PD-1 is a member of the immunoglobulin superfamily induced on activated immune cells and functions as a negative regulator that prevents excessive stimulation and autoimmunity (8). The protein comprises an extracellular IgV-type domain, a single transmembrane helix and a cytoplasmic tail that contains the immunoreceptor tyrosine-based inhibitory motif (ITIM) and the immunoreceptor tyrosine-based switch motif (ITSM) required for inhibitory signaling (18) (Figure 1A). PD-1 engages two physiological ligands, PD-L1 and PD-L2, which are inhibitory B7 family members expressed on diverse cell types (2, 3). On antigen-presenting cells, PD-L1 can be induced by lipopolysaccharide and GM-CSF, whereas on T and B lymphocytes it increases following T-cell receptor activation (2). In tumors, PD-L1 is upregulated by interferon-γ and other inflammatory cues from T cells, NK cells and myeloid cells, contributing to immune evasion (5, 19). PD-L2 expression is detected on subsets of dendritic cells and specific tumor cell lines (3, 20, 21). Soluble PD-L1 (sPD-L1) has been associated with poor responsiveness to PD-1/PD-L1 blockade, although its biology remains incompletely defined (22–24); see Supplementary Table S2 for induction and cellular sources and sPD-L1 correlates.

The extracellular domain of PD−1 is formed by nine β−strands connected by eight loops, which adopt a β−sandwich Ig-like topology along with a flexible loop located at the N−terminus (25). It also features several N−glycosylation sites, which can influence the binding of specific antibodies (26).

PD-L1 (CD274) is a type I transmembrane B7 family ligand comprising an extracellular N-terminal IgV-type domain that forms the PD-1 binding interface, a C-terminal IgC2-type domain, a single-pass transmembrane helix, and a short cytoplasmic tail with no known signaling motifs (Figure 1B). Its expression is inducible by interferon-γ and inflammatory cues across antigen-presenting cells and tumor cells, and it can also be upregulated on activated lymphocytes. At the structural level, PD-1 engages PD-L1 in a 1:1 complex dominated by contacts on the PD-1 GFCC′ sheet and CC′ loop and the PD-L1 IgV face (27); biophysical studies indicate a moderate-affinity, entropy-favored interaction, whereas PD-1 binds PD-L2 with higher affinity and a larger enthalpic contribution (28). These ligand-specific differences, together with heterogeneous and inducible PD-L1 expression in tumors, shape the pharmacology of PD-1/PD-L1 blockade and contribute to context-dependent clinical effects. For induction patterns and sPD-L1 correlates see Supplementary Table S2. Representative PD-1:PD-L1 complex structures are listed in Supplementary Table S4, and the N-terminal IgV domain of PD-L1 follows the canonical nine-stranded β-sandwich fold with connecting loops (29).

Structural studies of PD-1 ectodomains and their complexes with ligands or antibodies, including apo PD-1 (3RRQ) and PD-1:PD-L1 complexes (6UMT, 4ZQK), show that the ectodomain adopts a two-sheet β-sandwich stabilized by a Cys54–Cys123 disulfide, and that ligand recognition centers on the GFCC′ sheet and CC′ loop (27, 30, 31). In these complexes, PD-1 and PD-L1 adopt a nearly orthogonal orientation, with an extended, predominantly hydrophobic interface supported by intermolecular hydrogen bonds and π–π stacking interactions that stabilize the PD-1–PD-L1 complex (27, 30). By contrast, PD-1 binds PD-L2 with higher affinity and a larger enthalpic component, reflecting additional contacts within the GFCC′ region (28) (see Supplementary Table S3 for residue-level and conformational notes). From a therapeutic perspective, these structural data localize most ligand contacts and high-affinity antibody epitopes to the N-terminal segment, CC′ loop, and GFCC′ sheet, which together form a relatively compact hot spot for PD-1:ligand engagement and drug binding (Figures 1C, D).

