Facts and Hopes: Immunocytokines for Cancer Immunotherapy

Cytokines are potent small molecules involved in cell–cell interactions and in mediating host immune responses. They play important functions in anticancer immunity through modulation of various cellular responses, including the regulation of immune proliferation, differentiation, and effector functions, thus making them longstanding targets for cancer therapy (1, 2). However, deploying natural cytokines clinically has been challenging. Despite being the first form of cancer immunotherapy to receive regulatory approval - IFNα in 1986 and IL2 in 1992 - there has been little subsequent progress in clinical development of cytokine-based therapeutics (3, 4). This may be attributable to several factors. The pleiotropic activity of cytokines affects multiple immune compartments, some without beneficial antitumor response. Furthermore, the precise mechanisms underlying tumor regressions with cytokine administration are generally unknown. Thus, biomarkers are not available to guide dose or scheduling. Finally, cytokines have traditionally been administered at extremely high levels in early studies. Such dosing may bypass or obscure normal physiologic antitumor mechanisms and lead to unwanted toxicity. To address these limitations, various modifications have been made to: (i) increase tumor targeting, (ii) enhance binding selectivity, and (iii) improve the pharmacokinetic profile of these agents (5).

In recent years, improved understanding of cytokine immunobiology and enhancements in protein engineering have sparked renewed interest in this arena (2, 6). “Next generation cytokines” have since been developed, including immunocytokines. Immunocytokines are antibody-based immunoconjugates, consisting of monoclonal antibodies or antibody fragments targeting a tumor antigen or component of the tumor microenvironment (TME) fused with cytokine payloads (Fig. 1; ref. 7). These agents seek to improve the therapeutic index of cytokines by using antibodies as vehicles for the targeted delivery of immunomodulatory agents within the local TME Fig. 2; ref. 8).

In this review, we provide an overview of the development of immunocytokines as novel cancer immunotherapies. We outline key factors in immunocytokine design, from tumor targeting components to molecular formats and cytokine payloads, emphasizing the clinical translation of such developments. Finally, we review challenges in the clinical development of these agents and summarize important future directions in the field.

One key component of an immunocytokine is the target antigen (9). Common targets have been tumor-associated antigens expressed on the cell surface of tumor cells. One early-target was epithelial cell adhesion molecule (EpCAM), a transmembrane glycoprotein overexpressed in various epithelial cancers (e.g., colon or prostate) and linked to tumor growth and metastases (10). Another target, disialoganglioside 2 (GD2), is a glycosphingolipid highly expressed on neuroectoderm-derived tumors (e.g., melanoma and neuroblastoma), with limited expression in normal tissues (11). Immunocytokines have also been developed to target HER2, CD20, and carcinoembryonic antigen (CEA), among others (Table 1). One limitation in targeting tumor-associated antigens (TAA) is that target expression can be transient and/or heterogeneous. Furthermore, only select patients may express the target of interest, potentially limiting populations most likely to benefit.

Various immunocytokines have also been directed against components of the extracellular matrix (ECM) of tumors (12). These include vascular antigens that are upregulated in cancer angiogenesis, including tenascin-C and extra domain A (ED-A) and B (ED-B) of fibronectin (13). Beyond targeting angiogenesis, immunocytokines have also been directed against: (i) components of collagen within tumor stroma, (ii) antigens exposed within necrotic tissues (e.g., DNA/histone complexes), and (iii) antigens expressed on T cells within the TME (e.g., programmed cell death 1; PD-1; refs. 14–17).

The molecular composition of immunocytokines may also have important implications for antitumor activity, affecting drug clearance rates and ability to penetrate and be retained within tumors (18).

