Soluble CTLA-4 as a context-dependent immune checkpoint: biology, biomarker potential, and therapeutic implications

Abstract

Cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) exists not only as a membrane-bound immune checkpoint but also as a naturally secreted soluble isoform (sCTLA-4), generated primarily through exon 3 skipping. This review focuses specifically on splice-derived sCTLA-4 and examines its biological and clinical significance. sCTLA-4 is shaped by immune state, cellular source, and tissue microenvironment. Mechanistically, sCTLA-4 can restrain B7/CD28-dependent immune activation, while recent preclinical evidence suggests a preferential effect on type 1 immune responses. In autoimmune and inflammatory diseases, longitudinal changes in sCTLA-4 often parallel changes in immune activity, although its performance as a disease-specific biomarker remains limited. In cancer, associations with tumor burden, prognosis, and treatment response are more variable and appear to depend strongly on the surrounding immune context. Thus, circulating sCTLA-4 may be better viewed as a state-dependent immunoregulatory signal than as a universal standalone biomarker. Current evidence does not clearly establish whether its elevation is causal, compensatory, or epiphenomenal. Key translational priorities include detection methods that distinguish sCTLA-4 isoforms and cellular sources, standardized longitudinal sampling, biomarker panels that integrate sCTLA-4 with defined immune phenotypes, and therapeutic strategies tailored to specific immune contexts while preserving peripheral tolerance.

Graphical Abstract

1. Introduction

Cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4, CD152) is the prototypic inhibitory immune checkpoint molecule. It is constitutively expressed at high levels on CD4+FoxP3+ regulatory T cells (Tregs) and is transiently upregulated on activated conventional T cells. CTLA-4 outcompetes CD28 for the shared ligands B7-1 (CD80) and B7-2 (CD86) and depletes B7 molecules from antigen-presenting cells through trans-endocytosis (1–3). Therapeutic blockade of CTLA-4 with ipilimumab established immune checkpoint inhibition as a therapeutic class. However, it also revealed a substantial burden of immune-related adverse events (irAEs). These safety concerns have motivated efforts to widen the therapeutic window through local delivery, sustained-release formulations, and rational combination with PD-1/PD-L1 inhibitors (4–8).

Less appreciated, but increasingly central, is a naturally occurring soluble isoform of CTLA-4 (sCTLA-4) generated through alternative splicing that removes the transmembrane-encoding exon 3. As sCTLA-4 contains the B7-binding IgV-like domain without a membrane anchor, it binds extracellular CD80/CD86. This limits CD28 co-stimulatory ligand access and modulates T-cell activation beyond the immunological synapse (9, 10). sCTLA-4 is detectable in the serum of healthy individuals and is markedly dysregulated in autoimmunity, chronic infection, malignancy, and aging. These observations have made it a candidate liquid-biopsy biomarker and a target in its own right (11–15).

Recent advances have renewed interest in the biological and clinical significance of sCTLA-4. A recent mouse genetic and in vitro study by Osaki and Sakaguchi reported that sCTLA-4 deficiency preferentially enhanced type 1 immune responses (16). In contrast, impaired function of membrane-bound CTLA-4 (mCTLA-4) produced an autoimmune phenotype biased toward type 2 immunity (16). This immunotype-selective pattern is only backed by a single preclinical study and remains a tentative model instead of a confirmed immunological mechanism. Further animal validation and large-scale human cohort verification are required. sCTLA-4 has also been associated with immune features related to resistance to PD-1/PD-L1 blockade in lung and gastrointestinal cancers (14, 17). In addition, germline CTLA-4 variants are linked to sCTLA-4 expression and clinical outcomes in several malignancies, suggesting a connection between host immunogenetics and disease progression (18, 19). Preclinical studies of engineered sCTLA-4 variants and CTLA-4 trafficking modulators have shown that modifying this axis can alter tumor control and immune-related toxicity (20, 21). Collectively, these findings support sCTLA-4 as a biologically active molecule relevant to immune regulation, longitudinal disease monitoring, and therapeutic targeting.

Based on these advances, we consider that sCTLA-4 is a context-dependent immunoselective regulator with therapeutic potential, rather than a passive serum biomarker. This review summarizes the production mechanisms and biological functions of sCTLA-4. It integrates clinical evidence around relevant biological questions rather than disease types and critically analyzes analytical and therapeutic strategies needed for clinical translation.

2. Literature search strategy

This narrative review was informed by a literature search of PubMed, Web of Science Core Collection, and Embase for studies published from 1999 to December 2025. Search terms included combinations of “soluble CTLA-4”, “sCTLA-4”, “CTLA-4 alternative splicing”, “soluble immune checkpoint”, “regulatory T cells”, “autoimmune disease”, “cancer”, “biomarker” and “immunotherapy”. Peer-reviewed original studies directly addressing sCTLA-4 generation, cellular sources, biological functions, clinical associations, detection methods, or therapeutic implications were prioritized. Relevant high-quality reviews were used to provide additional context and identify primary studies. Duplicate publications, conference abstracts without sufficient data, non-peer-reviewed reports, studies lacking direct relevance to sCTLA-4, and articles with insufficient methodological or outcome information were excluded. Reference lists of eligible publications were also screened to identify additional pertinent studies.

3. Molecular generation and regulation of sCTLA-4

3.1. Gene architecture and alternative splicing

The human CTLA-4 gene comprises four exons encoding, respectively, the signal peptide, the ligand-binding immunoglobulin V-like domain, the transmembrane region, and the cytoplasmic tail. Alternative splicing that skips exon 3 yields a transcript that retains the high-affinity CD80/CD86-binding domain but lacks the membrane anchor. The translated product is secreted as a soluble glycoprotein (9, 10). sCTLA-4 transcripts are detectable in non-stimulated human T cells, indicating that basal alternative splicing contributes to a constitutive low-level soluble pool even in the absence of immune activation (10). A portion of secreted sCTLA-4 is also encapsulated into extracellular vesicles and exosomes. The biological implications of this property, including potential tissue-targeted delivery and an extended functional half-life, have not been fully elucidated to date (9).