Ligand engagement triggers inhibitory signaling by enabling TCR-proximal phosphorylation of ITIM and ITSM, recruitment and activation of Src-homology region 2-containing protein tyrosine phosphatase-2 (SHP-2) via its tandem Src homology 2 (SH2) domains and subsequent dephosphorylation of CD28/TCR pathway nodes, which attenuates PI3K–AKT and RAS–ERK signaling and suppresses T-cell effector function (32, 33). SHP-2 is generally considered the principal mediator of PD-1 inhibition, although context-dependent compensation by SHP-1 has been reported (34, 35); Supplementary Table S4 lists the Protein Data Bank (PDB) accessions and primary article references. Because ITIM and ITSM are the primary conduits of PD-1-mediated inhibition, effective extracellular blockade needs to sufficiently disrupt ligand binding to prevent phosphorylation of these motifs and subsequent SHP-2 recruitment.

Therapeutic antibodies block PD-1 by binding the IgV-type domain and occluding the ligand interface that is centered on the N-terminal segment, CC′ loop, and GFCC′ sheet (Figure 1C). Nivolumab engages the N-terminal segment and CC′ loop, while pembrolizumab recognizes an extended epitope across the GFCC′ sheet, yielding near-complete coverage of the PD-L1-binding surface (26, 36, 37). By targeting these ligand-contacting regions, antibody binding stabilizes PD-1 in a state incompatible with ligand engagement, thereby preventing ITIM/ITSM-dependent signaling and releasing T-cell inhibition (30, 36).

Relative to small molecules, therapeutic antibodies and non-antibody formats differ in epitope coverage, binding kinetics, distribution and pharmacology; conclusions about superiority require context-specific data rather than general class assumptions (38–40).

Immune-related adverse events (irAEs) are a major constraint of immune checkpoint inhibitor (ICI) therapy; PD-1/PD-L1 inhibitors are associated with a spectrum of cardiovascular irAEs, notably autoimmune myocarditis, although infrequent, carry disproportionate morbidity and mortality (42). Disinhibition of PD-1/PD-L1 and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) pathways breaks peripheral tolerance and can precipitate T-cell cross-reactivity against cardiac autoantigens (molecular mimicry) (43). In fatal cases with combined PD-1 and CTLA-4 blockade, TCR sequencing has demonstrated shared clonotypes in the tumor and myocardium, consistent with immune-mediated myocarditis (43). Preclinical evidence shows that loss of PD-1 or CTLA-4 signaling produces lethal myocarditis in mouse models (44, 45).

At the antigenic level, cardiac and vascular toxicity has been linked to T-cell responses against cardiomyocyte and endothelial antigens (46–51). Dysregulated cytokine signaling, particularly TNF-α, IFN-γ, IL-1β, IL-6 and IL-17, contributes to myocardial inflammation, endothelial activation/dysfunction and atherothrombotic risk (52–54). Emerging clinical and experimental data also support targeting the interleukin-6 (IL-6)-Janus kinase 2 (JAK2)-signal transducer and activator of transcription 3 (STAT3) axis (e.g., tocilizumab) in steroid-refractory cases (55–57). Multiple cohorts and imaging studies indicate an increased incidence of atherosclerotic cardiovascular events and measurable progression of vascular inflammation/plaques during ICI exposure, consistent with plaque destabilization and thrombotic complications in susceptible patients (58–61).

There are no established prophylactic strategies for immune-related adverse events (irAEs) induced by immune checkpoint inhibitors (ICIs). Interleukin-10 (IL-10), an immunoregulatory cytokine, has been proposed as a theoretical adjunct to PD-1/PD-L1 blockade to temper off-target inflammation while sustaining antitumor immunity; mechanistic and early clinical work with pegylated IL-10 (pegilodecakin) shows CD8+ T-cell activation and feasibility with anti-PD-1 therapy, yet randomized phase 2 trials in non-small-cell lung cancer did not improve efficacy and reported higher toxicity versus ICIs alone, so there is no guideline-endorsed prophylactic role (62, 63). Current guidelines emphasize reactive rather than preventive management of irAEs, with systemic glucocorticoids as first-line therapy once toxicities occur (64–66), and discourage routine glucocorticoid premedication except for patients with a documented history of infusion reactions (67).

