TCR Clonotype Tracking Across Treatment Timepoints: A Study Design Guide for Longitudinal Immune Monitoring Research

T cell receptor clonotype tracking — the longitudinal monitoring of individual TCR sequences across multiple timepoints — has become one of the most informative approaches for understanding how the immune system responds to immunotherapy. Immune repertoire sequencing approaches that enable this type of longitudinal analysis have advanced substantially, making clonal tracking accessible to a broader research community. Unlike static immune profiling at a single timepoint, longitudinal TCR tracking reveals the clonal dynamics that define treatment response: which T cell clones expand after checkpoint blockade, whether CAR-T cells persist beyond the initial infusion, and how the T cell repertoire evolves during combination therapy or vaccination.

For researchers designing studies that incorporate TCR clonotype tracking, the experimental choices made before sample collection often determine whether the resulting data can answer the intended questions. Timepoint selection, sampling strategy, sequencing depth, and metadata documentation each introduce variables that propagate through the analysis pipeline. This guide organizes the key design considerations for longitudinal TCR tracking studies, drawing on methodological advances published in 2024 and 2025, to help research teams build robust study designs that generate actionable immune monitoring data.

Why Longitudinal TCR Tracking Matters

TCR clonotype tracking leverages the unique rearranged TCR sequence carried by each T cell clone as a molecular barcode that can be quantified across timepoints. This approach provides a direct window into the clonal dynamics that underlie immunotherapy response and resistance. Unlike bulk immune cell counting or surface phenotyping alone, TCR tracking answers mechanistic questions: Are tumor-reactive T cell clones expanding in the blood after treatment? Do exhausted T cell clones persist or contract under checkpoint blockade? Which clonotypes from a CAR-T product remain detectable at six months?

A landmark 2024 study by Wang and colleagues, using paired single-cell RNA and TCR sequencing in 36 stage IV melanoma patients, demonstrated that checkpoint blockade induces distinct waves of clonal T cell responses peaking at 3, 6, and 9 weeks. Combination therapy produced greater magnitude of clonal responses than either single agent, and anti-CTLA-4 drove robust expansion of progenitor exhausted CD8+ T cells that synergized with anti-PD-1. These temporal dynamics — the timing, magnitude, and composition of clonal waves — are invisible in cross-sectional designs and illustrate why longitudinal TCR tracking is essential for understanding mechanism of action.

A 2025 single-cell meta-analysis by Shorer and colleagues, encompassing 767,606 T cells across 460 samples from 163 ICI-treated patients spanning six cancer types, further validated the predictive value of longitudinal TCR dynamics. A robust expanded CD8+ T cell transcriptional signature — including CXCL13, GZMK, RBPJ, and HLA-DQA1 — consistently differentiated responders from non-responders. Clones shared between tumor and blood were more abundant in non-responders, suggesting that the tissue distribution of clonal expansions carries independent biological information. The meta-analysis achieved AUC values up to 0.85 for response prediction, demonstrating that longitudinal TCR data can contribute meaningfully to biomarker development.

Establishing a True Baseline Before Treatment

The baseline timepoint is the reference against which all subsequent clonal dynamics are measured, yet it is frequently the most underspecified element of longitudinal TCR study designs. A true baseline should capture the pre-treatment repertoire under conditions that minimize confounding variables: the sample should be collected before any therapeutic intervention, processed using the same protocol as follow-up samples, and documented with sufficient metadata to account for potential confounders such as concurrent medications, recent infections, or vaccination history.

Multiple pre-treatment samples, collected days or weeks apart, provide the most robust baseline because they allow researchers to estimate the within-subject variability of the TCR repertoire in the absence of treatment. This is particularly important given that the peripheral T cell repertoire is not static — even in healthy individuals, clonal frequencies fluctuate due to environmental exposures and homeostatic proliferation. A 2024 statistical framework from the University of Cambridge developed by Blundell demonstrated that detecting non-neutral TCR clone behavior from longitudinal data requires accounting for this baseline variability. In healthy individuals tracked over one year, most clonal fluctuations fell within expected bounds, and the framework successfully distinguished vaccine-associated clonal expansions from background noise in a glioma neoantigen vaccine trial.

For studies where multiple pre-treatment samples are not feasible, a single baseline sample collected immediately before treatment initiation, paired with careful documentation of any recent immune-modulating events, provides a practical alternative. The key requirement is that the baseline and follow-up samples are processed identically — differences in cell isolation, cryopreservation duration, or nucleic acid extraction method between baseline and on-treatment timepoints can introduce technical variation that confounds clonal dynamic measurements.

Balancing Timepoint Resolution and Practical Feasibility

The number and spacing of timepoints determine what types of clonal dynamics can be resolved. Studies with too few timepoints may capture that a clone expanded but cannot determine when the expansion began, how long it persisted, or whether it represents a sustained response.

