The authors present the long-term follow-up (>25 years) data from 1 of the original prospective, randomized trials that compared adjuvant chemotherapy with expectant management in patients with high-grade, localized osteosarcoma. In addition, the value of pathologic necrosis induced by a single cycle of neoadjuvant chemotherapy was analyzed as a predictive marker of disease-free and overall survival.
Fifty-nine patients with high-grade, localized osteosarcoma were enrolled in a prospective trial that was performed between 1981 and 1984 at the University of California-Los Angeles (UCLA). Patients were randomized to receive either adjuvant chemotherapy or observation after surgical resection. Long-term outcomes, follow-up, and pathologic review of all available histologic sections were performed.
The 25-year disease-free survival rate was 28% for patients who received adjuvant chemotherapy compared with 15% for the untreated patients (P = .02). The overall survival rate at 25 years was also significantly higher for treated patients versus untreated patients (38% vs 15%; P = .02). Tumor necrosis >90% after a single round of chemotherapy was a statistically significant predictor of overall survival and disease-free survival for patients who received adjuvant therapy (164 months vs 65 months [P = .04] and 141 months vs 14 months [P < .01], respectively).
Several landmark studies during the 1970s and 1980s demonstrated the efficacy of adjuvant chemotherapy in improving the survival of patients with osteosarcoma. Before that, the overall 5-year survival rate for patients with localized, high-grade osteosarcoma was <20%.1 Today, with modern multimodality therapy combining systemic chemotherapy and advanced surgical resection, the cure rate for this patient population exceeds 70%.2
One of these landmark studies was the prospective, randomized trial performed by Eilber et al from 1981 to 1984 that compared adjuvant chemotherapy with no treatment after resection in patients with high-grade, operable osteosarcoma.3 That study was discontinued after 2 years, because those receiving immediate adjuvant chemotherapy had a statistically significant improvement in disease free-survival (DFS) (55% vs 20%; P < .01). In addition, the study also provided evidence that adjuvant chemotherapy improves patient survival (overall survival (OS) (80% vs 40%; P<.01). Although short-term DFS improvement with adjuvant chemotherapy was demonstrated in the original study, maintenance of the survival benefit has not been reported. In the current study, we provide the 25 year follow-up on the original randomized cohort from the study by Eilber et al in order to better delineate the incidence, natural history, and risk factors predictive of relapse and death in these patients with osteosarcoma.
MATERIALS AND METHODS
Patients and Treatment
One hundred twelve patients with primary bone tumors were evaluated between 1981 and 1984, and 78 were determined to have disease pathologically consistent with osteosarcoma. Nineteen were excluded because they had radiographic evidence of metastatic disease on presentation, low-grade osteosarcoma, or osteosarcoma associated with Paget disease. After informed consent was obtained, the 59 patients with stage IIA/IIB, intramedullary, high-grade osteosarcoma were enrolled onto the trial. Thirty-two patients were randomized to receive adjuvant chemotherapy, and 27 were randomized to no adjuvant therapy (Fig. 1). Age, sex, primary tumor site, and mean follow-up for each group are listed (Table 1).
Table 1. Demographics of the Treatment and Control Groups
No. of Patients (%)
Abbreviations: T-10B, high-dose methotrexate and doxorubicin plus combined bleomycin, cyclophosphamide, and actinomycin-D.
Age: Median [range], y
Procedure, extremity lesion, N = 55
Initial treatment for all patients included a single round of chemotherapy and radiation, as described in the original article.3 Four weeks after the completion of radiation, the patients underwent surgical excision of the tumor. The decision to proceed with limb salvage versus amputation was based on the surgical oncologic principles of the time: tumors of the tibia, large size, or in patients aged <10 years were considered indications for amputation. Forty-four patients underwent limb salvage, and 15 patients underwent amputation.
Patients who were randomized to receive adjuvant chemotherapy were treated on the T-10B protocol (high-dose methotrexate and doxorubicin plus combined bleomycin, cyclophosphamide, and actinomycin-D),4 as described in the original report.3 Patients in the treatment group who developed a recurrence received salvage chemotherapy with doxorubicin 60 mg/m2 and cisplatin 120 mg/m2 every 28 days for 5 months. Surgical resection of tumor recurrence and/or pulmonary metastases were performed after the last course of chemotherapy.