Cardiotoxicity, most notably myocarditis, remains a safety bottleneck. Recent studies support early diagnostic pathways incorporating biomarkers and imaging, and suggest benefit from high-dose pulse glucocorticoids for confirmed ICI myocarditis (68, 69), alongside surveillance strategies to detect subclinical myocardial injury in high-risk groups (70–72).

Monoclonal Abs exhibit limited penetration into solid tumors, which impedes achieving therapeutically effective concentrations in the tumor microenvironment (73–75). This limitation reflects a combination of the binding-site barrier, elevated interstitial fluid pressure and a dense extracellular matrix that together restrict macromolecular diffusion and retention (76, 77). Heterogeneous intratumoral distribution may contribute to treatment failure and the emergence of acquired resistance mechanisms (78).

Solid tumors commonly display poor vascularization and hypoxic niches that further compromise intratumoral distribution of PD-1/PD-L1 antibodies (79, 80). The binding-site barrier is particularly relevant, as antibodies rapidly engage PD-L1 at the tumor-immune interface, limiting penetration into deeper tumor regions (81, 82). In addition, PD-L1 expression within solid tumors is often heterogeneous and can be induced by interferon-γ (IFN-γ), creating immune-inflamed “hot spots” that sequester antibody binding and leave other tumor compartments underexposed (83–85).

By 2025, additional pharmacokinetic aspects of this limitation have been clarified. Anti-drug antibodies (ADA) against anti-PD-1 mAbs are increasingly recognized as contributors to accelerated clearance, reduced systemic exposure, and diminished response probability, supporting selective therapeutic monitoring of exposure and immunogenicity in patients with suspected early loss of drug activity (86, 87). Organ-specific barriers are particularly pronounced in central nervous system disease, where the blood-brain barrier limits mAb penetration and the need for steroids can further blunt benefit; early data on intrathecal anti-PD-1 administration for leptomeningeal metastases suggest feasibility and acceptable safety but rest on a limited evidentiary base that warrants prospective evaluation (88). From a delivery perspective, 10-minute nivolumab/pembrolizumab infusions and subcutaneous pembrolizumab (with hyaluronidase), which has shown non-inferior PK to intravenous dosing, may ease resource burden and expand access but do not overcom the fundamental diffusion and binding-site barriers that constrain antibody penetration into solid tumors (89–91).

Malignant cells employ multiple mechanisms to evade immune surveillance, of which PD-1/PD-L1 signaling constitutes only one axis of suppression (92). Notably, even tumors with a high tumor mutational burden (TMB) or microsatellite instability-high/deficient mismatch repair (MSI-H/dMMR) phenotype – features generally associated with heightened immunogenicity – can manifest primary or acquired resistance to PD-1 (92). Contributing mechanisms include upregulation of compensatory inhibitory checkpoints such as T-cell immunoglobulin and mucin-domain containing-3 (TIM-3), lymphocyte-activation gene-3 (LAG-3) and T-cell immunoreceptor with Ig and ITIM domains (TIGIT) (93–95), disruption of antigen-presentation machinery through loss of MHC-I or β2-microglobulin (96, 97), and expansion of immunosuppressive cellular subsets, including tumor-associated macrophages (TAMs), myeloid-derived suppressor cells (MDSCs) and regulatory T cells (Tregs) (98).

