Background: Other than metastases at diagnosis and histological response to preoperative chemotherapy, there are few reliable predictors of survival in patients with osteosarcoma. Microscopic vascular invasion (MVI) has been identified in the resection specimens of patients with osteosarcoma. However, it is unknown whether the MVI in resected specimens is associated with worse overall survival and higher cumulative incidence of local recurrence or metastasis in a large cohort of patients younger than 40 years with high-grade localized osteosarcoma.
Questions/purposes: (1) Is MVI associated with worse overall survival and higher cumulative incidence of events (local recurrence or metastasis) in patients younger than 40 years with high-grade localized osteosarcoma? (2) What clinical characteristics are associated with MVI in patients with high-grade localized osteosarcoma?
Methods: A total of 625 patients younger than 40 years with primary high-grade osteosarcoma between 1997 and 2016 were identified in our oncology database. We included patients younger than 40 years with primary high-grade osteosarcoma who underwent definitive surgery and preoperative and postoperative chemotherapy. The minimum follow-up period was 2 years after treatment. Patients with the following were excluded: metastasis at initial presentation (21%, n = 133), progression with preoperative chemotherapy precluding definitive surgery (6%, n = 38), surgery at another unit (2%, n = 13), lost to follow-up before 2 years but not known to have died (3%, n = 18), and death related to complications of preoperative chemotherapy (1%, n = 4). A retrospective pathologic and record review was conducted in the remaining 419 patients. The median follow-up period was 5 years (interquartile range [IQR] 3 to 9 years). The overall survival of the entire group (n = 419) was 67% [95% CI 63 to 72] at 5 years. Of the 419 patients, 10% (41) had MVI in their resection specimens. The Kaplan-Meier method was used to estimate overall survival. The cumulative incidence of events captured the first event of either metastasis or local recurrence. This analysis was completed with a competing risk framework: deaths without evidence of local recurrence or metastasis were regarded as a competing event. Clinical and histological variables (sex, age, tumor site, tumor largest dimension, surgical margin, chemotherapy-induced necrosis, type of surgery, histologic type of tumor, type of chemotherapy regimen, pathologic fracture, and MVI) were evaluated using the log-rank test or Gray test in the univariate analyses and Cox proportional hazard model or Fine and Gray model in the multivariate analyses.
Results: After adjusting for other factors, multivariate analyses showed that the presence of MVI in resection specimens was associated with worse overall survival and higher cumulative incidence of event (hazard ratio 1.88 [95% CI 1.22 to 2.89]; p = 0.004 and HR 2.33 [95% CI 1.56 to 3.49]; p < 0.001, respectively). A subgroup analysis demonstrated that the relationship between MVI and survival applied only to patients with a poor response to chemotherapy (less than 90% necrosis; overall survival at 5 years, MVI [+] = 24% [95% CI 11 to 39] versus MVI [-] = 60% [95% CI 52 to 66]; p < 0.001 and cumulative incidence of events at 5 years, MVI [+] = 86% [95% CI 68 to 94] versus MVI [-] = 54% [95% CI 46 to 61]; p < 0.001). The MVI (+) group had a higher proportion of patients with a poor response to chemotherapy (85% [35 of 41] versus 53% [201 of 378]; p < 0.001), involved margins (15% [6 of 41] versus 5% [18 of 378]; p = 0.021), and limb-ablative surgery (37% [15 of 41] versus 21% [79 of 378]; p = 0.022) than the MVI (-) group did.
Conclusions: MVI is associated with lower overall survival and higher cumulative incidence of local recurrence or metastasis, especially in patients with a poor histologic response to preoperative chemotherapy. Future studies in patients treated for osteosarcoma should consider this observation when planning new trials.
Level of evidence: Level III, therapeutic study.