Annals of Oncology 8: 765-771, 1997.© 1997 Kluwer Academic Publishers Printed in the Netherlands.

Original article

Long-term follow-up and post-relapse survival in patients with non-metastatic osteosarcoma of the extremity treated with neoadjuvantchemotherapy*

S. Ferrari,1 G. Bacci,1 P. Picci,1 M. Mercuri,1 A. Briccoli,1 D. Pinto,1 A. Gasbarrini,1 A. Tienghi2 &A. Brach del Prever3

xBone Tumor Department, Istituto Orlopedico Rizzoli, Bologna; 2Medical Oncology Division, Ravenna; iPaediatric Department, University ofTorino, Torino, Italy

Summary

Background: Most of the studies of the treatment of non-metastatic osteosarcoma of the extremity have reported resultsin terms of probability of survival up to five years with aminimum follow-up of less than two to three years. Definitionof reliable indicators of prognosis and predictive factors forsurvival require mature data derived from a long-term survivalanalysis.

Patients and methods: A review of 127 patients with non-metastatic osteosarcoma of the extremity, treated betweenMarch 1983 and June 1986, was performed. The treatmentprotocol consisted of primary chemotherapy with MTX (ran-domization to high vs. moderate dosages) and CDP followedby surgery. Postoperatively, patients with <60% tumor ne-crosis received ADM and BCD; those with tumor necrosis2t 60% < 90% (Fair Responders FR) were given MTX, CDPand ADM. Up to January 1984, patients with tumor necrosis> 90% received MTX and CDP only, and after then they weregiven the same treatment as for FR. A multivariate analysis totest predictive factors for survival was performed.

Results: With a median follow-up of 134 months (range114-153), the 12-year DFS was 46%. A good histologic re-

sponse, an LDH baseline value within the normal range, andthe use of high-dose MTX were positive predictive factors forDFS. With a median time of observation for survivors of 130months, the 12-year overall survival was 53%. None of thepatients who relapsed with local or distant recurrences otherthan lung metastasis are now alive. Patients with a relapse-freeinterval longer than 24 months had a significantly better post-relapse survival than those with a shorter relapse-free interval(40% vs. 7%; P = 0.0159). All of the patients who were notsurgically treated had disease progression and died within 40months after the first recurrence. The surgically-treated pa-tients had a 30% post-relapse survival probability.

Conclusions: In non-metastatic osteosarcoma of the ex-tremity, chemotherapy-induced tumor necrosis, the baselineLDH serum value and the use of HDMTX are significantpredictive factors for DFS. The relapse-free interval and thepossibility of metastasectomy are significant factors condition-ing the post-relapse survival.

Key words: high-dose methotrexate, lactic dehydrogenase,metastasectomy, neoadjuvant chemotherapy, non-metastaticosteosarcoma of the extremity, tumor necrosis

Introduction

Randomized and non-randomized studies [1-6] havedemonstrated the primary role of chemotherapy in thetreatment of non-metastatic osteosarcoma of the ex-tremity. However, most of these studies reported theirresults in terms of probability of survival up to five yearscalculated on study populations whose minimum fol-low-up was less than two to three years, and only for theMulti-Institutional Osteosarcoma Study (MIOS) havethe results been updated [7].

Analysis of these data yielded useful information onthe efficacy of different chemotherapy treatments, butdetermination of dependable indicators of prognosis

* Presented at the 21st Congress of the European Society for MedicalOncology, 1-5 November 1996, Vienna, Austria.

and predictive factors for survival require mature dataderived from a long-term survival analysis.

Our objective was to update the results obtained in astudy population of 127 patients with non-metastaticosteosarcoma of the extremity, treated with neoadjuvantchemotherapy between March 1983 and June 1986 andto evaluate the predictive factors for survival.

The early results of this study (IOR/OS/I), updatedto December 1988, were published in 1990 [8].

Patients and methods

The criteria of eligibility for the study were the following: biopsy-proven classic high-grade osteosarcoma of the extremity, age youngerthan 50, absence of metastasis, no prior history of cancer and no priorchemotherapy or surgical treatment for the bone lesion.

766

IOR/08/1

Pr«OD»ratlv8 chamothtripy

MTX COP MTX COP

For randomization, patients were stratified according to the proxi-mal (upper third of femur or humerus) or distal location of the tumor.

The second aim of the study was to evaluate postoperative chemo-therapy treatments in relation to the histologic response to primarychemotherapy.