Immunosuppressive myeloid cells in the tumor microenvironment are particularly prominent mediators of resistance. Cells such as tumor-associated macrophages (TAMs) and myeloid-derived suppressor cells (MDSCs) markedly constrain the efficacy of immunotherapy (99, 100), particularly in metastatic castration-resistant prostate cancer (mCRPC) (101, 102). TAMs, often skewed toward an M2-like phenotype, promote tumor progression by enhancing angiogenesis, tissue remodeling and suppression of cytotoxic T-cell responses, whereas MDSCs comprise heterogeneous immature myeloid populations – monocytic and granulocytic – that inhibit antitumor immunity through arginase-1, iNOS, reactive oxygen species and immunosuppressive cytokines (103). Across malignancies, including mCRPC, higher TAM/MDSC abundance correlates with resistance to PD-1/PD-L1 therapy (102). Phenotypic and functional heterogeneity complicates therapeutic targeting of these myeloid compartments (104). Recent syntheses highlight TAM/MDSC-directed and myeloid checkpoint/trafficking axes, underscoring the complexity imposed by intra-/inter-subset heterogeneity (105, 106).

Tumor-associated platelets. An additional and mechanistically distinct pathway of immune escape involves tumor-educated platelets, also referred to as tumor-associated platelets (TAP). Sustained antigenic stimulation reprograms platelets toward a pro-tumorigenic phenotype, characterized by heightened expression of adhesion molecules and secretion of bioactive mediators that facilitate tumor progression (107, 108). Upon tumor cell intravasation, TAP rapidly form heteroaggregates with circulating cancer cells, shielding them from cytotoxic lymphocytes while simultaneously providing a reservoir of adhesive proteins that promote vascular arrest and metastatic colonization (109–111). This platelet-mediated cloaking represents a formidable barrier to durable immunotherapeutic efficacy (112, 113). Collectively, these mechanisms generate a profoundly inhibitory tumor microenvironment that undermines T-cell effector function despite checkpoint inhibition (92).

Within conventional PD-1/PD-L1 monoclonal antibodies, one intuitive strategy to overcome the limitations outlined above has been to embed them in combination regimens rather than to change the antibody format itself. n this section, we focus on clinical combinations of standard PD-1/PD-L1 IgG4 antibodies with other agents (for example, CTLA-4 blockade, chemotherapy, targeted therapies and myeloid modulators) and summarize where they partially mitigate pharmacokinetic, toxicity or resistance constraints and where important gaps remain. Some limitations can be addressed within antibody-based combination regimens (see Supplementary Table S5 for an at-a-glance summary of late-phase combinations, key endpoints, and safety notes). Rational combinations address non-overlapping checkpoint biology and can counter selected intrinsic or adaptive resistance, albeit with toxicity trade-offs. In melanoma, anti-CTLA-4 plus anti-PD-1 delivers greater and more durable benefit than anti-PD-1 alone at 10 years, aligning with the division of labor between priming and effector phases (10).

In routine practice, intensifying immune activation with dual checkpoint blockade (e.g., CTLA-4 plus PD-1) increases the rates and grades of immune-related adverse events and does not consistently outperform PD-1 monotherapy for overall survival across NSCLC settings; benefits remain program- and subgroup-specific (131). In gastrointestinal oncology, a 2025 systematic review reported that pembrolizumab monotherapy reduces adverse events yet does not improve overall survival, whereas pembrolizumab plus chemotherapy improves overall survival without a significant increase in overall toxicity versus chemotherapy alone (132). Taken together, these data support a context-dependent risk-benefit calculus: dual-ICI regimens often elevate immunotoxicity and warrant careful patient selection, while ICI–chemotherapy combinations can deliver survival gains with acceptable incremental toxicity when dosing, timing, and comorbidity are considered (see Supplementary Table S5 for trial-level outcomes and irAE notes).

Pharmacology imposes additional design limits. Long antibody half-lives and sustained PD-1 receptor occupancy mean that discontinuation cannot rapidly abort immune-mediated toxicities; contemporary PK/PD analyses document prolonged PD-1 occupancy even at reduced or infrequent dosing, strengthening the rationale for formats with tighter exposure control (133). These pharmacokinetic constraints highlight intrinsic limits of IgG-based PD-1/PD-L1 antibodies and motivate exploration of compact scaffolds and rationally designed small-molecule PD-1/PD-L1 inhibitors to mitigate size-limited diffusion (134), alongside nanomedicine-based localization to confine checkpoint effects to the tumor microenvironment (135).

