Outcome Prediction of Pulmonary Metastasectomy ...

World J Surg (2013) 37:1973–1980 DOI 10.1007/s00268-013-2022-9

Outcome Prediction of Pulmonary Metastasectomy Can Be Evaluated Using Metastatic Lesion in Osteosarcoma Patients Isao Matsumoto • Makoto Oda • Tsuyoshi Yachi Hiroyuki Tsuchiya • Yoh Zen • Go Watanabe

•

Published online: 6 April 2013 Ó Socie´te´ Internationale de Chirurgie 2013

Abstract Background We investigated whether molecular prognostic factors should be evaluated in specimens of the primary or the metastatic lesion and if the prognosis after initial pulmonary metastasectomy can be predicted based on evaluation of metastatic lesion specimens in osteosarcoma patients. Methods This retrospective study included 29 osteosarcoma patients with pulmonary metastases (19 males, 10 females; age 21 ± 10 years). Molecular prognostic factors were the levels of vascular endothelial growth factor type A (VEGF-A), VEGF type C (VEGF-C), and Ki67. Primary and pulmonary metastatic lesions could be compared in 18 patients regarding the values of marker expressions and the prognosis after initial pulmonary resection. Finally, the prognosis of all 29 cases was compared according to the molecular markers of the metastatic lesions. Results Evaluation of the metastatic lesions reflected the prognosis after pulmonary metastasectomy more than that of the primary lesions. In the metastatic lesions, positive expression of VEGF-A (n = 15), VEGF-C (n = 2), and Ki67 (n = 15) was associated with a significantly poorer prognosis (p = 0.0013, 0.0001, and 0.037, respectively).

I. Matsumoto (&) M. Oda T. Yachi G. Watanabe Department of General and Cardiothoracic Surgery, Kanazawa University, 13-1 Takara-machi, Kanazawa 920-8641, Japan e-mail: [email protected] H. Tsuchiya Department of Orthopedics, Kanazawa University, Kanazawa, Japan Y. Zen Section of Diagnostic Pathology, Kanazawa University Hospital, Kanazawa, Japan

No patients with positive expression of both VEGF-A and Ki67 (n = 7) survived more than 5 years after the initial pulmonary resection. All patients who had negative reactions to both VEGF-A and Ki67 (n = 6) were alive at the end of the study. Conclusions Molecular prognostic factors should be investigated in specimens of the metastatic lesion. Combined evaluation of VEGF-A and Ki67 and of VEGF-C using pulmonary metastatic lesion specimens in osteosarcoma patients effectively reflects survival after pulmonary metastasectomy.

Introduction Advances in chemotherapeutic strategy and refinement of surgical indications and approaches have significantly improved the prognosis for osteosarcoma patients [1, 2]. The 5-year survival has reached 40–70 % for all osteosarcoma patients, although 30–50 % of patients still die due to pulmonary metastases [2]. Thus pulmonary metastases have a major impact on the prognosis of osteosarcoma patients. Previous reports have asserted that surgical removal of pulmonary metastases in osteosarcoma patients offers favorable results [2–5]. In several reported series, the 5-year survival rate after pulmonary resection was 20–40 %, and patients in whom the metastases were removed surgically had a better prognosis than the patients with metastases who did not undergo surgery [1–5]. Some clinical prognostic factors were found to be important in determining the indications for pulmonary resection among patients with osteosarcoma [3–6]. However, most had little clinical value for the surgeon when non-surgical treatment options were not available. In contrast, other investigators have advocated aggressive

123

1974

resection of pulmonary metastases because such clinical prognostic factors were not associated with the outcome [2, 5]. Recently, several molecular prognostic factors have been reported to be predictors of survival and relapse [6–8]. Most of these molecular factors were evaluated in primary osteosarcoma lesions. Tumor cells of various malignant grades coexist in the primary osteosarcoma lesion [9]. In addition, because most patients undergo chemotherapy before pulmonary metastasectomy, the metastatic tumors may have cells that survived through chemotherapy. Therefore, the behavior of the tumor cells of the metastatic lesion is not necessarily the same as that of those in the primary lesion. In the present study, we investigated whether molecular prognostic factors should be investigated in specimens of the primary or the metastatic lesion. We also evaluated whether the prognosis after repeat resection of pulmonary metastases can be predicted based on evaluation of metastatic lesion specimens from osteosarcoma patients.