Differential exon usage can generate several CTLA-4 isoforms. The main forms are full-length membrane CTLA-4 and soluble CTLA-4, which results from exon 3 skipping. In mice, additional rare isoforms have been reported. These include ligand-independent CTLA-4, which lacks the exon 2–encoded B7-binding domain but retains inhibitory signaling capacity, and a minor isoform containing only exons 1 and 4 (22, 23). However, these isoforms are mainly documented in mice. In humans, only full-length CTLA-4 and sCTLA-4 have been reliably detected at the protein level (9, 24). Because this review focuses on the secreted isoform, the following discussion centers on sCTLA-4.

This species difference is important for translational interpretation. Phenotypes generated by manipulation of the murine CTLA-4 locus may reflect the combined effects of sCTLA-4 and murine-specific isoforms. This is particularly relevant when knockout or editing strategies alter exons or regulatory elements shared by several transcripts. Consequently, changes in T-cell activation, autoimmunity, or tumor immunity cannot automatically be assigned to sCTLA-4 alone. At the same time, the canonical exon 3-skipped sCTLA-4 isoform is conserved in mice and humans, so mouse models remain mechanistically informative when the intervention is demonstrably isoform-specific. Their translational relevance is therefore conditional rather than absolute. Findings should be strengthened by transcript/protein-level confirmation of the affected isoform and validation in human cells or clinical cohorts (16, 22–24).

3.2. Transcriptional, epigenetic, genetic, and post-transcriptional control

sCTLA-4 expression is controlled at multiple regulatory layers. Transcriptionally, CTLA-4 activity is shaped by DNA methylation and histone modifications (25). Pharmacological dampening of T-cell activation, such as calcineurin inhibitor therapy in renal transplantation, reduces both CTLA-4 transcription and circulating sCTLA-4 levels. This indicates that systemic T-cell activation state is a major upstream determinant of the sCTLA-4 pool (26). However, the splice-factor program that determines exon 3 skipping in defined T-cell subsets remains incompletely resolved. SR proteins and hnRNPs have been proposed as candidate mediators through which non-coding CTLA-4 variation could influence splicing. However, direct evidence for an sCTLA-4 splicing program specific to effector Tregs or dependent on FOXP3 is currently lacking (18, 19).

Germline variation in CTLA-4 constitutes a distinct, clinically relevant regulatory layer. The CT60 and Jo31 variants, originally characterized in autoimmunity, alter sCTLA-4 serum concentrations and disease susceptibility across multiple ethnic populations (27–29). More recently, rs231726, rs11571317, and rs13384548 have been associated with disease risk, tumor burden, and stepwise sCTLA-4 elevation in pancreatic ductal adenocarcinoma (PDAC) and HCV-related hepatocellular carcinoma (HCC) (18, 19). The mechanistic interpretation of these associations remains incomplete. Most disease-associated CTLA-4 SNPs lie in non-coding regions and could in principle alter sCTLA-4 production through several non-exclusive routes. One route is altered affinity of cis-acting splicing enhancers or silencers for SR proteins and hnRNPs. Other possibilities include changes in mRNA stability mediated by the 3′ UTR or altered looping between enhancers and promoters in T cells and Tregs. Direct evidence linking specific SNPs to defined splicing outcomes is still limited, and this remains an important mechanistic gap that bridges human genetics and sCTLA-4 biology.

Post-transcriptional control of alternative splicing is itself responsive to the microenvironment. Transforming growth factor-β2 (TGF-β2), enriched in many solid tumors, potently induces sCTLA-4 secretion from T cells and correlates with serum sCTLA-4 in melanoma (30). These findings are consistent with TGF-β2-driven sCTLA-4 induction that may reinforce local immune suppression. However, a true reciprocal feedback loop has not yet been demonstrated. Hypoxia and hypoxia-inducible factor-1α (HIF-1α) signaling have been speculatively linked to a shift in CTLA-4 splicing toward the soluble isoform. However, direct evidence that HIF-1α regulates CTLA-4 pre-mRNA splicing is lacking, so this possibility remains conjectural (31). Inflammatory cytokines and paracrine factors from tumor-associated stromal cells may further modulate sCTLA-4 release, both directly and through extracellular vesicles (32).

3.3. Cellular sources and microenvironmental accumulation

Although sCTLA-4 can be produced by multiple cell types, effector Tregs are reported to be a major source under both basal and inflammatory conditions (16, 33). This enrichment is consistent with the constitutively high CTLA-4 expression of FoxP3+ Tregs. It also aligns with evidence that T-cell activation state regulates CTLA-4 transcription and the circulating sCTLA-4 pool (1–3, 26). However, the molecular basis for preferential soluble-isoform production in effector Tregs remains incompletely defined. Exon 3 skipping is subject to post-transcriptional regulation, but a Treg-restricted or FOXP3-dependent splice program has not been demonstrated. Thus, current evidence supports effector Tregs as an enriched source without establishing a unique lineage-specific mechanism of sCTLA-4 generation. Functionally, sCTLA-4 produced by Tregs can extend suppressive activity beyond direct cell contact by acting as a diffusible paracrine signal. Activated conventional CD4+ and CD8+ T cells contribute additional sCTLA-4 in proportion to their activation state and the surrounding cytokine milieu (10, 33).

A more recent and clinically relevant observation is that non-lymphoid cells can also contribute to sCTLA-4. sCTLA-4 expression has been detected in normal, dysplastic, and neoplastic oral and oropharyngeal epithelia (34). This suggests that epithelial tumor cells may seed sCTLA-4 in the tumor microenvironment (TME) independently of T-cell infiltration. Such production could promote immune escape at the earliest pre-infiltrative stages of carcinogenesis.

The accumulation of sCTLA-4 in tissue compartments varies according to disease type and the local immune environment. Serum sCTLA-4 is elevated in glioma and correlates with tumor grade (35). Pre-treatment serum sCTLA-4 independently predicts recurrence after radiofrequency ablation in HCV-related HCC (36). In malignant pleural mesothelioma, pleural-effusion sCTLA-4 varies with clinicopathological features but is not a standalone diagnostic or prognostic marker. TGF-β2 and TGF-β1/3 provide stronger signals in that setting (37). Mesothelioma therefore illustrates a more general principle. sCTLA-4 should be interpreted within tissue-specific soluble-checkpoint networks rather than as a universal, isolated marker.