The 2024 Cancer Cell study by Wang and colleagues provides a useful reference point for timepoint density. With sampling at approximately 3, 6, and 9 weeks during combination checkpoint blockade, the study identified distinct waves of clonal expansion that would have been missed with a single on-treatment timepoint. Anti-CTLA-4-driven progenitor exhausted T cell expansion peaked early, while anti-PD-1-driven differentiation of these cells into more exhausted states occurred at later timepoints. A study design sampling only at week 3 would have captured the expansion but not the differentiation trajectory; sampling only at week 9 would have missed the expansion entirely.

For studies where frequent sampling is constrained by patient burden or cost, a pragmatic approach is to concentrate sampling density during the period when clonal dynamics are expected to be most informative. For immune checkpoint inhibitor studies, this typically means weekly or biweekly sampling during the first 6–9 weeks of treatment, with less frequent sampling during the maintenance phase. For CAR-T therapy studies, sampling immediately before and after infusion (days 0, 7, 14, 28) captures the initial expansion and contraction dynamics, with monthly sampling thereafter to assess long-term persistence.

The decision framework balances biological resolution against patient burden, cost, and feasibility. A minimum of three post-baseline timepoints is recommended for any study that aims to model clonal trajectories rather than simply detect clone presence or absence.

Tumor Tissue and Peripheral Blood as Sampling Strategies

The choice between tumor tissue and peripheral blood as the sampling compartment depends on the research question, but the most informative longitudinal TCR studies increasingly incorporate both.

Peripheral blood offers practical advantages for longitudinal tracking: it can be sampled repeatedly with minimal patient burden, enabling high-density timepoint collections. Whole blood or PBMC samples provide sufficient T cell numbers for deep repertoire sequencing at each timepoint, and the blood repertoire captures systemic immune responses that may not be detectable within the tumor microenvironment. In a 2025 study by Cheloni and colleagues of CAR-T-treated lymphoma patients from the ZUMA-1 trial, longitudinal tracking of the peripheral TCR repertoire revealed that durable responders showed robust clonotypic expansion of native cytotoxic T cells post-infusion — findings that were accessible only through serial blood sampling and could not have been obtained from a single tumor biopsy.

Tumor tissue sampling provides complementary information that blood alone cannot access. The intratumoral TCR repertoire is enriched for tumor-reactive clones, and the dynamics of these clones within the tumor microenvironment profiling may differ substantially from their representation in blood. A 2025 study by Sentís and colleagues tracked T cell clonotypes across tumor and blood compartments over 12 months in a pediatric patient with malignant rhabdoid tumor, demonstrating that durable ICI-induced clonal expansion was detectable in both compartments but with different kinetics. Tumor-reactive TCRs identified from the tumor microenvironment could be tracked longitudinally in blood, enabling non-invasive monitoring of the antitumor response.

For teams with access to both sample types, deep TCR profiling of tumor tissue at limited timepoints (baseline biopsy, on-treatment biopsy where clinically indicated) paired with serial blood sampling at higher density captures both site-specific and systemic T cell response dimensions.

Sequencing Depth Requirements for Clonal Detection

Sequencing depth in TCR repertoire studies determines the sensitivity for detecting low-frequency clones and the precision of clonal frequency estimates. Unlike genomic DNA sequencing where depth requirements are well-standardized, TCR sequencing depth depends on the number of T cells sampled, the diversity of the repertoire, and the specific study objectives.

A major methodological advance in 2025 — TIRTL-seq, published in Nature Methods — provides a practical reference point for depth considerations. TIRTL-seq generates approximately 1 million unique paired alpha-beta TCR clonotypes from approximately 30 million T cells per sample at a cost under $2,000. In longitudinal validation, TIRTL-seq detected 980 CD8+ and 213 CD4+ clones expanding in the acute phase and 446 CD8+ and 134 CD4+ clones contracting in convalescence, demonstrating the impact of sequencing depth on the statistical power of longitudinal tracking.

Key considerations for matching sequencing depth to study objectives:

• Paired alpha-beta sequencing — provides complete receptor information and enables specificity prediction; TIRTL-seq identified 1,247 TCR-alpha partner chains compared to only 510 by 10x Genomics single-cell TCR sequencing in the same samples
• Single-chain (beta) sequencing — at equivalent cost achieves greater depth and higher sensitivity for rare clones; suitable for tracking the most abundant expanded clones (1–5 million reads per sample)
• Rare clonotype detection — tracking tumor-reactive T cells in blood or persistent CAR-T cells at late timepoints requires deeper sequencing or paired-chain approaches
• Common expanded clones — vaccine-induced or therapy-expanded clones typically reach frequencies above 0.1% of the repertoire, where moderate-depth single-chain TCR sequencing is often sufficient

A practical heuristic is to determine the minimum clone frequency that needs to be reliably detected, then calculate the required cell input and sequencing depth accordingly. BCR and TCR sequencing services that offer both paired and single-chain options allow researchers to match depth to their specific tracking objectives.

