Circulating T-lymphocyte subsets as biomarkers for immune checkpoint inhibitors in solid tumors.

Introduction
Immune checkpoint inhibitors (ICIs) have revolutionized the treatment of many solid tumors by unleashing T-cell-mediated anti-tumor immunity. Antibodies targeting checkpoint molecules such as programed cell death protein 1/programed death-ligand 1 (PD-1/PD-L1) and cytotoxic T-lymphocyte antigen-4 (CTLA-4) can produce durable remissions in cancers like melanoma, lung cancer, renal cell carcinoma, and others by reinvigorating exhausted T cells and enhancing cytotoxic T-lymphocyte activity [1]. Despite these advances, only a subset of patients derive significant benefit from ICIs, while others experience primary resistance or even hyperprogression. Established biomarkers, notably tumor PD-L1 expression and mismatch-repair deficiency or high tumor mutational burden, help predict response in some settings, but they are imperfect. For example, roughly half of non-small cell lung cancer (NSCLC) patients with PD-L1 combined positive score (CPS) ≥ 50 do not respond to anti-PD-1 therapy [2]. Moreover, obtaining tumor tissue for biomarker assessment can be invasive and may not capture the spatial and temporal heterogeneity of the tumor microenvironment (TME). There is thus an urgent need for robust, non-invasive biomarkers that can more reliably stratify patients and guide immunotherapy decisions.
Liquid biopsies of blood can be repeated over time and may reflect the overall immune landscape of a patient, rather than a single biopsy site. Among liquid biopsy approaches, profiling circulating T lymphocytes has emerged as one of the most promising strategies for predicting and monitoring ICI outcomes. Effective PD-1/PD-L1 blockade not only reinvigorates T cells already inside the tumor, but also depends on continuous recruitment of new T cells from the circulation to maintain the anti-tumor response [3], recent data show that patients responding to ICI often exhibit therapy-responsive T-cell subsets in peripheral blood, and that new T-cell clones from the blood can infiltrate tumors to sustain tumor control. Additionally, some tumor-infiltrating T cells may egress back into the bloodstream, meaning certain circulating T-cell populations can mirror the functional state of T cells in the TME [4, 5]. Monitoring circulating T cells offers valuable insights into a patient’s immune status and can help predict the efficacy of immunotherapy. Recent advances in flow cytometry have facilitated the detection of markers associated with lymphocyte differentiation, function, and activation status, which are strongly correlated with the efficacy and prognosis of tumor immunotherapy. T-lymphocyte fine typing, based on distinct molecular phenotypes, categorizes T cells into functional subgroups, differentiation subgroups, and functional phenotypes. This review focuses on identifying peripheral blood lymphocyte markers predictive of ICI therapy efficacy, aiming to provide valuable insights for researchers in the field.

T-lymphocyte subsets classification
The classification of T-lymphocyte subsets based on molecular phenotypes encompasses multiple analytical dimensions, which are fundamentally grounded in cellular lineages, differentiation stages, and functional phenotypes. Cellular lineage subgroups include cytotoxic T cells (CD8+T cells), helper T cells (CD4+T cells), Type I natural killer (NK) T cells, and γδ+T cells [6]. Differentiation subgroups, reflecting their immunological stage, encompass naïve T (TN), effector T (TEFF), memory T (TM), precursor exhausted T (TPEX), and terminally exhausted T (TEX) cells [7]. The relationships between T-cell differentiation subsets are shown in Figure 1. Functional phenotypes categorize T cells based on their specific roles in the immune response, such as activation, proliferation, senescence, and exhaustion.
Activation and proliferation phenotypes reflect the functional state of T cells, including their activation status, proliferative capacity, and functional activity, while senescence and exhaustion phenotypes indicate a decline in T cell reserves, reduced proliferative and survival potential, shortened lifespan, and diminished or anergic effector functions. These functional phenotypes are closely associated with predicting the efficacy of ICI [8, 9]. Proliferation phenotypes are assessed by the expression of Ki-67, a marker elevated during the G1-M phase of the cell cycle and play a pivotal role in regulating cell division. Activation phenotypes are commonly evaluated using markers such as human leukocyte antigen (HLA)-DR and CD38 [8]. Senescence phenotypes, frequently observed in terminally differentiated effector memory T cells (TEMRA), are characterized by the loss of CD28 and CD27, coupled with the expression of CD57 and KLRG1, signifying immune aging and a shortened lifespan. Exhaustion phenotypes are distinguished by high expression of inhibitory receptors, reflecting reduced proliferation, survival, and effector functions, with markers such as programed death-1 (PD-1), cytotoxic T-lymphocyte antigen 4 (CTLA-4), lymphocyte-activation gene 3 (LAG-3), and T-cell immunoglobulin domain and mucin domain-3 (TIM-3) being prominently expressed in exhausted T cells.

