Incidence, epidemiology and treatment results of osteosarcoma in Finland – a nationwide population-based study

Abstract

Background. Patients diagnosed with osteosarcoma in Finland during 1991–2005 were retrospectively analyzed in a nationwide, population-based study. We focused on the incidence, treatment and outcome of osteosarcoma patients. We also evaluated the value of known prognostic parameters. Material and methods. Osteosarcomas were retrieved from the files of the national Finnish Cancer Registry. Only patients with histologically confirmed osteosarcoma were included in the analysis. Histological review was performed. Results. The study consists of 144 osteosarcoma patients with a mean follow-up of 9.8 years for survivors. Mean annual incidence of histologically confirmed osteosarcoma was 1.8 new osteosarcomas per million. The 10-year sarcoma-specific survival for the whole population was 63% and 73% for patients with local disease at presentation. Overall limb-salvage rate was 73% and local control was 84% for patients with a peripheral tumor. Development of local recurrence and major deviation from the chemotherapy protocol were significant adverse factors for sarcoma-specific survival in multivariate analysis. Conclusion. The present nationwide and population-based study is our second report of treatment and prognosis of osteosarcoma in Finland. With modern chemotherapy the prognosis of local osteosarcoma has improved in Finland from 47% during 1971–1980 and 65% during 1981–1990 at five years to the present 73% during 1991–2005 at 10 years. The 10-year sarcoma-specific survival of 73% is excellent and comparable to results reported with contemporary treatment protocols in high-volume centers. However, improvement in limb-salvage rate and local control probably requires centralization of treatment of this rare disease.

Background

Osteosarcoma, the most frequent primary malignant bone tumor, is a tumor of the adolescence as approximately 60% of tumors occur in patients under 20 years of age. Patients older than 60 years form approximately 10% of the patient population [1,2].

Prognosis of non-metastatic osteosarcoma has improved dramatically since the invention of preoperative and postoperative chemotherapy in the 1970s. Currently the five-year overall survival rate is 75% to 77% for patients diagnosed with primary non-metastatic extremity osteosarcoma [3,4]. A nationwide osteosarcoma study from Norway with a non-selected study population including also metastatic disease at presentation and all primary tumor locations showed a 48-month survival of 47% during 1975–1977 and 25% during 1953–1974 [5].

For the present study, data for patients diagnosed with osteosarcoma in Finland during 1991–2005 were retrieved from the records of the Finnish Cancer Registry. The most important aim of was to study survival over study period and how the surgical and oncological goals were achieved in a nationwide and population-based material. Further, our aim was to describe the clinical features of osteosarcoma at presentation and their prognostic significance, the accuracy of histologic diagnosis and classification. Finally, we calculated the annual population-based incidence of osteosarcoma in Finland.

Material and methods

Patient data

Present series consists of 176 patients diagnosed with osteosarcoma in Finland during 1991–2005, retrieved from the files of the nationwide population-based Finnish Cancer Registry. The Finnish Cancer Registry was founded in 1952 and covers the whole population of Finland with approximately 5.4 million inhabitants. Since 1961, the reporting of newly diagnosed cancers to the Cancer Registry has been compulsory. The registry covers more than 99% of the solid tumors diagnosed in Finland [6]. Detailed clinical data was collected from the patient files. During the study period surgical treatment for paediatric patients was performed in four of five university hospitals and chemotherapy was administered in all five. Adult patients received surgical treatment and chemotherapy in all five university hospitals. The study was approved by the Joint Ethics Committee of Helsinki University Central Hospital and by the Ministry of Health and Social Affairs.

