Abstracts of Papers Presented at the Connective Tissue Oncology Society 4th Annual Scientific Meeting, 12–14 November 1998, Vancouver, BC, Canada

s of Papers Presented at the Connective Tissue Oncology Society 4th Annual Scienti® c Meeting, 12± 14 November 1998, Vancouver, BC, Canada Scienti® c Program Committee Members: Brian O’ Sullivan, Canada (Chairman) Colin Cooper, UK (Basic Science Program Chairman) Alberto Azzarelli, Italy Nicola Baldini, Italy

Incidence lymph node metastases 2.1% , distant metastases 2.3% . Treatment: 318 pts curative intent (81% ), 41 pts palliative (11% ), 32 pts no treatment (8%); older pts received more often palliative or no treatment at all. C onclusio n: Incidence of STS is signi® cantly increasing with age, whereas, in the m eantime, older pts more often rece ive no treatm en t. It is there fore expected that a substantial proportion of pts with STS cannot be treated in scienti® c studies, since m ost protocols exclude pts above 65 years. Slow accrual for such studies is therefore expected. Most patients are treated with curative intent (81% ). Initial lymph node or distant m etastases are rare (2%). EORTC STB SG, EORTC Data Centre, 1200 B russels, B elgium Doxorubicin (dox) still appears to be one of the most active drugs in the treatment of soft tissue sarcomas. However, treatment duration is limited due to cumulative cardiotoxicity. It is therefore important to test anthracycline analogs. At present only few analogs have been evaluated in studies with adequate number of patients. In two studies of the EORTC STBSG we have tested whether epirubicin (epi) is an alternative to standard dose dox in the treatment of chemonaive patients with advanced soft tissue sarcoma. The present report gives the ® nal results of these studies. In the ® rst study 210 patients were randomized to receive either dox or epi both at a dose of 75 m gm ± 2 given as bolus injection at 3-week intervals. Median age was 54 years (18± 80 years) and Karnofsky perform ance score 90 (50± 100). No difference in median survival (41 weeks after dox vs 48 weeks after epi) and duration of response (45 weeks after dox vs 77 weeks after epi) was found. Of 167 evaluable patients 36 (22%) had an objective tumor response (10 CR, 26 PR), and the response rate was slightly in favour of dox (25% vs 18% ) but at the expense of m ore toxicity (leucopenia, alopecia, nausea/ vomiting). These data indicate that epi may be less toxic than dox when administered in equimolar doses, and could suggest that increasing the epi dose m ay lead to a greater antineoplastic effect with acceptable toxicity. M oreover, cardiac toxicity may be related to the peak concentration of the drug, and lower toxicity could therefore be expected with fractionated as compared with bolus injections. In view of this, we initiated our second study comparing singleagent standard dose dox with that of two schedules of high dose epi. A total of 334 patients received dox 75 m gm ± 2 day 1, epi 150 mgm ± 2 day 1 or epi 50 mgm ± 2 day ± 1 , days 1, 2 and 3, all given as bolus injection at 3 week intervals. M edian age was 52 years (19± 70 years) and WHO performance score 1 (0± 2). Of 314 evaluable patients 45 (14% ) had an objective tumor response (8 CR, 35 PR). There were no differences between the three groups. Median time to progression for the groups dox, epi-150 and epi-50 was 16, 14 and 12 weeks and m edian survival 45, 47 and 45 weeks, respectively. Neither progression-free nor overall survival differed between the three groups. After the 1st cycle of therapy two patients died of infection and one due to cardiovascular disease, all on epi. Both dose schedules of epi were more myelotoxic than dox. Cardiotoxicity ( grade 3) occurred in £ 2%. Based on these studies we can conclude, that epi regardless of schedule and dose is not an alternative to dox in the treatment of patients with advanced soft tissue sarcomas. In addition, the results illustrate that the data from small studies of single institutions should always be confirmed by large multi-institutional studies before being taken for granted.

A . M . Cassoni, J. S.W helan
The M iddlesex Hospital, UCLH, London,W 1N 8A A, UK Fifty-two patients with EW ING S/PNE T were treated with hyperfractionated radiotherapy with planned gaps, concurrent with intensive chemotherapy.Thirty were treated within the EICESS study, eleven in the preceding E2 protocol. All patients requiring radical radiotherapy as part of their initial m anagement were included. Results: acute skin toxicity ; grade 3 extrem ities, 5 pts (19% ) but in all with perineum in ® eld (14 pts). recall reaction; 2 pts. treatm ent deliver y; radiotherapyÐ delay of 5± 7 days in 3 pts (16% ); chem otherapyÐ om ission of anthracycline, planned 5 pts, due to acute toxicity 3 pts, total 15% . Conclusion: the majority of patients can be treated with concurrent chem oradiation without m odi® cation of the chem otherapy regimen. Patients with perineum in the field however may need modi® cation of overall radiotherapy time or concurrent drugs. Late toxicity in spinal cord remains a concern.

Th. van Dalen, A . N. van Geel, H . J. H oekstra , F. van Coevorden, A. H . H ennipm an
Dutch Soft Tissue Sarcom a Group,Ser vaasbolwerk 14,3512 N K Utretcht,The N etherlands B ackground: Liposarcomatous tumours in the retroperitoneal space are rare.The anatomic localization permits expansion of these neoplasms with symptoms developing only at a late stage of the disease. The resultant giant retroperitoneal m asses comm only lead to a defeatist attitude among the treating surgeons. The case histories of patients with ª giantº liposarcomas in the retroperitoneal space were studied.
M eth ods: In a retrospective nation-wide study data were collected on 45 patients treated in the Netherlands between 1988 and 1993 for primary giant retroperitoneal liposarcom as. Patients with tum ours weighing m ore than 1 kilogram or measuring m ore than 10 cm in diameter were included. Median age was 53 years; there were 22 men (49% ).
C onclusion s: Despite the large dimensions of ª giantº retroperitoneal liposarcomas radical resection is feasible in m ost patients resulting in reasonable long-term survival. Aggressive surgical therapy is justi® ed in patients with prim ary retroperitoneal liposarcomas and patients with isolated locoregional relapse. Introduction : Gastric stromal sarcomas (GIST) represent only about 1± 3% of all gastric malignancies. Patients generally lack speci® c symptoms. Despite advances in radiological diagnostics and endoscopy including endosonography, preoperative diagnosis of these tumors is rare. Because of the rarity of these tumors there is a lack in understanding their natural history and the best approach to their treatment. Therefore, we analyzed retrospectively our series of patients with this malignancy. M eth ods: Twenty patients with GIST underwent surgery at the University-Hospital of Hamburg-Eppendorf from 1979 to 1995. Resection quality was classi® ed according to the UICC . In reviewing the histopathological slides tumor grade (G1± 3) was determined according to the grading system for soft tissue sarcomas of Enzinger and Weiss, while classi® cation of tumors were perfor med according to Fletcher and Franquem ont.
R esults : Symptoms were unspeci® c. Exact preoperative diagnosis was difficult because of subm u-cosal tum or growth and was correctly perform ed in one patient. Operations varied from excision of the gastric wall to extended gastrectomy (pancreas, spleen, partial liver resection). In 16 cases (80 %), tum or could be resected with wide m argins (R0). In 4 cases, tum or was found at the resection line (R 1). One patient died because of postoperative com plications. After a m edian follow-up of 69 m onths, 12 patients with R 0 resections (including tum ors with poor differen tiation or in® ltration of surrounding organs) lived tum or-free, and two died due to other causes. All four patients with R 1 resections died because of tum or disease within 40 m onths. There was a m ean survival time of 59 months and a 5-year-survival rate of 69 %.
D iscussion: Our series shows that not all gastric malignancies have a dismal prognosis, but that gastric stromal sarcomas have a favourable follow-up if resected with wide margins at initial surger y.