World J Surg (2013) 37:1973–1980

of the hilar and mediastinal lymph nodes was performed when the lymph nodes were enlarged. Disease-specific survival (DSS) after the initial pulmonary resection was used for comparison of prognoses. We defined DSS as the time from the initial pulmonary resection until death due to primary disease or the last follow-up. Disease-free interval (DFI) was defined as the time from the resection of the primary tumor until the first development of pulmonary metastasis. The values of typical molecular prognostic factors, including the expression of vascular endothelial growth factor type A (VEGF-A), VEGF type C (VEGF-C), and Ki67, were assessed in the primary lesions and metastatic lesions. Tissue samples were collected from the pulmonary tumors that were the largest of the non-necrotic tumors at the initial pulmonary metastasectomy. Primary tumor samples that were obtained before chemotherapy were available for 18 patients. Primary and pulmonary metastatic lesions in these patients were compared regarding marker expressions and the DSS after the initial pulmonary resection. Finally, the prognoses of all 29 cases were compared according to the molecular markers.

Patients and methods The Committee for Medical Ethics, Kanazawa University approved this retrospective study. Individual consents were obtained from the patients to use their resected specimens for research. This study included 29 osteosarcoma patients with pulmonary metastases who were treated at our institution between January 1978 and June 2007. More than 5 years have passed since the pulmonary surgery of the patients. The primary tumor was located in the femur (13 patients), tibia (12 patients), or humerus (4 patients). Patients with sacrum-originated osteosarcoma were excluded. All patients underwent surgery for the primary lesion, they were all exposed to chemotherapy before and/or after the surgery at the Department of Orthopedics in our institution and had no relapse of the primary lesion. The patients basically received a unified multiagent chemotherapeutic regimen that included cisplatin, adriamycin, and caffeine [10]. The surgical strategy for pulmonary metastasectomy was as follows: when the patient had bilateral pulmonary metastases, we employed bilateral simultaneous metastasectomy. We introduced video-assisted thoracic surgery (VATS) with 8- to 10-cm thoracotomy to pulmonary metastasectomy in 1997 for both uni- and bilateral cases. Before introducing VATS, a median sternotomy approach or clamshell incision was performed for bilateral simultaneous metastasectomy. During surgery to remove the metastases, all visible and palpable nodules were resected and reviewed histopathologically to confirm the diagnosis of metastasis. Sampling

123

Immunohistochemical assessment of VEGF-A, VEGF-C, and Ki67 We assessed the angiogenesis factors immunohistochemically (IHC) using the polyclonal antibodies for VEGF-A and VEGF-C. The primary antibodies used were the rabbit polyclonal antibody to VEGF-A (Santa Cruz Biotechnology, Santa Cruz, CA, USA) diluted 100-fold, and the goat polyclonal antibody to VEGF-C (Santa Cruz Biotechnology) diluted 100-fold. After reviewing the hematoxylin and eosin (H&E)-stained slides of the tumor specimens, blocks at the edge of the tumor area were selected. Paraffinembedded tumor tissues were cut into 4 lm thick sections and deparaffinized. IHC staining was performed using the labeled streptavidin–biotin method, as previously described [11]. Angiogenesis activity was scored as the ratio of VEGF-A- and -C-stained tumor cells in which cytoplasmic staining was positive. In addition, proliferative activity was assessed by IHC using the monoclonal antibody Ki67, which detects the proliferation-associated antigen Ki67. Mouse monoclonal anti-human Ki67 antigen (Clone MIB1; Dako, Glostrup, Denmark) was diluted 100-fold, and the IHC staining was performed using the same method used for VEGF-A and -C. Proliferative activity was scored as the ratio of Ki67stained tumor cells in which nuclear staining was positive [12, 13]. To assess VEGF-A, VEGF-C, and Ki67, all fields of the sections were scanned at low (940) and high (9400) power. Then, the most strongly stained areas in which non-

World J Surg (2013) 37:1973–1980

ossified tumor cells were best stained were chosen. Molecular factors were evaluated using anonymously prepared slides to avoid identification of the patients. Statistics The mean ± standard deviation values are shown, and equality of means was analyzed with the unpaired t test. The Pearson correlation test was used to investigate the relations of expressions of molecular markers between primary and pulmonary metastatic lesions. Survival curves after pulmonary metastasectomy were calculated using the Kaplan–Meier method and compared using the log-rank test. A uni- and multivariate Cox proportional hazards model was used to examine the prognostic values of molecular markers and to estimate hazard ratios (HRs) and confidence intervals (CIs). The accepted level of significance was p \ 0.05. All analyses were performed with PASW Statistics 18 software (SPSS, Chicago, IL, USA).