Cellular lineage subgroups and immune therapy efficacy evaluation
T cells are categorized into functional subgroups, including cytotoxic T cells (CD8+T cells), helper T cells (CD4+T cells), Type I NK T cells, and Tregs. Type I NK T cells are characterized by CD3+CD16+CD56+Vα24+Vβ11+ or CD3+CD16+CD56+Vα24+Jα18+, while Tregs are identified by CD4+CD25+CD127−/low or CD3+CD4+CD25+FoxP3+ [9]. Tregs play a critical role in immune suppression through various pathways, promoting immune tolerance and enabling tumor cells to evade immune surveillance. In patients with NSCLC treated with PD-1 inhibitors, elevated levels of peripheral blood Tregs are associated with poor treatment outcomes, with non-responders exhibiting significantly higher Treg levels [10]. Conversely, in patients with malignant melanoma, baseline levels of peripheral blood Tregs have been positively correlated with the efficacy of anti-CTLA-4 therapy [11]. Although functional subgroup proportions and absolute counts provide valuable insights into immune status, their broad categorization limits the sensitivity and specificity of ICI efficacy prediction. This review focuses on two more precise classifications, T-cell differentiation subgroups and T-cell functional phenotypes, as predictive indicators for ICI efficacy in solid tumors.

T-cell differentiation subgroups and immune therapy efficacy evaluation
T cells play a fundamental role in the immune response and serve as key mediators in cancer immunotherapy. Their differentiation into naïve, effector, and memory subgroups is now understood to play a crucial role in determining the efficacy of immune checkpoint inhibitors. The activation and function of these subgroups directly affect treatment outcomes, making it essential to understand their dynamics. By evaluating T-cell differentiation, clinicians can gain valuable insights into tumor responses and optimize personalized treatment strategies, thereby improving the precision and effectiveness of immunotherapy.

T-cell differentiation and subgroup classification
T-cell differentiation categorizes these cells into three main types: TN, TEFF, and TM cells [7]. TN cells are mature T cells that have not yet encountered antigen stimulation. They develop in the thymus and migrate to peripheral lymphoid tissues, such as the spleen and lymph nodes, where they can recognize antigens and initiate immune responses. TN cells are characterized by their longevity and potential to differentiate into effector cells [12]. They express CD45RA and lymph node homing markers, including L-selectin (CD62L) and C-C chemokine receptor type 7 (CCR7), which facilitate their circulation to lymphoid tissues [13]. These markers direct TN cells from endothelial venules into lymphoid tissues, where they encounter MHC signals presented by dendritic cells (DCs), subsequently differentiating into TEFF and TM cells. TN cells represent the immune system’s reserve capacity for mobilization. Upon antigen exposure, TN cells transition by downregulating CD45RA and upregulating CD45RO [13].
Terminally differentiated effector T cells are highly specialized effector cells that execute the immune response. They express KLRG-1 and CD57 but lack CD27, CD28, and homing markers, underscoring their terminal differentiation [14]. TM cells arise either from TEFF cells or directly from TN cells after antigen exposure. It has been established that TM cells express CD45RO and adhesion molecules such as CD44, enabling long-term survival. Following pathogen clearance, most TEFF cells (90–95%) undergo apoptosis, leaving TM cells as a heterogeneous population [15]. TM cells are further divided into stem cell memory T (TSCM), central memory T (TCM), and effector memory T (TEM) cells [13]. TCM cells, characterized by high proliferative capacity and interleukin-2 (IL-2) production, predominate in secondary lymphoid tissues and express CD62L and CCR7, whereas TEM cells lack homing markers, exhibit lower proliferative capacity, and primarily secrete effector cytokines such as IFN-γ [15]. TEMRA are characterized by the loss of CD28 and CD27 coupled with re-expression of CD45RA, displaying signs of cellular senescence [12]. The classification, phenotypes, and functions of T-cell differentiation subsets are summarized in Table 1, using CD8+ T cells as a representative example.