T. F. Langwieler, C. Zor nig, M . Peiper
Dept. of Surger y, University Hospital Hamburg-E ppendorf, 20246 Hamburg , Germany Introduction : Soft tissue sarcoma (STS) of the colon and rectum are rare mesenchymal neoplasm. Consequently, analysis of small series of patients concer ning tum or characteristics as well as their treatment and how these fa ctors in¯uen ce recu rrence and patien ts sur vival is particularly interesting.
M eth ods: Four consecutive patients with STS of colon and rectum operated on at our institution between 1987 and 1994 were included. In this retrospective study we analysed patient data, surgical therapy, tumor characteristics as well as follow-up.
Results : There were 4 patients with a m ean age of 49 (range 41± 58) years, 100% m ale. The STS were located in the sigm oid (n=1), rectum (n=2) and colon descendens (n=1 ). All tum ors were resected with wide m argins (R0). Two tumors were highly differen tiated (G 1), 1 m oderately (G 2) and 1 tum or was grade G3. Postoperative radiation therapy an d chem otherapy were adm inistered in the patient with the poorly differen tiated tum or. 2 patients developed tum or recurrence: in 1 patient 16 m onths after com plete resection peritoneal m etastasis without local recurrence occurred while the second patient developed liver m etastasis 9 m onths after com plete tum or resection. Two patients died after 18 and 29 m onths, respectively, d u e t o m et as t at ic t u m o r d i se a s e. T h e o t h er tw o patients are without evidence of disease after 45 and 61 m onths.
C onclus ion: The prognosis of sarcomas of colon and rectum is poor. Complete surgical excision is the optimal therapy. Never theless, adjuvant therapy might be favourable and we therefore consider this treatment in the future even for patients with tumor resection. Objective: To analyse prognostic factors in a large multicentric series of retroperitoneal sarcoma.
M ethods: A series of 165 patients (pts) with primary retroperiton eal sarco m a (retroS TS) reg istered in the FNCLC C sarcoma group data was analyzed. Pts referred for tumors recurrences were not considered. M edian age was 54 years (16± 82) and the sex-ratio was 0.9. The m edian follow-up was 47 months. In only 6% of cases, the tumor was <5 cm ; 31 % showed eviden ce of vasculoner vous involvement; 5% were N+ and 12% were M + at referral. MFH, liposarcoma and leiomyosarcoma represented 66% of histological types; tumor grade was: G1 17% , G2 40% , G3 43% . 150 pts had surgery, with complete tumor rem oval in 65% . Radiotherapy was given to 92 pts and chemotherapy to 77 pts. At the end of ® rst treatment, 118 pts (72% ) were considered as free of disease. Univariate and m ultivariate analysis of prognostic factors for Complete Remission (CR), Overall Survival (OS), Local Recurrence Free Interval (LRFI), Metastatic Free Interval (M FI) and Disease Free Interval (DFI) were done.
Results: 5-years OS was 45.7% and 5-year DFI was 28% , with a 5-year LRFI of 31% and M FI of 66% . A diseasefree status at completion of treatment was correlated with m etastases at entry, tum or grade and vasculonervous involvement. At multivariate analysis of the entire group, OS was independently correlated with grade (p=.0007) initial metastatic status (p=.005) and vasculonervous or bone involvement (p=.04). For 145 pts M 0 at entry, DFI was correlated with tumor grade (p=.0016 ), complete surgical excision (p=.0001) and radiotherapy (p=<.0001); MFI was correlated with tumor grade (p=.0074) and a non liposarcoma histology (p=.03); LRFI was correlated with adju vant radiotherapy (p<0.0 00 1), com plete excision (p=0.0001) and tumor grade (p=0.005). C onclusio n: If surgery and a complete tumor rem oval rem ains the cornerstone of the treatment in retroSTS, radiotherapy may contribute to local control. M oreover, although local control rem ains essential, a high risk of metastatic disease was apparent.

Riccardo Fer racini, M ar ina M arta no, N icola Baldini, Katia Scotlandi, Jay Wunder, Eric M asterson
Institute for Cancer Research, St. Prov. 142, Km 3.95, 10060 C andiolo, Italy Overexpression of the hepatocyte growth factor receptor (M et/HGF receptor), a transmem brane tyrosine kinase encoded by the M et proto-oncogene, has been involved in transformation and invasive behaviour of human carcinomas and sarcomas.We have previously found that bone sarcomas express high levels of HGF receptor while in some cases the ligand was co-expressed with the receptor, activating an autocrine loop. In this study we analyzed 35 bioptic samples of benign bone tumors for the expression of the M et protooncogene. The lesions included osteoblastomas, chondroblastomas, non ossifying ® bromas, giant cell tumors and desmoplastic ® brom as of bone. The snap frozen samples were tested by imm unohistochemistry and Western blotting with anti-M et antibodies or with RT-PCR for M et and HGF m RNA. About 50% of all cases scored positive for M et expression. Although non statistically relevant, we report a trend of M et positivity in samples from recurrent or locally aggressive lesions. Sporadic co-expression of the M et receptor and ligand (HGF) was demonstrated. The possible role of the M et receptor in the pathogenesis of benign bone neoplasms suggests an early involvement of the oncogene in sarcoma transformation.This is in contrast to the biology of carcinom as where the M et oncogene is ampli® ed as a late event in the neoplastic progression.

Istituti Ortopedici Rizzoli, 40136 B ologna , Italy
This study includes a continuous series of 103 patients with extra-abdominal desmoid tumor who were treated at the same institution between 1970 and 199 6. Twenty patients were excluded from the analysis because of an insufficient follow-up. Of the remaining 83 patients, 35 had a newly diagnosed lesion, whereas 48 presented with a local recurrence after being surgically treated elsewhere. Patients were treated only with surgery (n=63) or with surgery plus radiation therapy (n=17) at doses ranging 3500 to 6600 cGy. Three patients did not undergo surgery, and were excluded from the analysis. No patient died of the disease, but 37 (46% ) developed local recurrences. M ore than three recurrences in 8 patients (10% ). Overall, relapses were not associated with age, sex, site, or tumor size, whereas a trend toward a higher risk of recurrence was observed when surgical margins were found to be inadequate on pathologic examination. The relapse rate was similar in patients treated only with surgery (28 cases, 45% ) and patients treated with surgery plus radiation therapy (7 cases, 41% ). Am ong the 35 patients who were ® rst treated at our institution, 13 (37% ) developed a recurrence. Again, no differen ce in the relapse rate was observed among patients undergoing surgery plus radiation therapy (2/5 cases, 40% ) and patients treated only with surgery (11/30, 37%), whereas inadequate (intralesional or marginal) surgical margins were associated with a higher incidence of relapse (7/15, 47% ) compared to patients treated with a wide surgical excision (7/20, 35% ). These data con® rm that the best chances to prevent the risk of local recurrence in extra-abdominal desmoid tumor are offered by a careful preoperative staging in order to obtain adequate surgical margins. Proper evaluation of the ® nal specimen is m andatory to con® rm the adequacy of surgery, although postoperative radiation therapy does not appear to be able to prevent the risk of local recurrence. Clinical observation should be considered for slowly growing recurrences.

R . Keus, F. van C oevorden
The N etherlands Cancer Institute, Antoni van Leeuwenhoek H uis, Plesmanlaan 121, 1066 CX Amsterdam, The N etherlands B ackground and Objectives: Aggressive ® bromatoses (desmoid tumors) are rare soft tissue neoplasms with a typical clinical behavior of frequent local recurrences and absence of distant spread. Other characteristics of this disease are an age peak between 30 and 40 years, fem ale predominance often with pregnancy associated abdominal wall lesions. Mesenteric lesions are frequently seen in patients with an APC gene m utation as in familial polyposis coli. Local treatment m odalities such as surger y and radiotherapy form the mainstay of the m anagement of these tumors. In patients where local treatments are not feasible or failed, systemic therapy by horm onal m anipu lation or chem o therapy regim ens is employed. The objective of this study was to determine the efficacy of local treatment m odalities and to estimate the need for systemic treatment in our center.
M ethod s: A retrospective review of 105 patients (72 fem ale, 33 m ale) in the period 1973 to 1997. M ean age was 33.1 years. Tumors were located in the head and neck in 11 patients, the extremeties in 50, abdominal wall in 24, mesenteric in 6 and other trunk in 12. Treatm ent consisted of surger y only in 21 patients, combined therapy with surgery and radiotherapy in 62, radiotherapy only in 6, isolated limb perfusion with Adriamycine and M elphalan in 4, chemotherapy in 3. Observation after surgery elsewhere was performed in 9 patients.
R esults: Overall there was 86% local control at 5 years. There was no difference in local control for primary vs. recurrent lesions. Local control by treatment m odality was 83% , 89% , 83% and 78% for surgery only, surgery and radiotherapy, radiotherapy only and other treatments, respectively. For the differen t tumor sites the were 100% , 81% , 88%, 67% and 100% in head and neck, extrem ities, abdominal wall, mesenteric and other trunk, respectively. No effect on local control of age, radiation ® eld size, radiation dose was seen. The in¯uence of surgical m argin on local control rate in the surgery only group was 92% vs. 50% and in the combined therapy group 100% vs. 85% for patients with negative vs. positive surgical margins. Severe complications were seen in 14 patients: radiation myelitis 1, nerve injury 4, vascular injury 2, bone fracture 2, intestinal ® stula 3, chronic infection 1, post-irradiation sarcoma 1. The outcome of chem otherapy in 3 primary and 3 recurrent lesions was in 1 C R, 2 PR, 1 NC and 2 PD. Tamoxifen was applied in 4 patients with recurrent lesions resulting in 1 CR, 1 NC and 2 PD. C onclusio ns: These results show that local treatment with surgery and/or radiotherapy is effective in 86% of patients with aggressive ® bromatosis. Only patients with extensive or m ultiple lesions not suitable for local treatment, or those who will be likely to suffer from unacceptable m orbidity from local treatment should be considered for future studies of more effective systemic treatment.
Chemotherapy resistance remains an important clinical problem in osteosarcoma. This prospective, m ulticenter study analysed the prognostic value of multidrug resistance gen e (M DR 1) expression in patients with extrem ity, non-metastatic, conventional, high grade osteosarcoma. Between 1989 and 1994, 123 patients received adriamycinbased multiagent neoadjuvant chemotherapy and locally curative surgery, and were followed for a m inimum of 24 months or until systemic recurrence. For each patient, a sample from the prim ary tumor was analysed for M DR1 expression level using a quantitative polymerase chain reaction-based assay.
M DR1 level was low in 44 tumors, intermediate in 39 tumors and high in 40. Of 123 patients, 46 (37% ) relapsed with metastases at a minimum 2-year follow-up. After un ivariate an alysis, only tum or size (p =0.0 007) and chem otherapy-induced tumor necrosis (p=0.031) were associated with a higher risk of metastases. There was no linear association between M DR1 expression level and outcome (p=0.62). Surprisingly, patients with Intermediate M DR1 levels had the best survival, compared to those with either Low or High M DR1 expression. This relationship remained unchanged even after adjusting for other variables. After multivariate analysis, tumor size was the only signi® cant predictor of outcome (p=0.004). As well, no relation ship existed b etween M D R 1 level a nd chemotherapy-induced tumor necrosis.
These results suggest that there is no simple linear relationship between M DR1 gene expression and systemic relapse in osteosarcoma. Patients whose tumors had High M DR1 expression did worse than those with Intermediate levels. On the other hand, Intermediate levels predicted for a better outcome than Low levels. If M DR1 expression is important in osteosarcoma, other m echanisms m ay play a more critical role. Studies using immunohistochem istry (IHC ) have suggested that elevated P-glycoprotein levels are associated with a poor outcome. These results are in contrast to our data which show that patients whose tumors had low M DR1 levels had a worse prognosis.To help clarify this, we are presently evaluating a subset of tumors in our study of P-glycoprotein status by IHC. M utation of the p53 gene might be another potential mechanism affecting outcome. M utant p53 has been suggested to cause tumor d rug resistan ce, an d m ay also stim u late the M D R 1 prom oter, while wild-type p53 m ay act as a repressor. The interaction between these two genes may be important, and is also being investigated.