Results There were 19 males and 10 females with a mean age of 21 ± 10 years (median 19 years). The mean maximum diameter of pulmonary metastatic tumors was 24.8 ± 17.0 mm (median 20 mm). The mean number of metastatic sites was 9.0 ± 15.8 (median 3). The mean DFI was 17 ± 29 months (median 8 months). In all, 45 pulmonary surgeries were performed among 29 patients. Twelve patients required two or more pulmonary resections because of pulmonary recurrence. Seven male and five female patients underwent multiple surgeries: ten patients had two pulmonary resections, and two patients had four pulmonary resections. The mean follow-up after the initial pulmonary metastasectomy was 47 ± 47 months (median 26 months). There were no operative or hospital deaths. After adjuvant chemotherapy following the pulmonary resection, two patients died likely due to the adverse effects of chemotherapy. The 1-, 3-, 5-, and 10-year DSS rates after the initial pulmonary resection were 77.9 %, 51.9 %, 39.9 %, and 39.9 %, respectively. The 1-, 3-, 5-, and 10-year overall survival rates after initial pulmonary resection were 72.4 %, 48.3 %, 37.1 %, and 37.1 %, respectively. There was no significant difference between DSS and overall survival (p = 0.725, HR 0.889; 95 % CI 0.453–1.743). Nine patients survived [5 years, and four patients survived [10 years. Extrapulmonary recurrence sites included the brain (five patients), bone (three patients), liver (one patient), and mediastinal lymph nodes (one patient). Among the six patients who underwent a second pulmonary resection within 6 months after their initial pulmonary

1975

resection, five died within 12 months following the second resection. Comparison of primary and pulmonary metastatic lesions The primary and pulmonary metastatic tumor samples of 18 patients were evaluated. The results of the comparison of VEGF-A and Ki67 expressions in primary and pulmonary metastatic lesions are shown in Fig. 1. The correlations of primary and pulmonary metastatic lesions in VEGF-A and Ki67 expressions had coefficients of -0.0429 (p = 0.866) and -0.0662 (p = 0.794), respectively. The values of the two markers did not correspond between the primary and pulmonary metastatic lesions. The mean/median values of the VEGF-A staining ratio in primary and metastatic lesions were 0.36 ± 0.28/0.32 and 0.52 ± 0.07/0.54, respectively. The mean/median values of Ki67 staining in primary and metastatic lesions were 0.36 ± 0.19/0.41 and 0.40 ± 0.28/ 0.46, respectively. The values of VEGF-A and Ki67 in the metastatic lesions tended to be higher than in the primary lesions, although there was no significant difference (p = 0.11 and 0.68, respectively). The comparison of DSS associated with VEGF-A is shown in Fig. 2a, b. The median value of VEGF-A in the primary lesion was used as the cutoff index, which is nearly identical to the cutoff value of VEGF-positive values employed by Kaya et al. [7]. A positive staining area of \0.32 was classified as ‘‘negative’’ and one of C0.32 as ‘‘positive.’’ VEGF-A positive expression in metastatic lesions was associated with a significantly poorer prognosis (p = 0.036, negative relative to positive: HR 0.153; 95 % CI 0.007–3.264). However, there was no significant difference in the survival between those with positive or negative expression in primary lesions (p = 0.083, negative relative to positive: HR 0.866; 95 % CI 0.215–3.487). The comparison of DSS associated with Ki67 is shown in Fig. 2c, d. The median value in the primary lesion was also used as the cutoff index for Ki67. The results were the same as for VEGF-A. The statistical values for primary lesions were p = 0.937, negative relative to positive (HR 0.972; 95 % CI 0.484–1.952), and those for metastatic lesions were p = 0.013, negative relative to positive (HR 0.325; 95 % CI 0.113–0.939). Regarding the positivity of the prognostic molecular marker prior to and following the introduction of VATS, there were 10 positive and 6 negative cases for VEGF-A, 7 positive and 9 negative cases for Ki67 prior to the introduction, and 8 positive and 5 negative cases for both VEGF-A and Ki67 following the introduction. Although there were more positive Ki67 cases following the introduction of VATS, there was no difference in the ratio for VEGF-A regardless of whether VATS was performed. VEGF-C expression was noted in the primary lesion of one patient and in metastatic lesions of two patients

