGrath BE, Kraybill WG. Primary extremity sarcoma: what is the appropriate follow-up? Ann Surg Oncol 2000;7:9 14. 13. Gadd MA, Casper ES, Woodruff JM, McCormack PM, Brennan MF. Development and treatment of pulmonary metastases in adult patients with extremity soft tissue sarcoma. Ann Surg 1993;218:705 12. 14. Billingsley KG, Burt ME, Jara E, Ginsberg RJ, Woodruff JM, Leung DH, et al. Pulmonary metastases from soft-tissue sarcoma: analysis of patterns of disease and postmetastasis survival. Ann Surg 1999;229:602 12. 15. Jablons D, Steinberg SM, Roth J, Pittaluga S, Rosenberg SA, Pass HI. Metastasectomy for soft tissue sarcoma. J Thorac Cardiovasc Surg 1989;97:695 705. 16. Van Geel AN, Pastorino U, Jauch KW, Judson IR, van Coevorden F, Buesa JM, et al. Surgical treatment of lung metastases: the European Organization for the Research and Treatment of Cancer-Soft Tissue and Bone Sarcoma Group study of 255 patients. Cancer 1996;77:675 82. 17. Verazin GT, Warneke JA, Driscoll DL, Karakousis C, Petrelli NJ, Takita H. Resection of lung metastases from soft-tissue sarcomas: a multivariate analysis. Arch Surg 1992;127:1407 11. 18. Casson AG, Putnam JB, Natarajan G, Johnston DA, Mountain C, McMurtrey M, et al. Five-year survival after pulmonary metastasectomy for adult soft tissue sarcoma. Cancer 1992;69:662 8. 19. Demetri GD, Delany TF. NCCN sarcoma practice guidelines. National Comprehensive Cancer Network. Cancer Control 2001;6:94 101. 20. Pollock RE, Brennan MF, Lawrence W. Soft-tissue sarcoma surgical practice guidelines. Oncology 1997;11:1327 32. 21. Detsky AS, Krahn M, Naglie G, Naimark D, Redelmeier DA. Primer on medical decision analysis. Part 2: tree structures. Med Decision Making 1997;17:126 35. 22. Cantor SB. Decision analysis: theory and application to medicine. Primary Care 1995;22:261 70. 23. Tengs TO, Adams ME, Plisken JS, Safran DG, Siegel JE, Weinstein MC, et al. Five-hundred life-saving interventions and their cost-effectiveness. Risk Anal 1995;15:369. 24. Gold MR. Cost-effectiveness in health and medicine. New York: Oxford University Press, 1996. 25. Seemann MD, Seemann O, Luboldt W, Bonel H, Sittek H, Dienemann H, et al. Differentiation of malignant from benign solid pulmonary lesions using chest radiography, spiral CT, and HRCT. Lung Cancer 2000;29:105 24. 26. Williard WC, Collin C, Casper ES, Hajdu SI, Brennan MF. The changing role of amputation for soft-tissue sarcoma of the extremity in adults. Surg Gynecol Obstet 1992;175:389 96. 27. Yang JC, Chang AR, Baker AR, Sindelar WF, Danforth DN, Topalian SL, et al. A prospective randomized study of the benefit of adjuvant radiation therapy in the treatment of soft tissue sarcomas of the extremity. J Clin Oncol 1998;16:197 203. 28. Pisters PWT, Harrison LB, Leung DHY, Woodruff JM, Casper ES, Brennan MF. Long-term results of a prospective randomized trial of adjuvant brachytherapy in soft tissue sarcoma. J Clin Oncol 1996;14:859 68. 29. Rosenberg SA, Tepper J, Glatstein E, Costa J, Baker A, Brennan M, et al. The treatment of soft-tissue sarcomas of the extremities: prospective randomized evaluations of (1) limbsparing surgery plus radiation therapy compared with amputation and (2) the role of adjuvant chemotherapy. Ann Surg 1982;196:305 15. 30. Lewis JJ, Leung DHY, Casper ES, Woodruff J, Hajdu SI, Brennan MF. Multifactorial analysis of long-term follow-up (more than 5 years) of primary extremity sarcoma. Arch Surg 1999;134:199 204. 31. Potter DA, Glenn J, Kinsella EJ, Glatstein E, Lack EE, Restrepo C, et al. Patterns of recurrence in patients with highgrade soft-tissue sarcomas. J Clin Oncol 1985;3:353 66. 32. Tierney JF. Adjuvant chemotherapy for localized resectable soft-tissue sarcoma of adults: meta-analysis of individual data. Lancet 1997;350:1647 54. 33. Fleming JB, Cantor SB, Varma DGK, Holst D, Feig BW, Hunt KK, et al. Utility of chest computed tomography in patients with T1 extremity soft tissue sarcomas. Cancer 2001;92:863 8. 34. Porter GA, Pisters PWT, Mansyur C, Bisanz A, Reyna K, Stanford P, et al. Cost and utilization impact of a clinical pathway in pancreaticoduodenectomy patients. Ann Surg Oncol 2000;7:484 9. 35. Binkin NJ, Koplan JP. The high cost and low efficacy of weekly viral cultures for pregnant women with recurrent genital herpes: a reappraisal. Med Decision Making 1989;9: 225 30. 36. Nettleman MD, Geerded H, Roy MC. The cost-effectiveness of preventing tuberculosis in physicians using tuberculin skin testing of a hypothetical vaccine. Arch Intern Med 1997; 157:1121 7. 37. Cantor SB, Mitchell MF, Tortolero-Luna G, Bratka CS, Bodurka DC, Richards-Kortum R. Cost-effectiveness analysis of diagnosis and management of cervical squamous intraepithelial lesions. Obstet Gynecol 1998;91:270 7.