C. J. Swallow, C. N. C atto n, B. O' Sullivan, J. Couture, R. A. Kandel
The Sarcom a Site Group, University of Toronto, C anada Introduction : The anatomical constraints of the retroperitoneal space make adequate resection and effective irradiation of RPS particularly challenging, and treatment outcome is far worse than for sarcomas at other sites.We have recently implem ented a new treatment protocol for RPS which involves aggressive en bloc total gross resection combined with escalated dose radiotherapy to the tumor bed using external beam radiotherapy (RT) and postoperative BT. Treatm ent outcome has been studied prospectively.
M eth ods: Over an 18 month period, 32 patients with primary or recurrent RPS were referred to our center. Of these, 18 were eligible for curative therapy with combined surger y and radiation. Sixteen patients completed all phases of therapy, while 2 who developed m etastases during RT were taken off protocol. Sixteen patients had a total gross resection. Fourteen were treated with RT and BT, and 2 patients treated for recurrence after prior RT had BT only. RT was given preoperatively (n=12) or postoperatively (n=2) with a parallel pair to a dose of 45 or 50 Gy in 25 fractions over 5 weeks. Planar pulsed dose rate BT was administered postoperatively at 0.5 G y/hr prescribed at 0.5 cm through temporary catheters implanted intraoperatively. The m edian total combined dose to the tumor bed was 70 (60± 77) G y. Outcome was assessed prospectively with toxicity scored according to the RTOG system.
M edian postoperative follow-up is 14 (4± 24) months. The highest acute RTOG score after RT was 2, in 6 patients. After BT, 5 patients required parenteral nutritional support for transient duodenitis or enteritis and had an acute score of 3. One patient developed respiratory complications requiring early termination of BT and transfer to the ICU, and 1 patient died of complications following aspiration pneumonia. In 13 evaluable patients, the late toxicity score is 0 in 10, and 3 in 2. At a median of 14 months, one patient has failed with out-of-® eld mesenteric seeding, while 12 patients remain disease-free.
Sum m ar y: This study demonstrates the feasibility of combining an aggressive resection with high doses of radiation to the tumor bed for primary or recurrent RPS. Resection of involved adjacent structures produced a resection rate of 100% . Tre atm ent-related m o r ta lity was 6% . Postoperative complications and acute radiation toxicity were m anageable in the m ajority of cases. Preoperative RT was very well tolerated, and the late radiation toxicity from combined RT and BT was signi® cant but acceptable. The efficacy of this approach will be tested in a phase II trial.

Pediatric Oncology B ranch, N ational C ancer Institute, National Institutes of H ealth, B ethesda, M D 20892, USA
Osteosarcoma (OSA) is the m ost common primary tumor of bone. Despite successful control of the primary tumor and adjuvant chemotherapy, relapse of OSA in the lungs occurs in 30% of patients within 5 years. A relevant animal model for OSA is needed to evaluate novel treatments and to identify key determ inants for lung m etastases. The objective of this work was to develop a model of OSA in an immunocompetant host, with prim ary tumor growth at orthotopic sites, spontaneous pulmonary metastases and varian ts d iffer ing in pu lm on ary m etastatic poten tial. Hypothesis generation in these studies was based on the prem ise that tumor growth and m etastasis are in¯uenced by the host, the microenvironment and the tumor.
M urine OSA cell lines (K-7 and K-12), derived from a spontaneous murine OSA (Schmidt et al), were delivered at ectopic and orthotopic sites to 4 wk female balb/c m ice. Sites for tumor cell line (or enzymatically digested tumor tissue) injection included subcutaneous, intramuscular (IM ), intratibial, and intrafem oral. Sites for tumor fragm ent implantation included proximal tibial bone¯aps or cranial tibial m uscle¯aps. Endpoints included primary tumor growth, bone invasion, radiographic change and incidence of lung metastases. The in¯uence of tumorbearing-limb amputation on lung m etastases was studied. L u n g m eta stases fro m K -7 tu m ors were su rg ically re-implanted to tibial bone¯aps. Repeating this lung to bone passage two tim es, yielded the K -7M 2 variant. Features of K-7M2 were evaluated as above.
From the experimental variables studied the optimal m odel system consisted of surgical implantation of tumor tissue at a tibial muscle¯ap or IM injection of enzymatically digested tum or followed by amputation of limbs bearing 1 cm tumors. using this m odel system, K-7 tumors developed in 12 days in 93% of m ice, growing to 1 cm in 19 days. K-12 tumors developed in 12 days in 87% of m ice, growing to 1 cm in 26 days. After am putation pulmonary m etastases occurred in 100% of K-7 mice compared to 58% of K-12 m ice. The median survival for K-7 mice was shorter than K-12 (61 days compared to 138 days; p<0.01. K-7M 2 had prim ary tumor growth similar to K-7, but earlier death from lung metastases (median survival 48 days).
This biologically relevant OSA m odel will offer a valuable tool for the evaluation of novel OSA treatment strategies and will improve our understanding of OSA metastases. This model is characterized by orthotopic prim ary tumor growth in an immunocompetant animal, spontaneous lung m etastases, and variants with a pulmonary m etastatic gradient from most aggressive to least aggressive (K-7M2, K-7, K-12). The osteosarcom a model developed will offer a valuable tool for the evaluation of novel treatment strateg ies for osteo sarco m a an d for the exam ination of determ inants of osteosarcom a m etastases. Studies are underway to identify genetic determinants that may account for differen ces in the m etastatic potential of the K-7, K-12, and K-7M 2 m odels. Results of this model characterization and the ongoing genetic studies will be presented.

The University of Texas M .D. Anderson Cancer Center, H ouston, Texas 77030, USA
We have previously shown that overexpression of the transcription factor E2F-1 results in apoptotic cell death in leiomyosarcoma cells lines in vitro. This cell death appears to be independ ent of the p53 status of the cells. We hypothesized that overexpression of E2F-1 in established tumors would result in growth inhibition in vivo.
M ethods: The leiomyosarcoma cell line SKLMS-1 was injected into the¯ank of Balb/c nu/nu mice at 5 3 10 6 cells per animal. Once tumors reached 5 mm size, tumors were injected with PBS (6 m ice), a recombinant adenovirus expressing the Luciferase reporter gene (Ad5Luc; 6 mice) or a recombinant adenovirus expressing E2F-1 (Ad5E2F; 7 mice). Ad5Luc and Ad5E2F were injected at a dose of 2 3 10 9 pfu every other day for a total of 13 injections.
Tumor growth was m easured in 3 dimensions and recorded as tumor volume (m m 3 ). Mice were sacri® ced when tumors reached 20 mm in size and assessed for apoptosis using in situ TUNEL assay. E2F expression was evaluated using immunohistochemistry.
R esults: We noted marked reduction in tumor growth in the A d5E2F treated tum ors over controls (P BS and Ad5Luc). Please see graph below.
Tum or growth in the Ad5E2F treated anim als was signi® cantly reduced as compared to Ad5Luc treated or PBS treated controls (P=<0.0 001) by ANOVA. There was complete tumor regression in two animals. In situ TUNEL dem onstrated apoptosis in Ad5E2F treated tum ors in 20± 30% of cells per high power ® eld examined. Overexpresssion of E2F was con® rm ed using immunohistochemistry and was predominantly nuclear.
C on clusio ns : A denovirus m ediated overexpression of E2F-1 leads to nuclear localization of E2F-1 with resultant apoptotic cell death in leiomyosarcoma. In vivo gene therapy utilizing E2F-1 results in signi® cant growth inhibition and may prove to be a useful strategy in the clinical management of leiomyosarcoma.