Predictive and prognostic value of circulating T cell differentiation subgroups in ICI therapy
Multiple studies have demonstrated a strong association between the proportion and absolute count of T-cell differentiation subgroups and ICI efficacy. Both baseline profiles and the substantial shifts in T-cell differentiation subsets following immunotherapy (Figure 2) corroborate their established roles as key predictors of ICI efficacy, underscoring their potential as clinical biomarkers. Helper CD4+T cells regulate CD8+T cell differentiation through various pathways, enhancing their anti-tumor effects during ICI therapy [30]. In patients with advanced NSCLC receiving ICI treatment, higher proportions and absolute counts of peripheral blood CD4+TN and CD4+TSCM were positively correlated with prolonged median progression-free survival (PFS). Moreover, an increase in the absolute count of CD8+TSCM cells was associated with improved PFS, with CD4+TN absolute counts (≥ 5.5 cells/ml) identified as an independent prognostic factor for PFS. However, no significant correlation was observed between PFS and the proportion or absolute count of CD4+T cells, CD8+T cells, CD8+TN cells, or CD8+TSCM [31]. Another NSCLC study revealed differences in the proportions of CD4+TN and CD4+TM functional and phenotypic subgroups between responders and non-responders. Specifically, baseline CD4+TN cells producing cytokines (IFN-γ, TNF-α, IL-17A) and PD-1+CD4+TM cells were the most significant predictors of response to anti-PD-1 therapy. Patients with higher proportions of these subsets experienced better efficacy, validated in an independent cohort [32]. Patients with higher percentages of IFN-γ+CD4+TN and IL-17A+CD4+TN cells had significantly prolonged PFS, while higher percentages of TNF-α+CD4+TN cells showed a positive trend without statistical significance. NSCLC patients with more functional CD4+TN cells tended to respond better to nivolumab [32]. Disease progression in NSCLC was marked by a significant decline in CD4+T cell counts, primarily reflecting a reduction in TN cells [33]. In melanoma and NSCLC, a high peripheral blood TCM/EFF ratio was positively correlated with tumor lymphocyte infiltration and prognosis [34]. PD-1 inhibitor therapy in melanoma resulted in increased activated CD4+TEM, with a similar trend observed for TCM [21].
Peripheral blood CD8+TEM cells play a significant role in predicting ICI efficacy. In a study involving stage IV malignant melanoma patients treated with PD-1 inhibitors, responders exhibited significantly higher proportions of CD8+TEM cells both before and after one treatment cycle compared with non-responders [35]. CD8+TEM cells can be further subdivided into four subtypes based on the expression of CD27 and CD28: TEM1(CD27+CD28+), TEM2(CD27+CD28−), TEM3(CD27−CD28−), and TEM4(CD27−CD28+) [36]. Among 137 advanced melanoma patients treated with ipilimumab, a baseline CD8+TEM1 cell proportion exceeding 13% was associated with longer overall survival (OS) (P = 0.029) and higher response rates (P = 0.01). TEM1 was identified as an independent prognostic factor for OS. Conversely, a baseline CD8+TEMRA proportion above 23.8% was negatively correlated with OS (P = 0.034) but had no impact on clinical response [36]. Besides, baseline CD45RO+CD8+T cell levels were predictive of ipilimumab response, with baseline levels ≤25% indicating non-response. However, no such association was found with pembrolizumab response [37].
In patients treated with nivolumab, those with controlled disease exhibited significant increases in the proportions of CD4+TEMRA cells, whereas no such increase was observed in patients with disease progression. Similarly, CD8+TEMRA proportions showed an upward trend in disease-controlled patients. Notably, activated CD4+ and CD8+TEMRA cells, identified by CD38 expression, were elevated in disease-controlled patients. Furthermore, all CD8+T cell subsets, except for TEM, exhibited increased activation in disease-controlled patients. PD-1 expression on both CD4+ and CD8+T cells decreased significantly, irrespective of treatment response [38].

T-cell functional phenotypes and immune therapy efficacy evaluation
T cells exhibit a range of functional phenotypes, including activation, proliferation, senescence, and exhaustion. The activation and proliferation phenotypes are indicative of the active functional state and proliferative ability of T cells, which are essential for mounting an effective immune response. In contrast, the senescence and exhaustion phenotypes are associated with a decline in T cell reserves, reduced proliferative and survival capacity, shorter lifespans, and diminished or anergic effector functions. These functional states of T cells are closely linked to the outcomes and efficacy of immunotherapies, underscoring their significance in clinical applications.

Activation and proliferation phenotype
The activation phenotype of T cells is typically characterized by the expression of co-stimulatory molecules and chemokine receptors, including CD137, CX3C chemokine receptor 1 (CX3CR1), CD39, HLA-DR, CD38, OX40, and C-X-C chemokine receptor type 5 (CXCR5). These markers are indicative of the functional activation of T cells in response to specific stimuli. In contrast, the proliferation phenotype is evaluated through the expression of Ki-67, a marker that is upregulated during the G1-M phase of the cell cycle and plays a crucial regulatory role in cell division and classification.

