[Paper Review] Identifying expanding TCR clonotypes with a longitudinal Bayesian mixture model and their associations with cancer patient prognosis, metastasis-directed therapy, and VJ gene enrichment

David Swanson, Alexander D. Sherry+3 more|arXiv (Cornell University)|Jan 8, 2026

T-cell and B-cell Immunology0 citations

TL;DR

The paper introduces a longitudinal Bayesian mixture model to identify expanding, contracting, and static TCR clonotypes over time, relates dynamic clonotypes to MDT and prognosis, and uses penalized log-linear modeling to study VJ gene enrichment.

ABSTRACT

Examination of T-cell receptor (TCR) clonality has become a way of understanding immunologic response to cancer and its interventions in recent years. An aspect of these analyses is determining which receptors expand or contract statistically significantly as a function of an exogenous perturbation such as therapeutic intervention. We characterize the commonly used Fisher's exact test approach for such analyses and propose an alternative formulation that does not necessitate pairwise, within-patient comparisons. We develop this flexible Bayesian longitudinal mixture model that accommodates variable length patient followup and handles missingness where present, not omitting data in estimation because of structural practicalities. Once clones are partitioned by the model into dynamic (expanding or contracting) and static categories, one can associate their counts or other characteristics with disease state, interventions, baseline biomarkers, and patient prognosis. We apply these developments to a cohort of prostate cancer patients who underwent randomized metastasis-directed therapy or not. Our analyses reveal a significant increase in clonal expansions among MDT patients and their association with later progressions both independent and within strata of MDT. Analysis of receptor motifs and VJ gene enrichment combinations using a high-dimensional penalized log-linear model we develop also suggests distinct biological characteristics of expanding clones, with and without inducement by MDT.

Motivation & Objective

• Motivate the analysis of TCR clonality dynamics (expanding/contracting vs static) in cancer.
• Develop a two-component longitudinal Bayesian mixture model to accommodate variable follow-up and missing data.
• Link dynamic clonotypes to clinical outcomes, baseline biomarkers, and treatment (MDT).
• Assess VJ gene enrichment patterns using a penalized log-linear framework to identify meaningful interactions.

Proposed method

• Propose a two-component Poisson-based mixture model with dynamic (lambda_ijk) and static (lambda_ij) rates for clonal counts, incorporating Gamma priors and hierarchical sharing of hyperparameters across patients.
• Fit the model with Hamiltonian Monte Carlo in STAN to obtain posterior probabilities of dynamic membership for each clone.
• Compare the Bayesian mixture approach to Fisher’s exact test/ Beta-binomial methods and emphasize MAR handling and flexible follow-up.
• Apply a penalized fully saturated log-linear model with L1 regularization to test for VJ gene enrichment interactions with Clonotype expansion status, selecting the penalty by cross-validation.

Experimental results

Research questions

• RQ1 Can a longitudinal Bayesian mixture model reliably classify TCR clonotypes as expanding, contracting, or static across irregular follow-up times?
• RQ2 Are expanding clonotypes associated with metastasis-directed therapy and with progression-free survival in prostate cancer?
• RQ3 What are the relationships between clonal dynamics and baseline biomarkers, including potential interactions with MDT?
• RQ4 Do expanding clonotypes show distinct VJ gene family enrichment patterns when accounting for clonal dynamism?
• RQ5 Do receptor motifs or sequence-level characteristics differ between MDT-induced and non-MDT-induced expanding clones?

Key findings

• MDT significantly increases the number of expanding clones in the baseline-followup analysis.
• Counts of expanding clones show prognostic associations with progression-free survival, varying by MDT strata and model specification.
• Penalized log-linear modeling reveals significant enrichment of certain V and J gene family combinations, and specific expansion-related interactions (e.g., TCRBJ02*TCRBV02..BV29) indicate non-random enrichment patterns.
• Baseline biomarkers (e.g., upr, il15, vegfa) show associations with clonal expansions or contractions, with some interactions (gmcsf*mip1b, il15*vegfa) emerging under penalized regression.
• Lorenz-curve analyses suggest clonal expansions tend to translate rather than scale in frequency shifts.
• MDT status modulates both the number of dynamic clones and their prognostic associations, highlighting confounding in interpreting expansion-PFS links.