# Development and validation of a predictive score for chemoresistance in high-grade osteosarcoma at baseline

## TL;DR

This study creates a predictive score to identify high-grade osteosarcoma patients likely to be chemoresistant before treatment begins.

## Contribution

A novel predictive score for chemoresistance in osteosarcoma using baseline clinical and MRI parameters is developed and validated.

## Key findings

- The model includes four predictors: age, metastasis, tumor volume, and pre-chemotherapy tumor necrosis.
- The score ranges from 0 to 8, with higher scores indicating a higher probability of chemoresistance.
- The model showed good discrimination (AuROC 0.76) and calibration.

## Abstract

Histological tumor necrosis is the current indicator for the response of osteosarcoma after neoadjuvant chemotherapy. Chemoresistant tumors require close monitoring and adjustment of treatment. We aimed to develop a prediction score for chemoresistance in newly diagnosed osteosarcoma patients underwent neoadjuvant chemotherapy.

Data from a registry-based cohort of high-grade osteosarcoma patients treated with neoadjuvant chemotherapy between January 2008 and October 2023 were used. Histological necrosis from surgical specimens was the reference standard. Clinical and MRI parameters at baseline were derived by risk regression analysis.

From 139 patients, 93 (66.91%) were classified as chemoresistant (histological necrosis <90%). The model included four predictors: age >40 years, initial metastasis, tumor volume (≤150 ml, > 150–400, or > 400 ml), and pre-chemotherapy tumor necrosis >50%. The AuROC of the model was 0.76 (95% CI 0.68–0.85) and well-calibrated. Internal validation using a bootstrap technique showed consistent AuROC results. The prediction score ranged from 0 to 8, with a score of 0–2 indicating low probability (positive LHR = 0.45) and a score of 3–8 indicating high probability (positive LHR = 2.56) of chemoresistance.

High-grade osteosarcoma patients with a prediction score of 3–8 have a high probability of chemoresistance. This score could be used for risk communication and tailoring management at diagnosis.