HLA-DR/CD38
HLA-DR is a Class II major histocompatibility complex (MHC) antigen that is constitutively expressed on B lymphocytes, monocytes, and macrophages, where it plays a critical role in antigen presentation to CD4+ T cells. While most T cells do not express HLA-DR under normal physiological conditions, a subset of activated T cells can upregulate HLA-DR during the later stages of an immune response. Nevertheless, the exact mechanisms underlying this upregulation and its functional significance in T cells remain poorly understood.CD38, another activation marker, is constitutively expressed on naïve T cells but is downregulated in resting memory T cells. Upon activation, CD38 expression is re-established, making the co-expression of HLA-DR and CD38 a valuable indicator of T-cell activation [39].
Although HLA-DR/CD38 are recognized markers of T-cell activation, the correlation between their elevated expression and the efficacy of ICIs remains controversial. Small-sample studies have found that the percentage of activated CD4+CD38+T cells and HLA-DR+CD38+NK cells post-treatment was significantly higher in non-responders compared with responders, suggesting a potential association with poor treatment outcomes [35, 40]. Similarly, other studies have observed an elevation of HLA-DR+/CD8+T cells in the peripheral blood of non-responders [41].These findings may suggest that even though these cells exhibit characteristics of immune activation, they might be rendered ineffective at clearing tumor cells due to the influence of immunosuppressive molecules abundant within the TME. The predictive utility of these cells for ICI efficacy warrants further investigation with larger sample sizes. Furthermore, this suggests that a single activation marker has low specificity for predicting treatment outcome. Consequently, it was proposed that a combination of CD8+TEM, CD16+CD56+CD38+HLA-DR+ NK, and CD4+CD38+T cells may collectively constitute an early predictive biomarker panel for therapeutic efficacy. By integrating the monitoring of activation status across different immune cell subsets, the accuracy of predicting immunotherapy response can be improved. It is worth noting that a post-treatment increase in the proportion of activated CD4 + CD38+HLA-DR+T cells may not only correlate with treatment efficacy but may also be associated with an increased risk of developing ICI-related adverse events [35].

Ki-67
Ki-67, a nuclear protein, serves as a key marker of the proliferative state of T cells and plays a crucial role in assessing ICI efficacy. Post-treatment changes in the proliferation status of PD-1+CD8+T cells are closely linked to therapeutic outcomes. In patients with advanced NSCLC and thymic epithelial tumors treated with ICIs, the proliferation index of PD-1+CD8+T cells (Ki-67 D7/D0), measured one week after therapy initiation, demonstrated a positive correlation with sustained clinical benefit and prolonged PFS [42]. Among NSCLC patients receiving PD-1 inhibitor therapy, 70% exhibited an increase in Ki-67+PD-1+CD8+T cells, with the majority of responses occurring during the first or second treatment cycle. This proliferative response was positively associated with therapeutic efficacy and is thought to reflect the activation of tumor-specific T cells. These proliferating CD8+T cells exhibited an effector-like phenotype, characterized by the expression of HLA-DR, CD38, and low levels of Bcl-2 (Bcl-2lo), as well as co-stimulatory molecules such as CD28, CD27, and inducible T-cell co-stimulator (ICOS). Furthermore, these cells co-expressed high levels of PD-1 and CTLA-4.
Notably, 70% of patients with disease progression lacked PD-1+CD8+T cell responses, whereas 80% of patients who experienced clinical benefit demonstrated PD-1+CD8+T cell responses within the first 4 weeks of treatment [43]. These findings highlight the significance of Ki-67+PD-1+CD8+T cells as biomarkers for monitoring early treatment response and predicting clinical outcomes.

Co-stimulatory molecules and chemokine receptors

CD137
CD137 (also known as 4-1BB and TNFRSF9) is a co-stimulatory receptor belonging to the tumor necrosis factor receptor family. It is expressed on activated CD8+ and CD4+ T cells, NK cells, DCs, eosinophils, and endothelial cells. Its ligand, CD137L, is primarily expressed on activated antigen-presenting cells (APCs) [44]. The activation of the CD137-CD137L pathway plays a pivotal role in immune responses by promoting T-cell proliferation, enhancing effector functions, preventing activation-induced apoptosis, supporting mitochondrial metabolism, and facilitating DNA demethylation of key genes in CD8+ T cells. In APCs, this pathway contributes to maturation and survival, thereby enhancing their antigen-presenting capabilities [45].
A study involving 66 patients with advanced tumors undergoing ICI therapy revealed that higher frequencies of CD3+CD137+T cells and CD3+CD8+CD137+T cells were associated with longer PFS [46]. Further analysis identified that CD137+PD-1+ and CD8+CD137+PD-1+T cell subsets were positively correlated with response to anti-PD-1 therapy and longer OS. Notably, CD8+CD137+PD-1+T cells exhibited greater proliferative capacity compared with CD8+CD137−PD-1+T cells, underscoring the existence of T-cell subsets with distinct activation states defined by CD137 and PD-1 expression [46]. The co-expression of CD137 and PD-1 identifies a population of highly activated lymphocytes with enhanced tumor-specific reactivity. In hepatocellular carcinoma, CD137+PD-1+T cells were found to possess transcriptional features associated with robust activation and proliferative potential [47]. Similarly, in stage III malignant melanoma patients receiving adjuvant therapy with ipilimumab and nivolumab, circulating CD8+CD137+T cells were strongly associated with improved PFS, further emphasizing their prognostic value [48].