A frequently overlooked analytical question follows directly from this cellular landscape. Most antibody-based platforms quantify sCTLA-4 as a single analyte but use antibodies against the immunoglobulin domain shared by splice-derived and transmembrane isoforms. Because isoform-specific high-affinity reagents are lacking, these platforms do not specifically resolve splice-derived soluble CTLA-4 from the membrane isoform (9). It remains unproven whether membrane CTLA-4 enters the circulation as a proteolytically shed ectodomain, as has been documented for PD-1, PD-L1, and LAG-3. This should therefore be considered a hypothesis instead of a confirmed source of soluble CTLA-4. To the extent that the measured signal is a mixture, conflating these species may attenuate the apparent strength of clinical correlations. These isoform-specific biological and analytical considerations are summarized in Figure 1.

Figure 1.

Isoform-resolved model of sCTLA-4 biology and interpretation. Treg, regulatory T cell; mCTLA-4, membrane CTLA-4.

4. Biological functions of sCTLA-4

For a long time, sCTLA-4 has generally been regarded as a monofunctional molecule that broadly inhibits T cell activation. Recent studies have advanced our understanding of sCTLA-4 biology along four interconnected dimensions. These dimensions include competitive B7 binding and growing evidence for preferential inhibition of type 1 immune responses. They also include regulatory effects on diverse immune cell populations and integration into the wider soluble checkpoint network.

4.1. Competitive B7 engagement as the core molecular mechanism

The best-characterized activity of sCTLA-4 is high-affinity engagement of CD80 and CD86, which competitively reduces their availability for CD28-mediated co-stimulation and raises the activation threshold of responding T cells. Unlike mCTLA-4, which operates predominantly at the immunological synapse, sCTLA-4 can act at a distance. It can therefore modulate immune responses in tissue compartments and draining lymph nodes that are spatially separated from the producing cell (33, 38). This ability to sequester B7 ligands outside the immunological synapse provides the mechanistic basis for its effects on T-cell differentiation, myeloid polarization, and soluble-checkpoint interactions discussed below.

Abatacept is a CTLA-4–Fc fusion protein approved for the treatment of rheumatoid arthritis and commonly utilized in experimental research as a soluble CTLA-4 mimic. Its full therapeutic effect stems from B7 sequestration, providing pharmacological proof of concept that competitive B7 engagement alone can mediate immunosuppression (39–42).

4.2. Immunotype-selective suppression

One recent preclinical study provided direct evidence that the preferential effect of sCTLA-4 on type 1 immunity depends on its interaction with B7 ligands. Recombinant sCTLA-4 suppressed naïve T-cell proliferation and Th1 differentiation, whereas a Y139A sCTLA-4 mutant with impaired CD80/CD86-binding capacity largely lost this inhibitory activity (16). Moreover, blockade of CD80/CD86 reproduced the suppression of Th1 differentiation, supporting B7 sequestration and consequent attenuation of CD28-dependent priming as a key mechanistic requirement. In contrast, under IL-4-driven polarizing conditions, Th2 differentiation was relatively preserved even at sCTLA-4 concentrations that markedly inhibited Th1 differentiation (16). These findings indicate that the apparent immunotype selectivity does not result from type 1-specific ligand affinity of sCTLA-4. Instead, it likely reflects differences in how type 1 and type 2 differentiation programs depend on B7/CD28 costimulation. Their surrounding cytokine milieus may further shape this selectivity.

Consistent with this model, sCTLA-4-deficient mice exhibited spontaneous expansion and activation of Th1, Th17, Tfh, and Tc1 populations, together with aggravated autoimmunity and defective inflammation resolution (16). By contrast, mice with impaired membrane-bound CTLA-4 but preserved sCTLA-4 showed predominantly type 2-biased immune manifestations (16). Tumor-model studies further support a preferential effect on type 1 effector cells, as tumor-derived sCTLA-4 restrains CD8+ T-cell proliferation, activation, and cytolytic function (38). Although not isoform-specific, dendritic-cell studies also indicate that CTLA-4 engagement can reduce IL-12-dominant antigen-presenting programs (43). It can also favor a more regulatory cytokine environment, which would be expected to disfavor type 1 polarization. Collectively, these observations support a provisional model in which sCTLA-4 preferentially restrains B7/CD28-dependent type 1 immune circuits while relatively sparing type 2 responses. However, because the defining immunotype-selective evidence currently derives predominantly from a single preclinical study, this model requires independent experimental replication and validation in human immune systems.

Building on these preclinical findings, we use an immune phenotype-based framework as a hypothesis-generating tool rather than a validated mechanistic model. Within this framework, the net biological effect of sCTLA-4 may depend on the prevailing immune state and tissue microenvironment. If the preferential restraint of B7/CD28-dependent type 1 responses observed in mice is conserved in humans, increased sCTLA-4 could limit excessive Th1/Tc1-driven inflammation. In selected autoimmune or sterile inflammatory settings, this effect could support immune homeostasis. Conversely, effective tumor control in some settings depends strongly on CD8+/Tc1-mediated type 1 immunity. In such tumors, elevated sCTLA-4 could contribute to immune suppression and may be associated with adverse clinical outcomes. In type 2- or myeloid-dominant environments, however, circulating sCTLA-4 may be less directly related to disease outcome and may instead reflect the broader composition and activation state of the local immune compartment. These interpretations remain hypothetical and require prospective human studies integrating sCTLA-4 measurements with immune-phenotype profiling.

4.3. Cellular targets beyond Tregs

4.3.1. Regulatory T cells

Although competitive B7 binding serves as a shared molecular mechanism, this pathway triggers markedly distinct downstream effects across various immune cell subsets. Tregs are an important cellular source of sCTLA-4 and also represent a central functional target (16, 33). sCTLA-4 secreted by Tregs can reinforce the suppressive capacity of Tregs via autocrine/paracrine pathways and block CD80/CD86 signals to prevent destabilization of the Treg phenotype (16, 44).

