Stem-like T (T_SL) cells, also known as precursors of exhausted T (T_PEX) cells originate from antigen-stimulated naïve T cells. But under the chronic or persistent antigen exposure (as in cancer or chronic infections), they emerge as a distinct subset within the exhausted T cell lineage. Their canonical and defining feature is the high expression of the transcription factor TCF1, encoded by Tcf7, a key regulator of T cell stemness and a downstream effector of the Wnt/β-catenin pathway (22, 30–32). Phenotypically, these cells are characterized by the co-expression of TCF1 and intermediate-to-high levels of PD-1 (TCF1⁺ PD-1⁺), which distinguishes them from both naïve T cells (TCF1⁺ PD-1⁻) and T_TEX cells (TCF1⁻ PD-1⁺). They typically lack or express low levels of markers associated with terminal exhaustion, such as TIM-3. Additional surface markers often found on T_PEX cells include CXCR5 (important for lymphoid follicle homing) and Slamf6 (Ly108 in mice) (33), while they exhibit low expression of immediate effector molecules like granzyme B (GZMB) and interferon-γ (IFN-γ) (5, 9, 34, 35). Details are summarized in Table 1.

Functionally, T_PEX cells are demarcated from other T cell subsets by their unique combination of properties. Unlike naïve T cells, which are antigen-inexperienced, T_PEX cells are generated post-activation and possess a poised, antigen-experienced state while retaining a multipotent capacity. Compared to SLECs, which are terminally differentiated for immediate cytotoxicity but undergo rapid contraction. T_PEX cells exhibit minimal immediate effector function but sustain long-term proliferative potential and self-renewal. They also differ from conventional memory T cells (e.g., central memory and effector memory T cells), which arise from acute, resolved infections and are maintained in a quiescent state for rapid recall (4, 36, 37).

T_PEX cells exist within the context of persistent antigen, are often part of the “exhausted” lineage, and their self-renewal is continuously engaged to replenish exhausted effector pools. Most critically, they are distinct from T_TEX, which are epigenetically fixed, dysfunctional, and possess negligible proliferative capacity (28). T_PEX cells serve as the primary reservoir that undergoes proliferative expansion in response to ICB, driving the replenishment of the effector T cell compartment, which is absent in T_TEX subsets (2, 14). Overall, these functional profiles including self-renewal under chronic antigen pressure, multilineage differentiation, and therapy-responsive proliferation, defines their unique role as the central regenerative engine of the antitumor T cell response. Details are summarized in Table 1.

This metabolic profile of T_PEX cells starkly contrasts with other T cells. SLECs are highly glycolytic and dependent on one-carbon metabolism (38, 39). Memory T cells contain relatively greater mitochondrial mass, and the mitochondria have a fused ultrastructure and a relatively higher spare respiratory capacity (SRC) (38–40). T_TEX cells exhibit severe mitochondrial dysfunction driven by PGC1α suppression and metabolic paralysis, such as decreased glycolytic activity and oxidative phosphorylation (OXPHOS). Whereas T_PEX cells display distinct metabolic profiles characterized by increased fatty acid oxidation (FAO) and mitochondrial SRC, which generates less reactive oxygen species, minimizing oxidative damage. Furthermore, they possess abundant, fused mitochondria, indicative of metabolic fitness. The unique metabolic wiring of T_PEX cells is thus integral to their self-renewal capacity, persistence, and readiness to proliferate upon demand (39, 41, 42).