Broadly, there are two major delivery platforms for immunocytokines: intact IgG or antibody fragment formats. Early generations of immunocytokines commonly featured IgG formats (7). This format has several advantages. Intact IgG has a long serum half-life, mediated by large size, which reduces renal clearance. Intact IgGs can also bind to the neonatal Fc receptor (FcRn), which prevents degradation by the lysosomal compartment, allowing them to be retained in circulation (19, 20). Full size IgG antibodies have two heavy chains and two light chains, both with variable regions. This dual variable region structure leads to higher avidity for a given target, resulting in higher rates of retention at the target (21). Nonetheless, the large size of the IgG format can also pose important challenges. For example, it may impede extravasation from vasculature and therefore penetration into tumor. Furthermore, the long-half life and tissue retention may lead to significant off-target toxicity (22).

To overcome limitations of IgG-based immunocytokines, various antibody fragment formats have been developed, including monovalent, F(ab′), divalent, F(ab′)₂, and single-chain variable fragments (scFv; ref. 23). A principal advantage of these formats is their smaller size, which may promote tumor penetration, while retaining antigen specificity of the full-length antibody (8). Nonetheless, smaller size and lack of an Fc region also serve to decrease serum half-life – although this may decrease off-target toxicity (8). Importantly, depending on the antibody format, the avidity for the target may also be decreased. To overcome this, other antibody fragment formats (e.g., dia- and triabodies) have been developed (24). These contain more antigen-binding moieties and thereby have increased avidity while retaining other advantages of antibody fragment formats (21).

To date, various cytokine payloads have been incorporated into immunocytokines. The most extensively studied payloads include IL2, IL12, and tumor necrosis factor (TNF). Below, we summarize preliminary data behind each cytokine payload, emphasizing the existing clinical data generated thus far.

IL2 is a 15-kDa proinflammatory cytokine produced by activated T cells (25). IL2 mediates its protean immunologic effects through binding to the IL2 receptor (IL-2R), which is comprised of α, β, and γ subunits. IL2Rα (CD25) confers high-affinity binding to IL2 and is essential for the expansion of FOXP3⁺ T regulatory cells (Tregs), while IL2Rβ (CD122), and IL2Rγ (CD132) subunits, expressed on natural killer (NK) cells and resting CD4⁺ and CD8⁺ T cells, mediate signal transduction (13, 26). At low-doses, IL2 preferentially binds to its high-affinity CD25 receptors on Treg cells leading to immune evasion (27, 28). At high-doses, IL2 therapy induced durable responses in a small subset of patients with renal cell carcinoma (RCC) and melanoma (4, 29). However, significant safety concerns, including potentially life-threatening organ dysfunction, capillary leak syndrome, and coma, have limited its use (30). Various strategies to improve upon initial IL2 formulations have since been explored, such as PEGylation to increase half-life, introducing targeted mutations to modify function (e.g., CD25-biased, CD25/CD122-biased formulations), masking IL2 against Treg binding, and developing IL2-based immunocytokines to improve antitumor activity and reduce toxicity (31, 32).

IL2 is the most extensively studied immunocytokine payload to date (Table 1), with various agents entering phase I–II clinical trials. Broadly, these agents can be categorized according to IL2 formulation (unbiased vs. biased) and targeting approach (cell surface antigens vs. ECM). Unbiased IL2 formulations are those that are not selective for the intermediate-affinity dimeric or high-affinity trimeric IL2R whereas biased IL2 agents selectively target either CD25, CD25/CD122, or CD122 alone (25).

Many clinical trials have been performed with IL2 immunocytokines targeting tumor antigens. For example, Hu14.18-IL2 is a humanized mAb covalently linked to two molecules of IL2 that targets GD2. In preclinical murine models of neuroblastoma, Hu14.18-IL2 showed antitumor activity, and early-phase clinical studies demonstrated it was relatively well tolerated with evidence of immune activation (33). However, clinical activity has been disappointing. In one phase II study of Hu14.18-IL2 that enrolled 14 patients with melanoma, the objective response rate (ORR) was only 7.1% (34). Likewise, in a phase II study that enrolled 13 neuroblastoma patients with measurable disease, no objective responses were observed (35).

