Cureit: An End-to-End Pipeline for Implementing Mixture Cure Models With an Application to Liposarcoma Data

Area of Science:

• Oncology
• Biostatistics
• Cancer Research

Background:

• Mixture cure models offer an alternative to Cox proportional hazards models for oncology studies with a cure fraction.
• Existing software for cure models lacks comprehensive tools for evaluation, reporting, and visualization.

Purpose of the Study:

• Introduce the cureit R package, an end-to-end pipeline for building and evaluating mixture cure models.
• Demonstrate the utility of cureit in analyzing liposarcoma patient data to understand subtype-specific survival outcomes.

Main Methods:

• Utilized the cureit R package to fit and evaluate mixture cure models.
• Assessed associations between liposarcoma histologic subtypes and disease-specific death (DSD).
• Defined liposarcoma subtypes as well-differentiated, dedifferentiated, myxoid, round cell, and pleomorphic.

Main Results:

• Compared to well-differentiated liposarcoma, myxoid (OR, 6.25) and round cell (OR, 16.2) subtypes showed higher DSD incidence.
• Dedifferentiated liposarcoma was linked to a delayed DSD (HR, 10.6), while pleomorphic liposarcoma exhibited significantly higher risk in both incidence and latency (P < .0001).
• Brier scores confirmed comparable predictive accuracy between cure and Cox models.

Conclusions:

• The cureit pipeline effectively fits, evaluates, and visualizes mixture cure models.
• The study provides insights into subtype-specific associations with disease-specific death incidence and latency in liposarcoma patients.