Quantitative and Morphology-Based Deep Convolutional Neural Network Approaches for Osteosarcoma Survival Prediction in the Neoadjuvant and Metastatic Settings

Clin Cancer Res. 2025 Jan 17;31(2):365-375. doi: 10.1158/1078-0432.CCR-24-2599.

Abstract

Purpose: Necrosis quantification in the neoadjuvant setting using pathology slide review is the most important validated prognostic marker in conventional osteosarcoma. Herein, we explored three deep-learning strategies on histology samples to predict outcome for osteosarcoma in the neoadjuvant setting.

Experimental design: Our study relies on a training cohort from New York University (NYU; New York, NY) and an external cohort from Charles University (Prague, Czechia). We trained and validated the performance of a supervised approach that integrates neural network predictions of necrosis/tumor content and compared predicted overall survival (OS) using Kaplan-Meier curves. Furthermore, we explored morphology-based supervised and self-supervised approaches to determine whether intrinsic histomorphologic features could serve as a potential marker for OS in the neoadjuvant setting.

Results: Excellent correlation between the trained network and pathologists was obtained for the quantification of necrosis content (R2 = 0.899; r = 0.949; P < 0.0001). OS prediction cutoffs were consistent between pathologists and the neural network (22% and 30% of necrosis, respectively). The morphology-based supervised approach predicted OS; P = 0.0028, HR = 2.43 (1.10-5.38). The self-supervised approach corroborated the findings with clusters enriched in necrosis, fibroblastic stroma, and osteoblastic morphology associating with better OS [log-2 hazard ratio (lg2 HR); -2.366; -1.164; -1.175; 95% confidence interval, (-2.996 to -0.514)]. Viable/partially viable tumor and fat necrosis were associated with worse OS [lg2 HR; 1.287; 0.822; 0.828; 95% confidence interval, (0.38-1.974)].

Conclusions: Neural networks can be used to automatically estimate the necrosis to tumor ratio, a quantitative metric predictive of survival. Furthermore, we identified alternate histomorphologic biomarkers specific to the necrotic and tumor regions, which could serve as predictors.

MeSH terms

  • Adolescent
  • Adult
  • Bone Neoplasms* / mortality
  • Bone Neoplasms* / pathology
  • Bone Neoplasms* / therapy
  • Convolutional Neural Networks
  • Deep Learning*
  • Female
  • Humans
  • Kaplan-Meier Estimate
  • Male
  • Necrosis / pathology
  • Neoadjuvant Therapy
  • Neural Networks, Computer*
  • Osteosarcoma* / mortality
  • Osteosarcoma* / pathology
  • Osteosarcoma* / therapy
  • Prognosis
  • Young Adult