Forecasting Complications in Chemotherapy-Induced Stem Cell Mobilization for Multiple Myeloma
Overview
In a cohort of 109 multiple myeloma patients undergoing chemotherapy-based stem cell mobilization (SCM), 97% achieved successful stem cell collection, but 69% experienced severe adverse events (AEs) requiring hospitalization. Machine learning models were developed to predict AE occurrence and timing, demonstrating high accuracy for some complications and enabling risk-stratified outpatient management.
Background
Autologous stem cell transplantation remains a cornerstone treatment for multiple myeloma. Traditionally, inpatient chemotherapy-based SCM is standard in Germany, but outpatient approaches could alleviate healthcare resource constraints. Predicting complications during SCM is critical to safely implementing outpatient protocols. This study analyzed safety outcomes and applied machine learning to forecast adverse events and optimize clinical management.
Data Highlights
Parameter
Value
Number of patients
109
Successful stem cell collection
97%
Severe adverse events requiring hospitalization
69%
ROC-AUC for predicting elevated creatinine
1.0
ROC-AUC for predicting neutropenic fever
0.67
Mean error in forecasting AE onset
~1 day
Key Findings
High success rate (97%) of stem cell collection in chemotherapy-based SCM for multiple myeloma.
Majority (69%) of patients experienced severe adverse events necessitating hospitalization.
Machine learning classification models accurately predicted certain AE types, such as elevated creatinine (ROC-AUC 1.0).
Prediction of neutropenic fever was less accurate (ROC-AUC 0.67), indicating challenges in forecasting this complication.
Regression models forecasted the timing of AE onset with a mean error of just over one day, aiding clinical scheduling.
Simulations suggest that a risk-stratified outpatient SCM protocol could reduce inpatient bed usage by at least one third without compromising patient safety.
Clinical Implications
These findings support the feasibility of implementing risk-adapted outpatient SCM protocols in multiple myeloma, potentially reducing hospital resource utilization. Accurate prediction of specific adverse events can guide monitoring intensity and timing, improving patient safety. However, limitations in predicting neutropenic fever highlight the need for cautious clinical oversight.
Conclusion
Data-driven risk assessment models enable safer outpatient chemotherapy-based stem cell mobilization in multiple myeloma, optimizing resource allocation while maintaining high safety standards. Further refinement of predictive tools is warranted to enhance management of challenging complications.
References
Rajkumar SV 2024 -- Multiple myeloma: 2024 update on diagnosis, risk-stratification, and management
Pompa A et al. 2016 -- Outpatient stem cell mobilization with intermediate-dose cyclophosphamide is a safe and effective procedure
Pompa A et al. 2021 -- Safety of outpatient stem cell mobilization with low- or intermediate-dose cyclophosphamide in newly diagnosed multiple myeloma patients
Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group 2012 -- KDIGO clinical practice guideline for acute kidney injury
Larsen K et al. 2020 -- Feasibility of outpatient autologous stem cell transplantation in multiple myeloma and risk factors predicting hospital admission