CX3CR1
CX3CR1 is a chemokine receptor widely expressed on various immune cells, including DCs, monocytes, macrophages, NK cells, and T cells. In CD8+T cells, its expression is closely linked to their differentiation state. CX3CR1+CD8+T cells exhibit an effector memory phenotype and express elevated levels of cytotoxic effector molecules, such as granzyme and perforin [49]. However, these cells express lower levels of CXCR3 and CD62L, which restricts their migration to the TME and allows them to remain in circulation following the initial immune response. In contrast, CX3CR1−CD8+T cells preferentially migrate to the TME to mediate anti-tumor responses [50]. CX3CR1 expression remains stable during the effector phase of CD8+T cells, sustained through the unidirectional differentiation of CX3CR1−CD8+T cells into CX3CR1+ cells. Unlike transiently upregulated molecules such as PD-1, 4-1BB, ICOS, and Ki-67, CX3CR1 expression progressively increases over time, particularly during ICI therapy. While CX3CR1+CD8+T cells express higher levels of Ki-67 compared with their CX3CR1− counterparts, Ki-67 expression is transient, whereas CX3CR1 expression persists and increases progressively. The CX3CR1+CD8+T cell subset in peripheral blood is enriched with tumor-specific T cells. During ICI therapy, the T-cell receptor (TCR) repertoire of CX3CR1+CD8+T cells closely mirrors that of CD8+TILs, positioning CX3CR1 as a valuable dynamic biomarker for monitoring responses to anti-CTLA-4 and anti-PD-L1 therapies.
Studies have revealed that CX3CR1+CD8+T cells exhibit higher PD-1 expression and increased transcription of KLRG1, indicating their role as effector T cells responsive to PD-1 inhibitors. These cells can infiltrate tumors and mediate anti-tumor effects [49]. Moreover, CX3CR1+T cells demonstrate resistance to chemotherapy-induced toxicity. Among these, CX3CR1+Granzyme B+CD8+T cells are particularly implicated in chemotherapy-related adverse reactions, as they expel chemotherapy drugs via the ABCB transporter protein. In metastatic melanoma patients who were initially non-responsive to PD-1 inhibitors, a significant increase in CX3CR1+CD8+T cells increased significantly after treatment with a combination of chemotherapy and PD-1 inhibitors, highlighting their potential as predictive biomarkers for the success of combination therapy. Similarly, in NSCLC patients, CX3CR1+CD8+T cells have been identified as predictive markers for the efficacy of chemotherapy combined with immunotherapy. An increase of more than 10% in the CX3CR1+ subset of circulating CD8+T cells, compared with baseline levels, is strongly correlated with treatment efficacy. This increase can be detected as early as 4 weeks, with an overall predictive accuracy of 85.7% at 6 weeks. Notably, a CX3CR1 score increase of at least 10% is associated with significant improvements in both PFS and OS, highlighting the utility of CX3CR1+CD8+T cells as robust biomarkers for predicting the efficacy of combination therapy in NSCLC [51].
CX3CR1, the receptor for the chemokine CX3CL1, was observed to increase peripheral expression on CD8+T cells during a study of bevacizumab and atezolizumab combination therapy for renal cancer. Concomitant increases in CX3CL1 and other chemokine levels within the TME suggest that CX3CR1 upregulation may enhance CD8+T cell infiltration into tumors, potentially implicating CX3CL1 in promoting antigen-specific T-cell migration. These findings indicate that changes in CX3CL1 levels could potentially serve as predictive markers for the efficacy of such combination therapies. However, this conclusion is based on a single study and requires further research to validate its accuracy and reliability as a predictor of therapeutic efficacy [52].
A study of 36 patients with NSCLC treated with anti-PD-1 antibodies found no association between baseline CX3CR1+CD8+T cell proportion and OS. However, responders demonstrated a significantly greater maximal percentage change in the CX3CR1+ subset post-treatment compared with non-responders, with differences observed as early as 3 weeks. Furthermore, an increase in the CX3CR1 score (defined as a ≥20% increase in the CX3CR1+ subset relative to baseline) strongly correlated with improved ORR, PFS, and OS [53]. CX3CR1 offers several advantages as a blood biomarker. Its unidirectional differentiation ensures irreversible expression on fully differentiated T cells. Besides, circulating CX3CR1+CD8+T cells persist, making them well-suited for monitoring treatment efficacy over time.

