Elicit: Restoring T-Cell Activity via PD-1 Blockade
How does PD-1 blockade restore T-cell activity in pembrolizumab-treated tumors?
PD-1 blockade restores T-cell activity by releasing PD-1-mediated suppression of preexisting exhausted tumor-infiltrating lymphocytes, enabling IFN-γ signaling that drives their clonal expansion, proliferation, and cytotoxic function.
Abstract
PD-1 blockade restores T-cell activity primarily through activation of preexisting tumor-infiltrating lymphocytes rather than recruitment of new T-cell clones. Across melanoma and head and neck cancer, restoration occurred via clonal expansion of exhausted CD8+ TILs with tissue-resident memory programs, increased CD8+ memory T cells, and enhanced granzyme B and Ki67 expression indicating proliferation and cytotoxic capacity. Mechanistically, this operated through release of PD-1-mediated suppression, enabling IFN-γ signaling pathways that drove T-cell proliferation, antigen presentation, and cytotoxic function. In head and neck cancer, restoration was further supported by enhanced K+ channel activity and Ca2+ signaling, which provided metabolic infrastructure for sustained T-cell activity. Restoration proceeded rapidly—within 5 weeks—through a biphasic pattern of immediate metabolic changes followed by durable functional enhancement. However, in pancreatic cancer, while PD-1 blockade successfully reactivated T cells as evidenced by increased cell cycle entry and IFNγ signature, concurrent upregulation of NF-κB signaling counteracted these benefits, demonstrating that successful restoration requires absence of alternative inhibitory pathways.
Baseline tumor immune composition determined restoration capacity across all cancer types studied. Responding tumors had higher baseline densities of CD8+, PD-1+, and PD-L1+ cells at tumor margins, preexisting ZNF683+CTX+ TILs, and T cell-inflamed gene expression signatures. The degree of T-cell restoration correlated directly with clinical outcomes, as proliferating intratumoral CD8+ T cells correlated with radiographic tumor reduction and enhanced functional measures associated with durable responses. These findings indicate that PD-1 blockade does not generate de novo antitumor responses but reinvigorates existing exhausted T cells, making adequate baseline T-cell infiltration a prerequisite for clinical benefit.
Methods
We analyzed 10 sources from an initial pool of 200, using 9 screening criteria. Each paper was reviewed for 8 key aspects that mattered most to the research question. More on methods
Records from Elicit search
n = 200
Papers screened using: Patient Population, Pembrolizumab Treatment, T-cell Functional Measurements, Sample Analysis Type, Study Design, Immunological Data Inclusion, Pembrolizumab-Specific Data, Human Clinical Samples, Cancer Patient Population
n = 200
Papers screened out
n = 190
Papers included for extraction
n = 10
Paper search
We performed a semantic search across over 138 million academic papers from the Elicit search engine, which includes all of Semantic Scholar and OpenAlex.
We ran this query: “How does PD-1 blockade restore T-cell activity in pembrolizumab-treated tumors?”
The search returned 200 total results from Elicit.
We retrieved 200 papers most relevant to the query for screening.
Screening
We screened in sources based on their abstracts that met these criteria:
- Patient Population: Does the study involve patients with any type of malignant tumor?
- Pembrolizumab Treatment: Does the study include pembrolizumab administered as monotherapy or combination therapy?
- T-cell Functional Measurements: Does the study measure T-cell functional parameters (such as proliferation, cytokine production, cytotoxic activity, or T-cell infiltration) before and/or after pembrolizumab treatment?
- Sample Analysis Type: Does the study include tumor tissue analysis or peripheral blood T-cell analysis?
- Study Design: Is the study a randomized controlled trial, cohort study, case-control study, case series with ≥5 patients, systematic review, or meta-analysis?
- Immunological Data Inclusion: Does the study report immunological or T-cell functional data (not solely clinical outcomes like response rates or survival)?
- Pembrolizumab-Specific Data: Does the study include pembrolizumab data (not focusing solely on other PD-1 inhibitors like nivolumab or cemiplimab without pembrolizumab data)?