In patients with lung adenocarcinoma, Connolly et al. (36) observed that the majority of T_PEX cells were present in non-metastatic lung-draining lymph nodes, which was in line with the findings in mice, where T_PEX cells were predominantly located in tumor-draining lymph nodes (TDLNs). T_PEX cells within the TDLNs present with high CCR7 expression, which is vital for the migration and positioning of T_PEX cells. The stromal cells in TDLNs could produce CCL19 and CCL21, which are the ligands of CCR7, and attract T_PEX cells to concentrate in the inner T cell zone (TCZ) (43) (Figure 2A). Mature dendritic cells (DCs) carrying tumor-generated antigens infiltrate the TCZ, providing specific signals that tune naïve T cells toward T_PEX cells (44, 45). Meanwhile, in the outer TCZ, DCs attract T_PEX cells to differentiate into effector cells by expressing CXCR3 ligands, CXCL9 and CXCL10, and IFN-I (Figure 2B). Thus, blockading of PDL1 on DCs could induce local expansion of T_PEX cells within the TDLNs, which further traffics to the tumor and induces effective immunity (16, 46, 47). Other studies also discovered that T_PEX cells in the TDLNs are the precursors of the tumor-specific T cells (37, 48) and play a vital role in maintaining persistent T cell responses (Figure 2C). Altogether, increasing data have shown that TDLNs as a reservoir of T_PEX cells are key sites where de novo antitumor responses are initiated.

T_PEX cells predominantly localize to the specialized perivascular niches at the tumor-stroma interface, while T_EFF cells and T_TEX cells infiltrate deeper tumor parenchymal regions, where they directly engage tumor cells. These perivascular niches provide specific signals that sustain T_PEX cell survival and stemness. Multiple studies have revealed a spatial correlation between T_PEX cells and antigen-presenting cells (APCs) in tumor tissues. In kidney, bladder, and prostate cancers, T_PEX cells localize preferentially to tumor regions densely populated by MHCII⁺ APCs, while T_TEX cells and T_TEX cells resided distally. This compartmentalization establishes functional perivascular niches that drive T_PEX cell clustering and facilitate cross-presentation of tumor antigens (15, 49).

Furthermore, the presence of perivascular niches correlated with enhanced tumor vascularization. In a melanoma, T_PEX cells are localized preferentially within perivascular regions adjacent to CD31⁺ endothelial cells in tumor tissue (22). Similar perivascular niches, enriched in T_PEX cells and DCs, have also been documented in colon cancer and pancreatic ductal adenocarcinoma (50). These niches are coordinated by CXCR6-CXCL16 and CXCR3-CXCL9/CXCL10 chemokine interactions within the tumor stroma, interactions critical for mediating immunotherapy responses. Crucially, immunotherapy induces the formation of these perivascular niches, and their abundance correlates with the magnitude of therapeutic response, supporting their functional importance in antitumor immunity (50).

Tertiary lymphoid structures (TLS) are ectopic immune cell aggregates exhibiting architectural parallels to the follicles of secondary lymphoid organs (51). These structures facilitate antigen presentation to lymphocytes, thereby supporting the initiation and regulation of adaptive immune responses. Within the TME, TLS is predominantly localized at peri-tumoral sites or along the tumor-stroma interface (52) (Figure 2C). Their presence correlates positively with enhanced intratumoral T cell infiltration and favorable patient prognosis (34, 53). In the context of patients with stage I–IV non-small-cell lung cancer (NSCLC), T_PEX cells are observed to be located in the TLSs rather than in the tumor parenchyma (54). Studies further demonstrate significantly greater T_PEX abundance in TLS-enriched tumors, suggesting TLS may function as an intratumoral reservoir for T_PEX cells (34). Conversely, T cells from TLS-deficient tumors exhibit pronounced exhaustion phenotypes, characterized by elevated co-expression of PD-1 and TIM-3 (52). While burgeoning evidence underscores the prognostic and immunological significance of TLS and associated T_PEX populations, the precise molecular mechanisms governing their spatial organization, functional interplay, and therapeutic contributions remain incompletely elucidated.