CD39
CD39 is an extracellular enzyme that, in conjunction with CD73, catalyzes the conversion of ATP to ADP and AMP, ultimately yielding adenosine, a molecule with potent immunosuppressive properties. Within the TME, CD39 plays a significant role; increased expression promotes tumor growth and serves as an independent poor prognostic factor [54]. However, emerging research reveals that CD39 has diverse functions depending on the cell type on which it is expressed [55]. CD39 is notably expressed on CD8+T cells within the TME, where it is associated with chronic antigen stimulation. These CD39+CD8+T cells have been linked to clinical benefits in several cancers [56, 57]. Furthermore, CD39 is often co-expressed with CD103, a marker of tissue-resident memory CD8+ T cells. CD39+CD103+CD8+T cells are considered to comprise true tumor-specific TILs [58]. In head and neck cancer, a high proportion of CD103+CD39+CD8+TILs correlated with better OS [56]. Conversely, CD39−CD8+TILs are defined as a subset lacking chronic antigen stimulation within the tumor site. In patients with epidermal growth factor receptor-mutated lung cancer, 50% lacked CD39+CD8+TILs, a characteristic closely associated with poor response rates to PD-1 antibody immunotherapy [59]. These findings underscore the multifaceted roles of CD39 in cancer immunity and its potential utility as a biomarker for predicting immune responses [60].
In addition to monitoring TILs, tracking changes in peripheral blood CD39+ CD8+ T cells has emerged as a promising marker for evaluating the efficacy of solid tumor immunotherapies [60]. Circulating PD-1+CD39+CD4+T cells are enriched with activated HLA-DR+, ICOS+, and proliferating Ki-67+ cells, indicating their active involvement in sustaining immune responses. In human papillomavirus (HPV)-induced malignancies, the PD-1+CD39+ population contains a high proportion of HPV antigen-specific T cells, and the proportion of circulating PD-1+CD39+CD4+T cells has been shown to predict clinical response in HPV-related tumors treated with ICIs [61]. Similarly, in microsatellite instability-high (MSI-H) metastatic colorectal cancer, patients demonstrated rapid clinical responses to anti-PD-1 therapy, which were associated with high levels of CD39 expression in proliferative CD8+T cells in peripheral blood [62]. These observations aligned with evidence of an expanding CD39+T cell population in the peripheral blood of patients who respond to anti-PD-1 treatment. Collectively, these findings highlight the role of CD39+T cells as early, blood-based indicators of tumor-specific CD8+ responses and underscore the potential of CD39 as a valuable biomarker [60].

OX40
OX40/OX40L signaling is critical for the formation and survival of memory CD4⁺ and CD8⁺T cells, while also suppressing the inhibitory functions and differentiation of Tregs [63]. A study investigating immune responses in patients with gastric cancer treated with PD-1 antibodies revealed that post-treatment proportions of LAG3⁺CD4⁺ and CD8⁺T cells, and OX40⁺CD4⁺ and CD8⁺T cells, were positively correlated with PFS. LAG3⁺T cells, which often co-express PD-1, can regain their anti-tumor functionality through PD-1/PD-L1 blockade, leading to improved prognostic outcomes. The proportion of OX40⁺/LAG3⁺T cells was identified as an independent predictor of treatment efficacy [64]. Furthermore, following anti-PD-1 therapy, the emergence and temporary expansion of new antigen-specific T cell clones has been observed, which may enhance anti-tumor responses via OX40/OX40L signaling [65]. In a phase Ib clinical trial involving solid tumors, patients with sustained clinical benefit from immunotherapy exhibited significantly higher levels of circulating CD4⁺PD-1⁺OX40⁺T cells compared with non-responders. This highlights the predictive potential of OX40⁺T cells in determining the efficacy of immunotherapy [66].

CXCR5
CXCR5+CD8+T cells constitute a specialized subset capable of homing to lymphoid follicles, where they play significant roles in viral, tumor, autoimmune, and alloimmune processes within lymphoid tissues. CXCR5 is recognized as a critical marker for CD8 cytotoxic follicular T cells [67]. Increased infiltration of CXCR5+CD8+T cells in tumors has been linked to better clinical outcomes. For example, in hepatocellular carcinoma, these cells are associated with lower risks of early recurrence and metastasis, while in pancreatic cancer, their presence correlates with prolonged disease-free survival [68, 69]. CXCR5+PD-1+CD8+T cells are found in both peripheral blood and lymph nodes, with a higher frequency in lymph nodes. These cells exhibit an early effector memory phenotype, characterized by high expression of memory markers such as CD127, KLRG1, granzyme K, Eomes, and Tcf1, and low expression of effector molecules such as T-bet and granzyme B. Although they can rapidly produce cytokines like IL-2, their proliferative capacity is limited.
In hematologic malignancies, particularly chronic lymphocytic leukemia (CLL), patients exhibiting poor responses to PD-1 inhibitors demonstrate a loss of memory phenotype within CXCR5+PD-1+CD8+T cells. This phenotypic shift is accompanied by reduced PD-1 expression and increased effector differentiation, characterized by decreased expression of granzyme K, Eomes, and Tcf1, concurrent with increased levels of granzyme B and T-bet. These cells also produce elevated levels of IFN-γ and TNF-α but exhibit diminished functional responses to stimulation, including CXCR5 downregulation and limited granzyme B upregulation [70]. These findings underscore the potential utility of CXCR5+CD8+T cells as predictive markers for immunotherapy efficacy and highlight their complex role in anti-tumor immunity.

Senescence phenotype and exhaustion phenotype
The senescent phenotype is typically observed in TEMRA cells. These cells lose expression of the key co-stimulatory molecules CD28 and CD27 while expressing the senescence markers CD57 and KLRG1, indicative of immune senescence and a limited survival time. In contrast, the exhaustion phenotype is characterized by high expression of inhibitory receptors, resulting in reduced proliferative capacity, diminished survival, and impaired effector functions. This phenotype is associated with elevated levels of heterogeneous inhibitory receptors, including PD-1, CTLA-4, LAG-3, and TIM-3, reflecting a state of T-cell dysfunction often observed during chronic antigen exposure.