IL2-based immunocytokines have also been explored within hematologic malignancies. For example, DI-Leu16-IL2, a humanized anti-CD20 antibody fused to IL2, has been explored in patients with CD20-positive non–Hodgkin's lymphoma (NHL). A phase 1 trial of DI-Leu16-IL2 following peripheral B-cell depletion with low-dose rituximab in patients with relapsed/refractory disease showed durable tumor regression or stabilization in 14 of 16 (87.5%) evaluable patients (36). Among five patients with objective responses, the median duration of response was 13 months, with one ongoing complete response observed at 11 months. This drug was generally well-tolerated and demonstrated higher permissible doses with s.c. administration than could be achieved with i.v. therapy (37).

Cergutuzumab amunaleukin (CEA–IL2v) and simlukafusp α (fibroblast activation protein; FAP-IL2v) are IL2 variant (IL2v)-based immunocytokines that target carcinoembryonic antigen (CEA) and fibroblast activation protein α (FAPα) tumor antigens, respectively (38, 39). These are CD122-biased IL2 agents fused to the C-terminus of their respective antibodies together with novel Fc mutations that eliminate FcγR and C1q-binding. Preclinically, these variants lack preferential induction of Tregs due to abolished CD25-binding but retain capacity to activate and expand NK and CD8⁺ effector T-cells through IL2Rβγ in the periphery and the TME (13). A phase Ib trial of CEA-IL2v in combination with atezolizumab has completed enrolment (NCT02350673), but results have not been reported to date.

In addition to tumor antigens, IL2 immunocytokines have also been developed against components of the ECM. For example, L19-IL2 is a fully human recombinant monoclonal antibody comprised of IL2 fused to the L19 diabody, where L19 targets an alternatively spliced isoform of fibronectin (ED-B domain) expressed in the setting of angiogenesis (14). In preclinical models, IL19-IL2 selectively accumulates in tumor vasculature and demonstrated antitumor activity with concomitant increases in T lymphocytes, NK cells, and macrophages, along with a corresponding increase in IFNγ (14). L19-IL2 has been studied in multiple different tumor types, both as monotherapy and in combination with other agents, including systemic therapy and radiotherapy (40–42). Among metastatic solid tumors patients enrolled in a phase I/II study of L19-IL2, a best response of stable disease was observed in 17 of 33 patients (51%; ref. 40). Among the metastatic RCC subgroup, 15 of 18 (83%) patients experienced disease stabilization, but no objective responses were observed. Given the modest efficacy of L19-IL2 monotherapy, this agent has also been explored in combination with dacarbazine in metastatic melanoma (43). In an initial single-arm phase II study, L19-IL2 showed an ORR of 28%, with one complete response. However, the subsequent randomized phase II portion of this trial was disappointing, with no significant difference in progression-free survival (PFS) between L19-IL2 and dacarbazine alone (44). Nonetheless, L19-IL2 (darleukin) is still being explored in ongoing clinical trials, outlined below.

Another approach to developing immunocytokines is to target immune cells directly. The IL2v antibody fusion protein RO7284755 targets programmed cell death 1 (PD-1), which is overexpressed on the surface of activated effector T cells. Preclinical data show high-affinity PD-1 binding allows IL2v to trigger a proliferation cascade to generate a distinct population of tumor antigen–specific CD8⁺ T cells (17). By docking in cis to PD-1, it also overcomes the need for CD25-binding (45, 46). PD-1-IL2v (RO7284755) is currently being studied both as monotherapy and in combination with the anti–programmed death ligand 1 (PD-L1) antibody atezolizumab for the treatment of immune checkpoint inhibitor (ICI)-responsive advanced cancers (NCT04303858).