Patients who developed recurrent osteosarcoma in the control arm (no adjuvant chemotherapy) received the T-10B protocol. Patients who had resectable tumors after chemotherapy underwent surgical resection of the metastases, and those who had unresectable disease had their chemotherapy changed to the doxorubicin and cisplatin salvage protocol outlined above.
Twenty-four of 27 patients who were randomized to the control arm did not receive postoperative chemotherapy; whereas 3 patients crossed over and received some postoperative chemotherapy, but none received the T-10B protocol. Twenty-eight of the 32 patients who were randomized to receive the T-10 protocol received the entire course. The 4 remaining patients who did not receive the entire course refused chemotherapy at some stage during the study.
Statistical analysis was performed using standard Kaplan-Meier survival analysis in an intent-to-treat format. An as-treated analysis was not performed. Univariate Cox regression analysis was performed to evaluate the effect of chemotherapy, age, sex, surgical procedure (amputation vs limb salvage), and location of primary tumor (femur vs other location).
Pathologic analysis was performed by a single, senior fellowship-trained musculoskeletal pathologist (S.D.N.). Tumor necrosis was assessed relative to the percentage of residual viable tumor in a well described manner consistent with that used to grade bone sarcomas.5, 6
With a mean follow-up of 2 years, the patients who received immediate adjuvant chemotherapy had a statistically significant improvement in DFS compared with those who did not receive chemotherapy (55% vs 20%; P < .01). Patients who received adjuvant chemotherapy also had a statistically significant improvement in OS (80% vs 48%; P < .01).
The median follow-up for surviving patients was 303 months (range, 206-325 months). The DFS rate for the T-10B adjuvant treatment group was 28% compared with 15% for the patients who did not receive adjuvant therapy (P = .018; log-rank test) (Fig. 2). The 25-year OS rate for patients in the T-10B arm was 38% compared with 15% for those who did not receive adjuvant chemotherapy (P = .023; log-rank test) (Fig. 3).
Two of 20 patients (10%) who developed metastatic disease or a recurrence in the T-10B group and 1 of 24 patients (4%) with metastatic disease or recurrence in the no adjuvant therapy arm were cured. The mean time to relapse was longer in the group that received adjuvant therapy compared with the no adjuvant therapy group, although the difference did not reach statistical significance (11 months vs 7 months; P = .13). For all patients who died of disease, the mean time to death was similar in both groups: 32 months in the T-10B group (median, 23 months; range, 13-102 months) and 33 months in the no adjuvant therapy group (median, 18 months; range, 4-149 months) (Table 2).
Table 2. Survival Outcomes for Treatment and Control Groups
T-10B, N = 32
Control, N = 27
Abbreviations: HIV/AIDS, human immunodeficiency virus/acquired immunodeficiency syndrome; SD, standard deviation; T-10B, high-dose methotrexate and doxorubicin plus combined bleomycin, cyclophosphamide, and actinomycin-D.
No. of patients cured/total no. (%)
Time to death, mo
Mean ± SD
32 ± 24
33 ± 40
No. of patients cured of metastatic or recurrent disease/total (%)
No. of patients who died of causes other than disease
1: Complications relating to transfusion-acquired HIV/AIDS
Time after surgery to relapse/metastatic disease, mo
Mean ± SD
14 ± 12
7 ± 7
Univariate analysis revealed that at the 5-year, 10-year, and 15-year time points; age; sex; and the procedure performed were not significant predictors of disease-specific survival. The addition of adjuvant chemotherapy, however, did lead to a significant improvement in disease-specific survival at all time points (Table 3). Disease-free survival curves for both groups up to 25 years are provided in Figure 2. It is noteworthy that no patients in either group had a recurrence at >5 years after his or her index surgery.
Table 3. Univariate Analysis of Predictors of Disease-Specific Survival
Total No./ No. of Events
Abbreviations: DSS, disease-specific survival; T-10B, high-dose methotrexate and doxorubicin plus combined bleomycin, cyclophosphamide, and actinomycin-D.