Istituto N azionale Tum ori,Via Venezian 1, 20133 M ilano, Italy
Pur pose: Retroperitoneal soft tissue sarcomas (RSTS) are generally large lesions, often recurrent and involving viscera, which are difficult to be treated properly with surgery. For postoperative radiation therapy (RT) to be effective, one should employ doses and ® elds which are unfeasible. In an effort to improve results, we decided to assess the feasibility of an approach including preoperative RT. In this way, RT ® elds can be tailored to the lesion, and the lesion itself works as a spacer able to protect viscera, while surgery could take advantage from a response to RT.
M ater ials and m eth ods: Since September ' 96 to December ' 97, 12 consecutive patients with extensive recurrent RSTS were treated with preoperative RT up to 50 Gy. Surgery was scheduled between the 30th and 60th day after the end of RT. The ® elds of RT were designed upon an isodose simulation CT scan, especially targeted to the portions of the lesion for which good m argins would have been difficult to be achieved, also considering whether colectomy or nefrectomy would have been performed or not. The end points were the overall feasibility of the treatment program, the intraop. and postop. complications, the pathologic response rate and the local control.
Results : Twelve pts. were treated: 8 liposa. (3 G1), 3 leiomiosa. (2 G1), 1 NOS sarcoma. One pt. received only 32 Gy for gastrointestinal intolerance. Radiological response was minimal in all pts. All the patients underwent surgery 32± 46 days later. In ten patients, the lesion was removed completely. Resections were extended to viscera, with 4 nefrectom ies and 6 colectom ies. No m ajor surgical difficulties were encountered, and the lesions always appeared well demarcated. No surgical contamination did occur. Postoperatively, 4 of 6 colectom ies were complicated by anastomotic ® stulas, and two required colostomy. One pt. died three months later due to abdominal complications. One pt. died due to disease recurrence. One pt. was reoperated for disease recurrence 10 mos later. Pathologic response was always consistent with post-RT alterations. Up to now (follow-up range 8± 20 mos), all the other seven patients are alive with no evidence of disease.
C onclusio n: Preoperative RT is feasible in RSTS. In the case a large bowel anastom osis is required, precautional colostomy or ileostomy should be considered.
Introduction : Doxorubicin (DX ) and high-dose Ifosfam ide (HDI) are two active drugs in the treatment of STS with a similar response rate. We are performing a phase II trial in ® rst line treatment with HDI-D X (scalating dose of HDI), in order to evaluate the activity and toxicity of this regime.
C onclusio ns: HDI-D X is an active regime in adult STS. The impact of haematological toxicity may limit the use of this schedule and it is too early to evaluate the cardiac and neurological toxicity.

Yorkshire Cancer Research Department of Clinical Oncolog y,Weston Park Hospital, Sheffield S10 2JS, UK
Soft tissue Ewing's sarcoma, peripheral neuroectoderm al tumours (PN ET) and Askin's tumours are small round blue cell tumours which characteristically have a reciprocal translocation between chromosome 11 and 22. As with other rare tumours, clinical trial data on the optimal management of these patients is limited. We review the results of 6 patients treated with conventional chemotherapy and compared these to 9 patients treated with standard induction regimens followed by high dose chem otherapy (HDC). 6 p atien ts (1 fe m ale, 5 m ale) received s ta nd a rd chemotherapy with IVAD (or similar regimen). M ean age at presentation was 23 years (range 16± 33 years). 2 had localised disease, 2 locally advanced and 2 metastatic disease. Additional radical radiotherapy was given to 5 patients and surger y was performed in the other. 4 patients have died at a median time of 284 days. Survival in the 2 rem aining patients is 1947 and 2006 days. 9 patients received IVAD (7 patients) or EVAIA (2 patients) induction chemotherapy. Mean age at presentation was 26 years (range 17± 39 years). 3 had locally advanced disease with the rem ainder having metastatic disease. Complete response (CR) was achieved in 2 patients and a good partial response (PR) was seen in all others. HDC in all patients was with Etoposide, C arboplatin and Melphalan and a further 3 CR were achieved. Additional radical radiotherapy was given to 6 patients. 6 patients have died after a median interval of 421 days. 3 patients are alive and in rem ission at 417, 666 and 884 days.
S oft tissu e Ew ing ' s sarcom a often resp on d s to conventional chem otherapy but tends to relapse, and the long term prognosis is poor. HDC may offer some survival advantage but m ulti-centre clinical trials of this and newer strategies will be vital in improving the outlook for these patients. That trial had a response rate (RR) of 6 of 13. Thus a phase II trial was initiated. The objective was to estimate the RR and de® ne toxicity of VP, IFOS, and G-CSF in newly diagnosed M OS.The study was closed September 15, 1997 having reached its accrual goal. A total of 43 patients were registered. Two patients were ineligible; 2 were not evaluable; and 39 were evaluable for response. Patients received infusions of 100 mg/M 2 /day of VP over 1 hour followed by 3.5 gr/M 2 /day of IFOS over 4 hours for 5 days total. G-CSF, was began on day 6.This was repeated in 3 weeks. Response was determ ined at week 6. Eligibility included: age £ 30 years, biopsy proven newly diagnosed, previously untreated M OS, and ECOG performance status £ 2. Twenty-eight of 41 (68% ) have m etastatic sites only in the lung and 13 (32% ) have synchronous MOS with m etastatic sites in other bones.Toxicity was evaluated in 38. The most serious toxicities are; grade 5 sepsis (death) in 1; grade 3 and 4 sepsis in 6; grade 3 and 4 infections (non-sepsis) in an additional 5. One patient died of congestive heart failure (CH F), and a second patient had grade 1 C HF. Grade 4 neutropenia occurred in 30 and Grade 4 thrombocytopenia in 18. Grade 3 Fanconi's Syndrome developed in 4 and one additional patient had grade 4 Fanconi's. Response information is available on 38 patients. Three had CR's; 20 PR's; 3 MR's; 10 NR; 2 PD. The RR (CR + PR) is 61% ± 8% (SE). We conclude that the combination of VP and high dose IFOS are very effective induction treatment for high risk M OS despite signi® cant associated myelosuppression sometimes complicated by infection, and renal toxicity requiring electrolyte replacement.

University of M innesota M edical School, M inneapolis, M N 55455, USA
There is interest in administering chemotherapy drugs in liposomes as a means of altering the therapeutic index of the drug. Liposomal doxorubicin is of particular interest since less cardiotoxicity has been observed, and there may be greater anti-tumor efficacy. Pegylated-liposomal doxorubicin (Doxil¿ ) is a unique form of liposomal doxorubicin in which the liposom es are coated with m ethoxypoly (ethylene glycol). The polymer confers useful properties including a diminished uptake by the reticuloendothelial system, leading to a much longer half-life in blood (~50± 60 hours), and a different toxicity pro® le than non-pegylated liposomes. We are perform ing a phase II study of Doxil¿ in refractory sarcoma. The patient population consists of patients that have failed doxorubicin, DTIC, ifosfamide, VP-16, and in some cases additional chemotherapy. The initial dose per course is 55 mg/m 2 every four weeks. Dose modi® cation based on m ucositis and hand-foot syndrom e (the m ain limiting toxicities) is performed following an algorithm that m odi® es both dose and interval between treatments. Treatm ent was generally well tolerated. Of the ® rst 29 evaluable patients, there were: 5 osteosarcomas, 7 gastrointestinal leiomyosarcomas, and 17 other soft tissue sarcomas. Eight had stable disease for 14, 7, 6, 3, 3, 3+, 2+, and 2+, m onths; 1 a mixed response, 3 a minor response for 3+, 4, an d 9 m o nths, an d 2 a par tial respon se (m aintained for 5, and 24+ months). These data suggest that pegylated-liposomal doxorubicin has signi® cant activity in refractory sarcoma.

J. Aronowitz, B. N evaldine, P. H ahn
State University of New York Health Science C enter, Syracuse, N Y 13204, USA Introduction : Little is known regarding the repair kinetics of radiation-induced DN A double strand breaks (DNAdsb), especially in normal tissue. W ith increasing interest in assaying normal tissue radiosensitivity, as well as in optim izin g tim e-d ose-fraction ation sch em es to m axim ize therapeutic ratio, we endeavored to develop techniques to quantify norm al ® broblast repair kinetics.
M aterials and m ethods: Non-transformed cultured ® broblasts from a normal human subject were studied. It was determined that term inal differen tiation occurred after 18 passages. Utilizing a system based on pulsed-® eld gel electrophoresis, we measured repair kinetics of cohorts of cells from the 7th and 15th passages following irradiation with 7.5 Gy.
R esults: Approximately 50% of DN Adsb were repaired within m inutes (fast repair), an additional 30% before the end of the ® rst hour, and m ost of the rem aining breaks were repaired during the second and third hours.The T ½ of slow repair during the ® rst hour was 40 minutes. There were no appreciable differen ces in the repair kinetics between ® broblasts from early (7th) and late (15 th) passages. C onclusio n: The repair kinetics of DNAdsb in human ® broblasts can be determined. In this culture of normal human ® broblasts, repair kinetics were not altered by the age in culture. University, N ew York, N Y [RN T, M LK, RLF], Lone Star Oncolog y, A ustin, TX [JC ], N ew York M edical College,Valhalla, N Y [ZD, HC, AM , CP], Thompson C ancer Sur vival Center, Knoxville, TN [TP, SM ]

, A lfacell Corporation, B loom ® eld, N J [KS, SM M ], USA
Systemic therapy for advanced, unresectable M M rem ains unsatisfactory. No single or combination drug regim en has been shown to prolong survival.We have conducted a clinical phase II trial of ONC (previously known as P-30 protein), a novel ribonuclease isolated from the eggs of the leopard frog (Rana pipiens), in patients with advanced unresectable M M . ONC had previously dem onstrated in vitro activity against human prim ary mesothelioma cell cultures. In this multicenter open label trial, single-agent ON C was given at a dose of 480 m g/m 2 I.V. weekly (as a 30-m in infusion) to 105 enrolled patients (85M :20F, perform ance status 0± 2) until disease progression or intolerable toxicity. 92 pts had prior surgery (none had complete resection), 17 prior radiotherapy, 39 prior chem otherapy. Involvement of the pleu ra/lung was n oted in 97 pts, including 17 with pleural+peritoneal disease; 8 pts had peritoneal disease only. 38 pts had regional lymph node involvement, and 33 had d istant m etastasis. Toxicities were reversible and manageable. 16 pts were terminated from study due to toxicity, 4 of which for grade 4 toxicities. As of 6/15/98, the estimated Kaplan-M eier m edian survival times (M STs) for the intent-to-treat group (105 pts) was 5.8 months (1 and 2 year survival, 34.3% and 20.0% ). Patients with confirm ed epithelioid histology (50/105 ) fared better (M ST 9.6 months, 1 and 2 year survival, 42% and 23.4% ) than pts with pure sarcomatoid histology (8/105 pts, M ST 1.9 months, 1 year survival 12.5% ). Age under 50 (17/105 pts, M ST 19.8 months), fem ale sex (20/105 pts, M ST 11.4 months) and performance status of 0 (29/105 pts, MST 18.4 m onths) were favorable criteria for survival. Our 1 and 2 year survival data compared favorably with those of analagous prognostic groups as de® ned by the Cancer and Leukemia G roup B (CALGB) among 337 pts treated by CALG B with different Phase II agents or regimens over a 10-year period. Our data indicate that ONC may be active in selected patients with MM . A randomized Phase III trial of ON C v. doxorubicin is now in progress.

