Variance component mixture modelling for longitudinal T-cell receptor clonal dynamics (2502.04553v1)

Abstract: Studies of T cells and their clonally unique receptors have shown promise in elucidating the association between immune response and human disease. Methods to identify T-cell receptor clones which expand or contract in response to certain therapeutic strategies have so far been limited to longitudinal pairwise comparisons of clone frequency with multiplicity adjustment. Here we develop a more general mixture model approach for arbitrary follow-up and missingness which partitions dynamic longitudinal clone frequency behavior from static. While it is common to mix on the location or scale parameter of a family of distributions, the model instead mixes on the parameterization itself, the dynamic component allowing for a variable, Gamma-distributed Poisson mean parameter over longitudinal follow-up, while the static component mean is time invariant. Leveraging conjugacy, one can integrate out the mean parameter for the dynamic and static components to yield distinct posterior predictive distributions whose expressions are a product of negative binomials and a single negative multinomial, respectively, each modified according to an offset for receptor read count normalization. An EM-algorithm is developed to estimate hyperparameters and component membership, and validity of the approach is demonstrated in simulation. The model identifies a statistically significant and clinically relevant increase in TCR clonal dynamism among metastasis-directed radiation therapy in a cohort of prostate cancer patients.