Selective hybrid or adjacent strategies illustrate how pharmacology and biology can be aligned within and at the margins of the antibody class. Early clinical proof-of-concept indicates that adding the JAK1/2 inhibitor ruxolitinib can re-sensitize classical Hodgkin lymphoma to PD-1 blockade after prior CPI failure, reinforcing the role of JAK-STAT tuning in overcoming resistance (136). The adenosine pathway provides another lever: PD-L1 blockade combined with A2A antagonism (ciforadenant) has shown feasibility and preliminary activity with biomarker ties (137). Preclinical microenvironment-aware delivery platforms, including platelet-targeted nanoparticles (112) that enhance anti-PD-1 activity, are summarized in Supplementary Table S6.

Overall, these combination strategies show that standard PD-1/PD-L1 antibodies can be leveraged more effectively in specific contexts, but they do not fully resolve the three core limitations of pharmacokinetics and penetration, immune toxicity or adaptive and microenvironmental resistance. This motivates the subsequent section, which systematically examines within-class engineering of PD-1/PD-L1 antibodies (Fc modification, bispecifics and conjugate-like designs) and contrasts their mechanistic advantages and trade-offs.

Fc-engineered monospecific PD-1 antibodies, such as FcγR-attenuated IgG4 formats exemplified by tislelizumab, aim to reduce FcγR-mediated clearance of PD-1⁺ effector T cells and mitigate Fc-driven toxicities while preserving checkpoint blockade. Their main advantages and remaining gaps relative to conventional IgG4 PD-1/PD-L1 mAbs together with those of checkpoint-targeting bispecific and conjugate or tumor-restricted PD-1/PD-L1 formats discussed below, are summarized in Table 1. Studies of FcγR interactions along the PD-1/PD-L1 axis have shown that Fc-engaging anti-PD-1 antibodies can paradoxically limit antitumor activity, in contrast to Fc-competent anti-PD-L1 or anti-CTLA-4 mAbs where depletion of suppressive myeloid or regulatory populations may be beneficial (138).

Bispecific antibodies (bsAbs) that bind PD-1/PD-L1 together with a second checkpoint or a costimulatory receptor are designed to co-localize checkpoint blockade and conditional engagement of immune cells within the tumor microenvironment, potentially increasing therapeutic efficiency while limiting off-tumor activity. Similarly to anti-PD-1 mAbs (138, 141), bsAbs frequently employ Fc effector-null (FcγR/C1q-silent) formats to minimize Fc-mediated toxicities without compromising target engagement (144).

One of the first PD-1/PD-L1 inhibitors was the (2-methyl-3-biphenylyl)methanol scaffold (Figure 3), developed by Bristol-Myers Squibb (BMS). The most extensively validated small-molecule mechanism is to occupy a hydrophobic pocket on PD-L1 and induce PD-L1 dimerization, sterically shielding the PD-1 interaction surface and thereby disrupting PD-1/PD-L1 signaling; crystallography and NMR establish this mode for BMS exemplars, including BMS-202 (1) and BMS-8 (2), and related series (14, 195, 196). At the residue level, PD-L1 dimerizers occupy a hydrophobic tunnel formed by Tyr56, Met115, Ala121 and Tyr123, with polar linkers engaging Asp122 and Arg125, a ‘dimer-lock’ that sterically occludes PD-1 (Figure 4).

The “BMS series” (Figure 5) are archetypal biphenyl dimerizers of PD-L1, and recent asymmetric biphenyl designs preserve the dimerization paradigm while opening new vectors for ADME and safety optimization (14, 197). Representative PD-L1–ligand crystal complexes and associated activity metrics are summarized in Table 3 (e.g., PDB 5NIU, 6R3K). Binding-affinity profiling highlights several lead PD-L1–targeting compounds clustering in the low-nanomolar range (Figure 6).