Laborator y of Oncology Research, Istituti Ortopedici Rizzoli, 40136 B ologna , Italy
Innovative treatment modalities are needed for Ewing's sarcoma (ES), a neoplasm with a disappointingly low survival rate despite the use of aggressive m ultimodal therapeutic approaches.We have previously shown the existence and the pathogenetic relevance of an autocrine loop, mediated by the insulin-like growth factor-I receptor (IGF-IR), that is crucial for survival and proliferation of ES cells in vitro. In this study, we report that the IGF-IR-blocking monoclonal antibody a IR3 m ay also signi® cantly inhibit ES cell growth in vivo. In particular, in almost one-half of animals tested, blockage of IG F-IR by a IR3 induced a complete regression of tumors developing after s.c. inocula-tion of ES cells, suggesting the value of IGF-IR as a specific target for novel therapeutic strategies. Also suram in, a drug that is able to interfere with growth factor binding to their receptors, inhibited both the tumorigenic and the metastatic ability of ES cells, therefore offerin g as a prom ising agent to be combined with conventional cytotoxic drugs for the design of more effective therapeutic regimens. M oreover, both a IR3 and suramin treatment increased the antitumor in v itro effe cts of ad riam ycin a nd vin cristin e, two conventional chemotherapeutic drugs with a leader action on ES, producing additive growth suppression of ES cell lines. This result was obtained both by simultaneous and sequential treatments. Analysis of the proliferation rate and of apo pto sis revealed that a IR 3 an tib od y did n ot substantially affect the blockage of cell cycle in G2/M phase induced by adriamycin, whereas it signi® cantly enhanced th e in du ction o f ap opto sis. A syn erg istic an d do sedependent effect was indeed observed with regard to the percentage of apoptotic nuclei after treatment with a IR3 and adriamycin, indicating that the speci® c blockage of IGF-IR deprives ES cells of an important tools preventing apoptosis induced by chemotherapeutic agents. In conclusion, we showed that the blockage of IGF-IR by neutralizing antibody or by suramin may signi® cantly inhibit the growth of ES cells both in vitro and in vivo, and that this treatment strategy greatly potentiates the antitumor activity of conventional chemotherapeutic drugs.

Eastern cooperative oncolog y group (E COG), Oslo, N orway
Soft-tissue sarcoma tumours account for approximately 1% of all malignancies. Although m ost patients present with apparently localised disease about 50% will die from subsequent m etastases. In 70% of all cases metastatic disease involves the lungs only. This observation has led to the concept of surgical treatment for lung metastases. By proper selection a 5-year survival of about 40% can be obtained by m etastasectomyalone. The present study aims to assess if preoperative chemotherapy with Doxorubicin and Ifosfamide improves survival of patients scheduled for metastasectomy for pulmonary m etastases.
The study was initiated in April 1996 as an intergroup stud y b etween S SG , EO RT C a nd E C O G . Patien ts undergoing radical m etastasectomy of 5 or fewer lung m etas tases w ill be ran d om iz ed to receive pre-metastasectomy chemotherapy (treatment #1) or no chemotherapy (treatment #2). Patients with treatment #1 will receive 3 chemotherapy cycles before and 2 cycles after metastasectomy provided if there is either a complete or partial response on CT scan or if there is a histological response grade III or IV. For grade I tumours (no necrosis) th e p atien t d o es n o t co n tinu e to p o st-op erative chemotherapy. Each participating institution has to select one of the two proposed chem otherapy regim en (standard: Doxorubicin 50 mg/m 2 +Ifo sfamide 5 g/m 2 or intensi® ed: D oxo r u bicin 75 m g/m 2 +Ifo sf am id e 5 g/m 2 +GCSF:granulocyte-macrophage colony-stimulating factor).
The principal end-point is overall survival. Secondary end-points are disease free survival and side effects. To detect an improvement of 15% in the 3-year overall survival, 340 patients need to be included in the study. Final analysis will be performed after observation of 190 deaths.
Stud y status: So far 5 institutions from 2 cooperative groups have entered 12 patients. A total of 12 EORTC investigators have obtained approval from their local ethical committee, and 3 additional centers have submitted the protocol for approval. From SSG a total of 16 centers have con® rm ed their interest in participation of the study. The ECOG has recently joined this intergroup protocol, and the SWOG is considering its participation.
Conclusions: The m ajor problem of this study has been a slow inclusion rate. As concluded in an intergroup m eeting last April 1998 in Leuven arranged by EORTC Soft Tissue and Bone Sarcoma G roup it is important that thoracic surgeons are informed of this protocol since they often see these patients primarily. Currently work is ongoing to recruit the participation of additional centers (SWOG and M SK).

The University of Texas M .D. Anderson Cancer Center, H ouston, Texas 77030, USA
Introduction: Mutations of the p53 tumor suppressor gene are the most frequent genetic abnorm ality in soft tissue sarcomas, a group of tumors frequently resistant to radiation. We have previously shown that wild type (wt) p53 status enhances radiosensitivity of human leiomyosarcoma cells. Consequently, we investigated the mechanisms by which wt p53 mediates this improved radioresponse.
M ethods: All experiments were conducted using a human leiomyosarcoma cell line SKLM S, with a m issense p53 m utation at codon 245, and derived stable transfectants expressing constitutive wt p53 (SKp53-2, SKp53-3), a tem perature-sen sitive p 53 m u tan t (SK A la-1), o r a neomycin-resistant marker gene (SKneo). After 0, 5, or 10 G y irrad iation, d istribution of cell cycle phases was determ ined by¯ow cytom etry, and apoptosis rate by TUNEL assay. DNA damage and repair following irradiation at these doses was quanti® ed by a PCR -based assay which displays decreased ampli® cation product when there is a damaged DNA template that cannot bind or extend PCR primers. In this case, the DNA template was an actively transcribed gene, glutathione-S-transferase p (GST-p ), whose gene product catalyzes glutathione conjugation to various electrophilic compounds including both carcinogens and anticancer agents. Damage to the GST-p gene presuma bly re¯ects the po ten tial d am age to o ther actively transcribed genes in this cell line. Cells were assayed at various intervals between 0 m in to 24 hrs after irradiation.
Results:There was minimal G1 and G2/M cell cycle arrest after irradiation in these cell lines. However, both the wt p53 containing radiosensitive cells (SKp53-2, SKp53-3) and mutant p53 radioresistant cells (SKLM S, SK neo) showed similar distribution of cell cycle phases after radiation. None of the cell lines had signi® cant apoptosis levels before or after treatment. The amount of initial DNA damage was comparable between wt and mutant p53 cell lines at both radiation doses. In contrast, wt p53 expressing sarcoma cells were markedly less able to repair radiationinduced damage compared to their m utant counterparts after both 5 and 10 Gy irradiation. 24 hr after 10 Gy irradia-tion, only 11± 15% of G ST-p gene remained intact in wt p53 cells, compared to 50± 100% in m utant p53 cells (p<0.05, t-test).
C onclusio n: These results indicate that p53 m ay have an additional important function to inhibit DNA repair in injured cells where downstream p53 pathways of cell cycle arrest and apoptosis are not as prom inent after cellular stress, thereby eliminating potential genomic damage from the population and enhancing the cytotoxic effect of radiation.These data support further investigation of p53 restorative strategies as a means to improve the therapeutic benefit from radiation in human sarcomas.

