Safety, pharmacokinetics, pharmacodynamics, and antitumor activity of cergutuzumab amunaleukin: a phase I study in patients with advanced and/or metastatic solid tumors

Abstract

Background

Cergutuzumab amunaleukin (CA) is a novel immunocytokine comprising an interleukin-2 variant moiety with abolished CD25 (interleukin-2 receptor α) binding, fused to a bivalent anti-carcinoembryonic antigen (CEA) monoclonal antibody. This first-in-human phase I study evaluated the maximum tolerated dose (MTD), safety, pharmacokinetics, pharmacodynamics, and antitumor activity of CA.

Material and methods

Patients had CEA-positive advanced and/or metastatic solid tumors that had progressed on standard-of-care treatment. The study consisted of two parts: patients received single-dose CA 0.1-6 mg (n = 5) in part I and multiple ascending doses of 10-40 mg every 2 weeks (q2w; n = 31) or 6-30 mg weekly (qw; n = 24) in part II. Patients with advanced renal cell carcinoma or melanoma (CEA-negative) were permitted in part II.

Results

Sixty patients were enrolled. Most common primary tumor sites in part I, part II q2w, and part II qw were the colon (40%, 44%, and 57%, respectively) and rectum (40%, 19%, and 14%, respectively). Four dose-limiting toxicities (DLTs) established an MTD of 30 mg q2w [grade (Gr)4 hypophosphatemia and thrombocytopenia at 40 mg; Gr2 capillary leak syndrome and Gr3 fatigue at 30 mg]. Dose escalation was halted in the qw regimen due to DLTs at 25 mg (Gr3 hypotension and thrombocytopenia). The most frequently reported adverse events were pyrexia (68%) and infusion-related reaction (52%). The pharmacokinetics of CA were consistent with target-mediated drug disposition, but approximate dose-proportional exposure was observed at 6-40 mg doses. In the blood, CA preferentially and significantly expanded CD8+ T cells and natural killer cells, but not regulatory T cells (T_reg). There were no objective responses; 6/53 (11%) evaluable patients had stable disease (median duration 4.5 months).

Conclusions

Consistent with its mechanism of action, CA had a manageable safety profile and expanded immune effector cells but not T_reg. Further development in combination with additional immunomodulatory agents is warranted.

Highlights

• •Cergutuzumab amunaleukin is an immunocytokine comprising an IL-2 variant fused to a bivalent anti-CEA-targeted antibody.
• •In solid tumors, the CA pharmacokinetic profile supported once-weekly or biweekly dosing with a manageable safety profile.
• •CA triggered effector immune cell expansion in the blood and showed signs of immune induction without T_reg expansion.
• •The study demonstrates a tolerable CA dose allowing treatment in the outpatient setting while producing immune activation.
• •This tolerable IL-2 variant has effective CD122/CD132+ cell engagement without CD25 binding and merits further development.

Introduction

Immunotherapies, such as immune checkpoint inhibitors (ICI), targeting immune surveillance-evading mechanisms, are currently in development for a wide range of cancers. However, not all patients benefit from ICI.¹ Such response heterogeneity may occur due to CD8+ T cell inactivity, which is driven by the tumor microenvironment (TME),^2,3 or lack of high mutational load/antigens.

Activation of the immune system through cytokine therapy, such as interleukin (IL)-2, may help overcome tumor resistance to ICI therapy by stimulating CD8+ T cell proliferation/activity. Endogenous IL-2 is essential in mediating antitumor response by promoting differentiation, proliferation, survival, and homeostasis of immune effector cells.⁴ High-dose recombinant IL-2 (aldesleukin) was the first approved immunotherapy for patients with metastatic renal cell carcinoma (RCC)⁵ and metastatic melanoma,⁶ having demonstrated durable responses in some patients. However, aldesleukin requires frequent infusions causing dose-related, systemic toxicities that may lead to organ failure [such as capillary leak syndrome (CLS)], and is therefore restricted to patients with excellent performance status in specialist settings.^7,8 The effectiveness of wild-type IL-2 therapy is also hampered by the promotion of regulatory T cells (T_reg), which express high levels of CD25 to constitute a high-affinity IL-2 receptor (IL-2R) and limit immune responses in the TME.^4,9, 10, 11, 12