Beyond biphenyls, multiple chemotypes have been reported, including pyrazolones with biophysical evidence of PD-1/PD-L1 disruption measured by microscale thermophoresis (MST) or Förster resonance energy transfer (FRET) (198), indanes as rigidified scaffolds with PD-1/PD-L1 activity (199), and symmetric phenyl-linked ligands with high PD-L1 affinity (200); contemporary reviews map the binding interactions and chemical space across these families (134, 201). Structure-guided rigidification, for example by moving from biphenyl to terphenyl scaffolds or by ring fusion, can strengthen pi-pi contacts with PD-L1 Tyr56 and open vectors for potency and pharmacokinetic optimization (134, 197). Two non-peptidic mechanistic classes dominate current efforts, namely PD-L1 dimerizers and steric blockers that occlude the PD-1 contact surface (198, 199).

Alternative small-molecule routes are also emerging. Certain ligands trigger PD-L1 internalization and degradation, potentially extending pharmacodynamic effects beyond plasma exposure, and cell-based studies show that ligand-induced dimerization can accelerate loss of PD-L1 from the membrane; structural analyses further rationalize how small molecules remodel PD-L1 surfaces to preclude PD-1 binding (205–208).

Relative to monoclonal antibodies, small molecules promise oral dosing, deeper penetration through fibrotic stroma and poorly perfused tumor regions, a short and controllable half-life for safety management (192), and scalable manufacturing with fine-grained medicinal-chemistry tuning (134). Disease-context data also suggest tumor-type-specific benefits, for example supportive biology for BMS-202 in glioblastoma models (20).

Clinically, no approved small-molecule PD-1/PD-L1inhibitors exist as of 2025, but several programs offer on-target pharmacodynamic evidence in humans. The PD-L1 ligand INCB086550 demonstrated target-engagement signatures and antitumor activity in preclinical and translational studies with early clinical readouts (209), whereas the oral PD-L1 inhibitor evixapodlin (GS-4224) reported PD-L1 engagement and expected immune pharmacodynamics in first-in-human testing; other oral agents such as MAX-10181 (see Figure 7) are progressing through phase-1 development (210, 211).

For combination strategies with antibody formats, small molecules can complement prolonged systemic control by bridging pharmacokinetic windows between infusions and diffusing into ‘cold’ regions that are less accessible to IgG, a rationale supported by preclinical synergy, for example an oral PD-L1 inhibitor MAX-10181 enhancing PD-1×CTLA-4 bispecific therapy in animal models, and by mechanistic data demonstrating distinct effects of dimerizers on PD-L1 membrane dynamics (159, 207). Notwithstanding these advances, the shallow and extended nature of the PD-1/PD-L1 interface and the need to stabilize PD-L1 dimers help explain why discovering potent, drug-like disruptors remains challenging. Disrupting this large protein-protein interface with drug-like molecules carries risks of insufficient potency, off-target cytotoxicity and discordance between biophysical readouts and T-cell functional rescue (196). Critiques of CA-170 (Figure 8) highlight a lack of direct PD-1/PD-L1 inhibition in independent systems (212) despite early clinical exploration, for example NCT02812875 (213), underscoring the need for rigorous on-target validation of dual-axis small molecules.

In summary, small molecules directly address delivery constraints intrinsic to antibody modalities while introducing tunable pharmacology and oral feasibility; success will depend on consolidating structural on-target validation, biomarker-driven pharmacodynamics and robust clinical efficacy that clearly differentiates these agents from established antibody checkpoints (134, 192). Near-term readouts from the programs listed in Table 4 will determine whether oral PD-1/PD-L1 agents merit broader integration as complementary modalities.