Histology

Original haematoxylin and eosin stained slides of biopsy specimens could be retrieved for histological re-evaluation for 162 (92.0%) of the 176 patients reported to Cancer Registry with osteosarcoma during 1991–2005. These 14 patients, for whom histological re-evaluation of the tumors was not possible due to missing specimens, were significantly older at diagnosis than the patients with re-evaluated tumors: mean (SD) 70.0 (16.7) years vs. 29.0 (18.7) years (p < 0.001). Furthermore, non-evaluated tumors were more often located axially, 64.3% vs. 29.6% (p = 0.008). Two of these patients were treated with curative intention: one patient has been disease-free for 15.5 years after a rib resection. The other patient with a maxillary tumor was treated with a wide resection and radiation therapy and has been disease-free for 8.4 years. Twelve patients were treated with palliative intention and they all died of the disease with the mean survival of four months (range 0.5–14 months).

The histological re-evaluation was performed by an experienced bone pathologist (T.B.) without any knowledge of either the previous pathologic assessment or of the patient's outcome. We excluded false histological diagnosis (eight), extraskeletal osteosarcoma (eight) or with a postmortem diagnosis only (two) leaving 144 patients (28 with metastatic disease at diagnosis) for final analysis. Tumor subtypes were classified according to present WHO criteria [1] and graded histologically according to the system described by Huvos [2]. Tumors were graded as low grade or high grade. Histologic response to neoadjuvant chemotherapy was recorded from the original pathology records and not re-evaluated. Histologic response was classified as good or poor according to the protocol in question.

Radiology and imaging

Due to the long time span studied, evolution of magnetic resonance imaging (MRI) techniques and the numerous radiology departments involved in imaging was not uniform and estimation of tumor volume not feasible. The presence of skip metastases was evaluated. Due to the heterogeneity of the imaging material − some available only on films − assessment of neurovascular involvement or tumor necrosis was not conducted. Chest computed tomography (CT)-scan and x-ray, bone scan, MRI or CT of the tumor area and x-ray of the entire bone involved were used for staging according to treatment protocols. Body CT or MRI was used for patients with an axial tumor. Shrinkage or growth of more than 20% of the soft tissue component was considered significant in the analysis of radiological response to neoadjuvant treatment.

Surgery

Tumor biopsies were taken surgically in the early part of the series and later with CT-guidance. Surgery was usually performed after three months of neoadjuvant chemotherapy. The aim was to resect the tumor with wide margins. Surgical margins were acquired from pathology reports and classified according to Enneking [7]. Margins were not re-evaluated in the present study. The biopsy tract was resected at surgery. The reconstructions were mainly performed with megaprostheses. Solitary pulmonary metastases were diagnosed with a chest CT-scan and were operated after chemotherapy.

Chemotherapy

Four trials were conducted during the study period and the number in parentheses indicates Finnish patients enrolled. From 1990 onwards Scandinavian Sarcoma Group (SSG) enrolled patients with non-metastatic osteosarcoma into the SSG VIII study (20 patients) [3]. The Italian/Scandinavian study ISG/SSG I study for patients with non-metastatic osteosarcoma was launched in 1997 (16 patients) [4]. The Italian/Scandinavian study ISG/SSG I included also a recommendation for the treatment of a first metastatic relapse [8]. Otherwise chemotherapy for metastatic relapse was heterogeneous. The ISG/SSG II study from 1998 on was for high- risk osteosarcoma patients with pelvic or metastatic disease at presentation (six patients, two with pelvic primaries and four with metastatic disease) [9]. The SSG enrolled classical osteosarcomas from 2001 to 2005 in the SSGXIV study (three patients) [10]. For the present study, major deviation from the chemotherapy protocol was arbitrarily defined as more than 25% of drug dose omitted.

Statistical methods

Metastasis-free survival (MFS), local recurrence- free survival (LRFS) and sarcoma-specific survival (SSS) were calculated according to Kaplan-Meier method. Differences in MFS and SSS rates of different subgroups were analyzed with the log-rank test for discrete variables and with Cox regression analysis for continuous variables. The level of significance was set at p < 0.05. P-values are two-sided. If the univariable test showed correlation between a descriptive variable and survival, then this variable was included in a backwards stepwise Cox proportional model for multivariable analysis. The PASW^® Statistics version 18 for Windows was used for all analyses.

