Objectives: Histological tumour necrosis is the current indicator for the response of osteosarcoma after neoadjuvant chemotherapy. Chemoresistant tumours require close monitoring and adjustment of treatment. Characteristics of tumours on baseline MRI may be able to predict response to chemotherapy. The aim is to identify which baseline MRI findings can help predict chemoresistant osteosarcoma.
Methods: Baseline MRI before giving neoadjuvant chemotherapy of 95 patients during 2008-2021 was reviewed by 2 musculoskeletal radiologists. Histological necrosis from surgical specimens was the reference standard. MRIs were reviewed for tumour characteristics (tumour volume, maximum axial diameter, central necrosis, haemorrhage, fluid-fluid level), peritumoural bone and soft tissue oedema, and other parameters including intra-articular extension, epiphyseal involvement, neurovascular involvement, pathologic fracture, and skip metastasis. The cut-off thresholds were generated by receiver operating characteristic curves which then tested for diagnostic accuracy.
Results: Two-third of patients were chemoresistance (histological necrosis <90%). Tumour volume >150 mL, maximum axial diameter >7.0 cm, area of necrosis >50%, presence of intra-articular extension, and peritumoural soft tissue oedema >6.5 cm significantly predicted chemoresistance, particularly when found in combination. Tumour volume >150 mL and maximum axial diameter >7.0 cm could be used as an independent predictor (multivariable analysis, P-value = .025, .045).
Conclusions: Findings on baseline MRI could help predicting chemoresistant osteosarcoma with tumour size being the strongest predictor.
Advances in knowledge: Osteosarcomas with large size, large cross-sectional diameter, large area of necrosis, presence of intra-articular extension, and extensive peritumoural soft tissue oedema were most likely to have a poor response to neoadjuvant chemotherapy.
Keywords: chemotherapy; magnetic resonance imaging; osteosarcoma.
© The Author(s) 2023. Published by Oxford University Press on behalf of the British Institute of Radiology. All rights reserved. For permissions, please email: journals.permissions@oup.com.