One patient was excluded from the analysis as DOO = Dies of Other Causes.
Histologic analysis of tumor necrosis was performed on specimens from 46 of the 59 patients (78%). The slides from 13 patients were not available for review, because they either were returned to the patient's home medical institution or were damaged in the Northridge earthquake of 1994. All histologic specimens were from the original resection, which was performed after a single round of chemotherapy.
For all 46 patients, tumor necrosis was not significantly predictive of survival. Patients who had tumor necrosis <90% survived for a mean of 97 months (range, 4-304 months), whereas patients who had tumor necrosis >90% survived for a mean of 114 months (range, 6-325 months; P = .64). However, when patients were stratified into adjuvant and no adjuvant arms, tumor necrosis >90% was a statistically significant predictor of both OS and DFS for the patients who received adjuvant therapy (OS: 164 months vs 65 months; P = 004; DFS: 141 months vs 14 months; P < .01).
Although adjuvant chemotherapy currently is a mainstay of our armamentarium to combat osteosarcoma, it was not always so. Several landmark studies, including that by Eilber et al and the Multi-Institutional Osteosarcoma Study (MIOS),7 indelibly changed treatment for osteosarcoma, because those investigators prospectively evaluated adjuvant chemotherapy in a randomized fashion among patients with localized, high-grade disease. However, many of those studies, including the studies noted above, were discontinued at relatively short follow-up, because statistically significant treatment benefits of chemotherapy were noted during predesigned, early data analyses. The survival benefit over a long period (ie, decades) was assumed from these studies, but never demonstrated. Moreover, a recent study by Berend et al suggests that the conclusions drawn from the Eilber et al and MIOS studies may be outdated, and postoperative adjuvant chemotherapy may have no survival benefit after neoadjuvant therapy and resection.8 It is in this context that analyzing the 25-year follow-up from the original cohort was performed to answer the question of whether survival benefits of adjuvant chemotherapy are maintained over time.
At a median follow-up of >25 years, the current study indicates that improvements in DFS and OS in patients with osteosarcoma who receive adjuvant chemotherapy are maintained with statistical significance. Although the original study was hindered by short follow-up, this study confirms that the survival benefit in this cohort is maintained over 25 years. Although the intellectual concept from the Berend et al study is compelling—aggressive preoperative chemotherapy should eradicate sensitive micrometastases and wide margins should clear the local tumor burden—our investigation suggests that adjuvant chemotherapy provided a sustained, statistically significant survival benefit despite preoperative chemotherapy and radiation.
An additional piece of information gleaned from this cohort is that no patients developed a tumor recurrence 5 years after initial resection. Although other studies have reported local recurrences up to 17 years after resection,9 our data suggest that local recurrence is at least a rare event after 5 years. However, this does not obviate the need for continued regular follow-up and surveillance after 5 years. Previous reports have indicated that 5% to 10% of patients with osteosarcoma who have survived at 5 years will experience a recurrence or a second malignant neoplasm after 5 years of follow-up.10, 11 Therefore, it is our institutional practice to follow these patients annually in perpetuity.
Although we conclude that adjuvant chemotherapy had a sustained survival benefit in this cohort, the cohort does not help address the issue regarding whether neoadjuvant therapy added to adjuvant therapy has a survival benefit over adjuvant therapy alone. Recent literature has suggested that, in patients who underwent surgical resection and received adjuvant chemotherapy, there was no difference in event-free survival among patients who received neoadjuvant therapy before resection.12 Because all of the patients in this cohort received neoadjuvant therapy, the study population is not suited to provide any insights into this important finding, and we will rely on future studies to address this question.