Results

Clinicopathologic data

The study material consists of 144 patients with osteosarcoma with the mean follow-up of 9.8 years for survivors. This yields the mean annual incidence of 9.7, i.e. 1.8 osteosarcomas per million. In the reclassification the osteoblastic subtype was the most common with 86 tumors (Table I). In 14 tumors subclassification was not possible. Four of these were postirradiation osteosarcomas. Five were situated in skull bones with both osteoid and chondroid formation. Five other tumors were clearly malignant osteoid producing tumors but shortage of sample material prevented subclassification.

Table I. Histological review of the 144 osteosarcomas included in the analysis by five-year periods.

	1991–1995	1996–2000	2001–2005	Total
Conventional	34	43	28	107
Osteoblastic	31	35	20	86
Chondroblastic	2	6	7	15
Fibroblastic	1	2	1	4
Telangiectatic	0	3	0	3
Parosteal	2	6	0	8
Periosteal	5	4	0	9
Low-grade central	1	2	1	4
Small cell	1	0	1	2
Not otherwise specified	3	6	4	13
Total	46	64	34	144

Increasing percentage of patients older than 40 years were diagnosed during the five-year time periods and the percentage of female patients also increased during the five-year time periods (Table II). Demographic data are depicted in Table III. Nine patients had a predisposing factor: five patients had a postirradiation sarcoma. Li- Fraumeni syndrome was present in two patients, RAPADILINO syndrome and retinoblastoma syndrome was each present in one patient, respectively. Fifty-six (39%) of the 144 tumors were located in the femur, 26 (18%) in the tibia and 20 (14%) in the humerus.

Table II. Demographic data by five-year time periods.

	1991–1995	1996–2000	2001–2005	p^a
Sex (M/F)	1.875	1.286	0.619	0.054
Age, mean (percentage over 40 years)	24 (10.9)	24 (17.2)	31 (32.4)	0.204 (0.020)
Axial/peripheral	0.314	0.280	0.545	0.332
Percentage of metastatic disease at diagnosis	19.6	15.6	26.5	0.434
Percentage of low-grade tumors	15.2	14.1	8.8	0.750

^aBy χ² test.

Table III. Description of tumour, patient and treatment characteristics of the 116 patients with local disease, and the corresponding estimates of 10-year metastases-free and sarcoma-specific survival.

Characteristics	No of patients	10-Year SSS	p^f	95% CI	10-Year MFS	p^f	95% CI
Time of diagnosis
1991–1995	37	64%	0.230	0.439–1.219	59%	0.048	0.36–0.99
1995–2000	54	76%			75%
2001–2005	25	80%			83%
Axial vs. peripheral
Axial	29	65%	0.144	0.27–1.22	72%	0.932	0.45–2.39
Peripheral	87	75%			71%
Sex
Male	60	68%	0.302	0.33–1.41	64%	0.052	0.24–1.02
Female	56	78%			79%
Age at diagnosis
Mean 25.5 years			0.078^g	1.00–1.04		0.976^g	0.98–1.02
Range 7.8–77.1 years
Tumor grade
Low	18	94%	0.07	0.95–3.57
High	98	69%
Extent of surgery^a
Limb-sparing	68	77%	0.527	0.26–1.99	72%	0.512	0.29–1.84
Amputation	18	72%			72%
Margin status^b
Intralesional	2	50%	0.667	0.41–1.78	100%	0.895	0.55–1.99
Marginal	76	75%			71%
Wide	28	86%			82%
Radical	4	50%			50%
Major deviation from chemotherapy protocol^c
Present	16	43%	0.009	1.28–7.77	49%	0.067	0.92–5.54
Absent	41	75%			71%
Histologic response to chemotherapy^d
Good	24	92%	0.013	1.22–22.65	83%	0.124	0.79–5.77
Poor	48	62%			62%
Development of local recurrence
Yes	21	21%	<0.001	4.25–18.69
No	89	88%
Formal inclusion in trial^e
Yes	41	53%	0.011	1.20–5.31	53%	0.034	1.04–4.18
No	57	80%			76%

CI, Confidence interval; MFS, metastases-free survival; SSS, sarcoma-specific survival.

