Introduction:
The aim of this study is to evaluate the significance of MRI findings, surgical margins, previous interventions and their effects on oncological and functional outcomes in parosteal osteosarcoma.

Methods:
Twenty-seven patients (8 male/19 female) operated with the diagnosis of primary parosteal osteosarcoma in our institution were retrospectively reviewed. The epidemiological data, biopsy method, misdiagnosis/improper interventions and delay in diagnosis were noted. Maximum circumferential and longitudinal extension, intramedullary involvement and neurovascular extensions in MRI sections were evaluated. Resection type (segmental intraarticular/ segmental intercalary/hemicortical), reconstruction type (biologic/non-biologic) and surgical margins were noted. Functional and oncological results at last follow-up were assessed.

Results:
The mean age was 31.6 (12-73) years; median follow-up was 63 (15-270) months. The most common site was distal femur. Percutaneous biopsies in elsewhere centers were related with increased number of re-biopsies and misdiagnosis/improper interventions (p<0.001,p=0.044). Intramedullary involvement rate was related with maximum circumferential extension and maximum longitudinal extension (p=0.006,p=0.005). The intramedullary involvement ratio of ≤10% suggested no recurrence or metastasis. Mean MSTS score was 81.1% (60-100%). Neurovascular involvement was related with metastatic disease, deep infections and complication related surgeries (p=0.017,p=0.002,p=0.005). The most common resection type was segmental intraarticular resection (63%). Hemicortical resections with biological reconstructions had the best MSTS scores (p=0.002). Higher maximum circumferential extension rate, maximum longitudinal extension, intramedullary involvement rate of the lesion and neurovascular involvement were related with lower MSTS scores (p=0.003,p=0.028,p=0.038,p=0.022). Five year overall survival was 95.5%, local recurrence-free survival was 77.2% and metastasis-free survival was 69.4%.

Conclusion:
The lesion’s extent of intramedullary involvement, neurovascular bundle proximity and maximum periosteal circumferential extension on MRI should be considered when planning the surgery. The evaluation of maximum circumferential extension on MRI is crucial for the resection margins. Hemicortical resection and biological reconstruction should be considered whenever possible.