IL12 is a pleiotropic proinflammatory cytokine primarily produced by antigen presenting cells (APC). Comprised of two subunits, p40 and p35, IL12 stimulates the effector function of activated T cells and NK cells and production of IFNγ (47). In various murine models, treatment with recombinant IL12 leads to enhanced innate and adaptive antitumor immunity and dramatic antitumor activity (48). However, in clinical trials, systemic administration of recombinant IL12 has produced disappointing results, marked by low therapeutic efficacy and significant systemic toxicity, including fatal inflammatory syndrome (8, 49). IL12-based immunocytokines have been developed as one approach to overcome this. NHS IL12 consists of two IL12 heterodimers fused to the human IgG1 NHS76 antibody that targets areas of tumor necrosis by binding exposed histone complexes in DNA (50). A phase I study of s.c. administration of NHS IL-12 in patients with metastatic or locally advanced epithelial or mesenchymal tumors demonstrated that this agent was well tolerated and increased T- cell receptor (TCR) diversity and tumor-infiltrating lymphocyte density, indicative of an antitumor immune response (51). Interestingly, reductions in prostate-specific antigen (PSA) levels were seen in five of eight evaluable patients with prostate cancer, but no objective responses were observed. Nonetheless, given evidence of immune activation with monotherapy, NHS IL12 is being explored in combination with other agents.

Preclinical data suggests that IL12 may be more effective as part of combination therapy (52, 53). Currently, NHS IL2 is being studied in combination with docetaxel in patients with metastatic prostate cancer (NCT04633252) and in combination with the anti–PD-L1 inhibitor, avelumab (NCT04756505). Other strategies in the development of IL12-based immunocytokines have focused on modifications in target antibodies (e.g., IL12–7NP2, which targets FAP), delivery systems (e.g., intratumor delivery), or novel combination strategies (e.g., histone deacetylase inhibitors; refs. 54, 55).

TNFα is a multifunctional cytokine that plays key roles in vascular permeability, inflammation, and apoptosis (2). TNFα acts on endothelial cells to increase vascular permeability and mediates hemorrhagic tumor necrosis (56). Systemic administration of TNFα has been limited due to significant adverse events including severe hypotension and organ dysfunction (57). Furthermore, studies of TNFα have shown minimal to no tumor response (57–59). TNFα has since been studied as a payload in immunocytokine development. L19-TNF is an immunocytokine comprised of the single-chain human monoclonal L19 antibody fused to TNFα (50). In a phase 1 trial of L19-TNF plus doxorubicin in patients with advanced cancers, primarily soft tissue sarcomas (STS), the combination was overall well tolerated and demonstrated preliminary evidence of antitumor activity (60). In an expansion cohort of 15 patients with STS, there was 1 complete response, 1 partial response and 7 patients with disease stabilization (60). A randomized phase II trial of L19-TNF plus dacarbazine versus dacarbazine alone in advanced STS is ongoing (NCT04733183). L19-TNF is also being studied in other tumor types including glioblastoma and melanoma (NCT03779230; refs. 61, 62).

Despite a decade of progress in immunocytokine development, no agent has yet gained regulatory approval, leaving important challenges for the field. Here, we outline potential strategies to improve upon existing immunocytokines, highlighting the need for: (i) improved antibody targets, (ii) enhanced antibody designs, (iii) improved drug delivery techniques, (iv) novel combination strategies, and (v) modifications in payloads (Fig. 3).

Lack of highly specific targets has been a fundamental limitation of immunocytokines. With recent progress in transcriptomic and proteomic profiling technologies, efforts are ongoing to elucidate new tumor targets (18). Efforts in antibody target discovery may also benefit from the recent explosion in antibody–drug conjugate (ADC) development, to facilitate identification of new cell surface, tumor-specific targets (63).

Selectively engineering modifications in the antibody component of immunocytokines may be another strategy to improve efficacy. For example, point mutations in the Fc tail have been suggested to enhance effector function (21) by increasing antibody-dependent cellular phagocytosis and cytotoxicity (64). Other antibody enhancement strategies have focused on modifying pharmacokinetics by manipulating posttranslational glycosylation (65, 66). Lastly, cross-linking cytokines to bispecific (e.g., BITEs or BiKEs) and tri-specific (TriKEs) immune engagers have also been developed, aiming to increase immune activation, recruit effector cell populations, and increase specificity for the tumor by targeting multiple antigens (2, 21, 67).

