AI-Quantified Body Composition and Survival Data for 136 Cancer Patients in ADC Trials

136 patients treated with antibody-drug conjugates in early-phase clinical trials between March 2015 and March 2023 were analyzed for survival outcomes. A deep learning software, Anthropometer3DNet, automatically quantified 3D body composition parameters from pre-treatment CT scans. The data shows higher subcutaneous adipose tissue was associated with longer progression-free survival (HR=0.67) and overall survival (HR=0.66).

Use Cases

• Predicting progression-free survival based on AI-quantified body composition parameters like subcutaneous adipose tissue.
• Modeling overall survival using features such as visceral adipose tissue and Royal Marsden Hospital prognostic score.
• Analyzing the association between total adipose tissue and patient time in protocol during treatment.
• Investigating the link between body composition metrics and all-grade toxicity events in patients receiving ADCs.

Strengths

• Includes data from 136 patients with detailed clinical and AI-derived body composition metrics.
• Provides specific hazard ratios and p-values for associations between body composition features and survival outcomes.
• Covers a defined time range of patient enrollment from March 2015 to March 2023.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect temporal and institutional bias inherent to a single-center retrospective study.

Provenance

Source

Retrospective analysis of patients treated at a single institution (author Matthieu Delaye).

Collection Method

Deep learning software (Anthropometer3DNet) automatically quantified body composition from pre-treatment CT scans.

Time Range

March 2015 to March 2023

Freshness

Last updated 2026-05-13 04:56:23; freshness should be verified.