TMBcat

TMBcat optimizes tumor mutation burden (TMB) stratification using a minimal joint p-value criterion to improve prediction of immunotherapy efficacy.

Key Features:

• Optimized Categorization: Employs a statistical framework to optimize TMB categorization across diverse cancer cohorts.
• Minimal Joint P-value Criterion: Implements a minimal joint p-value criterion to determine optimal classification thresholds.
• Multidimensional Endpoint Analysis: Applies to multidimensional endpoints, enabling joint analysis of multiple therapeutic outcomes.
• Fault-Tolerance to Measurement Error: Provides fault-tolerance to TMB measurement errors to enhance robustness of categorizations.
• Non-Linear Relationship Detection: Detects non-linear relationships between TMB and immunotherapy efficacy, noting that higher TMB does not always confer greater benefit across carcinomas.
• Flexible Stratification Thresholds: Identifies multiple classification thresholds to enable flexible differentiation of patient prognosis across cancer types.

Scientific Applications:

• Retrospective Validation in Specific Cohorts: Validated on retrospective cohorts of 78 non-small cell lung cancer patients and 64 nasopharyngeal carcinoma patients treated with anti-PD-(L)1 therapy, confirming non-linear TMB–immunotherapy outcome relationships.
• Cross-study Cohort Validation: Further validated on an assembled cohort of 943 patients from 11 published studies to support insights for therapeutic selection and treatment strategies.

Methodology:

Applies a statistical framework implementing a minimal joint p-value criterion to optimize TMB categorization across multidimensional endpoints and incorporates fault-tolerance to TMB measurement errors.