BATPC: A biology-anchored tumor progression classification for interpreting multidimensional progression phenotypes...

Area of Science:

• Oncology
• Radiology
• Biomedical Informatics

Background:

• Conventional RECIST criteria oversimplify tumor progression, labeling diverse biological patterns as 'progressive disease'.
• The lack of a standardized vocabulary for tumor progression hinders clinical decision-making, trial interpretation, and real-world evidence utilization, especially with multimodal monitoring and immunotherapy.
• Existing methods fail to capture the heterogeneity of tumor progression, impacting treatment efficacy and patient outcomes.

Purpose of the Study:

• To develop the Biology-Anchored Tumor Progression Classification (BATPC), a novel taxonomy for describing tumor progression.
• To integrate multimodal signals including anatomical, metabolic, immune, and temporal data for a comprehensive assessment.
• To establish a standardized, interpretable vocabulary for clinically actionable tumor progression phenotypes.

Main Methods:

• Developed BATPC, a rule-based taxonomy integrating anatomical, metabolic/functional, immune/microenvironmental, treatment-context, and temporal signals.
• Scored each domain on a 0-3 rubric, using deterministic rules to map patterns to ten canonical progression phenotypes.
• Constructed a conceptual synthetic cohort to illustrate BATPC's interpretation of heterogeneous multimodal signals.

Main Results:

• BATPC differentiated ten biologically coherent progression phenotypes, including indolent drift, metabolic escape, immune flare/pseudoprogression, oligoprogression, sanctuary-site progression, functional collapse, and hyperprogression.
• The framework resolved patterns missed by size-based criteria, offering interpretive granularity across multiple dimensions.
• Preliminary intra-rater reliability showed 90% exact phenotype agreement (κ=0.88) on a synthetic cohort.

Conclusions:

• BATPC reframes tumor progression as a multidimensional biological process, providing a structured, interpretable vocabulary.
• The framework generates clinically actionable insights invisible to conventional endpoints and is compatible with EHRs, real-world evidence, and clinical trials.
• Future validation using real-world datasets and prospective imaging cohorts is planned.