Clonal Supply, Local Maintenance, and the Immunologic Hand-Off During Checkpoint Blockade
Tertiary lymphoid structures and tumor-draining lymph nodes are often discussed as if they sit on a single immune conveyor belt. This review argues for a more exact framing: the same responding clone may be primed in a node, expanded in blood, supported in a TLS, and functionally read out in the tumor, but those roles cannot be assumed to collapse into one another. The distinction matters because checkpoint blockade changes timing, trafficking, and local survival signals at once. When those pieces are examined separately, the biologic picture becomes easier to confuse; when they are traced together, the source of response, the site of maintenance, and the site of effector activity can be assigned more carefully.
A central theme is clonal provenance. In experimental systems, tumor-antigen-specific TCF1-positive CD8 T cells reside in tumor-draining nodes and can generate tumor-infiltrating progeny. Depleting MYB-dependent nodal progenitors reduces infiltration and checkpoint benefit in the tested models, while blocking lymphocyte egress with fingolimod lowers the tumor-associated precursor pool. Those observations support a continuing external supply of cells during therapy rather than a one-time priming event. At the same time, they do not prove that every productive response remains node-dependent once a TLS is established. A smaller late effect of node removal can reflect prior priming, previously recruited cells, or partial local autonomy, not a single mechanism.
Checkpoint blockade shifts the burden of proof onto timing. Early removal of tumor-draining nodes reduced efficacy more than later removal in mouse experiments, while late restriction of lymphocyte egress had little impact on TLS number after formation in a separate stage-resolved model. Early restriction of lymphocyte egress impaired TLS formation; later restriction did not erase established structures. This is the kind of divergence that makes stage matter more than anatomy alone. In the murine lung-cancer model examined by Mattiuz and colleagues, depletion of type 1 conventional dendritic cells after TLS establishment disrupted germinal centers and adaptive immune output, showing that maintenance still depends on active cellular support even after a structure is formed.
Why TDLNs Still Matter After the First Wave of Priming
Antigen delivery into the draining node remains the first gate in the process. Altered ACKR4 expression reduced dendritic-cell migration, nodal T-cell expansion, and PD-1 blockade efficacy in experimental tumors. Disrupting sentinel lymphatic drainage or CCR7-positive dendritic-cell migration similarly impaired cross-presentation and infiltration during sequential immunoradiotherapy. These experiments do not describe the same cell type in every setting; they show that antigen delivery, dendritic-cell trafficking, and the responding lineage all shape whether a treatment-period reservoir is built at all.
The node does more than initiate activation. Paired studies show that nodal activation and tumor entry can be separated. In one mouse model of PD-L1 blockade, inhibiting CXCR3-dependent trafficking preserved much of the nodal response while reducing tumor infiltration and benefit. In another system, selective resection of a mapped dominant draining node or systemic egress inhibition impaired anti-PD-1 efficacy in orthotopic mouse bladder cancer. The biological implication is straightforward: a node may remain immunologically active yet fail to feed the tumor with effective descendants if trafficking is interrupted.
Human data point in the same direction, although with different limitations. Prior extensive dissection of non-metastatic draining nodes in recurrent biliary tract cancer was associated with a shorter duration of immunotherapy benefit, and stem-like cells in archived nodes were associated with outcomes. Archived tissue can show that a reservoir existed, but it cannot measure ongoing export during treatment. That gap is where receptor sequencing, longitudinal sampling, and spatial profiling become essential.
TLSs as Sites of Support, Not Automatic Independence
Intratumoral TLS studies add a second layer of complexity. In untreated renal cell carcinoma, selected TLS-associated CD8 T cells recognized autologous tumor antigens and retained a stem-like transcriptional program. In neoadjuvant-treated breast cancer, progenitor-exhausted T cells interacted with B cells that supported proliferation and effector differentiation. These findings make TLSs look like local support stations, not passive aggregates of lymphocytes. The local environment can sustain proliferation, antigen recognition, and differentiation without requiring every replenishment event to originate inside the tumor.
Local support, however, is not the same as local autonomy. A TLS can maintain reactive cells while still depending on external replenishment of new clones, ongoing dendritic-cell input, or help from a node that continues to seed the response. The review’s framework separates cellular replenishment from local support and from distal antibody-mediated effects because the same treatment period can include all three. In glioma models, cervical nodes supported B-cell and plasma-cell responses whose circulating antibodies promoted macrophage-mediated tumor-cell phagocytosis. Those data show that nodal influence can reach the tumor through soluble products rather than direct cellular migration through a TLS.
B cells deserve particular attention in this setting. TLS maturation, germinal center activity, and plasma-cell differentiation are not decorative features; they create the conditions for antigen-specific humoral output and shape how checkpoint therapy is executed locally. When cDC1s are depleted after TLS establishment, germinal centers and adaptive output weaken, showing that TLS function remains biologically contingent. A structure can persist while its immune productivity falls.
Clonal Identity Across Compartments Requires Paired Sampling
Shared T-cell receptor sequences across paired TDLN and tumor specimens provide direct molecular evidence of clonal continuity. In breast cancer, sentinel nodes and autologous tumors share highly expanded T-cell clonotypes, and functional assays identify tumor-reactive cells among them. In paired single-cell transcriptional and TCR analyses, precursor-like and differentiated states can be traced across nodes and tumors. Defined nodal memory-like cells can generate Tpex-like progeny in blood and tumors in the tested system. Those are not interchangeable findings. Sequence overlap shows shared ancestry; lineage tracing shows directionality; function shows whether the clone is doing useful work at the destination.
The same logic applies to B-cell receptor work. Shared clonotypes across compartments can imply a common lineage, but they do not by themselves show where the clone was maintained, where it expanded, or whether antibodies rather than cells carried the dominant effector role. The review repeatedly returns to this point because the literature mixes cellular and humoral routes, sometimes in the same disease model.
What This Framework Changes for Bedside Interpretation
The practical consequence is sharper interpretation of checkpoint response. If a patient responds, the response may reflect nodal priming, ongoing export from TDLNs, local support inside TLSs, or distal antibody-mediated effects. If a patient fails, the bottleneck may be antigen delivery, precursor maintenance, egress, local survival, or effector routing. Those are different failures and they do not lead to the same therapeutic fix.
This review does not claim a universal sequence from node to TLS to tumor. It argues for a testable hierarchy of dependence based on timed perturbation, paired compartment sampling, and clonal tracing. The remaining uncertainty is not whether TDLNs and TLSs matter; both clearly do. The harder question is which compartment is supplying the relevant clone at a given moment, which site is keeping it alive, and which route is actually responsible for the antitumor effect under checkpoint blockade.