1994, The British Journal of Radiology, 67, 129-135 Lung metastasectomy sarcoma in patients with soft tissue 1 M H ROBINSON, MD, MRCP, FRCR, 2 M SHEPPARD, FRCPATH, 3 E MOSKOVIC, MRCP, FRCR and 4 C FISHER, MD, FRCPATH department of Clinical Oncology, Weston Park Hospital, Whitham Road, Sheffield S10 2SJ, UK, department of Histopathology, Brompton Hospital, Fulham Road, London SW3 6HP, UK and Departments of 3 Radiology and "Histopathology, Sarcoma Unit, Royal Marsden Hospital, Fulham Road, London SW3 6JJ, UK Abstract Metastases develop in 50-60% of patients with high grade soft tissue sarcomas despite primary treatment. Between 1970 and 1990, 189 patients with soft tissue sarcoma referred to the Royal Marsden Hospital Sarcoma Unit developed lung metastases as the sole first site of disseminated disease. 44 of these 189 cases have been treated by pulmonary metastasectomy. In an attempt to determine which patients benefit from this surgery the medical records, radiology and pathology of these cases have been reviewed. Both the overall 5 year survival (70% vs 19%) and the subsequent survival from the time lung metastases developed (52% vs 7.5%) of those selected for thoracotomy were better than in the 145 patients not undergoing surgery. On multivariate analysis, survival and control of lung disease following resection were not related to the number of metastases resected, completeness of excision, use of adjuvant chemotherapy or presence of bilateral disease. The most important factor was the use of lung resection itself as treatment, for which the risk of death was 0.2 compared with those not having metastasectomy. Age less than 40 years and primary tumour in a lower limb site were also factors associated with a reduced risk of death. Pulmonary metastasectomy should be considered in selected patients with soft tissue sarcoma after primary local cure. However, without collection of prospective data on all patients developing lung metastases or a randomized trial the true role of lung metastasectomy will never be clarified. Metastases develop in 50-60% of patients with high grade soft tissue sarcomas and without treatment are almost invariably fatal [1]. The lung predominates as the site of first metastasis. In 1964 Clagett demonstrated a 23% 5 year survival following resection of metastatic sarcoma [2]. Results similar to those seen in osteosarcoma have been reported by others [3-5]. Survival following pulmonary metastasectomy has been correlated in some series [4, 5] with the length of the disease free interval following treatment of the primary tumour. Patients with a disease free interval of less than 12 months appear to do particularly badly. Other workers have associated prognosis with tumour doubling time [7], number of metastases and resectability [6-8]. However, the results in the literature are not consistent [8] which makes it difficult to detail an appropriate treatment policy for patients developing lung metastases. The patterns of metastasis in a group of patients referred to the Royal Marsden Hospital Sarcoma Unit Received 24 May 1993 and in revised form 12 July 1993, accepted 26 July 1993. Reprint requests: Dr M H Robinson, YCRC Senior Lecturer Clinical Oncology, Weston Park Hospital, Whitham Road, Sheffield S10 2SJ. with soft tissue sarcoma were documented and a detailed study of those undergoing subsequent pulmonary metastasectomy was performed. To determine the value of this procedure, the survival of this selected group was compared with the survival of a cohort of patients not offered metastasectomy. Patients and Methods All patients 293 patients referred to the Royal Marsden between 1970 and 1990 with a diagnosis of soft tissue sarcoma have developed metastatic disease (other than lymph nodes). Of these, 189 had lung metastases as the first and only evidence of metastatic spread. Patients with lung metastases Lung metastases were usually diagnosed radiogically by chest radiography, linear tomography or computed tomography (CT). Since 1985, CT of the lungs has been carried out on referral. Clinical follow-up after treatment of the primary tumour was carried out at 2 monthly intervals for the first year, 3 monthly for the second, 6 monthly until the fifth, and yearly thereafter. Routine chest radiographs and chest CT were carried out for the first 5 years, with a plainfilmat each visit and chest CT 3 monthly for the first year, 4 monthly for the second, 6 monthly for the third and yearly thereafter. Vol. 67, No. 794 129