PoiloD»r«llva chomothrapy

AOU

9

ADM BCD

UTX

13

CDP

14

ADM

ADM

17

BCD

MTX

21

ADM

CDP

22

BCD

ADM

25

ADM

MTX

28

BCD

CDP

30

ADM

33

ADM

• M k

BCD

Figure 1. IOR/OS/I. Chemotherapy regimen.Abbreviations: MTX - methotrexate, random 7.5 g/m2 vs. 750 mg/m2;CDP - cisplatin 120 mg/m2 over 72-hour continuous infusion (i.a. inthe preoperative phase, i.v. postoperatively); ADM - doxorubicin 45mg/m2/daily for two days, BCD - bleomycin 15 mg/m2/daily, cyclo-phosphamide 600 mg/m2/daily, dactinomycin 600 ug/m2/daily fortwo days; GR - good responder patients (tumor necrosis ^ 90%); FR- fair responder patients (tumor necrosis ranging between 60-89%);PR - poor responder patients (tumor necrosis < 60%).Since February 1984 good responder patients had the same treatmentof fair responder patients.

The pre-treatment patient evaluation comprised medical history andphysical examination, complete blood cell count, liver function tests,and determination of serum creatinine, electrolytes, serum alkalinephosphatase (SAP), and lactic dehydrogenase (LDH).

The primary tumor was evaluated with plain roentgenograms andcomputed tomography (CT) scans; angiograms were performed ifindicated by the choice of surgical procedure.

Computed tomography or whole lung tomography and bone scin-tigraphy were used to exclude metastases.

The type of surgery (amputation or limb salvage) as well as the typeof reconstruction for patients treated for limb salvage (prosthesis,allogTaft, vascularized graft) were chosen according to location andextent of the tumor, and the patient's age and desired life style.

In any case, for conservative surgery, it was mandatory that thepreoperative staging assure achievement of adequate surgical margins.

Surgical margins of the resected specimen and histologic responseto primary chemotherapy were assessed at the study Institution.

Surgical margins were classified as radical, wide, marginal andintralesional according to Enneking et al.'s criteria [9].

The histologic response to primary chemotherapy was rated 'good'(equal to or more than 90% tumor necrosis), 'fair' (60%-89% tumornecrosis) or 'poor' (less than 60% tumor necrosis) and evaluatedaccording to a method previously reported [10].

After chemotherapy completion, patients were followed every twomonths for two years, every three months during the third year, everyfour months during the fourth year, and then every six months, withradiographs of the stump or the resected limb, and radiographs of thechest.

The diagnostic and surgical procedures and the follow-up werecentralized at the study Institution.

The chemotherapy treatment was administered in the Chemo-therapy Division of the Rizzoli Institute. Two Oncologic Centres (thePaediatric Divisions of the Universities of Bologna and Torino) coop-erated for chemotherapy treatments.

IOR/OS/I was a randomized study projected to compare the effi-cacy of high-dose methotrexate (HDMTX) versus moderate-dosemethotrexate (MDMTX) in a chemotherapy regimen consisting ofcisplatin (CDP), doxorubicin (ADM), bleomycin, cyclophosphamideand dactinomycin (BCD).

Chemotherapy

The chemotherapy regimen of IOR/OS/I is outlined in Figure 1.Preoperatively, patients received two courses of MTX and CDP.

MTX was given intravenously (i.v.) on day 1; patients were random-ized to receive HDMTX (7.5 g/m2 in a six-hour infusion) or MDMTX(750 mg/m2 in a 30-minute infusion).

In both cases, citrovorum factor rescue was started 24 hours afterthe beginning of MTX infusion. In patients who were given HDMTX,hydration was performed during the first 24 hours after administrationof the drug according to Rosen et al.'s guidelines [11].

On day 7 CDP (120 mg/m2) was intraarterially (i.a.) delivered in a72-hour continuous infusion.

A second course of MTX and CDP was repeated after a 14-dayinterval

The start of post-operative chemotherapy was scheduled for oneweek after amputation or three weeks after a limb-salvage procedure.

According to the histologic response to primary chemotherapy,patients received different treatments.

Patients with a good histologic response (GR) were given two cyclesof MTX and CDP IV, with the same doses and schedule as in thepreoperative phase.

Patients with a fair histologic response (FR) were given three cyclesof MTX and CDP i.v., with doses and schedule as in the preoperativephase and four courses of ADM (45 mg/m2/day for two days).

Patients with a poor histologic response (PR) were treated with fivecycles of ADM (45 mg/m2/day for two days) and BCD (bleomycin 15mg/m2/day, cyclophosphamide 600 mg/m2/day, and dactinomycin600 ug/m2/day for two days).

After February 1984, good responders received the same treatmentas fair responderts because of four early relapses among the first 15good-responder patients.

