In osteosarcoma, treatment guidelines recommend standard chemotherapy regardless of severity of disease. Treatment individualization will minimize risk of failure and adverse effects, specially in patients who have good prognosis. Therefore, there is a pressing clinical need to develop a risk adapted strategies and to adjust chemotherapy to prognostic factors.

Aim: to asses usefulness of Classification and Regression Tree Analysis (C& RT) for stratifying patients with localised osteosarcoma to the risk groups according to clinical and biological markers.

Material and methods: 100 patients with localised osteosarcoma were included, aged 5–23 years (mean 14), with extremity localisation of the primary tumour. Follow up – at least 5 years since a date of diagnosis. We analysed clinical prognostic factors (tumour size, pathological fracture, alkaline phosphatase, age), histological prognostic factors (% of viable tumour cells after pre-operative chemotherapy, subtype of osteosarcoma and its aggressiveness) and biological factors (expression of VEGF, Ki-67, Topoisomerase II alpha and P glycoprotein). The expressions of proteins were measured immunohistochemically in biopsy samples. C& RT model included all described above factors.

Results: C& RT analysis revealed that the most important prognostic factors in localised osteosarcoma were: VEGF, Topoisomerase II alpha and tumour size. This markers were included into the risk classification and three risk groups were proposed: with poor prognosis (n=13) – 5 year OS 31%, moderate (n=57) – 5 year OS 63% and with good prognosis (n=30) – 5 year OS 93%), P=0.000.

Conclusion: C& RT is useful method for stratifying patients with osteosarcoma to risk groups. The stratification should include biological and clinical prognostic markers.