CD28
CD28 is a key co-stimulatory molecule expressed during the early stages of T-cell differentiation. TN and TCM cells express high levels of CD28, with TCM cells exhibiting higher expression than TN cells. In contrast, TEM and TEMRA cells lose CD28 expression, marking a state of immune senescence. In NSCLC patients with lung metastases, peripheral blood CD8+CD28+T cells are positively correlated with improved efficacy of stereotactic body radiotherapy [71]. The effectiveness of anti-PD-L1/PD-1 therapies relies on CD28 expression for robust anti-tumor T-cell responses [72, 73]. Responders to PD-1/PD-L1 inhibitors demonstrate significantly higher counts of circulating CD8+CD28+T cells, which predict treatment response with a sensitivity of 0.689 and a specificity of 0.714. Higher CD8+CD28+T cell counts are also associated with longer median PFS and OS, although they increase the risk of grade 3–4 immune-related adverse events [74]. Conversely, peripheral blood CD4+CD28−T cells have been identified as potential markers for hyperprogressive disease (HPD). In NSCLC patients, the proportion of CD4+CD28−T cells after treatment was significantly higher in HPD patients compared with non-HPD progressors and responders [75]. A separate study of 144 lung cancer patients receiving ICI therapy found that high counts of CD28−CD57+ killer cell lectin-like receptor G1 (KLRG1)+CD8+T cells were associated with poor prognosis. These senescent-like cells, with reduced proliferative capacity and diminished IL-2 production, represent potential markers for poor ICI therapy outcomes [76].

PD-1
PD-1 is known for its negative regulatory role on T cells, but its expression can also serve as a marker of T-cell activation and a predictor of PD-1 inhibitor efficacy. In patients with NSCLC treated with PD-1 inhibitors, studies have shown that low peripheral blood counts of NK cells and PD-1+CD8+T cells are associated with poor PFS and OS [77, 78]. Moreover, TCR diversity in PD-1+CD8+T cells prior to immunotherapy is a key determinant of treatment efficacy, with higher diversity correlating with better responses and longer PFS [79].
Differentiation subgroups within the PD-1+CD8+T cell population offer further insights into the efficacy of PD-1 inhibitors. For instance, NSCLC patients with high baseline proportions of PD-1+ early effector memory CD8+ T cells (TEEM; CD28+CD27−CD45RO+) and PD-1+ effector CD8+T cells (TE; CD28−CD27−CD45RO+) were associated with sustained benefit from treatment. Following anti-PD-1 therapy, PD-1+TEEM cells displayed characteristics of early responders, constituting the primary component of circulating PD-1+Ki67+CD8+T cells during expansion [80]. PD-1+TIGIT+ double-positive T cells are particularly relevant as markers of T-cell activation and predictors of PD-1 inhibitor efficacy. One month post-anti-PD-1 therapy, higher proportions of circulating PD-1+TIGIT+CD8+T cells correlated positively with clinical response and prolonged OS. These double-positive T cells are enriched with highly active, tumor-specific cells, newly emerging T cell clones, and CXCR5-overexpressing T lymphocytes [67].
PD-1+CD4+T lymphocytes, in addition to CD8+T cells, have demonstrated significant predictive value in immunotherapy. In NSCLC patients, a positive correlation has been observed between the baseline proportion of PD-1+CD4+T lymphocytes and survival following tumor vaccine therapy [81] and ICI therapy [82]. This positive correlation has also been observed in prostate cancer patients undergoing immunotherapy [81]. In malignant melanoma patients, PD-L1+CD4+ and PD-L1+CD8+T cells are strongly associated with improved outcomes following immunotherapy [48]. The ratio of CD8+PD-1+ to CD4+PD-1+T cells is also a key predictor of ICI therapy efficacy in NSCLC patients [83].
While PD-1 is commonly recognized as a negative regulator of immune activity and a marker of immune exhaustion, its expression on T cells, particularly at baseline, often predicts better outcomes with anti-PD-1 immunotherapy. Following PD-1 inhibitor treatment, a decline in both the proportion and absolute number of PD-1+T cells is typically observed.