Statistical analysis

The disease-free survival (DFS) was calculated from the first day ofchemotherapy until recurrence (local or distant) or until the last follow-up examination. Overall survival (OS) was calculated from the first dayof chemotherapy until death of the disease or the last follow-up.

The survival curves were calculated according to the Kaplan-Meier method and compared by means of the log-rank test.

The Cox proportional hazards regression analysis was used formultivariate analysis to test predictive factors for survival.

The frequency of the distribution of different parameters wascompared among groups of patients by means of the chi-square test.Significance was set at P < 0.05.

Results

Between March 1983 and June 1986, 242 patients withosteosarcoma were registered at the study Institution,and 170 of them were eligible for the study.

Thirty-two of the eligible patients declined to enterthe study and were treated with immediate surgery fol-lowed by adjuvant chemotherapy.

Eleven of the 138 patients who entered the study werenot evaluable.

Five patients refused surgery and moved to anotherhospital for local treatment with hyperthermic antiblas-

767

Table 1. IOR/OS/I: clinical characteristics of cvaluable patients.

Age (years)

Sex

Serum alkalinephosphatase

Serum lacticdehydrogenase

Site

Tumor volume

Histology

Median 15 (range 5-43)

MaleFemaleNormalElevatedNormalElevatedFemurTibiaHumerusOther< 150 ml3*150 mlOsteoblasticChondroblasticFibroblasticTeleangiectaticOthers

72 (57%)55 (43%)53 (42%)J 4 (58%)81 (64%)46 (36%)68 (53%)34 (27%)15(12%)10(8%)53 (42%)74 (58%)97 (77%)15(12%)3 (2%)8 (6%)4 (3%)

tic perfusion with CDP (300 mg/m2); all of them died ofuncontrolled disease. Five patients refused chemother-apy after the first cycle of MTX (three cases) or aftersurgery (two cases). Three of these patients underwentlimb-salvage procedures and two amputatations. Lungmetastases developed in these patients 10 to 24 monthsafter sugery. One patient died of a pulmonary embolismimmediately after a limb-salvage procedure.

The remaining 127 eligible patients were evaluable forthe survival analysis.

Sixty-seven patients were randomized to receiveHDMTX and 60 to receive MDMTX.

The clinical features of the evaluable patients arereported in Table 1.

Surgery and histologic response to primary chemotherapy

One-hundred twenty-six patients underwent surgery.Progression of disease and lung metastasis developed

in one patient in the HDMTX group during the pre-operative phase of treatment. This patient was movedto another hospital for lung metastasectomy and localtreatment with hyperthermic perfusion with CDP: hedied six months later of uncontrolled disease.

Ninety-one patients (72%) underwent limb-salvagesurgical procedures, three (2%) had a rotation-plastyand 32 patients (26%) had amputations.

Adequate surgical margins (radical or wide) wereachieved in 113 patients (90%): in 101 the margins werewide and in 12 were radical. In 13 patients (10%) surgicalmargins proved to be inadequate (marginal nine patients,intralesional four patients).

Of the 126 patients who underwent surgery, 66 (52%)had good histologic responses (> 90% tumor necrosis)to primary chemotherapy; 45 (36%) had a tumor ne-crosis ranging from 60% to 89% (fair responders) and 15patients (12%) had poor histologic responses (tumornecrosis <60%).

According to the dose of MTX, good histologic re-

OVERALL SURVIVAL

53K.I67 12?)

« % (5ft 127)

DISEASE-FREE SURVIVAL

Figure 2. IOR/OS/I protocol. Disease-free survival and overall sur-vival of evaluable patients.

sponse was observed in 41 (62%) of the 66 patients whoreceived HDMTX and in 25 (42%) of the 60 patientstreated with MDMTX (P < 0.04).

Disease-free survival

With a median follow-up of 134 months (range 114-153),58 of the 127 evaluable patients (46%) were continuouslyfree of disease, and 69 patients (54%) had distant orlocal recurrences.

The median time to recurrence was 14 months (range2-96).

Sixty-three patients (50%) relapsed with metastasesonly: in 58 cases (92%) the first site of metastases werethe lungs, in 4 (6%) the skeleton, one patient developed(2%) epidural metastases.

Local recurrence developed in six of the 126 operatedpatients (5%) and in four cases it was associated withlung metastases.

The median time to local recurrence was 10 months(range 6-28).

Local recurrences were observed in 2 (1.7%) of113 patients with adequale surgical margins and in 4(30.7%) of 13 patients with inadequate surgical margins(P < 0.001).

According to the tumor necrosis, the incidence oflocal recurrence was 1.5% (1 of 66) in GR patients, 6.6%(3 of 45) in FR patients and 13.3% (2 of 15) in PRpatients (P = 0.147).

