Evaluating comorbidity scoring systems for flumatinib therapy in chronic myeloid leukemia: a machine learning and SHAP-based predictive analysis - Report - MDSpire
Advertisement
Evaluating comorbidity scoring systems for flumatinib therapy in chronic myeloid leukemia: a machine learning and SHAP-based predictive analysis
Clinical Report: Assessing Comorbidity Scoring Systems for Flumatinib Treatment
Overview
This study evaluates the predictive value of comorbidity scoring systems in chronic myeloid leukemia patients treated with flumatinib. The Cumulative Illness Rating Scale for Geriatrics (CIRS-G) demonstrated the strongest predictive performance for major molecular response.
Background
Chronic myeloid leukemia (CML) management has evolved significantly with the introduction of tyrosine kinase inhibitors (TKIs) like flumatinib. Comorbidities in older patients can complicate treatment outcomes, making it essential to understand how different comorbidity scoring systems can predict responses to therapy. This study aims to clarify which scoring system best predicts molecular responses in CP-CML patients receiving flumatinib.
Data Highlights
Model
AUC
Incremental Value (ΔAUC)
p-value
XGBoost
0.852
0.078
0.006
Key Findings
The XGBoost model achieved an AUC of 0.852 for predicting major molecular response.
Integrating CIRS-G into the baseline model significantly improved predictive performance.
A CIRS-G score ≥ 8 may severely compromise therapeutic efficacy.
Advanced age and severe comorbidity were associated with lower predicted probabilities of response.
Reduced dose interventions correlated with altered model predictions in specific subpopulations.
Clinical Implications
Clinicians should consider the CIRS-G score when assessing treatment plans for older CP-CML patients on flumatinib. Understanding the impact of comorbidities can help tailor therapy to improve patient outcomes and manage potential risks effectively.
Conclusion
CIRS-G is the most effective comorbidity scoring system for predicting molecular responses in CP-CML patients treated with flumatinib. Machine learning techniques enhance the understanding of complex interactions between comorbidities and treatment efficacy.