4.3.2. CD8+ T cells

Beyond Tregs, CD8+ T cells constitute the target with the greatest clinical relevance. Kennedy et al. demonstrated that tumor cell-derived sCTLA-4 suppresses CD8+ T cell proliferation in vitro, accelerates tumor growth and experimental metastasis in vivo. Characteristically, it restrains intratumoral CD8+ T cells in a non-cytotoxic state instead of directly inducing apoptotic depletion (38). Isoform-specific anti-sCTLA-4 antibody blockade reverses such suppression, enhances the activation and cytolytic capacity of intratumoral CD8+ T cells, and facilitates tumor control (38). Genetic animal studies conducted by Osaki and Sakaguchi further support the selective regulation of Tc1 cells by sCTLA-4, as mice deficient in sCTLA-4 spontaneously accumulate activated Tc1 cells (16). This observation remains preclinical and requires independent validation in human studies.

4.3.3. Dendritic cells, monocytes, and macrophages

Myeloid cells are not merely passive platforms presenting B7 molecules but actively participate in sCTLA-4-related immune regulation. Upon maturation induced by lipopolysaccharide (LPS), Poly(I:C), or inflammatory cytokines, human monocyte-derived dendritic cells (DCs) upregulate both membrane-bound and soluble CTLA-4 (43, 45). In contrast, immature DCs barely express CTLA-4. Engagement of CTLA-4 on DCs shifts cytokine secretion from IL-12 toward IL-10, impairs antigen-presenting capacity, and endows DCs with a partial tolerogenic phenotype (43). Accordingly, DCs are susceptible to paracrine sCTLA-4 and can additionally secrete sCTLA-4 themselves during maturation.

In monocytes and macrophages, tumor-derived sCTLA-4 promotes the accumulation of CD206+ M2- like macrophages within tumors, whereas macrophages from mice lacking sCTLA-4 spontaneously polarize toward an M1 phenotype (16). Tumors with high sCTLA-4 expression remodel myeloid composition, accompanied by decreased numbers of intratumoral F4/80⁺ macrophages and Ly6C^hi monocytes (38). These findings extend the regulatory spectrum of sCTLA-4 from lymphoid effector cells to myeloid polarization, indicating that macrophage phenotypic skewing is an underrecognized component of sCTLA-4-driven immune remodeling.

4.3.4. NK and B lymphocytes

Western blot and quantitative real-time PCR have detected both membrane and soluble CTLA-4 isoforms in human NK cell lines. Activated murine and human NK cells upregulate CTLA-4, which suppresses IFN-γ production by NK cells upon stimulation with mature DCs and attenuates NK effector functions (46, 47). Nevertheless, it remains unclear whether sCTLA-4 directly modulates NK cells independent of T cell pathways. NK cell depletion experiments by Kennedy et al. showed that sCTLA-4-mediated suppression of tumor cytotoxicity persisted in the absence of NK cells (38). This suggests that CD8+ T cells, rather than NK cells, are the major effector population restrained by sCTLA-4 in this model.

B lymphocytes also transcribe sCTLA-4 mRNA and express low levels of the corresponding protein (9, 33). sCTLA-4-deficient mice exhibit elevated production of autoantibodies and IgE (16), yet direct evidence supporting an intrinsic B cell regulatory function of sCTLA-4 is still lacking.

4.3.5. Non-immune sources

As described in Section 3.3, epithelial tumor cells (34) and certain non-lymphoid cells under specific conditions contribute to the pool of circulating sCTLA-4. In vivo animal studies have directly shown that tumor cell-derived sCTLA-4 regulates CD8+ T-cell function and macrophage polarization (16, 38). These findings confirm that non-Treg-derived sCTLA-4 is biologically active rather than an inert byproduct.

4.4. Integration within the soluble checkpoint network

sCTLA-4 does not function in isolation but participates in an interactive soluble checkpoint network, in which the PD-1/PD-L1 axis acts as its primary partner. In colorectal cancer, concurrent elevation of sCTLA-4 and soluble PD-L1 generally predicts markedly poorer prognosis. Mechanistically, sCTLA-4 neutralizes CD80/CD86-mediated co-stimulation, whereas PD-L1 directly suppresses effector T cells by recruiting SHP-2 to deliver inhibitory signals (48, 49). The PD-1/PD-L1 axis also sustains Treg viability via the PI3K–Akt–mTOR and adenosine pathways and upregulates PD-L1 expression on Tregs (44, 50). TGF-β-related signaling mediated by sCTLA-4 can converge with the above regulatory programs responsible for Treg stabilization (51).

5. Clinical relevance of sCTLA-4

sCTLA-4 is a multidimensional signal rather than a universal standalone biomarker. It correlates with disease activity, tumor burden, immune-cell composition and activation state, host genetics and treatment response, yet such associations vary markedly across diseases and clinical settings. Therefore, sCTLA-4 should be interpreted within its biological context. Table 1 summarizes available clinical evidence grouped by core biological questions.

Table 1.

Summary of sCTLA-4 as a clinical biomarker across diseases, organized by the dominant biological question addressed.