123

1976

World J Surg (2013) 37:1973–1980

Fig. 1 Results of the comparison of VEGF-A and Ki67 expressions in primary and pulmonary metastatic lesions. Solid black lines difference [0.1, metastatic lesions (M) [ primary lesions (P). Gray lines difference [0.1, P [ M. Dotted lines difference up to 0.1

Staining ratio of VEGF-A Median: 0.32

Staining ratio of Ki67

Median : 0.54

1

1

0.9

0.9

0.8

0.8

0.7

0.7

0.6

0.6

0.5

0.5

0.4

0.4

0.3

0.3

0.2

0.2

0.1

0.1

0

Median : 0.41

Median : 0.46

Primary lesions

Metastatic lesions

0

Primary lesions

a

Metastatic lesions

b

Probability Survival


negative (n=5)

negative (n=9) positive (n=9)

positive (n=13)

VEGF-A: Primary lesions

VEGF-A: Metastatic lesions Time in months

Time in months No. at risk

9 9

5 5

5 3

3 2

1 1

1 1

1

(negative) (positive)

No. at risk

c

5 13

4 6

4 4

4 1

2

2

1


d



negative (n=8)

negative (n=9)

positive (n=9)

Ki67: Primary lesions

No. at risk

9 9

5 5

5 3

2 3

2

2

1

positive (n=10)

Ki67: Metastatic lesions

Time in months (negative) (positive)

No. at risk

8 10

6 4

6 2

5

2

2

1

Time in months (negative) (positive)

Fig. 2 Prognosis associated with VEGF-A (a, b) and Ki67 (c, d) expressions in primary and pulmonary metastatic lesions

123

World J Surg (2013) 37:1973–1980

including the former patient. The degree of staining was less than 10 % in all specimens. In summary, the markers’ values in pulmonary metastatic lesions reflected survival after pulmonary metastasectomy more than did those in primary lesions. Therefore, the prognoses of all 29 cases were compared using the markers in metastatic lesions. Comparison of prognosis in all cases using the molecular markers The prognosis of all 29 cases was compared according to the molecular markers. The median value of VEGF-A in pulmonary metastatic lesions was used as the cutoff index. A positive staining area of \0.56 was classified as ‘‘negative’’ and an area of C0.56 was deemed ‘‘positive.’’ There were 15 patients who had positive reactions to VEGF-A. Positive expression of VEGF-A was associated with a significantly poorer prognosis (Fig. 3a; Table 1). VEGF-C staining was noted in only two patients—who died during the sixth and seventh months after pulmonary resection. Thus, for the evaluation criteria of VEGF-C, no matter how slight, any degree of positive staining in the tumor was considered ‘‘positive’’ expression. Patients with positive VEGF-C staining had a significantly poorer survival (Fig. 3b; Table 1). Metastases to the subaortic lymph nodes were confirmed in one of those patients. For the comparison of prognosis regarding Ki67 staining, the median value was employed as the cutoff index. By this criterion, 15 patients had ‘‘positive’’ reactions (i.e., C0.33 positive staining for Ki67). Those patients had a significantly poorer prognosis (Fig. 3c; Table 1). For both VEGF-A and Ki67, there was no definite trend to the prognosis of borderline cases around the cutoff value. On multivariate analysis, VEGF-A and Ki67 were significant factors portending a poor prognosis (Table 2). Seven patients had positive expressions of both VEGFA and Ki67, and six patients had both negative expressions. No patients with positive VEGF-A and Ki67 survived more than 5 years after the initial pulmonary resection. On the other hand, all of the patients with both of these markers negative were alive (maximum 163 months) at the end of the study. There were significant differences between patients with both markers negative, those with other combinations in which one factor was positive and one negative, and others with both positive (Fig. 4). Patients who had both markers negative had a significantly better prognosis by the Bonferroni correction (p = 0.0004).