M H Robinson, M Sheppard, E Moskovic and C Fisher Table I. Factors assessed by multivariate analysis for independent prognostic value for survival following development of lung metastasis Table II. Number of lung metastases found at resection Number of metastases Number of patients Age: 40-50, 50-60, and > 60 years Site: lower limb, upper limb, girdle, truncal, head and neck, retroperitoneal Histology Grade Size: < 5 cm, 5-10 cm, > 10 cm Depth: superficial vs deep History: < 1 year vs > 1 year Lung metastasis resection performed vs not Time to metastasis from diagnosis 1 2 3 4 5 6 14 15 Unknown 22 6 The histopathology of the resected metastases, medical records, chest radiographs and CT scans of 44 patients subjected to pulmonary metastasectomy (since 1980) have been reviewed. Each patient had histologically confirmed pulmonary metastases. 28 men and 16 women (median age 32 years) have been studied. The majority of primaries in this group were in the extremity or girdle (35). Malignant fibrous histiocytoma (10), synovial sarcoma (8) and sarcoma of unspecified histology (6) were the commonest histological types. Tumours were categorized as high, intermediate or low grade. Most were high grade (37) with seven low grade and no intermediate grade tumours. The mean length of history prior to diagnosis of the primary was similar in those 44 patients undergoing metastasectomy compared with those 145 that did not (316 vs 322 days). Treatment policy for the management of lung metastases was not consistent prior to 1987. However, patients with solitary lung lesions or a long disease free interval with no evidence of recurrence at other sites were considered suitable for metastasectomy. In 1987 this treatment policy was formalized in that surgery was offered to patients with four or fewer lesions and a disease free interval of over 1 year. After confirmation that surgery was technically possible, these patients were first given two courses of chemotherapy with either adriamycin or ifosfamide to evaluate response. Surgery was then performed. Four further courses were given postoperatively to those in whom a partial response to chemotherapy was seen. Only 12 patients were entered into this protocol. The tumour doubling time of resected lung metastases was calculated for those patients for whom radiology was available for review. Actuarial curves for time to metastasis, survival and freedom from disease were determined by the method of Kaplan and Meier and compared using the logrank test [9, 10]. A Cox stepwise regression model using the Biomedical Statistics package (BMDP) was used to assess the individual impact of all potential prognostic factors on survival following diagnosis of lung metastasis [11]. Table I details the factors entered into the multivariate analysis. Results Time to metastasis Median time to lung metastasis from diagnosis was greater in those undergoing metastasectomy than those who did not (674 vs 391 days; p = 0.017). In those undergoing surgery 10 patients had recurred locally prior to developing lung metastases, three recurred at the same time and four afterwards. In only one case was there another site of active disease at the time of thoracotomy. This was a synchronous metastasis in the liver which was also treated surgically. The mean number of lung metastases detected radiologically prior to resection was two. 11 patients had metastases in both lungs. Pulmonary surgery Complete resection of known pulmonary lesions was achieved at thoracotomy in all but five patients. In four cases this was because of pleural disease. The majority of patients had a wedge resection of the metastases. Bilateral thoracotomy was performed via a median sternotomy in seven cases and by a two stage operation in two cases. The surgeon was unable to palpate the lesion in the second lung in one case (successfully treated with radiotherapy), and another patient died before the second operation. Pneumonectomy was required in two patients. The mean number of metastases found at surgery was two (range 1-15) (Table II), with 22 patients having only one. However, on nine occasions more lesions were present than were discovered pre-operatively and on three occasions fewer. Thefindingof more metastases at operation than had been diagnosed radiologically preoperatively occurred in only one patient with a solitary metastasis radiologically compared with 6/11 with bilateral disease. The size of the 126 lesions resected ranged from 2 to 85 mm. Six patients had documented surgical complications. Most were minor apart from one pulmonary embolus and one post-operative death related to tumour embolus into the middle cerebral artery. Tumour doubling time The doubling time of pulmonary metastases prior to resection was calculated from chest radiography or CT 130 The British Journal of Radiology, February 1994