Disease category	Disease	sCTLA-4 direction	Sample/Design	Sample size (n)	Correlation	Clinical utility/question addressed	Ref.
Autoimmune diseases	SLE	↑ Elevated	Serum; cross-sectional	(52) 100 SLE + 40 HC (53) 79 SLE + 40 HC (54) 104 SLE	IFN-α activity; disease activity	Disease activity	(52–54)
Autoimmune diseases	Rheumatoid arthritis	↑/context-dependent	Serum; cross-sectional + intervention	(55) 56 RA + 32 HC (56) 95 RA + 30 OA + 50 HC (paired blood/SF n=37) (57) 36 RA (58) 200 RA + 200 HC (59) 24 RA	Disease activity (variable); MTX response	Disease activity + treatment prediction	(55–59)
Autoimmune diseases	Graves/TED	↑ Elevated	Serum; longitudinal	(29) 93 GO + 93 HC (60) 78 GD/TED + 40 HC; follow-up n=41	TED severity; treatment response	Disease activity	(29, 60)
Autoimmune diseases	Multiple sclerosis	↑ in relapse	Plasma; cross-sectional	(61) 21 MS + 22 NMO + 18 HC	Relapse vs remission	Disease activity	(61)
Autoimmune diseases	Immune cytopenia (paediatric)	↑ active; ↓ remission	Serum; longitudinal	(12) 47 patients + 47 HC	Disease activity	Real-time pharmacodynamic monitoring	(12)
Tumors	Glioma	↑ Elevated	Serum; cross-sectional	(35) 50 glioma + 50 HC	Tumor grade (positive)	Tumor-burden indicator	(35)
Tumors	HCV-related HCC	Stepwise ↑	Serum; cross-sectional	(19) 75 HCC + 75 HCV + 75 HC (36) 88 HCC	High-risk CTLA-4 genotypes; tumor burden	Host genetics + tumor burden	(19, 36)
Tumors	Pancreatic ductal adenocarcinoma	↑ Elevated	Serum; cross-sectional	(18) 125 PDAC + 125 HC	Grade; CA19-9/CEA; rs231726/rs11571317	Host genetics + tumor burden	(18)
Tumors	Colorectal cancer	↑ Elevated	Serum; cross-sectional	(49) 131 CRC	Overall survival (negative); PD-L1 co-elevation	Tumor immune state	(49)
Tumors	Melanoma	↑ Elevated	Serum; pre-treatment	(30) 32 melanoma + 27 HC + 40 SLE (PBMC subset n=12) (62) N/A (perspective) (63) 62 mixed advanced cancers	Anti-CTLA-4 antibody efficacy (negative)	ICI response prediction	(30, 62, 63)
Tumors	NSCLC	Panel-dependent	Plasma; pre/on-treatment	(14) cohorts A/B: 50/149 (tPD-L1 subset n=121); D/E: 42/43	sPD-L1 + sCTLA-4 with PD-1/PD-L1 non-response	ICI response prediction	(14)
Tumors	Gastrointestinal cancer	↑ in resistance	Serum; pre/on-treatment	(17) 36 advanced gastric cancer	sCTLA-4 + CRP; anti-PD-1 resistance	ICI resistance stratification	(17)
Tumors	Bladder cancer	Variable	Serum/multi-omic	(64) TCGA-BLCA n=406; multi-cohort validation	GNG4 co-expression; treatment-modality response	Chemo vs immunotherapy selection	(64)
Tumors	Malignant pleural mesothelioma	Variable	Pleural effusion; cross-sectional	(37) 79 MPM + 34 fibrinous pleuritis	Weaker than TGF-β isoforms	Negative case; panel-dependent	(37)

↑, higher sCTLA-4 concentrations relative to the comparator group (healthy controls, or the inactive/remission state within the same cohort); ↓, lower sCTLA-4 concentrations relative to the comparator group. Sample sizes are reported by cohort where applicable; cohort-specific n values are shown for multi-cohort analyses. Reference 62 is a narrative perspective article without a primary patient cohort and is included for mechanistic interpretation only. SLE, systemic lupus erythematosus; HC, healthy controls; RA, rheumatoid arthritis; OA, osteoarthritis; SF, synovial fluid; MTX, methotrexate; GO, Graves' orbitopathy; GD, Graves' disease; TED, thyroid eye disease; MS, multiple sclerosis; NMO, neuromyelitis optica; HCC, hepatocellular carcinoma; PDAC, pancreatic ductal adenocarcinoma; CRC, colorectal cancer; PBMC, peripheral blood mononuclear cells; NSCLC, non-small-cell lung cancer; tPD-L1, tissue PD-L1; sPD-L1, soluble PD-L1; CRP, C-reactive protein; TCGA-BLCA, The Cancer Genome Atlas bladder urothelial carcinoma cohort; GNG4, G-protein subunit gamma 4; MPM, malignant pleural mesothelioma; ICI, immune checkpoint inhibitor; N/A, not applicable.

5.1. sCTLA-4 as an index of disease activity in autoimmune and inflammatory conditions

Across autoimmune and inflammatory disorders, current evidence is more consistent with sCTLA-4 as a state-dependent marker of immune activity than as a disease-specific biomarker. In systemic lupus erythematosus (SLE), elevated serum sCTLA-4 correlates with interferon-α activity and composite disease-activity scores (52–54), supporting an association with the intensity of immune activation. Evidence from longitudinal or within-disease comparisons provides stronger support for dynamic monitoring. Circulating sCTLA-4 decreases in parallel with clinical improvement in Graves’ disease and thyroid eye disease (TED) (29, 60). It is also higher during relapse than remission in multiple sclerosis (61) and falls substantially during remission in paediatric immune cytopenia (12). The convergence of these findings across clinically distinct disorders suggests that sCTLA-4 may reflect a shared immunoregulatory state associated with active disease rather than a pathology-specific signal. However, most available studies are observational and use heterogeneous designs. Therefore, these associations do not establish causality or show that sCTLA-4 provides independent clinical information beyond established measures of disease activity.

Two interpretive challenges complicate this picture. First, the rheumatoid arthritis (RA) literature is genuinely heterogeneous. Some studies report elevated sCTLA-4 in active disease (55, 56), while others find a negative or null association with disease-activity indices (57). This heterogeneity is unlikely to be explained by sampling alone. Plausible contributors include disease duration (early versus established RA), background therapy (untreated versus DMARD-experienced cohorts), and assay differences in epitope specificity. The enrolled disease subset, including seropositive versus seronegative RA, may also contribute. Higher baseline sCTLA-4 has been associated with improved response to methotrexate (57), suggesting that within RA, sCTLA-4 may report a particular immunoregulatory phenotype rather than disease intensity per se.

Second, causality is not established. Elevated sCTLA-4 in active disease has been interpreted as a reactive, IFN-α-driven homeostatic response that dampens overactive T-cell programs (54). Alternatively, it may act as a pathogenic species that competitively displaces endogenous mCTLA-4 from B7 ligands and thereby weakens negative feedback signaling (13, 65). These two interpretations are not mutually exclusive. sCTLA-4 might begin as a compensatory response and become maladaptive under chronic conditions, yet the two models yield opposite predictions about therapeutic targeting. Cross-sectional studies cannot adjudicate between them. Longitudinal trajectories of sCTLA-4 across flares, treatment-induced remissions, and Mendelian-randomization analyses leveraging CTLA-4 polymorphisms are required to clarify direction of effect.