B. M .Wehrli,V. L. For nasier
Vancouver General Hospital, Vancou ver, B C and Wellesley Central Hospital, Toron to, Ontario, Canada B ackground: Since 1995, all new oncology patients at the Princess M argaret Hospital (PM H), prior to their receiving treatment, are required to have their pathology reviewed by a designated PM H consultant pathologist. This review is performed both to con® rm the existing diagnosis and to ensure consistency in grading and staging of tum ors. Through an examination of this review process, we hope to gauge the demographic scope of our consult service, to ascertain the degree of efficiency and effectiveness of our existing consultation protocol and to determ ine speci® c areas which could be improved.
D esign: A retrospective review of 50 cases, received by the soft tissue an d bone review patholog ist between February 27, 1998 and July 30, 1998, was performed. Each case consisted of the consultant's report, and either the original pathology report or a letter of request for consultation, as well as any glass slides available for review. Criteria exa m in ed in clud ed : ho sp ita ls o rig in ating con s ults, turnaround time, availability of original pathology report, the nature of m aterial sent for consult purposes, the nature of the gross and microscopic descriptions of specimens from the originating hospitals, and the concordance of initial diagnosis with the consultant diagnosis.
R esults: The 50 consultation cases originated from 28 different hospitals, with all cases originating within Ontario with the exception of 6 cases, 2 of which originated in Q uebec, an d o ne case each in Newfo un dland , N ew Brunswick, Saskatchewan, and Alberta. Turnaround time ranged from one calendar day to 104 calendar days, with an average of 10 days. Of the 50 consultation cases, 40 (80%) were accompanied by copies of the original pathology reports and 10 (20% ) were not. The specimens described in the dictation reports consisted of 15 resections, 20 excisional/incisional biopsies, 4 core biopsies, and one ® ne needle aspiration biopsy. M ost gross and m icroscopic descriptions of the specimens received from the originating ho spitals were adequ ate. Fifteen (41% ) o f 37 cases demonstrated diagnostic concordance, whereas 22 (59% ) of 37 cases did not dem onstrate such concordance. Of these 22 cases in which a diagnostic discrepancy was evident, 7 (32%) represented major non-concordance. In 5 of these 7 m ajor non-concordant diagnoses, the referring diagnosis was malignant and the consultant diagnosis was benign, and in the rem aining 2 cases, the referrin g diagnosis was benign and the consultant diagnosis was malignant. The remaining 15 (68% ) of 22 non-concordant diagnoses were m inor in nature, with all dem onstrating concordance concerning the benign or malignant nature of the specimen.
Conclusion :The consultant's bone and soft tissue consultation practice receives varying specimens from a widespread area. The non-concordant diagnostic rate between referring pathologists and the consultant is considerable. Patients who are diagnosed with sarcomata, should have their pathology reviewed by a bone and soft tissue consultant prior to radical therapy.

K. M . Lee, H . D. Suit, I. J. Spiro
Pur pose: C T-guided technology can be used to enhance precision of dose delivery in brachytherapy (BRT). This paper is a review of 26 CT-guided BRT perfor med on 22 patients with tumors at skeletal sites.
M eth ods and m aterials: The records of the 22 patients were retrospectively reviewed with regards to patient characteristics, histological diagnosis, treatment, BRT technique, tumor control, palliative bene® t and complications. BRT was performed either intraoperatively or percutaneously with the use of CT imaging for placement of catheters.
Results: Between 1993 and 1998, 22 patients were treated with C T-guided B RT to 26 anatom ical sites, m ainly involving the axial skeleton. Histological diagnoses included chordoma, osteosarcom a, chondrosarcoma, adenocarcinoma and other connective tissue tumors. Nine patients presented with prim ary tumor while 13 had recurrent tumors including 11 with metastatic disease. Seven patients had previous irradiation at the BRT sites. Of the 15 BRT procedures perfor med with curative intent, there were 3 local failures including 1 with distant metastasis, 2 regional recurrence and 3 distant relapses. Of the 11 BRT procedures perfor med with palliative intent, there were 2 local failures also with distant m etastasis, 1 regional recurrence and 4 distant relapses. The main palliative bene® t was de® nite pain relief after 17 (71% ) BRT procedures. The m ain BRT-related complications were wound problems seen after 5 (19% ) BRT procedures. C onclusio n: C T-guided BRT can be used in the m anagement of tumors at skeletal sites, in particular the axial skeleton where a high tumor dose can be delivered with a marked dose gradient through the spinal cord.This relatively new technique rem ains und er further evaluation and undergoes constant ® ne-tuning.

IRSG of the C hildren's C ancer Group and Pediatric Oncolog y Group, UT M D Anderson C ancer C enter, Houston , TX 77030, USA
In 1972, m embers of the pediatric oncology cooperative groups form ed the IRSG with support from the National Cancer Institute to study rhabdomyosarcoma (RM S) and undifferentiated sarcoma (UDS) in previously untreated patients less than 21 years of age. Since then, 3 successive protocols have shown improved overall 5-year survival rates, from 55% , in IRS-I to 71% in IRS-III (P<0.001;J C lin Oncol 1995;13 :610). IRS-IV (1991± 1997) tested a randomized comparison of vincristine, actinomycin D, and cyclophosphamide (VAC, with C=2.2 gm /m2/course) vs. VAI (ifosfam ide, 9 gm /m 2/course) vs. VIE (etoposide, 500 m g/m2/course) for patients with localized disease. Conventional radiotherapy (XRT; 50.4 Gray) was compared to hyperfractionated XRT (59.4 Gray) for patients with localized, gross residual sarcoma. Preliminary results should be available in 11/98. Proposals for IRS-V therapy are based on stratifying patients into 4 categories according to the probability of failure: low-risk, interm ediate-risk with em bryonal (EM B) RM S, intermediate-risk with alveolar (ALV) RM S or UDS, and high-risk patients with metastases (Stage 4) at diagnosis. Tumor specimens will be subjected to cytogenetic and molecular analyses, in order to ascertain whether prognostic subgroups can be identi-® ed among the patients with EM B and ALV RMS and UDS. The general strategy is to design therapy based on the likelihood of cure. The Table shows the four categories of patients, the 3-year failure-free survival (FFS) and overall survival estim ates, and the types of treatm en t to be administered.

The University of Texas M .D. A nderson Cancer Center, Hou ston, Texas 77030, USA
Pur pose:To evaluate the therapeutic value of resection alone, radiation alone or combined modality therapy for desmoid tumors.
Patie n ts an d m eth od s: On e hun dred an d eighty-® ve consecutive patients with desmoid tumors treated at our institu tion b etween the yea rs 1965 an d 1994 were retrospectively reviewed and form the cohort of this analysis. Surgery alone was the treatment for 122 patients, surgery and radiation for 46 (combined modality), and radiation alone for 21. There were 108 women and 81 men. Patients' ages ranged from 1± 81 years, with a m ean and median of 31 and 29 years, respectively. Size ranged from 1± 28 cm, with a m edian of 8.4 and 7 cm, respectively. One hundred and four patients (55%) presented after one (54 patients) or m ore (50 patients) local recurrences. M argin status for the surgery alone group was m icroscopic-positive in 40 (33% ), m icroscopic-negative in 78 (64% ), and grosspositive in 4 (3%). Margin status for the combined m odality gro up was m icro scop ic-p os itive in 33 (72% ) an d m icroscopic-negative in 13 (23% ). All patients in the radiation alone group were treated for gross disease. M edian follow-up was 9.4 years.
Results: The 5-and 10-year overall actuarial relapse rate was 30% and 33% , respectively. Uncorrected survival rates were 96%, 92%, and 87% at 5, 10, and 15 years, respectively. For the patients treated with surgery alone the actuarial recurrence rate was 38% at 10 years. The 10 year recurrence rates for the m argin-negative and margin-positive patients were 27% and 54% (p=0.003), respectively. For the combined modality patients the 10-year actuarial recurrence rate was 25% . Among 13 m argin-negative patients the 10 year recurrence rate was 15% , whereas 33 m arginpositive patients had a 10 year recurrence rate of 31% (p=0.5). The difference in recurrence rates for patients with microscopically positive m argins treated with surger y alone versus combined m odality was signi® cant (p=0.007). On multivariate analysis only age >3 0 and combined m odality treatment (versus surger y alone) independently correlated with a decreaed rate of recurrence. For patients treated with radiation alone the 10-year recurrence rate was 24% .
Conclusion: W ide local excision with negative pathologic margins is the treatment of choice for most desmoid tumors. Function sparing resection is appropriate since adjuvant radiation abrogates the adverse impact of positive margins. Unresectable disease should be treated with de® nitive radiation with the expectation of excellent disease-free survival.

