Purpose: High-dose ifosfamide (HD-IFO) remains an effective regimen for advanced bone and soft tissue sarcomas, but predictors of long-term benefit are poorly defined. This study evaluated clinical outcomes and prognostic factors using machine learning-assisted modeling in sarcoma patients treated with HD-IFO at a high-volume academic center.
Methods: We retrospectively analyzed 26 patients with histologically confirmed bone or soft tissue sarcoma who received HD-IFO (≥ 12 g/m2 per cycle) between 2015 and 2025. Progression-free survival (PFS) and overall survival (OS) were estimated by the Kaplan-Meier method and compared across RECIST response categories using log-rank testing. Prognostic factors were identified using Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression with leave-one-out cross-validation. The top three variables were entered into multivariable logistic regression to estimate odds ratios (ORs) for OS > 24 months.
Results: Median PFS and OS from start of HD-IFO was 6.6 months (95% CI 4.4-9.8) and 24.7 months (95% CI, 14.7-34.2), respectively. Patients with progressive disease (PD) had significantly shorter OS than those with partial response (PR; p = 0.0047) or stable disease (SD; p = 0.0485). LASSO identified intervention prior to progression, prior tumor control ≥ 12 months, and absence of metastases as the strongest predictors for OS > 24 months. In multivariable analysis, intervention prior to progression (OR 24.18, 95% CI 1.81-1001.27, p = 0.037) and prior tumor control ≥ 12 months (OR 25.39, 95% CI 2.1-1008.9, p = 0.030) independently predicted OS > 24 months.
Conclusion: HD-IFO provides durable disease control in selected sarcoma patients, particularly those with sustained prior tumor control and intervention prior to progression.
Keywords: Bone sarcoma; High-dose ifosfamide; Machine learning-assisted modeling; Soft tissue sarcoma.
© 2026. The Author(s).