- Human Clinical Samples: Does the study use human tumor samples from pembrolizumab-treated patients (not solely in vitro cell line studies without clinical samples)?
- Cancer Patient Population: Does the study focus on cancer patients (not solely healthy volunteers or non-cancer populations)?
We considered all screening questions together and made a holistic judgement about whether to screen in each paper.
Data extraction
We asked a large language model to extract each data column below from each paper. We gave the model the extraction instructions shown below for each column.
- Study Population:
Extract patient and tumor characteristics relevant to PD-1 blockade and T-cell restoration, including:
Cancer type and stage
Sample size
Baseline T-cell infiltration status or immune phenotype
Prior treatments that could affect T-cell function
Patient demographics if relevant to immune response
Pembrolizumab Protocol:
Extract pembrolizumab treatment details including:
Dosing regimen and schedule
Treatment duration or number of cycles
Combination with other therapies (chemotherapy, radiation, other immunotherapies)
Timing of sample collection relative to treatment (baseline, on-treatment, post-treatment)
T-Cell Activity Measures:
Extract all measures of T-cell activity and function that were assessed for restoration by PD-1 blockade, including:
Specific T-cell functions measured (cytotoxicity, proliferation, migration, cytokine production, etc.)
T-cell phenotypes and activation markers (CD8+, memory subsets, exhaustion markers like PD-1, activation markers)
Functional assays used (flow cytometry panels, functional tests, gene expression)
Baseline vs. post-treatment changes in these measures
Restoration Mechanisms:
Extract mechanistic findings about HOW PD-1 blockade restores T-cell activity, including:
Molecular pathways identified (NF-κB, IFN-γ signaling, etc.)
Cellular mechanisms (reversal of exhaustion, memory cell expansion, etc.)
Ion channel or metabolic changes affecting T-cell function
Gene expression changes related to T-cell restoration
Any proposed causal mechanisms for the restoration process
Predictive Biomarkers:
Extract biomarkers or signatures that predict or correlate with T-cell restoration by pembrolizumab, including:
Baseline immune signatures or gene expression profiles
T-cell phenotype markers that predict response
Changes in immune cell ratios (CD8/Treg, etc.)
Early indicators of T-cell reinvigoration
Any validated or proposed biomarker assays for restoration
Anatomical Patterns:
Extract where T-cell restoration occurs and any location-specific differences, including:
Tumor tissue vs. peripheral blood vs. other anatomical sites
Differences in restoration patterns between locations
Migration or trafficking of restored T-cells
Comparison of intratumoral vs. circulating T-cell changes
Temporal Dynamics:
Extract timing and kinetics of T-cell restoration, including:
How quickly restoration occurs after pembrolizumab initiation
Duration of restoration effects
Early vs. late changes in T-cell activity
Persistence of restoration after treatment
Any time-dependent patterns in restoration mechanisms
Clinical Correlations:
Extract relationships between T-cell restoration and clinical outcomes, including:
- Correlation between degree of T-cell restoration and treatment response
- Differences in restoration patterns between responders vs. non-responders
- Association between specific restoration mechanisms and survival outcomes
- Whether restoration predicts long-term treatment benefit
Results
Characteristics of Included Studies
All 10 sources examined T-cell restoration mechanisms following pembrolizumab treatment across diverse cancer types. The studies varied in cancer type, sample size, treatment protocols, and analytical approaches.
