• Cytokines
• Clinical Trials as Topic

Interleukin-2 (IL-2) was discovered as a T-cell growth factor in 1976.1 In addition to interferon-alfa, it was among the first cytokines produced by recombinant technology and tested in clinical trials in the early 1980s. In the 1990s, a high-dose (HD) regimen of human recombinant IL-2, aldesleukin (Proleukin), received regulatory approval for the treatment of metastatic renal cell carcinoma (mRCC) and metastatic melanoma. In melanoma, HD IL-2 produced a 16% objective response rate (ORR) (6% complete response (CR) and 10% partial response (PR)).2 In mRCC, the ORR was 14% (5% CR and 9% PR).3 In both studies, patients with CR had durable benefit. However, the HD IL-2 regimen was associated with moderate to severe toxicity including hypotension requiring fluid or pressor support and vascular leak syndrome (VLS), resulting in dose-limiting pulmonary edema, renal failure, transaminitis, diarrhea, and altered mental status. Because of challenges in administration of HD IL-2 and the subsequent development of more effective agents, IL-2 clinical use is limited today.

For more than a decade after the first reports of clinical antitumor activity, many clinical trials of IL-2 were initiated to assess activity and toxicity of different doses and schedules and combinations with other anticancer agents, across a broad range of malignancies. In addition, various agents were combined with IL-2 to mitigate mechanisms of toxicity while attempting to preserve antitumor activity. To our knowledge, none of the many efforts to improve and expand activity or reduce toxicity led to a regulatory approval for standard of care clinical use.

The immunobiology of IL-2 and its receptor is complex and has been covered extensively in multiple prior reviews.4 From the perspective of clinical development, one of the greatest surprises came from the IL-2 gene knockout mouse, which had the phenotype of bowel autoimmunity and increased activated T-cells and B-cells.5 It was then recognized that IL-2 was critical for development and maintenance of T-regulatory cells (Treg) and that Treg expressed the high affinity trimeric IL-2 receptor (the alpha chain, CD25, in addition to the beta and gamma chains of the receptor).6 This new understanding of IL-2 biology led to the hypothesis that IL-2 antitumor effects by activation and expansion of effector T-cells were constrained by concurrent agonist effects on immune inhibitory Treg.

In the clinic, the antitumor effects of IL-2 were surpassed by the far more effective immune checkpoint inhibitors (ICI), particularly antagonists of the programmed death receptor 1 and its ligands (PD1-PD-L1 pathway). Two independent paths of investigation led to a resurgence of interest in clinical development of IL-2. The first was a demonstration that wild type IL-2 could combine with anti-PD-L1 to clear infection in a difficult mouse model of lymphocytic choriomeningitis virus (LCMV).7 In this model, IL-2 was shown to expand a subset of exhausted virus-specific CD8 T-cells and improve their responsiveness to antigen despite a concurrent increase in blood and tissue Treg. The second was promising antitumor activity of a pegylated IL-2, NKTR-214 (bempegaldesleukin) in preclinical mouse models.8

NKTR-214 is an engineered aldesleukin prodrug with six pegylated surface lysines that extend its half-life and reduce binding to the alpha chain of the IL-2 receptor.8 9 After administration, serial shedding of PEG chains increases binding to the IL-2 receptor. Because of its limited activation of the high affinity trimeric IL-2 receptor (IL-2Rαβγ) expressed on Tregs, NKTR-214 was predicted to reduce IL-2’s preferential expansion of Treg compared with the expansion and activation of T cells and NK cells, which only express the intermediate affinity dimeric IL-2 receptor lacking the α subunit (IL-2Rβγ). In mouse models, NKTR-214 increased Treg in peripheral tissues but depleted Treg in the tumor microenvironment through a mechanism that involved CD8 T cell production of interferon-gamma and tumor necrosis factor-α.10 In addition, direct binding of IL-2 to the trimeric IL-2 receptor on endothelial cells had been proposed as a potential mechanism of IL-2 induced VLS, and also supported a non-α receptor binding IL-2 design to reduce toxicity.11 Preliminary clinical data from NKTR-214 in combination with anti-PD-1 appeared promising in patients with immunotherapy naïve melanoma. Subsequently, the immense creativity of a modern generation of protein engineers and capital investment were unleashed to create many second-generation IL-2 molecules which are either in the clinic or soon-to-enter clinical development.

At the 2022 European Society for Medical Oncology (ESMO) Congress, the first phase 3 trials of NKTR-214 in combination with nivolumab were reported in melanoma and renal cell carcinoma.12 13 Both trials were negative and no improvement over standard of care was observed leading to discontinuation of NKTR-214 development. In patients with metastatic melanoma, the ORR, median PFS, and median overall survival for NKTR-214+Nivo vs Nivo alone were 27.7 vs 36% (p=0.03), 4.17 vs 4.99 months (HR=1.09, p=0.398), and 29.67 vs 28.88 months (HR=0.94, p=0.636).