The 12-year probability of DFS was 46% (Figure 2).At univariate analysis no significant differences in termsof DFS were found according to age (comparisons wereperformed according to different age groups: ^15 vs.<15, 5*21 vs. <21, = 30 vs. <30), site {femur vs.humerus vs. tibia vs. other), size (=150 ml vs. < 150 ml)or histologic subtype (osteoblastic vs. chondroblastic vs.telangiectatic vs. others) of the tumor.

The histologic response to primary chemotherapy(Figure 3) influenced the DFS, with a 12-year DFS of61% for GR patients, 35% for FR patients and 13% forPR patients (P = 0.0004).

According to the baseline serum value of alkaline

768

Table 2. IOR/OS/I: predictive factors of disease-free survival. Coxproportional-hazards regression analysis.

Figure 3. IOR'OS'I protocol. Disease-free survival and primarychemotherapy-induced necrosis.Abbreviations: GR - good responder patients (tumor necrosis^ 90%); FR - fair responder patients (tumor necrosis ranging be-tween 60-89%); PR - poor responder patients (tumor necrosis< 60%); log-rank P = 0.0004. GR vs. FR: log-rank P = 0.0022. GRvs. PR: log-rank P = 0.0003. FR vs. PR: log-rank P - 0.2823.

Figure 4. IOR OS I protocol. Disease-free survival and baselineserum lactic dehydrogenase (LDH) level. Log-rank P = 0.0013.

Figure 5. IOR/OS/I protocol. Disease-free survival and methotrexatedose log-rank. P = 0.0633.

phosphatase, patients with normal value by age had a12-year DFS higher than the one obtained in patienswith high SAP level (56% vs. 39%; P = 0.0577).

According to the serum LDH level (Figure 4), pa-tients with LDH lower than 460 U/l had better DFSthan patients with higher levels (55% vs. 29%; P -0.0013).

Variable B coefficients Rate ratio 95% CI P-value

Tumor necrosis>90% 1<90% 0.9863 2.68 1.61-4.46 0.0001

LDHNormal 1High 0.9236 2.52 1.53-4.13 0.0003

MTXHigh dose 1Moderate dose 0.4963 1.64 1.01-2.66 0.0441

SAPNormal 1High 0.4525 1.57 0.94-2.62 0.0841

Abbreviations: LDH - lactic dehydrogenase; SAP - serum alkalinephosphatase.

The DFS curves according to MTX dose (Figure 5)showed an advantage for patients who received HDMTXin comparison with those treated with MDMTX (52%vs. 38%; P = 0.0633).

A multivariate analysis including the factors that, inaddition to univariate analysis, appeared to be predic-tive of DFS (tumor necrosis, SAP, LDH, and MTXdose), was performed (Table 2).

The chemotherapy-induced tumor necrosis (< 90% vs.^90% tumor necrosis) was the most important pre-dictive factor for DFS (P = 0.0001) followed by base-line LDH serum level (Js460 U/l vs. <460 U/l; P =0.0003). The MTX dose revealed a statistical signifi-cance (P - 0.0441) whereas SAP lost the predictivesignificance for DFS it had appeared to have at theunivariate analysis.

Overall survival and post-relapse survival

With a median time of observation for survivors of 130months, the 12-year overall survival was 53% (Figure 2)for the 127 evaluable patients (48% (67 of 138) includingpatients who entered the study but were considered non-evaluable because of major protocol violations).

None of patients who relapsed with local recurrenceor with secondary disease other than lung metastases isnow alive.

The actuarial post-relapse survival was 19% at twoyears and 13% at 10 years after the first recurrence.

The time to relapse significantly influenced the post-relapse survival.

Patients with a relapse-free interval longer than 24months had a 40% eight-year survival, whereas thosewith a shorter relapse-free interval had a 7% eight-yearsurvival^ = 0.0159).

While the post-relapse outcome of all patients hasbeen recorded, in only 50 cases (72.5%) has the treat-ment after relapse been recorded.

Twenty-four patients (48%) whose metastases wereconsidered resectable underwent surgical treatment; infour patients it was combined with chemotherapy, and inone with radiation therapy.

769

Due to the site or the extent of the secondary disease,surgical treatment was not considered feasible in theremaining 26 patients (52%), all of whom were treatedwith chemotherapy and/or radiation therapy.

The chemotherapy planned for relapsing patients wasbased on the first-line treatment: those patients who hadpreviously received M-DMTX were given HDMTX, GRand FR patients received BCD and the GR patients whowere not given ADM were treated with this drug.

The surgical treatment influenced the post-relapsesurvival.