LAG-3/TIM-3
Co-inhibitory checkpoint receptors are critical regulators of T-cell responses. LAG-3, TIM-3, and TIGIT are considered next-generation immune checkpoint receptors. LAG-3 is reportedly expressed on activated T cells, B cells, NK cells, and DCs, exerting inhibitory regulatory effects [84]. As a structural homolog of CD4, LAG-3 transmits inhibitory signals through motifs in its cytoplasmic tail, which dissociate phosphorylated Lck from CD4 or CD8 co-receptors, thereby limiting TCR signaling. LAG-3 is co-expressed with other checkpoint receptors in tumors and is associated with poor prognosis. While monotherapy targeting LAG-3 has demonstrated limited efficacy, its combination with PD-1 inhibitors significantly enhances tumor control [85]. TIM-3, expressed on various immune cells, binds ligands such as GAL-9, phosphatidylserine, high-mobility group box 1, and cancer/testis antigen-associated cell adhesion molecule-1. TIM-3 inhibits TCR signaling by recruiting phosphatases CD45 and CD148 and serves as a marker of terminal exhaustion of CD8+ T cells in chronic viral infections and cancer [85]. A study on advanced NSCLC revealed that ICI responders had higher baseline percentages of PD-1+, CTLA-4+, TIM-3+, and PD-L1+CD4+TM cells compared with non-responders. Patients with elevated baseline PD-1+, PD-L1+, CTLA−4+, or TIM-3+CD4+TM cells experienced better prognoses and longer PFS, highlighting the predictive value of these molecules for ICI efficacy [32]. In contrast, increases in LAG-3+ and TIM-3+T cells post-immunotherapy were observed in progressive disease (PD) patients with gastric cancer. These cells represent exhausted T-cell populations that fail to respond to PD-1 therapy [64].
The contradictory findings from these studies suggest a critical nuance: although LAG-3 serves as an exhaustion marker, its initial expression on the cell surface may be contingent upon prior T-cell activation. Specifically, elevated LAG-3 expression on TILs has been correlated with a favorable prognosis across various tumor entities. Furthermore, evidence indicates that CD8+T cells that co-express PD-1, TIM-3, and LAG-3 are characterized by both tumor reactivity and neoantigen specificity [64].This leads to the hypothesis that LAG-3-positive T cells are typically PD-1 expressing, and that PD-1/PD-L1 blockade can effectively restore the anti-tumor function of this tumor-specific exhausted T-cell population, thereby translating into improved clinical outcomes. Consequently, the LAG-3 positive T cells linked to prognostic improvement following anti-PD-1 therapy likely include a tumor-specific exhausted subset that is amenable to reactivation PD-1/PD-L1 inhibition. Conversely, LAG-3-positive T cells detected at the time of PD may exhibit simultaneous expression of multiple inhibitory receptors, suggesting a state of hyper-exhaustion that renders them unresponsive to PD-1 blockade. This complex interaction underscores that integrated biomarker detection is essential for ICIs efficacy prediction.

Other molecules
CD226 (DNAM-1) is an adhesion molecule that enhances NK cell and CD8+T cell-mediated cytotoxicity. Its ligands, CD112 and CD155, are commonly expressed on tumor cells [86]. Preclinical studies have demonstrated that the TME promotes the differentiation of CD226+CD8+T cells into CD226−CD8+T cells, which suppresses CD8+T cell functionality and diminishes the efficacy of tumor immunotherapy. These CD226−CD8+T cells exhibit altered TCR signaling, reduced effector molecule expression, and ineffectiveness during ICI therapy [87]. CD226 expression is crucial for the efficacy of anti-PD-1 therapy, and the accumulation of CD226−CD8+T cells is hypothesized to limit ICI effectiveness. In patients with acute myeloid leukemia, a distinct subset of peripheral blood CD8+T cells, characterized by PD-1+TIGIT+CD226− expression, has been associated with CD8+T cell dysfunction and poor clinical prognosis, suggesting its potential as a prognostic biomarker [88].

Conclusions
Monitoring peripheral blood T-cell subsets offers significant value in evaluating immunotherapy efficacy, providing critical insights into the status and function of systemic immune cells, thereby enabling predictions of ICIs treatment efficacy and response. A primary advantage of this approach is its accessibility, which facilitates dynamic, real-time tracking of immune changes throughout the therapeutic course. However, the clinical application of this strategy faces substantial hurdles. Firstly, existing guidelines and consensus documents are insufficient, failing to fully encompass the highly diverse and complex T-cell subsets relevant to predicting ICIs response. Secondly, the intricate nature of the TME and its dynamic interaction with circulating T lymphocytes pose a major challenge to relying on single circulating T lymphocyte markers, as the predictive performance of a solitary indicator is generally low.
Future research must address these limitations across multiple dimensions to achieve breakthrough progress. Constructing predictive systems through the combined detection of multiple circulating T-cell subsets is critical for enhancing predictive efficacy and promoting wider clinical adoption. Simultaneously, a key research direction involves integrating multi-parameter flow cytometry data, clinical baseline information, cytokine levels, and various multi-omics data to build comprehensive predictive models. The emergence of artificial intelligence (AI) offers robust solutions for these models, particularly given its capability to process high-dimensional, multi-modal data, making AI essential for deeply fusing multi-omics and multi-dimensional information to construct a synergistic and panoramic prediction system [89]. Lastly, establishing a large-sample database for the circulating T-lymphocyte subsets is of paramount value for promoting the rigorous clinical validation and subsequent application of this detection technology.