Immunocytokines are a novel modality of cancer immunotherapy combining the tumor-targeting ability of tumor-specific antibodies with the broad-based immune-modulatory activities of cytokines. To improve the efficacy and safety of IL-2 therapy, various immunocytokines have been developed by fusing wild-type IL-2 to tumor-targeting antibodies.^13,14 To date, none of these molecules has advanced beyond phase II trials, potentially due to toxicity, rapid systemic clearance, short half-life, preferential activation of T_reg over immune effector cells, or inadequate tumor targeting.¹⁵ Several cytokine-based drugs other than IL-2 have been engineered to improve the therapeutic index of their respective wild-type molecules, including PEGylated versions of IL-10 (pegilodecakin, AM0010) and an IL-15 superagonist (ALT-803).¹⁶

Cergutuzumab amunaleukin (CA) is a novel immunocytokine comprising a single moiety of a variant form of IL-2 (IL-2v) fused to a bivalent carcinoembryonic antigen (CEA)-targeted antibody.¹⁵ CA recognizes a membrane-proximal CEA target epitope but not soluble CEA. CA is unable to bind to IL-2Rα (CD25) present on endothelial cells and T_regs, but retains its intermediate-affinity binding with IL-2Rβγ expressed on CD8+ T cells, CD4+ T cells, and natural killer (NK) cells.¹⁵ In preclinical studies, CA showed superior tumor targeting and reduced clearance versus wild-type IL-2 and strongly expanded peripheral and tumor tissue NK cells and CD8+ T cells, skewing the T cell compartment in favor of CD8+ over CD4+ cells (including T_regs).¹⁵ Results of an exploratory immune-positron emission tomography study in patients with solid tumors demonstrated that Zn-89-labeled CA preferentially targeted CEA surface expression-positive (CEA+) tumors, with a trend toward dose-dependent CEA-mediated tumor accumulation.¹⁷

Here, we present the findings of a phase I study of single-agent CA in patients with metastatic/unresectable CEA+ solid tumors who had progressed on standard-of-care treatment. A detailed biomarker analysis from this study has been reported separately.¹⁸

Materials and methods

Study design

This first-in-human, open-label, dose-escalation phase I study (ClinicalTrials.gov: NCT02004106; BP28920) was conducted at nine centers across seven countries (Denmark, Finland, France, the Netherlands, Spain, Switzerland, and the United States). The study consisted of two parts (Supplementary Figure S1, available at https://doi.org/10.1016/j.esmoop.2026.107697). Part I evaluated the safety of a single intravenous (IV) infusion of CA at low doses (0.1, 0.5, 1.5, 3, or 6 mg; maximum dose planned in part I) in a single-patient cohort. Each patient was observed for 14 days before the dose was escalated based on safety observations and stopping rules [i.e. grade ≥3 treatment-related adverse event (AE), excluding hematologic AEs or infusion-related reactions (IRRs)].

Part II of the dose escalation was a multiple ascending dose study in which IV CA monotherapy was given initially once every 2 weeks (q2w; 10, 20, 30, or 40 mg) or in the later cohorts weekly (qw; 6, 10, or 20 mg) including intrapatient dose escalation (i.e. with the starting dose increased on the second or third administration: 20/25 mg or 20/20/30 mg). Each cycle was 14 days for q2w and 7 days for qw. Each patient was observed for 21 days for assessment of any dose-limiting toxicity (DLT). The first patient in each cohort and regimen was observed for safety for at least 1 week before enrolling additional patients into the same cohort. Part II employed a modified-continual reassessment method with overdose control based on DLTs observed at each dose and to establish the maximum tolerated dose (MTD) of each schedule.¹⁹ When intrapatient dose escalation was used, dose selection was informed by mathematical modeling of pharmacokinetics and tumor uptake, derived from a previously published imaging study.^17,20