^aIn 86 extremity tumor patients having operation. ^bIn 110 patients having operation. ^cIn 57 patients with sufficient data. ^dAccording to protocol in question; recorded in 72 cases. ^eHigh-grade tumors. ^fLog-rank-test, except for ^gCox analysis.

Preoperative assessment and surgery

Radiological review was performed only for patients with local disease (n = 116) at presentation because our primary aim was to study local disease at presentation. Full sets of diagnostic MRI scans of 65 patients and CT-scans of 13 patients were available for review. Four patients of the 78 patients with radiological review had a skip metastasis. Three of the four skip metastases were missed during primary diagnostics. One patient with a skip metastasis was treated according to the protocol for local disease and is therefore included in the study. A scan after preoperative chemotherapy was available for review only for 45 patients. In 19 (42%) patients the soft tissue extension of the tumor grew more than 20% and in seven (16%) patients the tumor shrank more than 20% during preoperative chemotherapy.

Number of patients treated surgically in each university hospital during the study period varied from 11 to 59. One university hospital treated surgically only one patient during the whole study period. Hundred and ten patients with local disease of the 116 patients (95%) underwent surgery. Ninety-three patients with surgical treatment had a preoperative incisional biopsy and 11 a needle biopsy but six patients had no histological preoperative diagnosis: one patient underwent surgery based on radiological findings and five patients received a malignant diagnosis after the first surgical procedure and were re-operated. Of these patients, 81 had surgery combined with chemotherapy, three with radiotherapy and seven with chemotherapy and radiation therapy.

Eighty-six of 87 patients with extremity sarcoma underwent surgery; one refused amputation. Limb-sparing surgery was performed in 68/86 (79%) patients. The remaining surgeries consisted of 15 amputations, two exarticulations and one rotationplasty. Two more amputations were performed because of infectious complications. During follow-up, three patients were treated with amputation or exarticulation for local recurrence lowering the total limb-salvage rate to 73%. Twenty-four of the 29 patients with axial tumors underwent surgery. Surgery was not feasible in five patients with axial tumor because of the tumor location and local spreading.

Tumor excision was completed in one operation in 94 patients. Of the remaining 16 patients, two needed more than two tumor operations. The postoperative margins were considered radical in four patients (4%), wide in 28 (25%), marginal in 76 (69%) and intralesional in two (2%) patients.

Further surgery due to infections and problems with soft tissue flaps was needed for 25 of 110 (23%) patients. Eight patients required a microvascular flap because of infection or necrosis. Eight patients had a revision because of necrosis or fistula and four patients received a skin graft. Two patients had surgery because of an occlusion in the flap vessels. Two patients required an autogeneous bone transplant and one patient required re-osteosynthesis.

Chemotherapy

Ninety-two patients of the 116 (79%) patients with local disease received chemotherapy. Only one of 18 patients with a low grade tumor received chemotherapy. Five patients received only postoperative chemotherapy, nine patients only preoperatively and the rest both preoperative and postoperative chemotherapy. Chemotherapy was planned according to (in parentheses patients not formally enrolled but chemotherapy protocol used as a base for treatment) the SSG II protocol among one (1) patient, SSGVIII protocol among 47 (27) patients, ISG/SSGI among 30 (14) patients, ISG/SSGII among two (0) patients and SSGXIV among 11 (8) patients. One patient received MTX/CDDP/DOX/BCD without any obvious protocol. Among eight patients aged 40–50 years chemotherapy was modified from SSGVIII in five patients and from SSGXIV in three patients.