The current study confirms the well known correlation between survival expectancy and tumor pathologic necrosis. In this trial, pathologic necrosis was assessed after a single cycle of chemotherapy and preoperative radiotherapy, in contrast to current practice, in which neoadjuvant chemotherapy is administered for 12 weeks before patients undergo surgery with the objective of local control. This suggests that response assessment could be made earlier, perhaps stratifying patients into good or poor responder categories much earlier. Although our group's more recent work has used combined positron emission tomography/comupted tomography (PET/CT) imaging to noninvasively assess early response,13 this study was structured so that histologic data would be available after a single dose of chemotherapy and radiation. The confirmation that early good responders (tumor necrosis >90%) have improved survival if chemotherapy is continued corroborates other clinical trials and our data with combined PET/CT scanning suggest that good responders will continue to do well with adjuvant chemotherapy. The finding that tumor necrosis was not predictive of survival if chemotherapy was discontinued confirms the belief that early response may be of value to tailor on-going treatment, and not to dictate the length of treatment. Identifying responders to treatment and continuing therapy based on this response remains a fundamental tenet of our treatment protocol for adult soft tissue sarcomas. Whether this is applicable to children, adolescents, and adults with osteosarcoma remains to be determined. Assessing how to manage nonresponders with alterations in treatment protocol is an area that continues to challenge us and requires further investigation. However, this study of early pathologic response to therapy contributes to the increasing body of evidence that responses to therapy can be evaluated much earlier in the treatment course. Future research hopefully will identify molecular markers in tumors that will distinguish responders from nonresponders before the initiation of treatment.
This study has several weaknesses that must be addressed. First, we acknowledge that the neoadjuvant chemotherapy used today has changed substantially from the regimen used in the study cohort, which may temper the applicability of conclusions drawn from these data. In fact, literature from the European Osteosarcoma Intergroup has suggests that a short, intensive course of doxorubicin and cisplatin may be as effective as the polydrug T10 regimen.14 Because the neoadjuvant regimen used in this study may not be as effective as current regimens, the adjuvant therapy in this cohort may demonstrate a distortedly large survival benefit. The current standard of care for osteosarcoma includes high-dose methotrexate (12 g/m2), rather than the low-dose that the patients in this study received, as well as the abandoning of bleomycin, cyclophosphamide, actinomycin-D, and vincristine because of unproven efficacy. This may explain the overall inferior survival in this cohort compared with the approximately 70% expected long-term survival for patients with localized osteosarcoma who received treatment today.
The current study has another weakness, in that 17 of 23 patients in the control group ended up receiving chemotherapy for their recurrence. This may have affected our ability to interpret the long-term results of the study as a demonstration of the efficacy of chemotherapy, and it offers an alternative conclusion suggesting the efficacy of early chemotherapy versus late chemotherapy (after recurrence). Interpretation of the results in that manner would serve as a validation of the work by Link,15 who demonstrated an OS overall survival benefit from chemotherapy at initial presentation as opposed to delaying chemotherapy until recurrence. Thus, the current study indicates that the survival benefit from the initiation of chemotherapy at first presentation of osteosarcoma is maintained at 25 years.
In addition, in this study, we did not evaluate tumor necrosis at the time of surgery. Although tumor necrosis has been well established as a prognostic marker,16 a survival benefit from intensifying or changing therapy based on tumor necrosis has not yet been proven. It is our hope that randomized, international, multicenter, collaborative trials, such as the European and American Osteosarcoma Study Group (EURAMOS) EURAMOS1 trial, which randomizes patients with osteosarcoma who have poor tumor necrosis to receive the addition of intensified therapy with ifosfamide and etoposide to the standard backbone arm of high-dose methotrexate, doxorubicin, and cisplatin, will answer this question.
In conclusion, the current results suggest that, in patients with high-grade, localized osteosarcoma who receive neoadjuvant chemotherapy and undergo surgical resection, adjuvant chemotherapy carries a statistically significant OS and DFS benefit out to 25 years. Tumor necrosis after just 1 round of chemotherapy and preoperative radiation was predictive of OS and DFS in patients who went on to continue receiving chemotherapy. Although prospective, randomized trials undoubtedly should form the backbone of our scientific understanding of the role of chemotherapy in osteosarcoma, a long-term, retrospective analysis like this provides key insight into the durability of survival benefits over time. A sustained survival benefit over 25 years provides strong evidence for the importance of adjuvant chemotherapy in our treatment of osteosarcoma.