All patients who were not surgically treated had dis-ease progression and died within 40 months after thefirst recurrence.

Thirty percent of the surgically-treated patients had a10-year probability of post-relapse survival.

Nine (37%) of 24 patients who underwent metasta-sectomy were alive, six of them free of disease after thefirst surgical procedure, two after a second thora-cothomy and one with uncontrolled disease after thethird thoracothomy.

Discussion

The results achieved with the IOR/OS/I protocol aresimilar to those obtained with similar protocols basedon MTX, BCD, CDP, ADM and derived from the T-10protocol [11].

In the CCG-782 study [12], patients were preopera-tively treated with HDMTX, vincristine (VCR) andBCD; the postoperative treatment consisted of a com-mon phase based on MTX, BCD and ADM, and asubsequent phase differentiated according to histologicresponse: patients with more than 95% tumor necrosis(good histologic response) were given MTX, BCD andADM while the others received a 'salvage chemother-apy' with CDP, ADM and BCD. Twenty-eight percent ofthe patients had good histologic response. An eight-yearevent-free survival of 53% and an eight-year OS of 60%,were obtained. Patients with good histologic responsehad significantly better prognoses than patients with apoor histologic response.

Saeter et al. [13] reported similar results with a similarprotocol: 17% of patients had good histologic responseto a primary chemotherapy based solely on HDMTX;the five-year relapse-free survival was 54% and the OSwas 64%. The histologic response to primary chemo-therapy proved to be an important prognostic factoralso in this study.

In the cooperative osteosarcoma study (COSS)-82study [14], 26% of patients preoperatively treated withHDMTX and BCD showed good histologic response;the four-year metastasis-free survival (MFS) was 49%;good-responder patients, postoperatively treated withthe same drugs preoperatively employed, had a four-yearMFS of 73%, whereas poor-responder patients who re-ceived a 'salvage chemotherapy' with CDP and ADMhad a four-year MFS of 41%.

In our study, patients were preoperatively treatedwith MTX and i.a. CDP. The percentage of patientswith good histologic response (^ 90% tumor necrosis)was 52% (62% in those patients who were randomized toreceive HDMTX). This high percentage of good histo-logic response, superior to that reported in the previousstudies, was obtained "because of the preoperative use ofCDP and its intraarterial administration.

Despite this high percentage of good responders toprimary chemotherapy, the 12-year DFS and OS were46% and 53%, respectively; for the patients who weretreated with HDMTX, the 12-year DFS was 52% andthe 12-year OS was 61%.

IOR/OS/I was in fact a randomized study to evaluatethe role of high-dose MTX in the treatment of non-metastatic osteosarcoma of the extremity.

After a multivariate analysis, a significant advantagefor patients treated with HDMTX (7.5 g/m2) was dem-onstrated in comparison with patients treated withMDMTX (750 mg/m2) (Table 2).

These results, achieved in a multidrug regimen, con-firm the importance of HDMTX in the treatment ofosteosarcoma.

A relationship between serum MTX concentrationand tumor necrosis was reported by Saeter et al. [13]; weobserved a close relationship between values of MTXserum peak and tumor necrosis in subsequent studies [15]on patients treated with HDMTX (8-12 g/m2). More-over, a relationship between MTX serum peak and prog-nosis was found by us [16] and other authors [17,18].

A significant influence of dose intensity of HDMTXon the outcome of patients with osteosarcoma was foundin the recently reported meta-analysis performed byDelepine [19].

The role of HDMTX was discussed after the publica-tion of the results of the first EOI study [20] in which twodifferent chemotherapy regimens were compared: onewas based on six cycles of CDP/ADM while the othercomprised four cycles of CDP/ADM plus four cycles ofHDMTX.

In terms of DFS, the first protocol was more effectivethan the second (five-year DFS: 57% vs. 41%, P - 0.02).

These results led to a new study [21] comparing achemotherapy protocol based only on CDP and ADMto a T-10 type chemotherapy regimen, with HDMTX,VCR, BCD, CDP and ADM. No differences were foundbetween the two protocols in terms of progression-freesurvival (39% vs. 40%), or of OS (55% for both protocols).

The oncologic results of this study were inferior tothose achieved in the previously-mentioned studies [12-14] and similar to those achieved in the moderate-doseMTX arm of our study (12-year DFS: 38%).

Even though they highlight the importance of CDPand ADM, with doses and schedules according to thosein the EOI study, in our opinion they do not support aproposal of chemotherapy treatment of osteosarcomabased on CDP and ADM only and excluding HDMTX,because of the demonstrated activity of HDMTX onosteosarcoma.

