Blood Immune Signatures Separate Relapsing From Non-Relapsing Melanoma Patients
Resected stage III melanoma still forces a difficult trade-off in the adjuvant setting. Anti-PD-1 therapy can lower recurrence risk, yet some patients relapse anyway while others develop immune-related toxicities serious enough to interrupt treatment. In this cohort of 39 patients treated with pembrolizumab after surgery, longitudinal blood profiling uncovered immune patterns that tracked with both tumor recurrence and adverse effects.
The study centered on a practical clinical question: can circulating immune cells and soluble checkpoints reveal which patients are headed toward benefit, relapse, or toxicity before the outcome becomes clinically obvious? The investigators sampled blood at baseline and again at 6 and 12 weeks, then used multiparametric flow cytometry and multiplex immunoassays to examine dendritic cells, monocytes, T cells, NK cells, iNKT cells, γδ T cells, soluble checkpoints, activation markers, and functional cytokine responses. The breadth of that approach matters. Anti-PD-1 therapy acts through a network, not a single cell type, and the biology of resistance often emerges from coordinated changes across innate and adaptive compartments.
Innate Immunity Carried Much of the Signal
The clearest message from the immune profiling was that innate and adaptive immunophenotypes diverged between non-progressing patients and those who relapsed during treatment. Dendritic cells and monocyte subsets emerged as central elements of that split, with distinct signatures associated with recurrence and with immune-related adverse effects. In melanoma, dendritic cells are not passive bystanders; they govern antigen presentation, T-cell priming, and the tone of the downstream response. If those cells drift toward dysfunctional or suppressive states, anti-PD-1 may release an exhausted immune system without restoring a coordinated antitumor program.
Particular attention fell on DC3 and monocyte subsets. Their prominence is biologically plausible because these cells sit at the intersection of antigen handling, inflammatory signaling, and myeloid suppression. Monocytes can support antitumor immunity in one setting and fuel immunosuppressive myeloid programs in another, including pathways that resemble myeloid-derived suppressor cells or pro-tumor macrophage differentiation. In a postoperative melanoma patient starting pembrolizumab, that balance may shape whether the immune system expands durable tumor control or drifts into ineffective inflammation.
A distinct immune phenotype did not just accompany relapse; it also appeared alongside toxicity. The same broad monitoring strategy identified immune profiles associated with adverse effects, suggesting that immune activation and immune misdirection may share overlapping cellular circuitry. Clinically, that raises an uncomfortable but important point: the immune signatures that predict successful checkpoint engagement may not be identical to those that predict tolerability.
Soluble Checkpoints Added a Second Layer of Risk Stratification
Among the most interesting candidate biomarkers were plasma ULBP3/4 and CD73. Those soluble markers were part of the predictive signatures linked to clinical outcome, and they fit neatly into the biology of checkpoint therapy resistance. ULBP family ligands participate in stress surveillance, while CD73 contributes to the generation of immunosuppressive adenosine. High adenosine tone can blunt effector-cell activity, weaken antigen-specific responses, and create a biochemical environment where PD-1 blockade alone may be insufficient.
The study also drew attention to newer checkpoints, especially Gal9/TIM3 and CD155/TIGIT. These pathways are increasingly relevant in cancer immunology because they reflect alternative inhibitory circuits that can take over when PD-1 is blocked. If the PD-1 axis is released but TIM-3 or TIGIT signaling remains active, T cells and NK cells may stay functionally constrained. The result is not necessarily a clean immunologic reawakening; it can be a partial response with persistent exhaustion, compensatory inhibition, or both.
That biological logic strengthens the case for combination strategies. The data point toward a future in which anti-PD-1 therapy in adjuvant melanoma may need to be paired with agents that target myeloid suppression, adenosine biology, or alternative checkpoints in selected patients rather than delivered as a uniform single-agent approach.
Functional Immune Readouts Mattered as Much as Cell Counts
A major strength of the work was its emphasis on immune function, not just frequency. The analysis included activation status, natural cytotoxicity receptor expression, transcription factors linked to helper T-cell polarization, and cytokine or chemokine secretion after stimulation. That design captures a crucial truth about checkpoint therapy: a normal-looking cell count can conceal a deeply altered immune program. A dendritic cell that cannot prime effectively, or a T-cell subset locked into exhaustion, is not clinically equivalent to a cell that retains cytotoxic or antigen-presenting competence.
The investigators also compared patients with and without adverse events inside the non-progressor group, then examined immune shifts in patients who progressed from baseline to progression time. This layered structure makes the findings more clinically useful than a simple responder-versus-nonresponder split. It suggests that different immune trajectories may accompany recurrence and toxicity, even within patients who initially appear to be doing well.
One sentence captures the bedside implication plainly: early peripheral blood changes may foreshadow both treatment failure and treatment harm.
Why This Matters for Stage III Melanoma Care
Adjuvant pembrolizumab is already standard therapy for resected high-risk melanoma, but standard use has outpaced biomarker development. Many patients receive the drug without any reliable way to estimate recurrence risk or the likelihood of immune toxicity. This study adds to the argument that peripheral blood can serve as a clinically accessible window into checkpoint response, especially when the analysis includes myeloid subsets, soluble checkpoints, and functional outputs together.
The cohort was modest and single-center, and the work remains exploratory rather than practice changing. Still, the direction is clear enough to matter. The immune network most likely to predict outcome in adjuvant melanoma is not limited to PD-1 expression on T cells. It spans dendritic cells, monocytes, NK-lineage cells, γδ T cells, soluble ligands, and alternate inhibitory axes that can redirect the immune response away from durable tumor control.
For clinicians, that means the biology of relapse after surgery may already be visible in blood weeks after treatment begins. For translational researchers, it provides a rationale to test biomarker-guided combinations in which adenosine signaling, TIM-3, or TIGIT are blocked alongside PD-1 in patients whose baseline or on-treatment immune profile suggests resistance. The unanswered clinical task is to determine which of these signatures remain stable in larger cohorts and which can be converted into a decision tool that changes management before relapse or severe toxicity declares itself.