Managing Technical Variation Across Timepoints

Technical variation between timepoints is one of the most underestimated risks in longitudinal TCR studies. Differences in sample collection, processing, storage, and sequencing between baseline and follow-up timepoints can introduce systematic biases that masquerade as biological clonal dynamics or obscure genuine signals.

The most critical source of technical variation is the sample type and processing protocol. PBMCs isolated by density gradient centrifugation, cryopreserved under standardized conditions, and extracted simultaneously for nucleic acid preparation minimize batch effects. A key decision is whether to use DNA or RNA as the starting material for TCR sequencing: DNA-based approaches capture the genomic TCR rearrangement and are not affected by transcriptional activity, while RNA-based approaches provide higher sensitivity for expressed TCR sequences but are influenced by cell activation state. In longitudinal studies, consistency in the chosen approach across all timepoints matters more than the specific choice itself.

Batch effects can also arise from sequencing runs performed at different times. Concurrent sequencing of all timepoint samples from the same patient — rather than sequencing timepoints as they are collected — eliminates inter-run technical variation and is strongly recommended. When concurrent sequencing is not possible, including a reference sample (such as a commercially available PBMC standard) in each run enables batch correction and quality monitoring.

The 2024 Cambridge framework by Blundell emphasized the importance of contamination control in longitudinal TCR data. Cross-contamination between samples sequenced in the same run can create false positives in clonal sharing analyses, and the study specifically developed a decontamination pipeline to address this issue. Metadata documentation — including collection date, processing personnel, reagent lots, and sequencing run identifiers — should be captured prospectively to enable investigation of unexpected technical variation during analysis.

Interpreting Clonal Expansion, Contraction, and Persistence

Once longitudinal TCR sequencing data are generated, the analytical challenge shifts to distinguishing meaningful clonal dynamics from technical noise and natural repertoire fluctuations. Several statistical frameworks have been developed specifically for this purpose.

The Cyclone algorithm, developed by Wang and colleagues for their 2024 Cancer Cell study, clusters T cell clones by their temporal trajectory patterns, enabling researchers to identify clones that expand early, expand late, contract, or remain stable across treatment timepoints. Cyclone was used to identify distinct waves of clonal responses to checkpoint blockade — a finding that emerged only through trajectory-based clustering rather than pairwise comparisons between individual timepoints.

Pavlova and colleagues (2024) developed VDJtrack, which uses a log-linear modeling framework to assess factors affecting clonal survival and expansion. The approach groups clonotypes by their frequency class (singletons, doubletons, expanded clones) and models recapture probability as a function of clone size and sample diversity. For detecting significantly expanded clones between two timepoints, the study applied edgeR — a negative binomial model originally developed for RNA-seq differential expression — with thresholds of log2 fold change ≥ 5 and adjusted p ≤ 0.01.

A 2025 Bayesian approach by Swanson and colleagues introduced a longitudinal mixture model that accommodates variable follow-up durations and missing timepoints — practical features that address real-world constraints in clinical studies. Applied to prostate cancer patients undergoing metastasis-directed therapy, the model identified significant increases in clonal expansions among treated patients that correlated with clinical outcomes.

For research teams without dedicated bioinformatics support, key interpretive principles include: focusing on clones that exceed a minimum frequency threshold at any timepoint (typically > 0.01–0.05% of the repertoire), requiring consistent direction of change across multiple timepoints rather than relying on pairwise comparisons alone, and validating unexpected clonal dynamics through independent assays such as tetramer staining or functional testing where possible.

FAQ

Key Takeaways for Longitudinal Study Design

• Baseline sampling establishes the reference for all clonal dynamics; multiple pre-treatment timepoints provide the most robust baseline, but a single well-documented sample immediately before treatment is a practical alternative
• Timepoint density determines the resolution of clonal trajectories; at least three post-baseline timepoints are recommended, with weekly or biweekly sampling during the period of expected maximal clonal activity
• Peripheral blood enables high-density non-invasive sampling and captures systemic immune responses; tumor tissue provides site-specific clonal information — combining both compartments yields the most complete picture
• Sequencing depth requirements depend on the minimum clone frequency that must be reliably detected; TIRTL-seq and similar deep profiling approaches enable detection of rare clonotypes that single-chain or low-depth sequencing may miss
• Consistent sample processing across all timepoints, concurrent sequencing of longitudinal samples from the same patient, and prospective metadata documentation are essential for minimizing technical variation
• Statistical frameworks including trajectory clustering (Cyclone), log-linear modeling (VDJtrack), and Bayesian mixture models provide complementary approaches for identifying biologically meaningful clonal dynamics
• For researchers new to TCR clonotype tracking, pilot studies with a small number of patients and frequent timepoints can establish feasibility and inform power calculations before scaling to larger cohorts

For Research Use Only. Not for use in diagnostic procedures.