| Study | Full text retrieved? | Cancer type | Sample size | Pembrolizumab protocol | Main T-cell measures | Study design |
|---|---|---|---|---|---|---|
| G. Oliveira et al., 2023 | No | Locally advanced HNSCC | 14 tumor biopsies (6 matched pre-post) | Two doses neoadjuvant | Single-cell analysis, TCR sequencing, CD8+ TILs with ZNF683 expression | Phase 2 trial |
| M. Rijnders et al., 2021 | No | Metastatic urothelial cancer | 56 patients | 200 mg IV every 3 weeks | Multiplex flow cytometry, Th1 cell density, PD-1+ CD4+ T cells | Prospective cohort |
| A. Ribas et al., 2016 | No | Melanoma | 53 patients, 102 biopsies | Not specified | Multicolor flow cytometry, CD8+ memory T cells | Primary study |
| A. Ribas et al., 2016a | No | Melanoma (at least some patients) | 53 patients, 102 biopsies | Not specified | Multicolor flow cytometry, CD8 T memory cells | Primary study |
| Lestat R. Ali et al., 2023 | No | Advanced PDAC | Not specified | Neoadjuvant chemoradiation + anti-PD-1 | Single-cell transcriptional profiling, TCR clonotype tracking | Comparative study |
| Alexander C. Huang et al., 2016 | No | Stage IV melanoma | 39 patients | Not specified | 16-parameter flow cytometry, granzyme B, Ki67 | Serial analysis |
| P. Tumeh et al., 2014 | Yes | Metastatic melanoma | 46 patients | 2 mg/kg Q3W, 10 mg/kg Q3W, or 10 mg/kg Q2W | Quantitative IHC (CD8, Ki67, granzyme B), TCR sequencing | Correlative study |
| Hannah S Newton et al., 2020 | Yes | HNSCC | Not specified | Single dose | Patch-clamp electrophysiology (K+ channels), Ca2+ fluxes, 3D chemotaxis | Clinical trial |
| M. Ayers et al., 2017 | Yes | 9 cancer types (melanoma, HNSCC, gastric, TNBC, bladder, anal, biliary, colorectal, ovarian) | 220 patients total | Not specified | Gene expression profiling (680-gene NanoString panel) | Multi-study analysis |
| M. Ayers et al., 2017a | No | 9 cancer types including melanoma and HNSCC | 220 patients (started with 19 melanoma patients) | Not specified | Gene expression profiling | Multi-study analysis |
The included studies spanned multiple cancer types, with melanoma and HNSCC being the most frequently studied. Sample sizes ranged from 14 tumor biopsies to 220 patients across multiple cancer types. Three studies had full text available, while seven were analyzed from abstracts only. Treatment protocols varied, with some studies using neoadjuvant approaches and others treating metastatic disease. Analytical methods ranged from flow cytometry to single-cell transcriptomics, electrophysiology, and gene expression profiling.
Mechanisms of T-Cell Restoration
Reversal of Exhaustion and Memory Cell Expansion
PD-1 blockade restored T-cell function primarily through reversal of exhaustion in preexisting tumor-specific T cells. In HNSCC, responding tumors contained clonally expanded exhausted CD8+ TILs with tissue-resident memory programs characterized by high cytotoxic potential (CTX+) and ZNF683 expression at baseline. Treatment activated these preexisting CTX+ZNF683+CD8+ TILs, paralleling loss of viable tumor. This revival of preexisting ZNF683+CTX+ TILs was identified as a major mechanism of response.
Across melanoma studies, CD8+ memory T cells emerged as the most prominent phenotype expanded intratumorally following PD-1 blockade. The frequency of T cells increased significantly in responding patients, with expansion specifically concentrated in CD8 T memory subsets. This expansion was accompanied by increased granzyme B and Ki67 expression in CD8 T cell subsets, indicating enhanced cytotoxic potential and proliferation. An early wave of proliferation gave rise to a pool of re-invigorated CD8 T cells.
In metastatic urothelial cancer, reversal of exhaustion was evidenced by decreased circulating PD-1+ CD4+ T cells following treatment. This was accompanied by expansion of memory T cells, indicated by increased 4-1BB+ CD28+ CD4+ T cells. Spatial reorganization also occurred, with juxtaposition of Th1 cells with CD8+ T cells and myeloid cells in metastatic lesions.
Molecular Signaling Pathways
Multiple molecular pathways mediated T-cell restoration. IFN-γ signaling emerged as a central driver across studies. PD-1 blockade induced an IFNγ signature characterized by IFN-γ-responsive genes related to antigen presentation, chemokine expression, cytotoxic activity, and adaptive immune resistance. This included upregulation of HLA molecules, immunoproteosome components, and checkpoint inhibitors like PD-L1 and PD-L2. Activation of downstream signaling molecules like STAT1 and CMKLR1 was also observed.