Peptides, and especially macrocyclic peptides (see Figure 9), offer geometrically adaptable contact surfaces capable of engaging the relatively flat PD-1/PD-L1 interface, where classical small-molecule drugs often struggle to achieve sufficient affinity and selectivity (214). Their relatively large, conformational pre-organized binding surfaces can mimic protein epitopes and more fully cover the shallow, hydrophobic PD-L1 patch than typical small molecules, which facilitates high-affinity engagement of this difficult protein–protein interface (215–217). Macrocyclization increases conformational pre-organization, proteolytic stability and cellular activity versus linear congeners, as shown for cyclic PD-L1 inhibitors that deliver multi-fold gains in PD-1/PD-L1 blockade and in vivo antitumor efficacy. Mechanistically, most macrocyclic peptides bind PD-L1 within hydrophobic pockets, occluding the PD-1-binding surface and preventing complex formation (218). Certain molecules can simultaneously disrupt PD-L1 interactions with PD-1 and with CD80 in cis, which may amplify immune activation (219). Alongside direct blockers, “degrading” strategies have emerged: stapled peptide PROTAC constructs drive PD-L1 depletion and reduce PD-L1 levels more effectively than the reference small-molecule inhibitor BMS-8 in cell systems (220). Consistent data show that lowering membrane PD-L1 is accompanied by enhanced T-cell activity, aligning with the pharmacodynamics rationale for targeted down-regulation (221). A parallel route employs PD-1-directed peptide ligands, rationally derived from PD-L1 contact regions and validated biophysically (SPR/ELISA) and in functional cellular assays, demonstrating targeted interception of the interface from the opposite side (214). Representative peptide and macrocycle programs are summarized in Table 5, and a structural comparison with a rigid small-molecule PD-L1 inhibitor is shown in Figure 10.

From a developability standpoint, PD-L1-focused macrocycles have progressed from concept to manufacturing-scale synthesis: process development has been reported for the macrocyclic inhibitor BMS-986189 at the hundreds-of-grams active-pharmaceutical-ingredient scale with solutions for stereocontrol and purity (222). The translational trajectory is further supported by macrocyclic PD-L1 radioligand tracers such as [18F]BMS-986229 (Figure 9): competitive target-engagement readouts and whole-body dosimetry have been demonstrated in non-human primates, and clinical positron emission tomography (PET) studies with this ligand have been initiated (223–225). Clinically, the field remains early: no peptide PD-1/PD-L1 inhibitors are approved, yet readiness for clinical formats is evidenced by PD-L1 radioligand imaging and the ability to scale macrocycles to pharmaceutical quality, whereas potent exemplars (for example, pAC65) still face preclinical species-specificity barriers (219, 222, 223).

In contrast to peptide macrocycles, small-molecule PD-L1 inhibitors are typically more rigid: widely used biphenyl and terphenyl series possess a core with restricted degrees of freedom, and ortho-substitution on one rings lock its orientation, favors ligand-induced PD-L1 dimerization snd enables very high affinity binding (204). Across published PD-L1 ligands, macrocyclic peptides usually span micromolar to high-nanomolar half-maximal inhibitory concentrations, whereas optimized small molecules can reach low-nanomolar or even picomolar potency; for example, macrocycles reported by Sun and colleagues are approximately 10³ to 10⁴ times less potent than terphenyl scaffolds described by Muszak and colleagues (204, 226). A representative structural comparison of a macrocyclic peptide PD-L1 inhibitor and a rigid terphenyl-based small-molecule PD-L1 inhibitor is shown in Figure 10.

Against this backdrop, peptide and macrocyclic ligands offer several physicochemical and pharmacologic advantages relative to small molecules. They are well suited to large and relatively flat protein-protein interfaces where key interaction hotspots are spatially dispersed, they can be engineered for modular conjugation or local delivery formats, and they often follow more predictable metabolic routes than classical xenobiotic small molecules (227). Macrocyclization further improves membrane permeability and resistance to proteolysis compared with linear peptides, which supports continued interest in macrocycles as an intermediate between large biologics and conventional small molecules (228). However, these advantages come at the cost of generally poor oral bioavailability, limited metabolic stability and lower intrinsic potency, so that chemical measures such as cyclization or stapling and parenteral administration are usually required (218, 229, 230).