770

Besides the use of HDMTX, in our study we foundtwo additional significant factors which are predictive ofDFS: the primary chemotherapy-induced tumor necrosisand the serum level of LDH.

With respect to rumor necrosis and DFS: patientswith good histologic response to primary chemotherapyhad 61% DFS, those with a tumor necrosis rangingbetween 60% and 89% (fair response) had a 35% DFS,whereas a 13% DFS was obtained in poor-responderpatients (P = 0.0004).

These results confirmed the predictive significance oftumor necrosis in osteosarcoma, as has been reported inseveral studies [12-14, 20-22] and they clearly indicatethe need of a differentiated and effective 'salvage chemo-therapy' for poor-responder patients.

The use of ifosfamide in the postoperative treatmentof poor-responder patients can be considered.

In fact, in our second study of neoadjuvant chemo-therapy for non-metastatic osteosarcoma of the extremity[23], the post-operative use of ifosfamide in patients whorespond poorly to a primary chemotherapy based onHDMTX, CDP and ADM allowed them the sameprobabilty of DFS as that of good-responder patients(five-year CDFS: GR 71%, PR 57%); similar results havebeen reported by Benjamin et al. [24].

The baseline serum level of LDH showed an inde-pendent prognostic significance for DFS (P = 0.0013):patients with normal LDH levels had 55% DFS, whereasthose with higher levels had 29% DFS.

Similar results have been reported by Meyers et al.[22] and Link et al. [7].

SAP level was a significant predictive factor for DFSat the univariate analysis only, but it lost predictivesignificance after multivariate analysis (Table 2). Thesedata are in contrast to those reported in the MSKCCexperience [22].

Other factors such as sex, tumor site and tumor size,which were considered significant predictive factors forsurvival by several authors [25—27] showed no signifi-cant importance in our study.

The long-term follow-up of our study populationshowed a sort of biphasic pattern of relapse.

The curve of DFS was characterized by a first phasecorresponding to the first two years after the start oftreatment, with a rapid decrease of probability of DFS(two-year DFS 54%) followed by a second phase inwhich the DFS curve slowly decreased to a plateaureached by the eight-year (four-year DFS 48%, six-yearDFS 47%, eight- and 12-year DFS 46%).

A similar pattern of relapse was described by theScandinavian Sarcoma Group [28] and CooperativeOsteosarcoma Study [29]: with a median follow-up of7.75 years, the plateau of DFS was reached at the eigthyear of observation, whereas in the Memorial Sloan-Kettering experience [22] the latest recurrence occurredin the ninth year of observation with the same follow-upof the previous studies.

Thus, there are possibilities of late relapse in non-metastatic osteosarcoma; patients must be informed ofthis, and a long-term follow-up recommended.

Moreover, the possibility of late relapse has to beconsidered in evaluating the early results of new chemo-therapy regimens.

The post-relapse survival was very poor, with two-and 10-year survivals of 19% and 13%, respectively.

A 25% two-year post-relapse survival was reportedfor the patients of the chemotherapy arm of the Multi-Institutional Osteosarcoma Study [7].

Saeter et al. [28] reported a projected five-year survivalrate of 24% from the first metastatic event in a popula-tion of 60 recurrent osteosarcoma; patients with localrecurrence were excluded from this analysis.

Three factors appeared to influence post-relapse sur-vival: the site of first recurrence, the possibility of appro-priate surgical treatment and the relapse-free interval.

None of patients who relapsed with local recurrenceor with metastases other than lung, survived.

The same results were reported by Ward et al. [30].In our study, a 30% 10-year survival was obtained in

patients who could undergo appropriate surgical removalof lung metastases.

These results are similar to those reported in otherstudies: Beron et al. [31] reported a five-year survival rateof 36% for 22 patients who underwent complete meta-stasectomy. Huth et al. [32] reported a 23% four-yearsurvival in patients who underwent metastasectomy,whereas none of the patients who were unable to under-go surgery survived. Ward et al. [30] reported a 23% five-year survival in patients who relapsed with lung metas-tasis only.

In our study, a relapse-free interval longer than 24months was an important prognostic factor for post-relapse survival (RFI ^ 24 months: eight-year survival40%, RFI <24 months: eight-year survival 7%, P =0.0159).

In the SSG experience [28] a relapse-free intervallonger than 21 months proved significant for survival.Ward et al. [30] found a better survival rate in patientswith a relapse-free intervals longer than eight months.

The role of chemotherapy in relapsing patients withosteosarcoma has not been defined.

In our study, a further chemotherapy treatment(ADM, HDMTX and BCD) was administered in early-relapsing patients without improvement of their prog-nosis; however it must be stressed that most of thesepatients could not be treated with surgery.

