Conditional survival patterns and individualized prognostic prediction in malignant peritoneal mesothelioma

Area of Science:

• Oncology
• Biostatistics
• Epidemiology

Background:

• Malignant peritoneal mesothelioma (MPM) is a rare, aggressive cancer with poor prognosis and limited treatment options.
• Low incidence hinders large-scale studies on survival outcomes and treatment impact, necessitating advanced analytical methods.

Purpose of the Study:

• To apply conditional survival (CS) analysis to assess survival trends in MPM patients.
• To explore the survival benefits of chemotherapy using propensity score matching (PSM).
• To develop and validate a nomogram for CS prediction in MPM.

Main Methods:

• Conditional survival (CS) analysis on SEER data (2000-2019) for 1,549 MPM patients.
• Machine learning algorithms (Random Survival Forest, LASSO regression) and Cox models for nomogram development.
• Propensity score matching (PSM) to evaluate chemotherapy's effect on survival.

Main Results:

• CS rates improved with time post-diagnosis, indicating a need for early intervention.
• The CS-based nomogram showed strong predictive performance (C-index: 0.701 training, 0.717 validation).
• PSM analysis suggested chemotherapy is associated with improved overall survival in MPM patients.

Conclusions:

• CS analysis offers a time-updated survival estimate, aiding patient counseling and follow-up strategies.
• The CS-based nomogram provides a valuable tool for individualized risk assessment in MPM.
• Chemotherapy may offer survival benefits, warranting further investigation for MPM treatment optimization.