Development and validation of prognostic models for anal cancer outcomes using distributed learning: protocol for the
Stelios Theophanous1, Per-Ivar Lønne2, Ananya Choudhury3
1Leeds Institute of Medical Research at St James's, University of Leeds, Leeds, UK. umsth@leeds.ac.uk.
The atomCAT2 study uses distributed learning to build prognostic models for anal cancer patients treated with chemoradiotherapy. This approach enables multi-centre collaboration while protecting patient privacy, aiming to improve treatment selection and clinical trial design.
Area of Science:
- Oncology
- Radiotherapy
- Epidemiology
- Biostatistics
Background:
- Anal cancer incidence is rising, posing challenges in improving treatment outcomes and reducing toxicity despite current chemoradiotherapy standards.
- Developing robust prognostic models for rare cancers like anal cancer is difficult due to data limitations, especially from single centers.
- Multi-center model development is hindered by data privacy regulations, necessitating innovative approaches like distributed learning.
Purpose of the Study:
- To develop and validate robust prognostic models for overall survival, locoregional control, and freedom from distant metastasis in anal cancer patients.
- To leverage distributed learning for multi-center data analysis, overcoming privacy barriers and data limitations.
- To build upon the proof-of-concept from the atomCAT1 study with a larger, multi-center cohort in the atomCAT2 study.
Main Methods:
- Retrospective multi-center cohort study involving 18 radiotherapy centers.
- Extraction and organization of patient data at each participating center.
- Development and validation of Cox proportional hazards models using distributed learning for identified prognostic factors and outcomes.
Main Results:
- Summary statistics will be calculated and exchanged between centers prior to modeling.
- Cox proportional hazards models will be developed and validated across centers for overall survival, locoregional control, and freedom from distant metastasis.
- Reported outcomes at specific timepoints and factor effect estimates will enable outcome prediction for future patients.
Conclusions:
- The atomCAT2 study will analyze a large, cross-institutional cohort of anal cancer patients treated with chemoradiotherapy.
- Findings will provide insights into current international clinical practice outcomes.
- The study aims to aid personalization and design of future anal cancer trials through improved patient risk stratification.
More Related Videos
06:46Competing-Risk Nomogram for Predicting Cancer-Specific Survival in Multiple Primary Colorectal Cancer Patients after Surgery
Published on: September 27, 2024
04:09Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma
Published on: October 10, 2018