The analysis of administered chemotherapy was possible for 57 patients (excludes four patients aged over 40 years). The chemotherapy was completed as planned for MTX in 49/57 patients (86%) and eight patients had a major deviation from the protocol (four liver reactions, two progression, one severe stomatitis, one acute kidney failure), for CDDP 35/56 (63%) and eight patients had a major deviation (two neutropenic infections, four progressions, one Fanconi syndrome, one case of tinnitus), for DOX 43/57 (75%) and seven patients had a major deviation (three neutropenic infections, four progression), for IFO 35/39 (90%) and one patient with a major deviation (neutropenic infection) and for VP-16 10/11 (91%) and no major deviations. Sixty-four patients received granulocyte colony- stimulating factor.

Of the eight patients aged over 40 years, three refused any further chemotherapy. No data on chemotherapy was available in two patients. One did not recover from surgery, one progressed during preoperative treatment and one patient had approximately 2/3 doses because of liver reaction and repeated infections. Four patients received granulocyte colony-stimulating factor.

Radiation therapy

Sixteen patients (13.8%) received radiation therapy. Doses varied from 12 Gy to 70.5 Gy, mean 57.3 Gy (two files missing). One patient was treated with radiation therapy preoperatively to make the tumor operable and six patients were treated with radiation but were not operated. Nine patients received radiation therapy postoperatively: seven patients with a jaw or skull tumor and a contaminated or marginal margin, one patient with a pelvic tumor and a marginal margin and one patient with a humerus tumor and uncertain margins.

Treatment of metastatic disease at presentation

Twenty-eight patients had metastatic disease at diagnosis, 18 (64%) patients with isolated lung metastases and five with isolated bone metastases. Others had metastases in several locations. Twenty-six patients had an incisional biopsy and two had a core-needle biopsy. Sixteen patients had the primary tumor removed. Seven patients were treated with curative intention. All these seven patients had a peripheral tumor, isolated pulmonary metastases and received chemotherapy combined with surgery. Chemotherapy was planned according to ISG/SSGII in four patients, ISG/SSGI in one patient, SSGVIII in one and Children's Cancer Group protocol in one patient. Of the seven patients two died of second metastases and others remain disease-free at 5.8–11.3 years.

Of the 21 patients with palliative treatment 10 received chemotherapy alone, six patients chemotherapy and removal of the primary tumor, two patients had surgery with chemotherapy and radiation therapy and one each: surgery, radiation therapy and radiation therapy with chemotherapy. The latter patient remains disease-free at 16.7 years after radiation therapy and chemotherapy.

Survival

There were no treatment-related deaths. The 10-year sarcoma-specific survival for the whole population was 63% improving by five-year periods without statistical difference (p = 0.586) (Figure 1). The 10-year sarcoma-specific survival for patients with local disease at presentation was 73% and 69% for patients with local high-grade disease at presentation and 58% for patients with high-grade tumor of any stage.

Figure 1. Sarcoma-specific survival by five-year periods for the whole study population: 1991–1995 (blue), 1996–2000 (green), 2001–2005 (brown).

Local recurrences

Local recurrence-free survival for 110 patients with local disease and surgery at 10 years was 80%. Rates were 66% and 84% for patients with axial tumor and with peripheral tumor, respectively. Fifteen of the 21 patients with local recurrence developed also distant metastases. Event-free survival was 68% at 10 years for patients with local disease at presentation (Figure 2). None of the four patients with skip metastases developed a local relapse.

Figure 2. Event-free survival for the 116 patients with local disease at presentation.

Metastatic disease

Ten-year metastasis-free survival for the patients with local high-grade disease at presentation (n = 98) was 66%. Isolated lung metastases were the most common pattern of first metastases seen in 20 patients.