The results for NKTR-214 have understandably called into question the potential viability of the many other engineered IL-2 molecules in clinical development. Was this a failure for IL-2 or a failure specific to NKTR-214 engineering? In a report based on a standard of care registry database, administration of HD aldesleukin to patients with disease progression after treatment with immune checkpoint inhibitors continued to show a comparable level of durable response as in the 1980s, with the caveat that most patients in this series had only received single agent anti-PD-1.14 In contrast NKTR-214 demonstrated no monotherapy objective responses in ICI-resistant/refractory patients enrolled in phase 1 studies, and in phase 3 studies, there was no numeric increase in response rate with NKTR-214+nivolumab compared with nivolumab alone.15 Additionally, NKTR-214 cytokine mediated toxicities were similar to aldesleukin, which precluded dose escalation. Although the dose taken forward for phase 2 and 3 trials was calibrated to allow safe outpatient administration, strict management guidelines to reduce agent-induced hypotension were required including intravenous fluid administration before each dose of NKTR-214.16 Additionally, in the phase 3, melanoma data presented at the 2022 European Society for Medical Oncology (ESMO) Congress, grade 3–4 treatment emergent adverse events were observed in 21.7% with NKTR-214+nivolumab vs 11.5% with nivolumab alone.

Like wild-type IL-2, NKTR-214 exposure in humans appears to be much lower than what was achieved in mouse tumor models. NKTR-214 activates and expands NK cells, and although a clear role for NK cells in the antitumor effects of IL-2 has not been established, increasing data support a major role for NK cells in mediating IL-2 toxicities.17–19 Therefore, NKTR-214’s failure in the phase 3 trial could be explained by inadequate exposure particularly within the tumor microenvironment, or inadequate expansion of tumor-specific T cells, or inadequate selectivity for tumor-specific CD8 T-cells over Treg. Another consideration is that CD25 binding may be important for efficacy. A recent paper showed that IL-2 binding to CD25 was required for its synergistic effect with PD-1 blockade in the difficult to-treat LCMV mouse model.20 Since October of 2022, two other non-alpha IL-2 variants, NL-201 and Thor-707 (SAR444245) were discontinued or deprioritized from clinical development.

Various protein engineering approaches have been used to improve the properties of IL-2 and could perhaps overcome the limitations presented by NKTR-214. In addition to development of other non-alpha IL-2 variants, prodrug molecules have been created which are activated specifically within the tumor microenvironment, and IL-2 variants have been attached to tumor targeting molecules such as fibroblast activation protein α. Non-alpha IL-2s have also been directed to specific T cell subsets with ‘bispecific’ antibodies, referred to as cis-targeting, for example, to CD8 T-cells or T cells expressing PD-1.21 22 A recent publication indicates that cis-targeting of non-alpha IL-2 to T-cells may bypass the requirement for CD25 binding.22 Finally, a novel IL-2 molecule with reduced binding to the gamma IL-2 receptor chain (and therefore preferential binding to cells expressing the alpha and beta chains) recently entered clinical trials.23 These approaches are expected to increase IL-2 effect within the tumor microenvironment and/or bias IL-2 away from interaction with NK cells±Tregs.

Perhaps the biggest hurdle for clinical development of all the novel IL-2 molecules is the lack of predictive biomarkers to identify patients who can respond to IL-2 alone or who require an IL-2 effect to enable response to immune checkpoint inhibitors. The actual clinical significance of IL-2-mediated effects on Treg remain undefined, and perhaps diminishes in importance if the specific IL-2 variant sufficiently expands tumor-specific CD8 T cells and/or restores the functionality of the tumor-specific exhausted CD8 T cell pool, particularly within the tumor microenvironment. In addition to biological design, optimizing the dose and duration of exposure for a particular IL-2 variant may be necessary for clinical success.

Despite the concerns, many of the IL-2 variant molecules are significantly differentiated from NKTR-214 to justify continued clinical development. Moreover, based on prior preclinical data, there are many opportunities to explore in the clinic that go beyond just combination of the novel IL-2 molecules with immune checkpoint inhibitors, for example, combinations with other cytokines, cell therapies, and bispecific CD3 (T cell) or NK cell engagers. Clinical data and correlative studies from ongoing and pending trials, which address drug exposure and confirm IL-2 immune modulatory effects in the tumor, are necessary to determine if we can indeed teach this old dog new tricks.