Highly potent small molecules, in turn, maximize binding affinity but are more prone to off-target interactions with other proteins and to the formation of potentially toxic metabolites; very low, subnanomolar to picomolar inhibitory concentrations can narrow the therapeutic window, because toxic doses may be reached at relatively low systemic concentrations even though short half-lives partly mitigate overdose risk (231). From a chemistry standpoint, complex multistep synthetic routes can also render some small-molecule series more expensive to develop at scale (231).

With respect to immunogenicity, pharmacovigilance analyses suggest that peptides and macrocycles may be somewhat more predisposed to eliciting allergic reactions than small molecules, although patient-specific factors remain dominant determinants (232). These complementary profiles have motivated interest in combination strategies in which monoclonal antibodies are paired with peptide or small-molecule PD-1/PD-L1 modulators, for example by using peptides or macrocycles for imaging or local modulation while antibodies provide sustained systemic blockade (233).

No approved peptide/macrocyclic PD-1/PD-L1 therapeutics as of 2025; clinical activity to date centers on imaging tracers and manufacturing readiness rather than therapeutic efficacy endpoints. Looking ahead, three avenues are prominent: (i) further optimization of macrocycles guided by PD-L1 structural data and interaction maps to lower polarity while maintaining affinity (218); (ii) development of degrading and multimodal peptide constructs (PROTAC or conjugates) to achieve sustained PD-L1 suppression (220); and (iii) integration with local delivery and PET tracers to personalize dosing and patient selection (224, 225).

Over the past decade, a set of less established approaches has emerged alongside reduced-size antibody scaffolds and small molecules, most notably decoy receptors built from soluble PD-1 ectodomains and nucleic acid aptamers against PD-L1 (Table 6). These modalities were conceived to bypass limitations of IgG formats and classical medicinal chemistry by offering alternative geometries and manufacturability routes (234). Development has been steady but limited relative to antibodies, with few human efficacy signals to date (235). Conceptually, PD-1 decoy receptors and PD-L1 aptamers are positioned as non-antibody PD-1/PD-L1 blockers that can support alternative manufacturing, imaging and local-delivery solutions in settings where full-length monoclonal antibodies or small molecules are constrained by size, tissue penetration or immunogenicity (16, 234, 236).

Soluble PD-1 decoys can sequester PD-L1/PD-L2 and thereby interrupt PD-1 signaling without using an antibody paratope (Table 6). Proof-of-concept includes a plant-produced PD-1-Fc that blocks PD-1/PD-L1 in vitro, underscoring manufacturability in non-mammalian systems and the feasibility of the decoy geometry (237). Mechanistic precedent also comes from high-affinity PD-1 variants that exhibited greater antitumor activity than anti-PD-L1 antibodies in specific murine models and dosing regimens, and that served as immuno-PET probes, illustrating high-avidity trapping of PD-L1 in vivo (234). By contrast, a clinical pilot of the PD-1-engaging ligand-based decoy AMP-224 (PD-L2-Fc), in refractory metastatic colorectal cancer was tolerable but produced no objective responses, highlighting the translational gap (235).

Aptamers are short single-stranded DNA or RNA ligands generated by systematic evolution of ligands by exponential enrichment (SELEX) and can be chemically modified and conjugated with ease, providing high affinity with low intrinsic immunogenicity (236). Recent selections yielded new DNA aptamers to PD-L1 with nanomolar affinity and functional blockade in cell systems, confirming tractable discovery and on-target activity (238). Pharmacokinetic limitations, notably rapid renal clearance, can be mitigated by increasing hydrodynamic size, for example via albumin conjugation, which improved in vivo antitumor efficacy of a PD-L1 aptamer (239). Aptamer-guided nanocarriers have also been engineered to respond to radiotherapy, enhancing reactive oxygen species production and macrophage reprogramming in tumor models (240). Beyond discovery tools, antagonistic aptamers such as PL1 restored T-cell function and inhibited growth of PD-L1-positive tumors in mice, reinforcing biological validity (241).