The frequencies of T_PEX cells are positively correlated with the presence of immunologically active structures, including TLS, perivascular areas, and TDLNs (50–54). These niches provide critical survival signals (e.g., IL7 and IL-15) and intermittent antigen presentation, which sustain T_PEX cell clustering and prevent terminal differentiation (51, 52, 54). Clinically, niche-rich tumors exhibit higher T_PEX cell frequencies and improved patient outcomes (68, 69). The absence of these structures correlates with lower T_PEX frequencies and a more exhausted T cell landscape (50–54, 68, 69). The TDLN serves as a critical extratumoral reservoir, maintaining a higher frequency of T_PEX cells that can be recruited to the tumor (36, 67). T_PEX cells continuously traffic between the TDLN and tumor, a process required for replenishing the intratumoral pool (36, 70, 71). Using FTY720 to block this egress could reduce T_PEX cell frequency and antitumor immunity (36, 70, 72).

The nutrient-depleted, hypoxic TME imposes metabolic stress that can also affect the T_PEX cells. A clear consensus indicates that T_PEX cells rely on oxidative metabolism (OXPHOS/FAO) for long-term persistence, unlike their glycolytic effector progeny (39, 40). Furthermore, chronic antigen exposure such as high levels of IFN-I (late phase), IL-2, and inflammatory signals, constantly depletes the T_PEX pool by driving differentiation, making their sustained frequency a balance between self-renewal and differentiation pressure (37, 73).

Additionally, the regulatory mechanisms governing T_PEX cell frequency described above also directly dictate the balance between tumor immune control and disease progression. A robust and well-maintained T_PEX pool, supported by functional niches and balanced cytokine signals (51, 52, 54), establishes a state of continuous immunosurveillance. This enables the adaptive immune system to dynamically respond to tumor evolution by generating T_EFF cells continuously. In this context, the immune system can control tumors in a state of long-term equilibrium or even mediate tumor regression, a hallmark of effective immunotherapy. Clinically, this is reflected in the strong association between high T_PEX abundance, the presence of TLS, and favorable patient outcomes across multiple cancer types (68, 69).

Conversely, the breakdown of T_PEX-supportive regulation is a pivotal event driving cancer progression. This failure can occur through several interconnected mechanisms: 1) Loss of supportive niches (e.g., absence of TLS, vascular abnormalities, and lymph node metastasis), leading to T_PEX depletion (50–54, 68, 69); 2) Overwhelming differentiation pressure from chronic inflammation and antigen load, which exhausts the progenitor reservoir (37, 73); 3) Metabolic sabotage within the TME, impairing the mitochondrial fitness of T_PEX cells (39, 40). The consequence is a collapsed regenerative engine for antitumor immunity. The T cell compartment becomes dominated by terminally exhausted, dysfunctional T_PEX cells, incapable of controlling tumor growth. This failure to replenish effector cells leads to diminished immune pressure, allowing for unchecked tumor expansion, evolution of antigen-loss variants, and eventual metastatic dissemination.

Consequently, the frequency of T_PEX cells is a balance between supportive signals and differentiation pressures that deplete the pool by driving terminal exhaustion. The dynamic regulation of T_PEX cells is a core modulator of anti-tumor immunity. Therapeutic strategies that successfully maintain, expand, or restore T_PEX cells, such as ICB, microenvironment modulators, or adoptive transfer of stem-like T cells, essentially work by restoring this critical regenerative capacity, thereby shifting disease progression from advancement to control.

A strong consensus exists around a core set of transcription factors (TFs) that are necessary and instructive for the stemness and persistence of T_PEX cells, including TCF1 (22, 31, 74), BCL6 (75, 76), ID3 (77, 78), FOXO1 (79–81), and EOMES (82, 83). Among these TFs, TCF1 is the foremost and crucial for T_PEX cell formation and maintenance (22, 31, 74). TCF1 establishes the transcriptional network required for T_PEX cell differentiation by promoting EOMES and BCL6 expression, while suppressing T-BET and BLIMP1 expression (75, 84). Consequently, TCF1-deficient CD8⁺ T cells result in the impaired maintenance of T cells response and poor efficacy of ICB in mouse models (85). Similarly, the deletion of BCL6 displays reduced T_PEX cell abundance and attenuates the long-term tumor control (75, 76, 84). In contrast, overexpression of TCF1, ID3, or FOXO1 in CD8⁺ T cells or in CAR-T cells could improve the persistence and tumor control (75, 76, 79, 81, 86). Another key regulator, MYB, is typically expressed in hematopoietic stem cells and in human stem-like T cells. Evidence shows that MYB-deficient T cells fail to respond to ICB (87), and Myb overexpression in CD8⁺ T cell enhanced the T_PEX cell formation and improved tumor control in mouse models (88, 89).