Saeter et al. [28] found that an 'adequate' chemo-therapy was a significant prognostic factor for survival,whereas other authors [7, 30, 32] were not able to demon-strate an influence of chemotherapy on post-relapsesurvival.

A promising activity of high-dose ifosfamide in re-current osteosarcoma has recently been reported [33,34]; immunotherapy [35] or endocrine therapy [36] couldbe novel and interesting modalities of treatment forpatients with metastatic osteosarcoma, nonetheless anadequate surgical removal of metastases is a necessarycondition for survival.

771

Acknowledgements

Supported by Rizzoli Research and CNR ('ACRO'project) funds.

The Authors thank Dr. Michela Versari for assistancein preparing the manuscript and Dr. Sandra Sottili forthestatistical analysis

References

1. Link MP, Goorin AM, Miser AW et al. The effect of adjuvantchemotherapy on relapse-free survival in patients with osteosar-coma of the extremity. N Engl J Med 1986; 314: 1600-6.

2. Eilber F, Giuliano A, Eckardt J et al. Adjuvant chemotherapy forosteosarcoma: A randomized prospective trial. J Clin Oncol 1987;5:21-6.

3. Rosen G, Marcove RC, Caparros B et al Primary osteogenicsarcoma: The rationale for preoperative chemotherapy and de-layed surgery. Cancer 1979; 43: 2163-77.

4. Jaffe N, Robertson R.Takue Yet al. Control of primary osteosar-coma with chemotherapy. Cancer 1985; 56: 461-6.

5. Winkler K, Beron G, Kotz R et al. Neoadjuvant chemotherapyfor osteogenic sarcoma: Results of a cooperative German/Aus-trian study. J Clin Oncol 1984; 2: 617-24.

6. Campanacci M, Bacci G, Bertoni F et al. The treatment ofosteosarcoma of the extremity: Twenty years' experience at theIstituto Ortopedico Rizzoli. Cancer 1981; 48: 1456-61.

7. Link MP, Goorin AM, Horowitz M et al. Adjuvant chemotherapyof high-grade osteosarcoma of the extremity. Updated results ofthe Multi-Institutional Osteosarcoma Study. Clin Orthop 1991,270: 8-14

8. Bacci G, Picci P, Ruggien P et al. Primary chemotherapy anddelayed surgery (neoadjuvant chemotherapy) for osteosarcoma ofthe extremity. The Istituto Rizzoli experience in 127 patientstreated preoperatively with intravenous methotrexate (high versusmoderate doses) and i.a. cisplatin. Cancer 1990; 65: 2539-53.

9. Enneking WF, Spanier SS, Goodman MA. Current conceptsreview. The surgical staging of musculoskeletal sarcoma. J Boneand Joint Sur 1980; 62A: 1027-30.

10. Picci P, Bacci G, Campanacci M et al. Histologic evaluation ofnecrosis in osteosarcoma mduced by chemotherapy. Regionalmapping of viable and nonviable tumor. Cancer 1985; 56: 1515-21.

11. Rosen G, Caparros B, Huvos AG et al. Preoperative chemo-therapy for osteogenic sarcoma. Selection of postoperative adju-vant chemotherapy based on the response of the primary tumorto preoperative chemotherapy. Cancer 1982; 49: 1221-30.

12. Provisor AJ, Ettinger LJ, Nachman JB et al. Treatment of non-metastatic osteosarcoma of the extremity with preoperative andpostoperative chemotherapy: A report from the Children's Can-cer Group. J Clin Oncol 1997; 15: 76-84.

13. Saeter G, Alvegard TA, Elomaa I et al. Treatment of osteosarco-ma of the extremities with theT-10 Protocol, with emphasis on theeffects of preoperative chemotherapy with single agent high dosemethotrexate: A Scandinavian Sarcoma Group study. J ClinOncol 1991; 9: 1766-75.

14. Wtnlcler K, Beron G, Delling G et al. Neoadjuvant chemotherapyof osteosarcoma: Results of a randomized cooperative trial(COSS-82) with salvage chemotherapy based on histologicaltumor response. J Clin Oncol 1988; 6: 329-37.

15. Bacci G, Ferrari S, Picci P et al. Methotrexate serum concentra-tion and histological response to multiagent primary chemother-apy for osteosarcoma of the limbs. J Chemotherapy 1996; 8: 472-8.

16. Ferrari S, Sassoli V, Orlandi M et al. Serum methotrexate con-centration and prognosis in patients with osteosarcoma ot theextremities treated with a multidrug neoadjuvant regimen. JChemother 1993; 5: 135-41.