5.2. sCTLA-4 in tumors: tumor burden, tumor immune state, and host genetics

5.2.1. sCTLA-4 as a tumor-burden indicator

Table 1 shows that sCTLA-4 tracks disease extent in some tumors, but this association should not be equated with tumor specificity. Increasing burden may enlarge both tumor-derived and immune-cell-derived sCTLA-4 pools and trigger compensatory immune suppression (16, 33, 34, 38). The circulating level therefore likely integrates tumor mass, immune activation, and negative feedback. Its value is more plausible for longitudinal monitoring alongside established burden markers than for independent staging.

5.2.2. sCTLA-4 as an indicator of the tumor immune state

The variable associations in Table 1 are consistent with context-dependent biology. Preclinical mouse-model data suggest that, in tumors dependent on type 1 immunity, sCTLA-4 may sequester B7 ligands and restrain CD8+ or Tc1 responses (16, 38). Whether this pattern applies consistently in patients remains uncertain. In Treg- or myeloid-dominant settings, sCTLA-4 may instead mark a broader suppressive network. This possibility helps explain why combinations with other soluble checkpoints or inflammatory mediators generally outperform sCTLA-4 alone. Spatial changes in the CTLA-4/CD28 axis in oral squamous cell carcinoma (66) further indicate that circulating values cannot fully represent tissue-level immune regulation.

5.2.3. sCTLA-4 as a reporter of host immunogenetics

Table 1 further indicates that host genetics may modify the sCTLA-4 signal. CTLA-4 variants could affect transcription, alternative splicing, or microenvironmental responsiveness (18, 19), so the same concentration may not have identical biological meaning across genetic backgrounds. Because direct variant-to-splicing mechanisms remain unresolved, genotype should currently be used for stratification rather than causal interpretation. Future studies should pair protein measurement with germline analysis and cross-ancestry validation.

5.2.4. The boundaries of sCTLA-4 as a tumor biomarker

Three limits of the current oncology evidence deserve emphasis. Most studies are single-center and cross-sectional, with small absolute differences between groups and substantial overlap of distributions; effect sizes for sCTLA-4 alone are rarely sufficient to drive clinical decision-making. The mesothelioma data (Table 1) illustrate that elevations of soluble checkpoints can be biologically real without being diagnostically useful (37). And studies establishing associations between CTLA-4 genotypes and sCTLA-4 levels in specific populations (18, 19) should not yet be generalized across ancestries without independent validation. These limitations do not diminish the conceptual relevance of sCTLA-4; rather, they argue for multi-analyte, multicenter, and longitudinal designs in the next generation of studies.

5.3. sCTLA-4 as a predictor of response to immune checkpoint and other therapies

The most clinically actionable application of sCTLA-4 lies in predicting and monitoring response to therapy. Two non-exclusive mechanisms have been proposed for how baseline sCTLA-4 modifies the activity of anti-CTLA-4 antibodies. On the one hand, sCTLA-4 may compete with mCTLA-4 for CD80/CD86 in the extracellular space. It can occupy B7 ligands that would otherwise become available for CD28 re-engagement after mCTLA-4 blockade. In this model, elevated sCTLA-4 sustains B7 ligand occupancy and dampens net CD28 co-stimulation. On the other hand, sCTLA-4 may absorb a fraction of the administered antibody as a circulating “decoy.” This could reduce the effective antibody concentration available to engage tumor-infiltrating Tregs and effector cells. These mechanisms predict different relationships between sCTLA-4 concentration, antibody pharmacokinetics, and clinical response, and the available human data do not yet distinguish them. Distinguishing the competition and decoy models will require paired measurement of sCTLA-4 concentration and therapeutic-antibody pharmacokinetics across the dosing interval. Measuring free versus sCTLA-4-bound antibody would provide a tractable correlative design for future trials. In melanoma, however, both models predict the empirically observed inverse relationship between baseline sCTLA-4 and clinical response to ipilimumab (30, 62, 67). The corresponding mechanisms are illustrated in Figures 2 and 3.

Figure 2.

Competitive B7-occupancy model for sCTLA-4-mediated attenuation of anti-CTLA-4 therapy. Under anti-CTLA-4 blockade, CD80/CD86 are expected to become available for CD28-mediated co-stimulation. Elevated sCTLA-4 may continue to occupy CD80/CD86 in the extracellular space, thereby limiting CD28 engagement and weakening T-cell activation.

Figure 3.

Circulating decoy model for sCTLA-4-mediated reduction of effective anti-CTLA-4 antibody exposure. Circulating sCTLA-4 may bind a fraction of administered anti-CTLA-4 antibodies before they reach the tumor microenvironment. This sequestration could reduce the amount of free antibody available to engage CTLA-4-expressing immune cells and attenuate anti-tumor immune activation.

Beyond CTLA-4-directed therapy, sCTLA-4 may also inform PD-1/PD-L1 blockade when measured within a multi-analyte panel that includes other soluble checkpoints or inflammatory markers. Oshima et al. reported that elevated sCTLA-4 combined with high C-reactive protein was associated with anti-PD-1 resistance and outperformed either marker in isolation (17). In NSCLC, the sPD-L1/sCTLA-4 combination helped discriminate responders from non-responders in immune-reactive tumors and complemented tissue PD-L1 immunohistochemistry (14). Across both settings, longitudinal sCTLA-4 kinetics, such as declining concentrations during effective therapy, appear more informative than single-time-point measurements (14, 17). This shift toward dynamic monitoring is methodologically important and is one of the strongest practical arguments for standardizing pre-analytical workflows.

sCTLA-4 may also have predictive or pharmacodynamic value outside ICI therapy, although the evidence remains exploratory. After chemoradiotherapy or radiotherapy for solid tumors, post-treatment elevation of sCTLA-4 has been associated with improved survival, possibly reflecting radiation-induced immunogenic cell death and a compensatory regulatory response (68). In bladder cancer, integration of sCTLA-4 with G-protein subunit gamma 4 (GNG4) expression may stratify patients likely to respond differentially to chemotherapy versus immunotherapy (64). In immune cytopenia, dynamic reductions in sCTLA-4 parallel response to immunosuppressive therapy, supporting its potential use as a real-time pharmacodynamic readout (12).

