Interpretable Prediction of Lymphopenia Following Neoadjuvant Chemoradiotherapy for rectal Cancer: An Analysis of the ARISTOTLE randomized phase 3 Trial.
Authors
Affiliations (8)
Affiliations (8)
- Department of Medical Physics and Biomedical Engineering, University College London, Gower Street, London WC1E 6BT, United Kingdom. Electronic address: [email protected].
- Department of Medical Physics and Biomedical Engineering, University College London, Gower Street, London WC1E 6BT, United Kingdom; University College London Hospitals NHS Foundation Trust, Radiotherapy Physics, 250 Euston Road, London NW1 2PG, United Kingdom.
- Leeds Institute of Medical Research, University of Leeds, Leeds, United Kingdom; Department of Medical Physics, Leeds Teaching Hospitals NHS Trust, Leeds, United Kingdom.
- Leeds Institute of Medical Research, University of Leeds, Leeds, United Kingdom.
- Department of Medical Physics and Biomedical Engineering, University College London, Gower Street, London WC1E 6BT, United Kingdom.
- University College London, Cancer Institute, London, United Kingdom.
- Mount Vernon Cancer Centre, The Radiotherapy Trials Quality Assurance Group, London, United Kingdom.
- Department of Molecular and Clinical Cancer Medicine, University of Liverpool, United Kingdom.
Abstract
This study aimed to develop a mechanistic model explaining the development of lymphopenia during neoadjuvant chemoradiotherapy and investigate how changes in lymphocyte counts relate to survival in patients with locally advanced rectal cancer. We retrospectively analyzed 2485 lymphocyte count measurements from 498 patients enrolled in the phase 3 trial ARISTOTLE across 73 centers. We developed a parametric exponential model to predict weekly lymphocyte counts. Segmentations of bone and blood vessel-rich organs were generated on computed tomography using an artificial intelligence segmentation tool, TotalSegmentator, for dose metric calculations. Covariates included in the prediction were age, gender, pretreatment lymphocyte counts, cumulative radiotherapy doses to bones, estimated blood dose from a dynamic blood flow model, and cumulative chemotherapy doses. Bootstrap Cox regression identified the cutoff for relative lymphocyte change from pretreatment to nadir (ΔL) linked to progression-free survival (PFS). The lymphocyte prediction model explained 65% of the variation in weekly lymphocyte counts in the independent test dataset (r<sup>2</sup> = 0.65, 95% CI 0.56-0.72), with a mean absolute prediction error of 0.24 (95% CI 0.21-0.29). The strongest predictor of lymphocyte counts was the radiation dose received by 95% of the estimated circulating blood volume, followed by treatment with capecitabine and irinotecan. The optimal cutoff threshold of ΔL was 0.75 ± 0.03. The categorized ΔL was significantly associated with PFS in multivariate Cox regression (hazard ratio: 1.24 [95% CI: 1.07-1.43], P < .005). This study identifies key treatment-related factors that contribute to lymphocyte depletion and the impact of categorized ΔL on PFS in locally advanced rectal cancer.