If TCF1, BCL6, and FOXO1 form the “brakes” on differentiation, a separate set of factors act as the “accelerator”, pushing T_PEX cell toward effector and exhausted fates.

One such regulator is BLIMP1, which promotes inhibitory receptor expression (e.g., PD-1, LAG-3) while directly repressing stemness genes (Tcf7, Bcl6, Ccr7, Sell, Cxcr5, Il-7r) (75, 90); reciprocally, TCF1 also repressed BLIMP1 (91). Consequently, BLIMP1 deletion expands the T_PEX pool and improves responses to immunotherapy (75). Similarly, T-BET is essential for forming effector subsets and can antagonize the exhaustion marker PD-1 expression (17, 92, 93).

Moreover, persistent antigen signaling established a robust causal link to terminal exhaustion via the induction of TOX. TOX is highly expressed in T_TEX cells and necessary for the full exhausted phenotype (9, 94, 95). Its absence preserves TCF1 expression and stem-like potential of T_PEX cells (75, 96). IRF4 exhibits kinetic complexity: it promotes early effector differentiation but under chronic stimulation cooperates with TOX to repress TCF1 and enforce exhaustion (18, 97).

The roles of ID2 and EOMES appear context-dependent. ID2 expressed in T_TEX cells and generally antagonizes TCF1/BCL6 in chronic LCMV infection (98–100), yet its loss in tumor model impairs T_PEX maintenance and anti-PD-1 response (101). EOMES is highly expressed in T_PEX cells but its deficiency reduces T_TEX formation while expanding the T_PEX compartment (163), suggesting a complex, stage-specific function that warrants further investigation.

Beyond transcription factors, cytokines critically shape T_PEX cell fate, though their necessity within tumors in vivo is often nuanced and context-dependent.

Homeostatic cytokines IL-7 and IL-15 are established promoters of the stem-like state, cooperating to instruct T_PEX cell differentiation (102–104) and used in culture to generate stem-like CAR-T cells (105, 106). Conversely, IL-2 predominantly drives effector differentiation by inducing BLIMP1 expression; engineered IL-2Rβγ agonists synergize with PD-1 blockade by expanding the T_PEX-derived effector pool (107). However, the role of IL-10 appears to be more complex. In a murine melanoma model, IL-10 signaling promoted T cell exhaustion and impaired antitumor responses (108). But PEGylated IL-10 treatment enhanced intratumoral CD8⁺ T cell expansion and effector function and was related to tumor regression in cancer patients (109, 110). Moreover, IL-10 administration metabolically reprograms T_TEX cells, enhancing antitumor immunity in mouse tumor models independently of T_PEX cell (39). Thus, further research is necessary to fully understand the precise effect of IL-10 on T_PEX cells.

The roles of TGF-β defy their traditional immunosuppressive labels. In tumor models, loss-of-function studies show TGF-β induced BCL6 expression in CD8⁺ T cells and was essential for maintaining the T_PEX pool by enforcing residency and limiting premature differentiation (111–113). Consistently, TGF-β treatment enhances the stemness of CAR-T cells and improve antitumor efficacy (114).

IFN-I promote T_TEX cells by antagonizing the formation and maintenance of T_PEX cells (21). In the TME, IFN-I and IFN-II contributed to T cell exhaustion and activated resistance programs in tumor cells that limit anti-tumor T cell responses (21, 115).