Discussion It is generally regarded that clinicopathologic prognostic factors play an important role in determining the

1977

indications for pulmonary metastasectomy in osteosarcoma patients. Previous studies have identified various independent risk factors for mortality from osteosarcoma after developing a pulmonary disease including primary tumor size, site, extension, or histology; surgical approach; location of pulmonary metastases; pleural disruption; resection margins; number or size of pulmonary metastases; complete or incomplete resection; and the time of identification of pulmonary metastasis [1, 3–5]. In particular, two independent factors that predict survival after pulmonary metastasectomy were DFI and the response of the primary tumor to preoperative chemotherapy [14, 15]. However, Putnam and Roth [5] reviewed the results of surgery for recurrent pulmonary metastasis and concluded that repeat resection can be accomplished safely and offers prolonged post-thoracotomy survival. More recently, Harting et al. [2] reported that measures of tumor burden have little meaningful influence on overall survival. They advocate considering repeat pulmonary resection for patients with recurrent metastases from osteosarcoma even for those with a short DFI. This strategy allows treatment of any remaining micrometastases via concomitant chemotherapy and enables patients with biologically favorable tumors the opportunity for prolonged survival. Not all osteosarcoma patients with pulmonary metastases benefit from repeat pulmonary metastasectomy, however. Our results demonstrated that some patients died from recurrence soon after repeated pulmonary resection. Therefore, more accurate prognostic factors as potential indicators for pulmonary metastasectomy are needed. Several molecular prognostic factors have been reported to be predictors of survival and relapse [7, 8, 16]. Most of these molecular factors were evaluated in the primary lesion. There are only a few studies that have investigated molecular factors based on specimens from metastatic lesions. Oda et al. [9] reported that there was no significant difference between primary and metastatic sites with respect to the expression of VEGF-A, although VEGF-A and Ki67 expressions in the primary and metastatic sites were non-uniform. Our study also demonstrated that the behavior of the tumor cells of the metastatic lesion was not necessarily the same as those of the primary lesion, and the values of pulmonary metastatic lesions are more likely to reflect survival after pulmonary metastasectomy than those of primary lesions. We believe that molecular prognostic factors should be investigated in specimens of the metastatic lesion when no local relapse is found in the primary lesion. The role of angiogenesis in osteosarcoma is still controversial. Some investigators have stated that overexpression of VEGF-A by osteosarcoma cells is associated with a worse prognosis, with increasing vascularity of the osteosarcoma [7, 8]. Other investigators have stated that the degree of

123

1978

World J Surg (2013) 37:1973–1980

a

b VEGF-A

VEGF-C



negative (n=14)

negative (n=27)

positive (n=2)

positive (n=15) Time in months

No. at risk

14

10

8

15

5

2

6

4

3

1

(negative)


(positive)

27

15

2

0

11

8

4

3

1


c


Ki67

negative (n=14)

positive (n=15)


14

10

7

5

15

5

3

1

4

3

1


Fig. 3 Prognosis in all cases using VEGF-A, VEGF-C, and Ki67 expressions in metastatic lesions

Table 1 Univariate analysis of prognostic molecular factors using specimens from metastatic lesions Factors

Number of patients

VEGF-A Negative

14

Positive

15

VEGF-C Negative

27

Positive

2

Ki67 Negative

14

Positive

15

p

HR (95 % CI)

0.001

0.162 (0.045–0.576)

0.0001

0.058 (0.008–0.421)

0.038

0.035 (0.130–0.995)

VEGF-A vascular endothelial growth factor type A, VEGF-C vascular endothelial growth factor type C, HR hazard ratio, CI confidence interval

123

microvessel density and VEGF-A expression is of no prognostic value, although these factors have a key functional role in tumor development, invasion, and metastatic spread [9, 13]. These conflicting results may have arisen because they had been evaluated in primary lesions. We also assessed proliferative activity using the Ki67 assay, which detects the proliferation-associated antigen Ki67. It is also reported that Ki67 overexpression could be used as a prognostic marker for the development of pulmonary metastasis in osteosarcoma patients [16]. Our results indicated that patients with negative expression of both VEGF-A and Ki67 had significantly better prognoses than those with other combinations of expression. The patients with both factors being positive had significantly poorer prognoses. Therefore, combined evaluation of VEGF-A and Ki67 can be a strong predictor of survival