17. Graf N, Winkler K, Betlemovic M et al. Methotrexate pharmaco-kinetics and prognosis in osteosarcoma. J Clin Oncol 1994; 12:1443-51.

18. Delepme N, Delcpine G, Corneille H et al. Dose escalation withpharmacokinetics monitoring in methotrexate chemotherapy ofosteosarcoma. Anticancer Res 1995; 15: 489-94.

19. Delepine N, Delepine G, Bacci G et al. Influence of methotrexatedose intensity on outcome of patients with high grade osteogenicosteosarcoma. Cancer 1996; 78: 2127-35.

20. Bramwell VHC, Burgers M, Sneath R et al. A comparison oftwo short intensive-adjuvant chemotherapy regimens in operableosteosarcoma of limbs in children and young adults: The firststudy of the European Osteosarcoma Intergroup. J Clin Oncol1992; 10: 1579-91.

21. Souhami RL, Nooy M, van der Eijken J et al. A randomized trialof two regimens of chemotherapy in operable osteosarcoma: Astudy of the European Osteosarcoma Intergroup (EOI) Proc AmSoc Clm Oncol 1996; 15: 520 (Abstr 1676).

22. Meyers PA, Heller G, Healey J et al. Chemotherapy for non-metastatic osteogenic sarcoma: The Memorial Sloan-KetteringExperience. J Clin Oncol, 1992; 10: 5-15.

23. Bacci G, Picci P, Ferrari S et al. Primary chemotherapy anddelayed surgery for non metastatic osteosarcoma of the extremity.Results in 164 patients preoperatively treated with high doses ofmethotrexate, followed by cisplatin and doxorubicin. Cancer1993; 72: 1216-26.

24. Benjamin RS, Patel SR, Armen Tet al. The value of ifosfamide inpostoperative neoadjuvant chemotherapy of osteosarcoma. ProcAm Soc Clin Oncol 1995; 14: 516 (Abstr 1690).

25. Petrilh AS, Gentil FC, Epelman S et al. Increased survival, limbpreservation and prognostic factors for osteosarcoma. Cancer1991; 68: 733-7.

26. Glasser DB, Lane JM, Huvos AG et al. Survival, prognosis, andtherapeutic response in osteogenic sarcoma. Cancer 1992; 69:698-708.

27. Bieling P, Rehan N, Winkler P et al. Tumor size and prognosis inaggressively treated osteosarcoma. J Clin Oncol 1996; 14: 848-58.

28. Saeter G, Hoie J, Stenwig AE et al. Systemic relapse of patientswith osteogenic sarcoma. Prognostic factors for long term surviv-al. Cancer 1995; 75: 1084-93.

29. Winkler K, Bielmg P, Bielack S. Die Chemotherapie des osteo-sarkoms. Z Orthop 1992; 130: 285-9.

30. Ward WG, Mikaehan K, Dorey F et al. Pulmonary metastases ofstage IIB extremity osteosarcoma and subsequent pulmonarymetastases. J Clin Oncol 1994; 12: 1849-58.

31. Beron G, Winkler K, Beck J et al. Prognosis after metastasis inosteosarcoma: experience from the COSS studies Contr Oncol1988; 30r 143-9.

32. Huth JF, Eilber FR. Patterns of recurrence after resection ofosteosarcoma of the extremity. Arch Surg 1989; 124: 122-6.

33. Chawla SP, Rosen G, Lowenbraun S et al. Role of high-doseifosphamide (HDI) in recurrent osteosarcoma. Proc Am Soc ClinOncol 1990; 9: 310 (Abstr 1201).

34. Picci P, Bacci G, Ferrari S et al. Salvage treatment with high-doseifosfamide and surgery for osteosarcoma patients relapsed withlung metastases. Preliminary results. Proc Am Soc Clin Oncol1996; 15: 459 (Abstr 1438).

35. Kleinerman ES, Jia SF, Griffin J et al. Phase II study of Lip-osomal Muramyl Tripeptide in osteosarcoma: the cytokine cas-cade and monocyte activation following administration. J ClinOncol 1992; 10: 1310-6.

36. Sekyi-Otu A, Bell R, Andrulis I et al. Insulin-like growth factor I(IGF-I) responsivity of human sarcomas: Correlation of IGF-Ireceptor gene expression measured by RT-PCr with IGF-I bind-ing capacity.Proc Am Soc Clin Oncol 1993; 12: 121 (Abstr 282).

Received 18 March 1997; accepted 18 June 1997.

Correspondence to:Dr. Stefano FerrariSezione di ChemioterapiaIstituto Ortopedico RizzoliVia Pupilli 140136 Bologna, Italy