Novel delivery techniques have also been explored to improve potency and reduce off-target toxicity as immunocytokines with unblocked and nonattenuated cytokines are likely to be associated with cytokine-related toxicities. One innovative strategy is the use of protease-cleavable epitope masking. In this approach, an epitope-masking molecule is linked to the antibody, which is cleavable by matrix metalloproteases specifically expressed at tumor sites. This essentially renders immune-active antibody moieties inactive until the tumor site is reached, thereby reducing systemic toxicity (68). One recent such example is WTX-124, which links IL2 to an inactivation domain that is cleaved by dysregulated proteases in the TME (69). In preclinical murine models, WTX-124 retained antitumor activity while minimizing IL2 effects in nontumor tissues.

“Activity-on-target cytokines” (AcTakines) are yet another emerging immunocytokine class with a novel delivery method. AcTakines replace the wild-type (WT) cytokine payload with a mutant form that has reduced affinity for its cognate receptor, thereby rendering it inactive in circulation. Upon cell-specific–binding through a targeting moiety, the cytokine regains full activity (70). This strategy was first studied to improve the therapeutic index of IFNα but has since been applied to many other cytokines including phase I trials with IL15 and IL21 (17).

Immunocytokines have been developed that carry multiple payloads, termed dual-cytokine immunocytokines. These are still in the preclinical stage but combinations such as IL2/TNF have shown promise (71, 72). In addition, coformulations of multiple immunocytokines have been explored. For example, Daromun is an intralesional therapy consisting of two immunocytokines (L19IL2 and L19TNF). In a phase II trial of Daromun that enrolled 20 efficacy evaluable patients with unresectable stage IIIC and IVM1a melanoma, the ORR of injected lesions was 55%. Interestingly, treatment responses were observed in 7 of 13 lesions that were uninjected, suggestive of a bystander effect (73). A phase III trial of intralesional Daromun as neoadjuvant therapy in patients with high-risk, resectable IIIB/C melanoma is underway (NCT02938299 and NCT03567889; ref. 74).

To augment the activity of immunocytokines, there is growing interest in developing combination strategies with chemotherapy, radiation, and immunotherapy partners. Given the success of ICIs across multiple tumor types, these agents represent appealing combination partners. Combinations of immunocytokines plus ICIs have shown encouraging results in preclinical models (52, 75). For example, L19-TNF in combination with the PD-1 inhibitor nivolumab eradicated tumors in sarcoma models (76). Recently, there have also been efforts to combine a PD-L1 inhibitor with the cytokine IL15 in a single novel immunocytokine, LH01. Interestingly, Shi and colleagues reported that LH01 demonstrated superior antitumor activity than the combination of anti–PD-L1 and IL15 (16). Clinically, ICIs are being investigated in combination with multiple other immunocytokines or engineered cytokines (Table 1).

Immunocytokines have the potential to improve the therapeutic window of cytokine-based therapeutics. Over the last decade, immunocytokine development has centered around antibody engineering, identification of antibody targets, and selection of optimal payloads. Although these studies established proof of principle in preclinical studies, clinical development has lagged. Moving forward, further refinements in drug delivery, payloads, and antibody constructs may help facilitate successful translation into the clinic. In parallel, efforts to identify robust combination strategies and ideal clinical opportunities to deploy such approaches will be important to build upon the advances in immunocytokine development.

A. Pabani reports grants and personal fees from Bristol-Myers Squibb; personal fees from AstraZeneca, Pfizer, Merck, Roche, and Canadian Agency for Drugs and Technologies in Health; and grants from Alberta Cancer Foundation outside the submitted work. J.F. Gainor reports personal fees and other support from Bristol-Myers Squibb, Merck, AstraZeneca, Moderna, and Blueprint; personal fees from Loxo/Lilly, Gilead, Pfizer, EMD Serono, iTeos, Amgen, Mirati, Nuvalent, Karyopharm, Beigene, Silverback Therapeutics, GlydeBio, Takeda, Curie Therapeutics, and Merus Pharmaceuticals; grants, personal fees, and other support from Genentech/Roche and Novartis; and other support from Jounce and Alexo outside the submitted work. In addition, J.F. Gainor's spouse is an employee with equity in Ironwood Pharmaceuticals.