World J Surg (2013) 37:1973–1980

1979

in only 3 % of patients during their lifetime [17]. Lymph node involvement is reported as a poor prognostic sign in osteosarcoma patients even for those who undergo systematic lymph node dissection [18, 19]. VEGF-C staining can be a useful parameter for predicting lymph node metastasis and the survival of osteosarcoma patients. We do not deny the option of repeat metastasectomy even if patients have positive VEGF-A and Ki67 or positive VEGF-C. Currently, we perform repeat metastasectomy for such patients when the patient can tolerate surgery. Although we have obtained the above results, we still often have no choice than to resort to surgery when there is no other effective therapeutic modality because most patients are young. Until the cutoff indices become more accurate, we believe it is necessary to retain the surgical option to avoid disadvantages especially to patients on the borderline. In this study, the cutoff indices of each molecular factor are not defined fully. The values of the cutoff indices may affect the strength of the molecular markers as prognostic factors. When it becomes possible to define cutoff indices more accurately, we can recommend considering our results as one of the factors not only for predicting prognosis but for decision-making about repeat pulmonary metastasectomy for osteosarcoma patients. To establish the utility of our methods, further prospective studies with a larger number of patients are needed. There are other limitations to our analysis. This study is a long-term retrospective examination concerning a small series of patients with surgically addressed osteosarcoma with pulmonary metastases treated at a single institution.

Table 2 Multivariate analysis of prognostic molecular factors using specimens from metastatic lesions Factors

Favorable

Unfavorable

p

HR (95 % CI)

VEGF-A

Negative

Positive

0.004

0.153 (0.042–0.558)

VEGF-C

Negative

Positive

0.093

0.418 (0.151–1.155)

Ki67

Negative

Positive

0.041

0.572 (0.335–0.977)

and relapse after pulmonary metastasectomy in osteosarcoma patients. The result of both negative staining of VEGF-A and Ki67 can be a good surgical indication and encouraging to patients and surgeons. If either VEGF-A or Ki67 staining is borderline, even patients with combinations other than both negative have the possibility of benefiting from surgery. No influence was observed in the above results concerning marker positivity regarding the change in the surgical procedure: although there were more positive Ki67 cases following the introduction of VATS as a less invasive technique, there was no difference in the ratio for positive VEGF-A regardless of whether VATS was performed. There are a few reports that evaluate VEGF-C in osteosarcoma patients. VEGF-C is a stimulator of vascular and/ or lymphatic endothelial cells. In this study, patients with positive expression of VEGF-C succumbed within 7 months after the initial pulmonary resection. Significantly, one of them had a metastasis to the subaortic lymph node. Lymph node metastases from osteosarcoma are rare because bones lack a lymphatic system. It has been noted

both negative (n=6)


Fig. 4 Survival for combinations of expression of VEGF-A and Ki67. Both positive patients with positive expression of both VEGF-A and Ki67. Both negative patients with negative expression of both VEGF-A and Ki67. Positive and negative patients with combinations in which one factor was positive and one negative

p=.014 p=.0004 positive and negative (n=16)

p= .013 both positive (n=7)


6

6

6

6

4

3

16

8

4

2

1

(positive and negative)

(both negative)

7

1

1

0

(both positive)

123

1980

World J Surg (2013) 37:1973–1980

Also, when examining the behavior of the tumor cells, not only primary tumors but also metastatic tumors may have varied degrees of malignancy. In addition, because the number of positive VEGF-C cases is low, further evaluation is necessary for clarifying whether positive VEGF-C is a strong prognostic factor.

Conclusions VEGF-A, VEGF-C, and Ki67 expressions in the primary sites did not correspond to those in the metastatic sites. We believe that molecular prognostic factors should be evaluated in specimens from the metastatic lesion in osteosarcoma patients. Our results demonstrated that evaluation of VEGF-A and Ki67 combination and VEGF-C using pulmonary metastatic lesion specimens effectively reflects survival. We believe that our method is easy and can be useful for predicting prognosis after pulmonary metastasectomy in osteosarcoma patients even when a specimen from a primary lesion is not available. Conflict of interest

None.