6. Challenges and opportunities in clinical translation

6.1. Limitation of detection method

Accurate quantification of sCTLA-4 is a prerequisite for its clinical translation, and the field is constrained by three overlapping analytical limitations. The first is sensitivity. Conventional ELISA platforms approach their lower detection limits at the concentrations typical of healthy individuals and patients in treatment-induced remission (67, 69). Digital ELISA achieves sub-attomolar detection through miniaturized bead-based reactions in femtoliter wells (70). Luminex-based bead arrays enable simultaneous profiling of soluble-checkpoint panels from small liquid-biopsy volumes (60, 71, 72). Surface-enhanced Raman scattering microfluidic biosensors achieve ultra-low detection limits for parallel checkpoint detection, although their clinical translation remains early-stage (71, 73). Importantly, improved sensitivity alone does not ensure greater biomarker validity or comparability across platforms. Assay reports should therefore specify the sample matrix, antibody epitope, calibration range, lower limit of quantification, and whether free or complexed sCTLA-4 is detected.

The second limitation is molecular specificity. Four CTLA-4 splice isoforms exist: full-length membrane-bound CTLA-4, sCTLA-4, liCTLA-4, and the exon 1/4 isoform. All four isoforms are detectable in mice, whereas only full-length CTLA-4 and sCTLA-4 are reliably detected at the protein level in humans. As the sole secreted CTLA-4 isoform, sCTLA-4 possesses only one splice variant generated via exon 3 skipping, which has been identified in humans, mice, rats and dogs (62, 74). Despite there being only one canonical sCTLA-4 splice variant, conventional antibodies cannot distinguish sCTLA-4 from other soluble CTLA-4 fragments in circulation. Consequently, measured signals may not originate purely from splicing-derived sCTLA-4. Therefore, isoform-selective immunoassays, mass spectrometry-based proteomics, and single-molecule sequencing of CTLA-4 pre-mRNA in defined cell populations (75) are required to convert theoretical differentiation of sCTLA-4 species into quantifiable clinical readouts.

This interspecies asymmetry also creates an analytical and model-design limitation. In murine studies, some antibodies, transcript assays, or genetic perturbations do not resolve liCTLA-4 and the exon 1/4 isoform. Such approaches may generate a composite CTLA-4 readout that is not directly comparable with the simpler human isoform repertoire. Preclinical conclusions about sCTLA-4 should therefore specify which CTLA-4 transcripts and proteins are altered or detected and should avoid extrapolating mouse phenotypes to humans without isoform-resolved validation.

The third limitation is pre-analytical. The Hayashi NSCLC study and the gastrointestinal cancer dataset of Oshima et al. show that sCTLA-4 measurements are sensitive to multiple pre-analytical factors (14, 17, 76, 77). These include sample matrix (serum versus plasma), platelet contamination, freeze-thaw history, antibody epitope, and whether the assay detects free or complexed sCTLA-4. These factors are particularly relevant when reported differences between clinical groups are modest, because analytical variation may contribute to apparent differences in effect size across cohorts. They also limit the transferability of numerical cutoffs derived from one study or platform to another. Future studies should therefore prospectively standardize sampling matrix, collection and processing procedures, storage conditions, sampling timepoints, calibrators, and analyte definitions. Without such harmonization, direct cross-study comparisons and meta-analyses will remain difficult, and universal diagnostic or predictive thresholds for sCTLA-4 cannot be reliably established.

Complementary technologies may help place circulating sCTLA-4 measurements into a broader biological context. Flow cytometry can relate circulating sCTLA-4 concentrations to checkpoint expression on defined T-cell populations (67, 78). Radiolabelled immune-checkpoint probes for PET imaging can assess checkpoint distribution non-invasively within tumors (79, 80). These approaches do not replace direct and molecularly specific quantification of sCTLA-4, but they may help clarify its cellular source and relationship to tissue-level immune regulation. Taken together, current analytical limitations suggest that sCTLA-4 measurements are presently more robust for standardized within-study comparisons, longitudinal monitoring, and multi-analyte profiling than for defining universally applicable standalone clinical thresholds.

6.2. Therapeutic strategies and translational considerations

sCTLA-4 has dual roles in promoting tumor immune escape and maintaining physiological immune tolerance. This creates a central therapeutic challenge: inhibiting its tumor-promoting activity while preserving its homeostatic functions. Although most emerging regulatory strategies do not directly target sCTLA-4 itself, they indirectly modulate it through other components of the CTLA-4 axis or resistance-related pathways. Their goal is to reduce tumor-associated CTLA-4-mediated suppression while limiting systemic interference with CTLA-4-dependent immune tolerance. Table 2 summarizes current and emerging therapeutic strategies involving sCTLA-4 or the CTLA-4 axis. These include clinically established CTLA-4-Fc-based immune modulation, protein-level targeting, intracellular trafficking modulation, delivery-based systems, and combination strategies. Most approaches beyond CTLA-4-Fc remain preclinical or in early translational development.

Table 2.

Therapeutic strategies involving sCTLA-4 or the CTLA-4 axis, grouped by evidence maturity.

Strategy level	Representative approach	Evidence maturity	Key translational issue
Approved immune modulation	Abatacept-like CTLA-4-Fc biology	Clinical use in autoimmunity; indirect relevance to sCTLA-4	May compromise anti-tumor immunity when used during ICI therapy
Protein-level targeting	Anti-sCTLA-4 antibodies; engineered soluble CTLA-4 mutants	Mostly preclinical	Must separate anti-tumor benefit from tolerance disruption
Trafficking modulation	LRBA-CTLA-4 disruption and CTLA-4 degradation	Preclinical/early translational	Effect on soluble versus membrane CTLA-4 requires clarification
Delivery-level targeting	Local delivery; TME-responsive antibody/nanoparticle systems	Preclinical to early clinical rationale	Requires biodistribution, toxicity, and pharmacodynamic endpoints
Combination strategies	Co-targeting CD47, PI3K-gamma, glycolysis, or epigenetic pathways	Conceptual to preclinical	Needs rational biomarker selection and toxicity monitoring

Evidence maturity was added to distinguish established clinical tools from exploratory or preclinical strategies.