Of the 20 patients with isolated lung metastases three patients received no treatment, one patient underwent metastasectomy only, three patients received chemotherapy only (one patient oral etoposide and two patients ifosfamide) and 12 patients were treated by both metastasectomy and chemotherapy. Six of the 12 patients had second-line chemotherapy according to ISG/SSGI first relapse protocol, one patient was not operated. One patient had high-dose melphalan with peripheral stem cell rescue. One patient received HDIFO, one MTX/DOX/BCD/IFO, one MTX/CDDP/DOX, one vinorelbine/CDDP/IFO/VP-16 and one paclitaxel/VP-16/IFO. A single patient with complete resection of bone metastases received chemotherapy according to ISG/SSGI first relapse protocol.

Six patients (18%) of 33 patients with metastatic relapse remain disease-free at follow-up after treatment for metastases (2.7–15.5 years).

Univariate and multivariate analysis of prognostic factors

Univariate analyses on prognostic factors for 10-year sarcoma-specific and metastasis-free survival are shown in Table III. In multivariate analysis, development of local recurrence and major deviation from the treatment protocol retained their significance as prognostic factors for sarcoma-specific survival (Table IV). In multivariate analysis both formal entry of trial and time of diagnosis lost their significance.

Table IV. Multivariate analyses of prognostic factors for metastases-free survival and sarcoma-specific survival.

MFS		SSS
Factor	P-value (95% CI)	Factor	P-value (95% CI)
Time of diagnosis	0.101 (0.380–1.089)	Histologic response to chemotherapy^b	0.255 (0.532–10.798)
Formal inclusion in trial^a	0.101 (0.890–3.689)	Formal inclusion in trial^a	0.179 (0.612–13.87)
		Major deviation from chemotherapy protocol^c	0.002 (1.994–21.151)
		Development of local recurrence	<0.001 (5.700–66.596)

CI, Confidence Interval; MFS, metastases-free survival; SSS, sarcoma-spesific survival.

^aHigh-grade tumors. ^bAccording to protocol in question; recorded in 71 cases. ^cIn 61 patients with sufficient data.

Patients without operative treatment of local tumor (n = 6).

Five patients with an axial tumor were not treated surgically: four patients with the primary tumor in the sacrum and one in pelvic bones. All five patients received radiation therapy (12 Gy to 69 Gy) and three patients received also chemotherapy. One patient remains disease-free at 13 years of follow-up. For the other four patients the sarcoma-specific survival varied between 4 and 62 months. One patient refused the amputation of a femoral tumor and developed pulmonary metastases: she was treated with a palliative amputation two months prior to death at five years.

Discussion

To our knowledge only five nationwide, population-based studies on clinical presentation and prognosis of osteosarcoma have been published [5,11–14]. However, a mandatory histological re-evaluation was only included in our previous study of osteosarcoma patients diagnosed during 1971–1990 in Finland [14]. The five-year survival rate was 53% during 1990–1994 in a study of all patients under 40 years with osteosarcoma in Britain [13]. Larsson and Lorentzon retrospectively studied all malignant primary bone tumors reported to the National Cancer Registry in Sweden during 1958–1968 but no survival rates were reported [11]. Harvei and Solheim conducted a nationwide, population-based retrospective review on 240 patients reported to the Cancer Registry in Norway for osteosarcoma diagnosed 1953–1977 [5]. No significant improvement in survival was seen during 1953–1975. However, introduction of adjuvant chemotherapy in 1975 was the probable cause for the improvement of the 48-month survival from 25% during 1953–1974 to 47% during 1975–1977. All patients with osteosarcoma in New Zealand during 1994–1999 were included in a retrospective survival analysis with a five-year survival of 44% [12].

In our previous study five-year overall survival rates were 47% and 65% during 1971–1980 and 1981–1990 for local disease, respectively [14]. In the present study of patients diagnosed during 1991–2005 the 10-year sarcoma-specific survival was 63% for the whole population and 58% for high-grade tumors of any stage. An improvement of treatment results was seen during the present study period but without a statistical significance. Improvement of chest CT- and bone scan imaging may serve as one explanation for improved prognosis of local disease at presentation as metastases are detected more precisely. In the largest series from a single-institution with 1458 patients diagnosed during 1982–2002 with all locations, all ages and all stages but excluding low grade tumors, the five-year overall survival for the entire cohort was 57% [15].