In melanoma, the release of the PD-1 immune checkpoint was associated with T-cell proliferation and increased effector function. The production of interferons by tumor-infiltrating CD8 cells induced PD-L1 expression, with increased pSTAT1 expression indicating activation of interferon signaling pathways. This led to enhanced proliferation (CD8/Ki67 double-positive cells) and cytotoxic function (granzyme B expression).
However, in pancreatic cancer, while PD-1 blockade successfully increased TCR clonotype entry into cell cycle and induced an IFNγ signature, reactivated T cells also exhibited robust NF-κB signaling. This NF-κB upregulation potentially counteracted the beneficial effects of anti-PD-1 reinvigoration, marking these T cells as non-productive contributors to antitumor immunity.
Ion Channel and Metabolic Mechanisms
In HNSCC, pembrolizumab enhanced T-cell function through increased K+ channel activity and Ca2+ signaling. Treatment increased KCa3.1 and Kv1.3 channel activity in both TILs and peripheral blood T cells. These ion channel changes were accompanied by improved Ca2+ fluxing abilities, which are crucial for T-cell cytotoxicity and chemotaxis. The molecular pathway involving increased K+ channel activity and Ca2+ fluxes enhanced T-cell cytotoxic potential and migratory ability. Importantly, reversal of exhaustion was demonstrated by improved chemotaxis in an adenosine-rich environment, indicating restoration of function even in immunosuppressive contexts.
T-Cell Clonal Dynamics
Memory cell expansion occurred through restricted TCR beta chain usage and clonal expansion, suggesting a more focused T-cell response. In HNSCC, single-cell analysis revealed that responding patients exhibited clonally expanded ZNF683+CD8+ T cells. Revival of nonexhausted persisting clones and clonal replacement were modest in responders, while nonresponders showed subsequent accumulation of highly exhausted clones. In pancreatic cancer, newly cycling clonotypes in blood matched activated T-cell clonotypes observed in the tumor, suggesting potential memory cell expansion or activation.
Predictive Biomarkers for T-Cell Restoration
Baseline Immune Signatures
Baseline immune phenotypes strongly predicted restoration capacity. In HNSCC, high numbers of CD103+PD-1+CD8+ T cells infiltrating pretreatment lesions predicted response. The presence of ZNF683+CD8+ T cells with high cytotoxic potential and tissue-resident memory programs at baseline distinguished responders from nonresponders. Conversely, nonresponder baseline tumor microenvironments exhibited relative absence of ZNF683+CTX+ TILs.
In metastatic urothelial cancer, responders had higher baseline fractions of CD4+ T cells expressing cosignaling receptors and higher baseline density of T helper-type 1 (Th1) cells (T-bet+ CD4+ T cells) compared to nonresponders.
Across melanoma studies, higher numbers of CD8+, PD-1+, and PD-L1+ cells at the invasive tumor margin and tumor center at baseline predicted response. A predictive model based on CD8 expression at the invasive margin was established and validated in an independent cohort. More restricted TCR beta chain usage, indicating a less diverse and more clonal T-cell population, also correlated with clinical response.
Gene Expression Profiles
A T cell-inflamed gene expression profile (GEP) emerged as a pan-tumor predictive biomarker across 9 cancer types in 220 patients. This preliminary IFN-γ signature (10 genes), expanded to a preliminary expanded immune signature (28 genes), and was refined to a final 18-gene profile. The T cell-inflamed GEP contained genes related to cytolytic activity, cytokines/chemokines, T cell markers, NK cell activity, antigen presentation, and immunomodulatory factors. This signature included IFN-γ-responsive genes related to antigen presentation, chemokine expression, cytotoxic activity, and adaptive immune resistance and was developed into a clinical-grade assay.