M ultidisciplinar y Sarcom a Clinic, University of M ichigan Com prehensive Cancer C enter, A nn A rbor, M I 48109, USA
M etastatic leiomyosarcoma, whether it emanates from the gastrointestinal tract, the uterus or other soft tissue or visceral sites, has proven difficult to treat with conventional modalities. New approaches are clearly worth evaluating. Several disparate lines of reasoning support investigations of imm unotherapy in this disease: Some form s of leiomyosarcoma are more frequent in immunosuppressed patients and m ay be associated with Epstein-Barr virus, providing a potential antigenic target. Human T cells are now known to be capable of recognizing antigens on human sarcoma cells. So m e patien ts w ith m etastatic leiom yosarco m a have relatively protracted courses and slowly progressive disease. From May 1997 through July 1998, we entered 8 patientsÐ ages 29 to 70Ð with metastatic leiomyosarcoma onto an adoptive imm unotherapy protocol involving vaccination with irradiated autologous tumor cells plus BCG followed 10 to 14 days later by removal of the regional lymph nodes draining the vaccine sites. These ª vaccine-primedº lymph node lymphocytes were sequentially cultured in vitro in anti-C D3 m on oclonal antibod y (OK T 3) and a low concentration of interleukin 2 (60 Cetus U/m l) for 2 to 3 weeks. Unless tumor progression supervened, activated lymphocytes were administered intravenously followed by moderate-dose interleukin 2 (180,000 Cetus U/kg t.i.d. for 15 doses). Patients manifesting stable or regressing disease were retreated with a second course of interleukin 2. To be eligible, patients had to have accessible or previously cryopreserved tumor and a life expectancy of at least 3 months.
Of the 8 patients, the primary site was gastric in 2, uterine in 2, and small bowel, renal, retroperitoneal and soft tissue in 1 each. All patients had measurable metastatic disease in liver (6 patients) or lung (2 patients) as well as soft tissue sites; no patient had exclusively soft tissue disease. Four patients (1 uterine, 3 gastrointestinal primaries) had received no prior chem otherapy, the remaining patients had all received doxorubicin alone or with ifosfamide. Two patients had rapid progression of tumor and were removed from study prior to the administration of interleukin 2. The rem aining 6 patien ts were all treated w ith activated lymphocytes plus interleukin 2 and tolerated the therapy well. Dose reductions were required in 2 cases because of renal or hepatic toxicity, no patients required ICU admission or vasopressors. Of the ® ve patients who have been assessed after their initial course of therapy, four met the study conditions for and received a second course of interleukin 2.
Adoptive immunotherapy with autologous tumor vaccineprimed lymph node lymphocytes and interleukin 2 is complex and resource-intensive, but feasible in patients with m etastatic leiomyosarcoma. Further investigations of this and other form s of imm unotherapy appear to be warranted in this disease.
A ckn ow led gem en ts: F u n d ed b y N IH gran ts 1R21CA 72034 and M 01-RR00042. M etabolism B ranch and Pediatrics B ranch, National Cancer Institute, N ational Institutes of Health, B ethesda, M D 20892, USA S yn ovial sarco m a (SS ) is an ag gressive, soft-tissu e m alignancy occurring prim arily in the extrem ities of adolescents and young adults. M ore than 70% of SS cases have been attributed to a t(X;18)(p11;q11) chromosomal translocation which fuses the SYT gene from chrom osome 18 to either the SSX1 or SSX2 gene on chromosome X. The resulting SYT-SSX1 and SYT-SSX2 fusion proteins are believed to function as aberrant transcriptional regulators. The fusion of SYT to SSX1 or SSX2 creates a unique peptide sequence at the breakpoint which is not expressed in norm al cells. Because the t(X;18)(p11;q11) translocation is associated with SS tumor cells, we proposed that the SYT/SSX fusion protein could serve as a tumor-associated antigen.To determine the immunogenicity of the breakpoint peptide sequence, two representative overlapping 17mer peptides (SS1 and SS2, both present in the two types of fusions) were synthesized, both of which span the breakpoint region. Immunization of C57B L/6 mice with SS1-pulsed spleen cells elicited an SS1-speci® c, CD8 + , class I-restricted T cell response, as seen by in vitro CTL , activity analysis. Efforts are ongoing to develop an animal m odel in which to assess the in vivo efficac y of the peptides for anti-tum or therapies. Application of this approach in novel immunotherapies depends upon the binding of these peptides to HLA molecules. Therefore, to assess binding of SS1 and SS2 to various HLA m olecules, these peptides were tested using T2 cells transfected with HLA m olecules and FACS analysis. The SS2 peptide speci® cally bound both HLA-B7 and HLA-B27 at 100 m M concentration, increasing expression of the class I HLA molecules about 2-fold. The SS1 peptide also speci® cally bo u nd H L A -B 7 at 100 m M con cen tration, in creasing H LA -B 7 express ion abo u t 3.5-fold. These results suggest that the SYT-SSX1 and SYT-SSX2 fusion proteins potentially encode an antigenic amino acid sequence which speci® cally binds human HLA molecules. This may be useful in the development of novel imm unotherapies for SS.

S. R. Patel, J. Jenkins, N. E. Papadopoulos, M . A. Burgess, C. Plager, P. W. T Pisters, B. W. Feig, K. H unt, A. Pollack, G. Zagars, R. E. Pollock, R. S. Benjam in
The Sarcoma Center, University of Texas M .D. Anderson C ancer C enter, 1515 Holcom be B lvd., Hou ston, Texas 77030, USA A limited number of chem otherapeutic agents have activity in soft-tissue sarcomas. It is therefore important to test newer agents for their activity in this disease. We are currently evaluating Gemcitabine, a nucleoside analog with activity in a variety of solid tumors, in a two-arm phase 2 study. Patients with recurrent/metastatic soft-tissue sarcomas who have received or refused standard chem otherapy with adriamycin and ifosfamide constitute one arm and patients with GI leiomyosarcoma irrespective of prior chemotherapy exp osu re (du e to their kn ow resista nce to stan da rd chemotherapy) constitute the other arm . G em citabine was given at a dose of 1000 m g/m 2 /wk 3 7 followed by 1 week off an d re-stag ing to assess respo n se. Patien ts with responding or stable disease were then continued at a dose of 1000 mg/m 2 /wk 3 3 followed by 1 week off, and cycles were repeated every 4 weeks until maximum response or progression. A total of 19 patients are evaluable for response, 11 G I leiomyosarcomas and 8 other STS. There were eight males and 11 females and the m edian age was 53 (28± 75) years. All patients had a PS for 0± 1. One patient out of 11 GI leio patients has demonstrated a minor response, while two patients out of the 8 other STS have achieved partial responses. One of the responders had a metastatic angiosarcom a to lungs and the other one had lung metastases from a leiomyosarcoma of uterine origin. Treatm ent was generally very well tolerated. Four patients experienced grade 3 neutropenia and three patients experienced grade 3 thrombocytopenia leading to delays in therapy without any other consequences. Two patients had a grade 3 increase in transaminase levels which was self-limiting. Two patients had grade 3 myalgias. This preliminary data is encouraging and suggestive of activity of Gemcitabine in soft-tissue sarcomas. Our study is continuing to accrue more patients to better de® ne the level of activity. ASTS refractory to doxorubicin (dox) and ifosfamide (ifo) are highly resistant to chemotherapy. The investigation of new drugs is therefore warranted in these pts. To establish the efficacy and safety of Raltitrexed (`Tomudex' ) in ASTS refractory to one or two lines of dox and ifo containing regimens, a phase II study was conducted between August 1997 and December 1997 in 8 centers of the EORTC STBSG group. Raltitrexed was given at 3 m g/m2/course as a 15 m in IV infusion every 3 weeks. 23 pts were included and 21 are evaluable for toxicity and response. M edian age was 54 (range 25± 73) with 15 males and 6 females. Performance status was 0 in 9 (43% ) pts, 1 in 12 (57% ) pts. 16 pts had previously received chem otherapy in m etastatic phase, 4 as adjuvant, and 1 both. The prim ary tumor was located in the trunk (n=11), in the limbs (n=7), in the head and neck (n=3). The predominant histology was leiomyosarcoma (n=7, 33%). Respectively 4, 13, 1, 1, and 2 pts received 1, 2, 4, 5, or 6 courses of Raltitrexed.The best response was stable disease in 4 (19%) pts, and disease progression in 17 pts (81% ), with a median time to disease progression of 6 weeks. The treatment was well tolerated with only 1 pt experiencing grade 4 neutropenia and thrombopenia (5%), 1 grade 3 nausea (5%), 1 (5%) lethargy, 1(5% ) headache, 1 (5%) asthenia. A single pt (5% ) experienced febrile neutropenia. We conclude that Raltitrexed is an ineffective treatment for ASTS failing conventional chemotherapy with dox and ifo. Tomudex is a trademark, a property of Zeneca Ltd.

The E ORTC Soft Tissue and B on e Sarcom a Group (STB SG), EORTC, 1200 B r ussels, B elgium
The characteristics of pts with ASTS still alive 5 years (yrs) after initial treatment with doxorubicin (DXR), i.e. long term survivors (LTS), were analyzed among the 1742 pts treated between 1976 and 1990 in trials of EORTC STBSG group. The m edian overall survival of this series was 11.3 m onths and the projected 5-years (yrs) survival was 8.6% . 39 pts were alive at 5 yrs among the 1308 uncensored pts. The percentages of females (69% vs 49% ), of grade 1 tumors (30% vs 6% ), and of pts with an initial performance status (PS) of 0 (49% vs 27% ) were superior in the LT S su bg rou p as com pa red to o th er p ts (p< 0.01). Histological subtypes were not signi® cantly differen t in LTS as compared to other pts. LTS pts less frequently had grade 3 tumors (35% vs 64%), and liver metastasis (5% vs 18% ) (p<0.01). Despite of these differences, LTS were observed in all categories of pts. A complete response (CR) to DXR was a major parameter correlated to 5-yrs survival: respectively, 18% (11/60) of pts in C R, 4% (9/215) of pts in partial response (PR), 2% (13/482) of pts in stable disease (SD), and 1% (3/372) of pts in progressive disease (PD) after DX R were alive at 5 yrs. Patients in CR, PR, SD and PD represented 31% , 25% , 26% , and 8% of LTS respectively. In multivariate analysis independent parameters correlated to CR and 5-yrs survival were similar to those correlated to overall survival in the same series, i.e. age, liver m etastases, PS, grade I tumors. In conclusion, LTS are observed in all prognostic subgroups of pts with ASTS treated with DXR, in particular among pts in CR, in whom 5-yrs survival is 18% . Achievement of C R should be the prim ary aim of chemotherapy. This work was supported by the Prix Pier re B ardoux . A clinically usefu l progno stication system sho uld be repro du cible an d give go od separation between two groupsÐ one with good and one with poor prognosis. We have recently proposed a system based on three negative prognostic factors large tumor size, vascular invasion, and microscopic tumor necrosis. Tumors which exhibit 2 or 3 f actors are catego rized as high -g ra de, th e o thers as low-grade. We have now tested this system for reproducibility both as regards classi® cation of necrosis and vascular invasion, and as regards prognostic strength related to grading.