Immune checkpoints are actually a normal part of the immune system with the role of preventing immune response from being too strong to destroy healthy cells in the body. As described before, when persistent antigen stimulation, T cells will undergo an exhausted phenotype with the increased expression levels of inhibitory receptors (e.g., PD-1 and CTLA4) (19). Notably, although T_PEX cells share some features of exhaustion, they retain proliferative capacity, self-renewal, and lineage plasticity, acting as the progenitor population that continually replenishes the T_TEX cells compartment, and responses to checkpoint blockade. Current ICB targeting CTLA-4 and PD-1 receptors have received positive outcomes (19, 116) and truly revolutionized the treatment of cancer patients with melanoma (117), breast cancer (118), lung cancer (119), and other cancers (120, 121).

Crucially, the mechanistic basis of ICB efficacy has been refined through detailed dissection of the T cell compartment. Accumulating evidence have suggested that the clinical benefit of PD-1/PD-L1 blockade is driven predominantly by the expansion and differentiation of T_PEX cells, rather than the functional restoration of T_TEX cells (5, 9, 23). T_PEX cells, characterized by TCF1⁺ PD-1⁺ expression, retain self-renewal capacity and serve as a proliferative reservoir. In contrast, T_TEX cells exhibit a fixed epigenetic and transcriptional state characterized by chromatin remodeling, TOX-driven transcriptional reprogramming, and metabolic exhaustion, rendering them refractory to reinvigoration (9, 122–124).

Preclinical studies established that T_PEX cells serve as the primary reservoir for tumor-specific T cell upon PD-1/PD-L1 inhibitors (36, 59). In murine models of chronic LCMV infection and tumors, PD-1 inhibitor induces the proliferation and differentiation of T_PEX subset into T_EFF cells, driving tumor control (11, 23). Clinically, the association between T_PEX cells and response to ICB is context-dependent and continues to be refined. Seminal work in melanoma demonstrated that higher frequencies of intratumoral T_PEX cells correlate with prolonged progression-free survival (PFS) and objective response to anti-PD-1 therapy, and that responding patients exhibit clonal expansion of these cells, replenishing the cytotoxic T cell pool post-treatment (22, 125, 126). These findings established T_PEX cells as a promising biomarker in this immunogenic cancer. However, subsequent studies across diverse tumor types and patient cohorts have revealed a more nuanced picture. While some reports corroborate a positive association with clinical benefit, others find correlations primarily with PFS rather than with objective response rates per se, and in certain contexts, the abundance of T_PEX cells alone does not robustly predict clinical outcomes (15, 49, 59, 127–129). These discrepancies may stem from differences in tumor immunogenicity, prior therapies, T cell sampling site (e.g., blood vs. tumor), and the precise phenotypic definition of the stem-like population. Therefore, while T_PEX cells are mechanistically crucial for sustaining antitumor immunity, their utility as a universal predictive biomarker requires further validation in specific cancer types and treatment settings.

Currently, rational combination therapies targeting T_PEX cell amplification have shown mechanistic and clinical synergies. Firstly, radiotherapy promotes immunogenic cell death, releasing DAMPs that activate dendritic cells by the cGAS-STING pathway. This cascade enhances T_PEX cell priming and recruitment to the tumor site and, when used in combination with anti-CTLA-4, amplifies the distal response (130, 131). Secondly, epigenetic reprogramming using DNMT inhibitors (such as azacytidine) or LSD1 inhibitors reverses T cell exhaustion by demethylating the Tcf7 enhancer. This sustains T_PEX to a stem-cell-like state, restoring anti-PD-1 reactivity in preclinical models (132, 133). Thirdly, chemotherapy (such as oxaliplatin) upregulates CXCL10 in tumor vessels via IFN-γ signaling. The resulting chemokine gradient drives T_PEX cell homing into tumors, enhancing the efficacy of ICB (134).