Early Indicators of Reinvigoration
Several early indicators predicted successful T-cell restoration. In metastatic urothelial cancer, decreased circulating PD-1+ CD4+ T cells at weeks 6 and 12 and increased 4-1BB+ CD28+ CD4+ T cells at week 12 indicated reinvigoration. Juxtaposition of Th1, CD8+, and myeloid cells in metastatic lesions also predicted response.
In melanoma, increased CD8/Treg ratio and upregulation of granzyme B and Ki67 in CD8 T cell subsets served as early indicators of reinvigoration. T cell subsets expressing T-bet, Eomes, PD-1, and other inhibitory receptors represented populations invigorated by anti-PD1 therapy.
In HNSCC, increased KCa3.1 and Kv1.3 activities in peripheral blood T cells, enhanced chemotaxis, evasion of adenosine control over chemotaxis, and increased Ca2+ fluxing abilities in activated peripheral blood T cells distinguished responders.
Expression of pSTAT1+ at the invasive margin was higher in responders to melanoma treatment, serving as an early indicator of interferon-driven T-cell reinvigoration.
Anatomical Patterns of T-Cell Restoration
Intratumoral vs. Circulating Compartments
T-cell restoration occurred in distinct patterns across anatomical compartments. Within tumor tissue, responding melanoma patients showed proliferation of intratumoral CD8+ T cells that directly correlated with radiographic reduction in tumor size. T-cell density increased in parallel at both the invasive margin and tumor center. The increase in CD8+ T cells was directly correlated with radiographic tumor reduction, and proliferating CD8/Ki67 double-positive cells were restricted to tumor parenchyma.
In HNSCC, restoration occurred within the tumor microenvironment through clonal expansion and activation of preexisting tumor-infiltrating lymphocytes. Response was associated with high numbers of CD103+PD-1+CD8+ T cells infiltrating pretreatment lesions. The primary anatomical pattern involved intratumoral changes rather than circulating modifications.
Across multiple cancer types, baseline intratumoral T cell infiltration predicted improved response to anti-PD-1 therapies. IFN-γ production by activated T cells within the tumor microenvironment coordinated immune responses and recruited additional T cells. Chemokines like CCL5 and CXCL9, produced in response to IFN-γ, recruited CD8+ T cells to the tumor site.
In peripheral blood, melanoma studies detected increased CD8/Treg ratio and invigoration of CD8 T cells. Unusual populations of CD56-expressing cells were detected in peripheral blood of melanoma patients with regressing disease.
Spatial Reorganization and T-Cell Trafficking
In metastatic urothelial cancer, responding patients showed spatial reorganization within metastatic lesions, with higher baseline density of Th1 cells that became localized near CD8+ T cells and myeloid cells upon treatment. This juxtaposition of immune cell types suggested enhanced cellular interactions facilitating antitumor responses. Changes in circulating T-cell subsets included decreased PD-1+ CD4+ T cells and increased 4-1BB+ CD28+ CD4+ T cells.
Evidence for T-cell trafficking between compartments came from pancreatic cancer, where PD-1 blockade increased flux of TCR clonotypes entering cell cycle in both blood and tumor. Newly cycling clonotypes in blood matched activated T-cell clonotypes in the tumor, indicating migration of tumor-recognizing T cells between these sites.
In HNSCC, pembrolizumab increased K+ channel activity and Ca2+ fluxes in TILs independently of treatment response, while peripheral blood T cells from responders exhibited unique enhancements including increased KCa3.1 activity accompanied by increased Kv1.3 and Ca2+ fluxes. Improved chemotaxis in peripheral blood T cells from responders facilitated better infiltration into the tumor microenvironment.
Temporal Dynamics of T-Cell Restoration
Rapidity of Restoration
T-cell restoration occurred rapidly after pembrolizumab initiation. In HNSCC, pathologic responses occurred after 5 weeks of PD-1 blockade, with activation of preexisting CTX+ZNF683+CD8+ TILs observed in the immediate postneoadjuvant setting.