Ninety-two percent of the original histologic samples were available and osteosarcoma diagnosis was confirmed in 90% of the tumors. Patients not available for review were significantly older and in most of these patients the tumor probably represented a metastasis from a different primary tumor instead of osteosarcoma. The mean annual incidence of osteosarcomas is 9.7, i.e 1.8 new osteosarcomas per million per year in Finland. Comparison to results from other groups remains difficult as cancer registries have different levels of ascertainment and coverage. Further, most reviews span a long period of time, and without the requirement for histologic review there might be a selection bias. Request for mandatory histology review itself – however – causes selection bias as 8% of original samples were unavailable leading to exclusion of these patients. Improvement in radiological techniques enables precise staging and patients with metastatic disease at diagnosis can be excluded from survival analyses. In a recent population-based review from northern England on osteosarcoma patients diagnosed during 1981–2002, the age-standardized incidence rate was 2.97 per million per year [16]. In the same analysis five-year survival rates were 61% and 62% for patient populations diagnosed during 1988–1994 and during 1995–2000, respectively. The study was limited to patients under 40 and the study group did not have access to staging data.

Preoperative chemotherapy together with limb sparing surgery and postoperative chemotherapy is the treatment of choice for osteosarcoma. Clinical trials have tried to optimize treatment results while minimizing side effects. During the study period 45 patients from Finland were formally enrolled in four trials. For patients not fulfilling the eligibility criteria chemotherapy was modified for age, liver and renal function and toxicity. In the multivariate analysis formal inclusion in a trial lost its prognostic adverse significance. Survival was similar in trial and non-trial patients in a national study by Stiller and colleagues [13]. Because of heterogeneity of the chemotherapy administered, major deviation from the chemotherapy protocol was arbitrarily defined as more than 25% of the drug in question omitted. MTX, CDDP and DOX had most major deviations, 14%, 14% and 12%, respectively. The trial regimen itself offers one possible explanation for these numbers, as ISG/SSGI protocol showed increased toxicity and the trial was closed prematurely [4]. Major deviation from chemotherapy protocol retained a statistical adverse significance.

Thirty-three (34%) patients of 98 patients with local high-grade tumor developed metastatic disease. An aggressive approach to treat pulmonary metastases with even repeated metastasectomies and chemotherapy may be curative in some patients [17,18]. Six of these patients (18%) in the present series remain disease-free at follow-up.

Local recurrence-free survival of 84% at 10 years for patients with extremity tumors is inferior to other series published. Primary limb-salvage surgery rate increased from 12% to 79% in Finland during 1971–2005 but it is still lower than in many series. Most probable cause of the high amputation rate is low surgical volume and inexperience with challenging oncological surgery: during the study period patients were operatively treated in all five university hospitals and surgical volumes were low varying from one to 59. Development of a local recurrence had an adverse prognostic effect on sarcoma-specific survival in multivariate analysis despite of aggressive treatment. Most of these patients also developed distant metastases.

Although reported to be a strong prognostic factor for disease-free survival, histologic response to neoadjuvant chemotherapy lost its statistical significance for sarcoma-specific survival in multivariate analysis in the present study [19,20]. Similar results were also published by Lewis and colleagues [21]. Several scales were used during the present study period to assess the degree of treatment-induced necrosis and it is therefore difficult to draw any profound conclusions.

With modern chemotherapy the prognosis of osteosarcoma in a series including tumors of all stages has improved in Finland to 63% at 10 years. Centralization of treatment would probably improve limb-salvage rate and local control.

Acknowledgements

MS was supported by a grant from Finska Läkarsällskapet and Finnish Medical Society Duodecim. The study was supported by EVO funds, Helsinki University Central Hospital and Finnish Cancer Society. The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

Declaration of interest: The authors report no conflict of interest. The authors alone are responsible for the content and the writing of the paper.