At the protein level, conformational epitope-specific anti-sCTLA-4 antibodies selectively neutralize sCTLA-4-mediated immunosuppression in defined antigenic contexts without triggering systemic autoimmunity in preclinical models (13). Engineered soluble CTLA-4 mutants can retain B7-1/B7–2 binding while selectively losing recognition by clinical anti-CTLA-4 antibodies. In preclinical models, these mutants suppressed irAEs while preserving anti-tumor efficacy (20). This offers a structural approach to the abatacept paradox, although clinical validation is still required.

At the intracellular trafficking level, LC427 is an oral small molecule that disrupts the LRBA–CTLA-4 interaction and promotes lysosomal CTLA-4 degradation. In preclinical models, it expands tumor-infiltrating CD8+ T cells and shows anti-tumor activity without inducing colitis-like irAEs (21). Although LC427 does not act directly on sCTLA-4, it offers an alternative route to increase intratumoral CD8+ T-cell activity. It may also be combinable with sCTLA-4-targeted approaches.

At the delivery level, tumor microenvironment-responsive systems, including reactive oxygen species-sensitive bispecific antibody–nanoparticle conjugates, are designed to confine sCTLA-4 blockade to the oxidative TME and limit systemic exposure (81, 82). Local intratumoral or intralymphatic delivery of low-dose anti-CTLA-4 agents has shown that systemic anti-tumor T-cell responses can be elicited without proportionate systemic toxicity (7), but isoform-specific delivery approaches remain early-stage.

Finally, combination-based therapeutic approaches can counter parallel immunosuppressive pathways without directly neutralizing sCTLA-4. Metabolic interventions include PI3Kγ inhibition to reshape macrophage activation (83) and hexokinase 2 inhibition to attenuate lactate-driven immune evasion (84). Epigenetic modifiers such as HDAC inhibitors (25) can also address resistance pathways that act in parallel to sCTLA-4-mediated suppression. These combinations require careful toxicity and pharmacodynamic assessment.

7. Conclusions and future perspectives

Soluble CTLA-4 occupies a paradoxical but increasingly intelligible position in immune biology. It is a regulated splice-derived checkpoint molecule with cell-type-specific production, proposed immunotype-selective function, measurable interactions with soluble PD-1/PD-L1 networks, and genetically influenced expression patterns. These features make sCTLA-4 simultaneously attractive and hazardous as a therapeutic target. Indiscriminate inhibition may impair inflammation resolution and peripheral tolerance. By contrast, context-selective modulation could increase intratumoral CD8+/Tc1 proliferation and cytolytic activity while limiting systemic disruption of CTLA-4-dependent tolerance. However, this therapeutic rationale depends on independent experimental replication and confirmation that the proposed immunotype-selective biology is relevant in humans.

Three priorities will determine whether these biomarker and therapeutic applications can be translated clinically. These span two horizons. The first two are near-term, presently achievable priorities in measurement and study design, whereas the third represents a longer-term, still largely hypothesis-driven therapeutic program. First, isoform- and source-resolved detection platforms must replace aggregate sCTLA-4 measurement, so that splice-derived sCTLA-4 and any membrane-derived species, together with individual splice variants, can be interpreted in clinical context. Second, sCTLA-4 must be embedded within multi-analyte, longitudinal liquid-biopsy panels with harmonized pre-analytical workflows. To date, its most clinically informative use has been as part of such panels rather than as a standalone marker. Third, therapeutic development should move from binary checkpoint blockade toward context-selective modulation. Candidate approaches include isoform-selective antibodies, engineered sCTLA-4 mutants, LRBA–CTLA-4 trafficking modulators, and TME-responsive delivery systems. These strategies should be evaluated with pharmacodynamic endpoints that capture sCTLA-4 kinetics, intratumoral B7 availability, Treg dynamics, and type 1/type 2 immune balance. Validation should explicitly span independent animal models and human cohorts.

Conceptualized this way, sCTLA-4 is not a passive serum marker alone but a context-dependent candidate point of intervention within the CTLA-4 axis. Modulating sCTLA-4 selectively may eventually widen the therapeutic window of cancer immunotherapy while preserving homeostatic immune programs that protect patients from treatment-related toxicity. Such modulation should be considered only where isoform, source, immunotype, genotype, and assay context warrant.

Funding Statement

The author(s) declared that financial support was received for this work and/or its publication. Beijing Research Ward Excellence Program (BRWEP, NO.BRWEP2024W102170100, BRWEP2024W102170116), Prevention and Control of Emerging and Major Infectious Diseases-National Science and Technology Major Project (2025ZD01906000, 2025ZD01906005), National Science and Technology Major Project for the Prevention and Treatment of Cancer, Cardiovascular and Cerebrovascular Diseases, Respiratory Diseases, and Metabolic Diseases (2024ZD0520504), the National Science Foundation of China (82274479, 82505285), Medical Innovation Variety and Platform Cultivation Project (Z251100004625042), the Excellent Youth Training Programme Of Beijing Ditan Hospital, Capital Medical University (ZZ202404).

Author contributions

XC: Writing – review & editing, Writing – original draft. LY: Writing – review & editing, Writing – original draft. XL: Writing – original draft. TJ: Writing – review & editing. JL: Writing – review & editing. ZY: Writing – review & editing, Funding acquisition.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. During the preparation and revision of this manuscript, the authors used a generative AI assistant (ChatGPT) to support language editing, copy-editing, and structural organization of author-written content. All scientific content, data interpretation, and conclusions were conceived, verified, and approved by the authors, who take full responsibility for the integrity and accuracy of the work. No generative AI tool was used to generate original data, figures, or references.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.