In metastatic urothelial cancer, changes in T-cell subsets were detected at weeks 6 and 12 after treatment initiation. Decreased PD-1+ CD4+ T cells appeared at weeks 6 and 12, while increased 4-1BB+ CD28+ CD4+ T cells emerged at week 12.
Melanoma studies demonstrated rapid restoration, with significant increases in CD8+ T cells at both invasive margin and tumor center occurring in parallel. Some patients experienced delayed responses, with initial increases in CD8+ cells at the invasive margin followed by mobilization into tumor parenchyma, suggesting a step-wise restoration process.
In HNSCC functional studies, restoration occurred immediately after pembrolizumab initiation, with increased KCa3.1 activity in responders appearing rapidly. Early changes included increased KCa3.1 activity and improved chemotaxis in responders.
Duration and Persistence
Effects on Kv1.3 and Ca2+ fluxes were prolonged and persisted after tumor resection in HNSCC responders. While the stimulatory effect on KCa3.1 was short-lived, Kv1.3 activity remained elevated in responders, indicating sustained restoration. Responders maintained chemotactic ability post-resection, while non-responders did not. After tumor resection, Kv1.3 activity increased in both responders and non-responders, but responders showed this effect earlier and with increased Ca2+ fluxing ability.
In melanoma, restoration effects were sustained with continued proliferation of CD8+ T cells during treatment. The increase in CD8+ T-cell density correlated significantly with tumor size reduction. Tumor responses were evaluated at 12 and 16 weeks, with some patients showing delayed responses meeting criteria for partial response at 36 weeks.
An early wave of proliferation (increased GzmB+Ki67+ cells) translated into a later increase in re-invigorated cells (GzmB+ cells), demonstrating temporal evolution of T-cell phenotypes.
Clinical Correlations
Response Prediction and Survival
The degree of T-cell restoration strongly correlated with clinical outcomes. In HNSCC, responding tumors had clonally expanded ZNF683+CTX+CD8+ TILs, while nonresponders lacked these cells. Revival of preexisting ZNF683+CTX+ TILs was a major mechanism of response, though the abstract did not specify long-term survival outcomes.
In metastatic urothelial cancer, decreased circulating PD-1+ CD4+ T cells and increased 4-1BB+ CD28+ CD4+ T cells associated with response. Responders had higher baseline fractions of CD4+ T cells expressing cosignaling receptors.
Across melanoma studies, statistically significant increases in T cells distinguished responders. Increased memory CD8+ T cells in responders contrasted with decreased CD4+ effector memory T cells in non-responders. CD8 T memory cells expanded in responders, while CD4 T effector memory cells decreased and CD4 T effector cells increased in non-responders.
Clinical responses correlated with immune changes in the tumor microenvironment, with changes in T-cell subsets potentially informing optimal treatment duration and cancer response timing. Proliferation of intratumoral CD8+ T cells correlated with radiographic reduction in tumor size. CD8 expression at the invasive margin served as a validated predictive model for treatment response, with tumor regression requiring pre-existing CD8+ T cells negatively regulated by PD-1/PD-L1.
In HNSCC, responders showed increased KCa3.1 and Kv1.3 activity and improved chemotaxis, with a unique pattern of increased Kv1.3 activity and Ca2+ fluxes distinguishing them from non-responders. Enhanced K+ channel activity and Ca2+ signaling contributed to improved antitumor capabilities and durable response, with these functions persisting over time.
The T cell-inflamed GEP correlated with clinical response to PD-1 blockade across multiple cancer types. Responders had higher expression of genes related to T-cell activity. Durable objective responses were expected to translate into increased survival compared to standard care. Baseline intratumoral T cell infiltration predicted improved response likelihood to anti-PD-1 therapies, with the T cell-inflamed GEP containing genes related to immune functions necessary for clinical benefit.
Synthesis
The available evidence reveals consistent mechanisms of T-cell restoration across most cancer types studied, with one notable exception requiring mechanistic explanation.