Lung metastasectomy in sarcomas.900-.800-.700-.600-.500-.400-.300-.100-0- LUNG RESECTION NO RESECTION 0 365 730 1095 1460 1825 2190 2555 2920 3285 3650 EOLLOW UP (days) Figure 1. Survival from diagnosis of primary tumour for patients developing lung metastasis as first sole site of metastasis according to whether they underwent metastasectomy. in 15 patients. This ranged from 35 to 2200 days (mean 350, median 225). Radiotherapy and chemotherapy 14 patients undergoing metastasectomy had received adjuvant chemotherapy; eight before thoracotomy, one after and five before and after the surgery. The majority of patients received single agents; six patients were given adriamycin and six ifosfamide. Other drugs used included methotrexate (two patients), cyclophosphamide (three), vincristine (three), carboplatin (one), VP16 (one) and vindesine (one). One patient received radiotherapy to the site of unresected disease in her other lung immediately after surgery. She was treated with a dose of 19.8 Gy given in 6.6 Gy fractions once a week over 3 weeks. 48 out of 145 patients not undergoing metastasectomy received palliative chemotherapy. Survival Survival of all patients. Median follow-up for survival from diagnosis of the primary lesion is 1433 days and, from the diagnosis of lung metastases, 297 days. Figures 1 and 2 compare the survival from diagnosis and the survival from the time of development of lung metastases, respectively, of the 44 patients in this series undergoing metastasectomy with that of 145 patients who did not. The survival of those selected for thoracotomy was better than that of those not undergoing surgery (p < 0.0001). A multivariate analysis was performed to assess the factors determining outcome following the development of lung metastasis for the 189 patients in this series. The factors entered into this analysis included site, histology, grade and time from diagnosis to the development of metastasis (Table I). The result is given in Table III. Only the use of lung metastasectomy and superficial primary tumour depth were significant independent factors for survival, both associated with a reduced risk of death. The metastasis free interval was not. Survival of those undergoing lung resection. Median survival following thoracotomy was not related to the length of the disease free period following treatment of the primary (Figure 3). Survival following resection was also not significantly Vol. 67, No. 794 131
M H Robinson, M Sheppard, E Moskovic and C Fisher I I L.100-0- LUNG RESECTION NO LUNG RESECTION 365 730 1095 1460 TIME FROM LUNG METASTASIS Figure 2. Survival from development of lung metastasis as first sole site of metastasis according to whether patients underwent metastasectomy. 1825 related to the number of metastases resected (Figure 4; p = 0.49) or to the use of adjuvant chemotherapy (p = 0.33). Patients with bilateral metastases and those with incomplete resection did not appear to do worse. Recurrence in lung 28 patients undergoing thoracotomy have subsequently developed further recurrent or metastatic tumour. 