P. G ustafson, M . A Ê ker m an, T. A . A lvega Ê rd
We selected 200 adult patients with STS of the extrem ity or trunk wall, 100 from the M usculoskeletal Tumor Center, L un d Swed en a n d 100 fro m th e Ins titut B ergo n ie Â , Bordeaux, France. All patients had been treated by surgery. The median follow-up for the 117 survivors was 10 (1.5± 27) years. All slides from all tumors were reviewed independently by three groups of pathologists for the presence or absen ce of vascular invasion and m icroscopic tum o r necrosis, without knowledge of the clinical course. Tumor size was considered same. The prognostic strength was compared using the grading obtained by the differen t pathologists. C oncordance in classi® cation was assessed by Kappa-analysis. Outcome related to grading was assessed by Kaplan-M eier technique. Concordance in classi® cation of vascular invasion, m icroscopic tum or necrosis, and grading was seen respectively in 156 (78% ), 154 (77% ), and 167 (84% ) of the 200 tumors. Based on the differen t observers grading, the cumulative 5-year m etastasis-free survival rate in the 200 patients varied for patients with low-grade tumors between 0.85 and 0.80 and for patients with high-grade tumors between 0.48 and 0.43. The kappavalue for grading between all three groups was 0.77.
Classi® cation of vascular invasion and m icroscopic tumor necrosis seems to have acceptable reproducibility. This so called SIN-system gave similar survival rates when used by different observers and applied to different series of STS patients. It gave good separation between patients into two groups with high or low risk for metastasis. M aterials and m eth ods: A retrospective review of 172 osteosarcoma patients diagnosed between 1987 and 1992 was perform ed. Forty patients had p-glycoprotein levels available. The majority of the osteosarcomas were stage II-B (33 patients), with the remaining seven being stage III. Tumor sites included 25 femora, 7 humeri, 5 tibiae, and one each of pelvis, radius and ® bula. The expression of P -glycop rotein by cu ltu red tum or cells fro m biop sy specimens was determined using immuno¯uorescent microscopy.

CA N W E EXP LA IN THE DIFFERE NT PROG NO STIC FAC TORS FO R RE SPO NSE AND S URV IVAL IN TH E VA RIO US HIS TO LO G ICA L SU BTYPE S O F ADVANCED S OFT TIS SU E S ARC OM A? A N ANA LYSIS OF 2185 PAT IENTS FROM TH E EO RTC S OFT TIS SU E AND BO NE
Results : M ore patients with detectable P-glycoprotein (12/18 or 67% ) developed m etastases as compared to those patients with undetectable P-glycoprotein (9/22 or 41% ; N.S.). Similarly, ten of 18 patients (56% ) with tumors expressing P-glycoprotein died of disease, while only 4 of 22 (18% ) with no detectable P-glycoprotein died (chisquare; p<0.02). Among the patients with stage III disease at presentation, two of three with P-glycoprotein expression died, while only one of four without P-glycoprotein died.
Conclusion: Expression of P-glycoprotein by tumor cells appears to be associated with an increased incidence of death in stage II and III osteosarcoma.
1.12). In comparison, the higher grade areas had uptake of 1.26± 1.64 cts/pixel (or 1.38± 1.47 max cts) of the contra lateral normal site. In addition, four other patients with chondroid lesions, diagnosed radiographically, have been followed with periodic clinical exam and roentgenogram s. All had T1201 scans with uptake noted to be equivalent to the con tra lateral site. A ll of these patien ts rem ain asymptomatic with no change in their radiographs during follow-up (range 5 mos.± 4 yrs.) T1201 uptake, m ediated through the Na-K ATPase dependent pump, is dependent not only on blood¯ow but also tumor of cell viability and activity. W ith a degree of heterogeneity seen in chondroid lesions, T1201 may help illucidate those tumors, and regions within tumors, that are m o re aggressive and therefore need m ore aggressive management despite a relatively low grade as seen on open biopsy. Between 3/94 and 10/97 20 women and 19 men, m edian age 51 yrs (27± 76 yrs), were treated for cure of their high grade primary extremity and limb girdle soft tissue sarcomas with preoperative chem otherapy and irrad iation. This involved two m o n th ly c ycles o f IM A P (ifo sf am ide 2,500 m g/m 2 + M ESNA 2,500 mg/m 2 daily for two days with m itomycin 4 /m g/m 2 , doxorubicin 40 m g/m 2 , and cisplatin 60 m g/m 2 added on the second day). Beginning six days before chem otherapy each m onth they received four days of GM -CSF (sargram ostim) 250 mcg/m 2 S.C. every 12 hours. This was then continued for 14 more days the day after chem otherapy was ® nished each month. External beam irradiation was begun at month three and con tinued ® ve days weekly for approxim ately 4,50 0± 5,000 cGy in ® ve weeks, accompanied by adjuvant doses of mitomycin 6 m g/m 2 , doxorubicin 30 mg/m 2 , and cisplatin 45 m g/m 2 given thrice (before, midway, and after irradiation.) De® nitive surger y accompanied by intraoperative electron irradiation or brachytherapy and/or further external beam treatment (to a total dose of 5,500± 7,000 cGy) was scheduled to occur after a one m onth rest interval. No postoperative chem otherapy was given. Sites of origin were thigh 23, pelvic girdle 5, leg 4, arm 3, shoulder girdle 2, and forearm 2. M aximal tumor diameters ranged from 1.6 to 30 cm (m edian 10 cm). Histologic types were malignant ® brous histiocytoma 20, synovial sarcoma 5, extraosseous osteosarcoma 4, malignant peripheral nerve sheath tumor 3, spindle cell 2, ® brosarcoma 2, and leiomyosarcoma 3. Surgery involved radical excision in two patients, wide excision in 22 patients (1 contaminated), marginal excision in 14 (1 contaminated), and no surgical excision in one patient (who developed brain m etastasis during preoperative treatment). Signi® cant complications included tissue necrosis requiring debridem ent and secondary closure (1), intralesional hemorrhage and necrosis requiring amputation (1), and pathologic fractures with infection requiring amputation (1). Limbs remained intact in 37 of the 39 patients. The prescribed irradiation was received in 36 cases at M ayo C linic; two patien ts rec eived preo perative irrad iation elsewhere; and one patient (who experienced massive intralesional hemorrhage and amputation) received none. Thirty four patients received two cycles of IMAP chem otherapy and ® ve received one cycle. Only eight patients received the full three cycles of M AP chemotherapy concomitant with preoperative irradiation; 12 patients received two cycles; 14 patients received one cycle; and 5 patients received none. W ith median follow up of 33 m onths (8± 51 m o.) four patients have died of m etastatic disease and a ® fth died N.E .D. in a m otorcycle accident. Two patients have experienced only local recurrence and a total of eight have experienced only m etastasis (including the four who died).

CH EM OTH ERAP
M a tu rin g K a p la n -M eier cu r ve s su gge st f avo rab le survival and tim e to m etastasis experien ce with this regim en.

EORTC D ata C enter and STB SG, B russels, B elgium
For the last 20 years, the (STBSG) has conducted successive clinical trials testing new agents and combinations in patients with non pretreated advanced soft tissue sarcoma. Initial results are generally con® rmed in large phase III randomized trials. The present work reviews the usefulness of this procedure, and compares the results of the con® rmation trials with those of the initial trials.
In 1982, a randomized phase II trial of Ifosfamide (IFOS) vs C yclophospamide showed a response rate of 25% in 38 patients treated with IFO S (5 g/m 2, 24 hr infusion, q 3 wks). In 1992, the same regim en used as the control arm in a randomized phase II trial showed 5% responses in 49 patients.
The original promising IFO S results encouraged the group to test, in 1984, IFOS 5 g/m2 combined with Doxorubicine (DOX) 50 mg/m2. In an initial non randomized study, a response rate of 34% was obtained in 203 patients. On this basis, the group em barked in 1985 on a randomized phase III trial, comparing this combination to DOX alone (75 mg/m 2); this study did not show any signi® cant advantage of the IFOS-DOX combination, where 28% responses were observed in 297 patients. The next step was to increase the dose of D OX to 75 mg/m 2 in the com bination, which was m ade possible by the add ition of growth factors. A pilot study with IFOS 5 g/m 2, DOX 75 m g/m 2 and Gm C SF conducted in 19 88 showed a response rate of 46% in 111 patients; in 1992, the group started a randomized trial com paring this regim en to the previous ª standard doseº regimen. Response rates were n ot statistically signi® can t, and substantially lower than those observed in the initial non randomized trials: 21% in the standard dose reg imen and 23% in the high dose regimen (to be com pared with the 34% and 46% of the initial pilot studies). Despite these large differen ces in response rates, no m ajor differen ces were observed in survival between all these trials.
These results underline the danger of historical control and small size studies in this disease. The pro® le of patients referr ed to ® rst line chem otherapy had probably ch ang ed over tim e, an d the n um ber o f histolog ical subtypes m ay be responsible for large heterogeneity in recruited patients. R andom ization and strati® cation for known prognostic factors does overcom e these problem s. R espon se rate is probably not a valid en d-point for comparing chemotherapy regimen in this disease. External review of response which is now a standard practice of the group m ay partially explain the decrease in response rates observed in our trials.