Studies on adoptive cell therapy (ACT) demonstrate that sustained antitumor immunity depends critically on the persistence of reinfused cells rather than their immediate cytotoxic capacity upon transfer because prolonged or excessive stimulation during T cell expansion is known to promote exhaustion and can compromise the functional potency of adoptively transferred cells (135). Consequently, contemporary ACT protocols prioritize generating less differentiated cell subsets over terminally differentiated populations to maximize antitumor efficacy in this context. Key strategies focus on culturing conditions that favor progenitor-like states. A pivotal approach involves the substitution of IL-2 with homeostatic γ-chain cytokines (e.g., IL-7, IL-15, or IL-21) during ex vivo expansion (136). Unlike IL-2, which can drive terminal effector differentiation and exhaustion, these cytokines promote homeostatic proliferation and help maintain or upregulate stem/progenitor-associated genes (such as Tcf7 and BCL6), thereby preserving a less differentiated, more persistent T cell product (103, 137, 138). Separately, the modulation of TCR signaling strength during activation (e.g., through altered peptide ligand or co-stimulation) is another critical lever to prevent over-stimulation and exhaustion, working in concert with cytokine conditioning to optimize T cell quality (9, 139). Additional strategies include augmenting stemness-promoting pathways like Notch signaling (140) and inhibiting transcription factors linked to terminal dysfunction (e.g., BLIMP1) (75).

Exogenous T cell therapies, particularly CAR-T immunotherapy, have established a new standard of care for several relapsed or refractory B-cell malignancies, including certain types of large B-cell lymphoma and B-cell acute lymphoblastic leukemia (141–144). Their application is actively being explored and is expanding into other well-defined hematologic and solid tumor settings. Notably, pre-infusion products enriched in CAR-T_TEX cell populations correlate with inferior outcomes, whereas stem- and memory-like phenotypes associate with higher response rates (145, 146). Although comprehensive clinical characterization of T_PEX phenotypes in CAR-T products remains limited, recent work identified PD-1⁺ TCF1⁺ CAR-T_PEX cells as predictors of improved clinical outcomes (147). Preclinically, engineered CAR-T models overexpressing T_PEX-associated TFs exhibit enhanced stem-like phenotypes, expansion potential, persistence, and therapeutic efficacy (81). Similarly, pre-existing TLS or APC-dense niches may be essential for generating and sustaining CAR-T_PEX phenotypes; thus, fostering these microenvironments may augment their persistence (2, 148). Furthermore, utilizing T_PEX cells and their molecular signatures as predictive biomarkers may optimize CAR-T clinical management.

Collectively, these findings underscore the paramount importance of preserving and augmenting T_PEX cells to enhance persistence and therapeutic outcomes in ACT, particularly in CAR-T immunotherapy.

Therapeutic vaccination targeting either shared tumor antigens or patient-specific neoantigen (neoAg) pools represents a promising strategy to activate antitumor T cell immunity (149, 150). Leading vaccination approaches aim to harness the self-renewal capacity, long-term persistence, and multilineage differentiation function of T_PEX cells through targeting of common tumor antigens or individualized neoAg (150, 151). Preclinical studies demonstrate that the efficacy of therapeutic vaccination depends critically on T_PEX cells; thus, enriching these populations during vaccination could theoretically enhance antitumor responses (22, 152). While vaccines initiate de novo T cell responses against tumors, functional exhaustion may limit their activity. Consequently, many clinical vaccine trials employ combinatorial approaches with ICB (153).

Accordingly, vaccines specifically designed to induce T_PEX cell populations have been developed and show potent synergy with ICB in tumor models (154, 155). Although clinical outcomes from tumor vaccine trials have yielded inconsistent results, expanding T_PEX cells represents a key consideration for improving future vaccine efficacy.

While the pivotal role of T_PEX cells in immunotherapy efficacy is well-established preclinically, translating these insights into clinical practice faces several challenges and opportunities.