21 patients have developed further lung metastases; 18 in the lung alone, one in combination with local recurrence, one with liver and bone disease, and one who initially relapsed locally has subsequently also relapsed in the lung. Eight patients have received no further treatment following disease relapse. Three patients have undergone a second thoracotomy. Of these two have died, one of whom had a further relapse in the lung. Five patients have had further surgery for logical relapse. Four patients have received chemotherapy followed by radiotherapy for lung disease, two chemotherapy alone and two radiotherapy alone. One patient who developed a solitary brain metastasis has undergone local resection and whole brain radiotherapy. The use of adjuvant chemotherapy was not associated with a reduction in further lung relapse on univariate analysis (p = 0.87). Neither the presence of bilateral metastases nor incomplete surgical resection (0.89) significantly increased the likelihood of further relapse in the lung. There were too few patients in the metastasectomy Table III. Significant prognostic factors for survival following the development of lung metastasis Factor Hazard for Relative to Hazard ratio P Metastasectomy Depth primary Operation Superficial None Deep 0.2(0.11-0.37) 0.28 (0.09-0.8) 0.0000 0.009 132 The British Journal of Radiology, February 1994
Lung metastasectomy in sarcomas.900-.800-.700.600.500.400.300.100-0 <= 1 YEAR OVER 1 YEAR <= 2 YEARS OVER 2 YEARS 365 730 1095 1460 1825 0 365 730 1095 1460 TIME FROM SURGERY TIME FROM SURGERY Figure 3. Survival according to time to metastasis; < 1 year vs > 1 year and < 2 years vs > 2 years not significant. 1825 group for a reliable Cox model to be calculated for either survival or lung recurrence. Discussion It has long been appreciated that many patients who die of lung metastases are otherwise disease free [12, 13]. The results in the literature suggest that the 5 year survival following lung metastasectomy for soft tissue sarcoma is better than that for resection of primary bronchogenic carcinoma. However, it is not clear which patients should be selected for this operation. Some authors believe that only patients with solitary lesions are suitable [14-16]. There is conflicting evidence for the importance of disease free interval in determining prognosis. Some authors have reported that disease free interval following primary treatment is an important prognostic factor and others that it is not [6, 17]. In this study no influence of disease free interval on survival was seen. A review of the practice at the Memorial Sloan Kettering Hospital suggested that results were best where complete resection of the lesions was achieved and were not influenced by either the number of metastases or the disease free interval [8]. A similar finding in osteosarcoma has been reported [18]. Pastorino reported 28 cases of soft tissue sarcoma lung metastases completely resected between 1970 and 1987 [19]. Their median interval to lung metastasis was 2 years compared with 13 months in this series, suggesting that there was some selection of cases considered for resection. Neither the number of metastases nor the disease free interval affected survival following resection in this group. Disease free survival following resection was 42% at 5 years. Another study reported the results of reoperation following second lung relapse; and suggested that the interval between first thoracotomy and relapse was the most important prognostic factor [20]. In general the results from most series are similar with 20-40% 5 year survival following this operation with a low operative mortality. However, the series are usually small and this treatment has not been offered as part of a randomized trial. The results seen in our study are similar to those seen elsewhere in the literature {e.g. see ref. 21). The problem remains as to how to improve Vol. 67, No. 794 133
M H Robinson, M Sheppard, E Moskovic and C Fisher 1 - J I i.800.500.900-.700-.600-.400-.300-.100 <=2 LUNG METASTASES >2 LUNG METASTASES 365 730 1095 DAYS FROM SURGERY Figure 4. Survival following lung metastasectomy according to number of lesions resected. 460 further the outlook for these patients. There is little evidence in the literature that the use of adjuvant chemotherapy with current drugs is of value in increasing disease free survival in these patients. The evidence from trials of its use as part of the primary treatment have indicated that it is of no value in prolonging survival [22]. Unfortunately, despite careful patient selection, 80% of patients undergoing metastasectomy will die from their disease, predominantly with lung recurrence. It is clear that an effective adjuvant therapy is required to improve these results. This is as yet not available. It is likely that the routine use of CT results in earlier diagnosis of lung metastases. However, CT may remain equivocal for a period during which a watch policy is instituted to confirm the diagnosis. This