Core Restoration Mechanism Across Responsive Tumors
In melanoma and HNSCC, PD-1 blockade consistently restored T-cell function through activation of preexisting tumor-infiltrating lymphocytes rather than recruitment of entirely new clones. This manifested as clonal expansion of exhausted CD8+ TILs with tissue-resident memory programs, expansion of CD8+ memory T cells, and increased granzyme B and Ki67 expression. The mechanism operated through release of PD-1-mediated suppression, enabling IFN-γ signaling that drove T-cell proliferation and cytotoxic function. This pathway was functionally supported by enhanced K+ channel activity and Ca2+ signaling in HNSCC, which provided the metabolic infrastructure for sustained T-cell activity.
Pancreatic Cancer: Reactivation Without Efficacy
In contrast to other cancer types, pancreatic ductal adenocarcinoma showed successful T-cell reactivation—increased TCR clonotype entry into cell cycle and IFNγ signature induction—without corresponding clinical benefit. This apparent contradiction is mechanistically explained by concurrent upregulation of NF-κB signaling in reactivated T cells. While other cancers showed isolated IFN-γ pathway activation driving effective antitumor responses, pancreatic cancer T cells exhibited both IF-γ and NF-κB signatures simultaneously. This dual activation pattern suggests that NF-κB signaling counteracted IFN-γ-mediated cytotoxic function, rendering the reactivated T cells “non-productive” despite their proliferation and tumor-recognizing potential. The mechanistic insight is that successful T-cell restoration requires not merely reversal of PD-1-mediated suppression, but also absence of alternative inhibitory pathways like NF-κB that can override restored cytotoxic capacity.
Baseline Immune Context Determines Restoration Potential
Studies using different analytical approaches consistently identified baseline tumor immune composition as the primary determinant of restoration capacity. Flow cytometry studies showed higher baseline CD8+, PD-1+, and PD-L1+ cells at tumor margins predicted response. Single-cell analyses revealed preexisting ZNF683+CTX+ TILs distinguished responders. Gene expression profiling demonstrated a T cell-inflamed signature at baseline correlated with benefit across 9 cancer types. These convergent findings from orthogonal methods indicate that PD-1 blockade does not generate de novo T-cell responses but rather reinvigorates existing ones. Tumors lacking adequate baseline T-cell infiltration cannot benefit from PD-1 blockade because there are no exhausted T cells to restore. The observation that revival of nonexhausted persisting clones and clonal replacement were modest further supports this interpretation.
Spatial and Temporal Coordination of Restoration
The restoration process exhibited coordinated spatiotemporal dynamics. Within tumors, CD8+ T-cell proliferation occurred in parallel at invasive margins and tumor centers, with some patients showing sequential patterns where margin increases preceded parenchymal infiltration. Between compartments, newly cycling clonotypes in blood matched activated tumor clonotypes, indicating trafficking of restored cells. Temporally, restoration proceeded through distinct phases: immediate increases in KCa3.1 activity and chemotaxis, early proliferation marked by GzmB+Ki67+ cells, and late emergence of re-invigorated GzmB+ cells. While early KCa3.1 effects were transient, Kv1.3 activity persisted after tumor resection, suggesting initial metabolic changes enabled sustained functional improvements. This biphasic pattern—rapid metabolic priming followed by durable functional enhancement—explains how checkpoint blockade can produce long-lasting responses from short treatment courses.
Clinical Translation: Necessary but Insufficient Conditions
The T cell-inflamed phenotype emerged as necessary but not always sufficient for clinical benefit. Baseline intratumoral T-cell infiltration improved response likelihood, with validated biomarkers including CD8 expression at invasive margins and the 18-gene T cell-inflamed GEP. However, even tumors with baseline T-cell infiltration could fail to respond if other suppressive mechanisms like NF-κB signaling or insufficient cytotoxic potential were present. Among responding tumors, the degree of restoration correlated with magnitude of clinical benefit: proliferating intratumoral CD8+ T cells directly correlated with radiographic tumor reduction, and enhanced K+ channel activity associated with durable responses. This dose-response relationship between T-cell functional restoration and tumor regression supports a causal rather than merely correlative relationship.