References 1. Tsuchiya H, Kanazawa Y, Abdel-Wanis ME et al (2002) Effect of timing of pulmonary metastases identification on prognosis of patients with osteosarcoma: the Japanese Musculoskeletal Oncology Group Study. J Clin Oncol 20:3470–3477 2. Harting MT, Blakely ML, Jaffe N et al (2006) Long-term survival after aggressive resection of pulmonary metastases among children and adolescents with osteosarcoma. J Pediatr Surg 41:194–199 3. Carter SR, Grimer RJ, Sneath RS et al (1991) Results of thoracotomy in osteogenic sarcoma with pulmonary metastases. Thorax 46:727–731 4. Putnam JB Jr, Roth JA, Wesley MN et al (1983) Survival following aggressive resection of pulmonary metastases from osteogenic sarcoma: analysis of prognostic factors. Ann Thorac Surg 36:516–523

123

5. Putnam JB Jr, Roth JA (1995) Surgical treatment for pulmonary metastases from sarcoma. Hematol Oncol Clin N Am 9:869–887 6. Clark JC, Dass CR, Choong PF (2008) A review of clinical and molecular prognostic factors in osteosarcoma. J Cancer Res Clin Oncol 134:281–297 7. Kaya M, Wada T, Akatsuka T et al (2000) Vascular endothelial growth factor expression in untreated osteosarcoma is predictive of pulmonary metastasis and poor prognosis. Clin Cancer Res 6:572–577 8. Charity RM, Foukas AF, Deshmukh NS et al (2006) Vascular endothelial growth factor expression in osteosarcoma. Clin Orthop Relat Res 448:193–198 9. Oda Y, Yamamoto H, Tamiya S et al (2006) CXCR4 and VEGF expression in the primary site and the metastatic site of human osteosarcoma: analysis within a group of patients, all of whom developed lung metastasis. Mod Pathol 19:738–745 10. Tsuchiya H, Yasutake H, Yokogawa A et al (1992) Effect of chemotherapy combined with caffeine for osteosarcoma. J Cancer Res Clin Oncol 118:567–569 11. Ohta Y, Nozawa H, Tanaka Y et al (2000) Increased vascular endothelial growth factor and vascular endothelial growth factorc and decreased nm23 expression associated with microdissemination in the lymph nodes in stage I non-small cell lung cancer. J Thorac Cardiovasc Surg 119:804–813 12. Cooper LS, Gillett CE, Smith P et al (1998) Cell proliferation measured by MIBI and timing of surgery for breast cancer. Br J Cancer 77:1502–1507 13. Bottini A, Berruti A, Bersiga A et al (2001) Relationship between tumour shrinkage and reduction in Ki67 expression after primary chemotherapy in human breast cancer. Br J Cancer 85:1106–1112 14. Kempf-Bielack B, Bielack SS, Ju¨rgens H et al (2005) Osteosarcoma relapse after combined modality therapy: an analysis of unselected patients in the Cooperative Osteosarcoma Study Group (COSS). J Clin Oncol 23:559–568 15. Bielack SS, Kempf-Bielack B, Delling G et al (2002) Prognostic factors in high-grade osteosarcoma of the extremities or trunk: an analysis of 1,702 patients treated on neoadjuvant Cooperative Osteosarcoma Study Group protocols. J Clin Oncol 20:776–790 16. Hernandez-Rodriguez NA, Correa E, Sotelo R et al (2001) Ki-67: a proliferative marker that may predict pulmonary metastases and mortality of primary osteosarcoma. Cancer Detect Prev 25:210–215 17. Jeffree GM, Price CH, Sissons HA (1975) The metastatic patterns of osteosarcoma. Br J Cancer 32:87–107 18. Pfannschmidt J, Klode J, Muley T et al (2006) Pulmonary resection for metastatic osteosarcomas: a retrospective analysis of 21 patients. Thorac Cardiovasc Surg 54:120–123 19. Kim SJ, Choi JA, Lee SH et al (2004) Imaging findings of extrapulmonary metastases of osteosarcoma. Clin Imaging 28:291–300