may mean that any advantage of earlier diagnosis is lost. It is clear from the literature that resection of pulmonary metastases is worthwhile for a selected group of patients with metastatic soft tissue sarcoma [2-4, 8]. A longer initial disease free interval and fewer lung metastases present at resection may increase the likelihood that the patient will remain disease free. However, the true value of this procedure will only become clear if patients are entered into multicentre randomized trials designed to answer this question. An alternative would be to document prospectively the outcome for all patients with soft tissue sarcoma developing lung metastases as the sole site of recurrence. Acknowledgments Dr M H Robinson is a Senior Lecturer supported by the Yorkshire Cancer Research Campaign. References 1. POTTER, D A, KINSELLA, T, GLATSTEIN, E ET AL, High grade soft tissue sarcomas of the extremities, Cancer, 58, 190-205 (1986). 2. CLAGETT, O T, ALLEN, T H, PAYNE, W P and WOOLNER, L B, The surgical treatment of pulmonary neoplasms: a 10-year experience. /. Thorac. Cardiovasc. Surg. 48, 391-400 (1964). 3. FELDMAN, P S and KYRIAKOS, M, Pulmonary resection for metastatic sarcoma, J. Thorac. Cardiovasc. Surg. 64, 784-799 (1972). 4. PUTNAM, J B, ROTH, J A, WESLEY, M N ET AL, Analysis of prognostic factors in patients undergoing resection of pulmonary metastases from soft tissue sarcomas, J. Cardiovasc. Surg. 87, 260-268 (1984). 5. VENN, G E, SARIN, S and GOLDSTRAW, P, Survival following pulmonary metastasectomy, Eur. J. Cardiothorac. Surg. 3, 105-110 (1989). 134 The British Journal of Radiology, February 1994
Lung metastasectomy in sarcomas 6. JABLONS, D, STEINBERG, S M, ROTH, J ET AL, Metastasectomy for soft tissue sarcoma, J. Thorac. Cardiovasc. Surg. 97, 675-705 (1989). 7. JOSEPH, W L, MORTON, D L and ADKINS, P C, Prognostic significance of tumour doubling time in evaluating operability of pulmonary metastatic disease, /. Thorac. Cardiovasc. Surg. 61, 23-32 (1971). 8. MARTINI, N and McCORMACK, P M, Pulmonary resection in sarcoma metastases, Dev. Oncol. 55, 197-200 (1988). 9. KAPLAN, E S and MEIER, P, Non-parametric estimation from incomplete observation, J. Am. Statist. Assoc. 53, 457-480 (1958). 10. PETO, R, PIKE, M C, ARMITAGE, P ET AL, Design and analysis of randomised clinical trials requiring prolonged observation of each patient, 11 analyses and examples. 19 The log rank test, Br. J. Cancer, 35, 1-39 (1971). 11. COX, D R, Regression models and life-tables, /. R. Statist. Soc. 34, 187-220 (1972). 12. FARRELL, J T, Pulmonary metastases: a pathologic, clinical, roentgenologic study based on 78 cases seen at necropsy, Radiology, 24, 444 (1935). 13. WILLIS, R A, The Spread of Tumours in the Human Body (Butterworth, London) (1952). 14. ALEXANDER, J and HAIGHT, C, Pulmonary resection for solitary metastatic sarcomas and carcinomas, Surg. Gynaecol. Obstet. 85, 129 (1947). 15. EHRENHAFT, J L, Pulmonary resections for metastatic lesions, Arch. Surg. 63, 326 (1951). 16. GLIEDMAN, M L, HOROWITZ, S and LEWIS, F J, Lung resection for metastatic cancer: 29 cases from the university of Minnesota and a collected review of 264 cases, Surgery, 42, 521 (1957). 17. MOUNTAIN, C F, McMURTREY, M J and HERMES, K E, Surgery for pulmonary metastasis: a 20 year experience, Ann. Thorac. Surg. 38, 323-330 (1984). 18. GOORIN, A M, DELOREY, M J, LACK, E E ET AL, Prognostic significance of complete surgical resection of pulmonary metastases in patients with osteogenic sarcoma: analysis of 32 patients, J. Clin. Oncol. 2, 425^31 (1984). 19. PASTORINO, U, VALENTE, M, GASPARINI, M ET AL, Lung resection for metastatic sarcomas: total survival from primary treatment, J. Surg. Oncol. 404, 275-280 (1989). 20. POGREBNIAK, H W, ROTH, J A, STEINBERG, SMet al, Reoperative pulmonary resection in patients with metastatic soft tissue sarcoma, Ann. Thorac. Surg. 52, 197-203 (1991). 21. FLY, M W, WOLTERING, G and ROSENBERG, S A, Aggressive pulmonary resection for metastatic osteogenic and soft tissue sarcomas, Ann. Thorac. Surg. 37, 123-127 (1987). 22. BRAMWELL, V H C, ROUESSE, J, SANTORO, A, BUESA, J ET AL, European experience of adjuvant chemotherapy for soft tissue sarcomas. Preliminary report of randomised trial of cyclophosphamide, vincristine, doxorubicin and dacarbazine. Cancer Treat. Symp. 3, 99-107 (1985). Vol. 67, No. 794 135