Patients

Eligible patients were aged ≥18 years with metastatic/unresectable CEA+ solid tumors. Enrollment of an additional cohort (3-6 patients) in part II with CEA surface expression-negative but IL-2–sensitive tumors (i.e. advanced RCC or cutaneous melanoma) was permitted. Patients had confirmed disease progression (PD) at baseline (using RECIST v1.1²¹), radiologically measurable and clinically evaluable disease, and an Eastern Cooperative Oncology Group performance status of 0/1. Patients with a history or clinical evidence of central nervous system (CNS) primary tumors or metastases and those with an active second malignancy (other than nonmelanoma skin cancer or cervical carcinoma in situ) were excluded. Additional eligibility criteria are provided in Supplementary Appendix S1 (available at https://doi.org/10.1016/j.esmoop.2026.107697).

CEA+ tumors were defined as those with ≥20% of tumor cell membranes staining with at least moderate intensity by immunohistochemistry (IHC). Staining for CEA was performed locally with a CEA31 mouse monoclonal IgG1 anti-CD66/CEACAM5 antibody (Cell Marque #760-4594, Ventana Medical Systems, Tucson, AZ) using an in-house validated procedure.¹⁷ CEA positivity was also retrospectively confirmed by central testing using the same assay.

Study endpoints

The primary objectives were to determine the MTD and describe the safety and pharmacokinetics of single-agent CA. Secondary objectives were to characterize the pharmacodynamics of single/multiple doses of CA on peripheral blood cells, determine changes in CA treatment-associated biomarkers, and obtain preliminary antitumor activity for single-agent CA. Determining the relationship between exposure-pharmacodynamics and clinical effects of CA at different doses was an exploratory objective.

Assessments

The MTD was defined as the dose that maximized the probability of a DLT being in the targeted toxicity interval of 20%-35%, subject to additional safety criteria.¹⁹ DLT was defined as the occurrence of prespecified grade ≥3 toxicity attributed to CA, with the exception of some acute IL-2 class-effect grade 3 toxicities resolving shortly after CA administration (e.g. fever and some other nonhematology abnormalities) within the 21-day DLT period. A detailed DLT definition is provided in Supplementary Appendix S2 (available at https://doi.org/10.1016/j.esmoop.2026.107697).

Safety was determined by AE reporting, graded according to the National Cancer Institute Common Terminology Criteria for AEs (v4.03); routine laboratory tests; vital signs; electrocardiograms; physical examination (including frequent body weight assessment, an indicator of CLS/edema); observation and monitoring of patients for IRRs; and determination of antidrug antibodies (ADAs). Because of the risk of developing CLS, in particular pulmonary edema, all patients had a chest X-ray at screening, before treatment, and during the study at any sign of pulmonary toxicity. Treatment with CA was interrupted until full resolution of pulmonary signs and symptoms.

Pharmacokinetic parameters for CA in serum following IV administration were determined via noncompartmental analysis using Phoenix WinNonlin v6.2 (Pharsight Co, Mountain View, CA) and included time to maximum concentration, maximum drug concentration (C_max), area under the serum concentration-time curve (AUC) from time of dosing to the last measurable concentration, AUC from time of dosing to the end of the dosing period (i.e. 336 hours with q2w dosing and 168 hours with qw dosing), time to last measurable concentration, and last measurable concentration.

Pharmacodynamic outcomes included the determination of the absolute count of immune cells (CD4+ and CD8+ T cells, NK cells, macrophages/monocytes, T_reg, and B cells) in the blood. Archival/fresh tumor tissue was analyzed centrally for programmed death-ligand 1 (PD-L1) expression using clone SP142 by IHC (Histogenex, Antwerp, Belgium). PD-L1 was scored on the area covered by PD-L1-positive tumor-infiltrating immune cells (IC), including macrophages, dendritic cells, and lymphocytes. Additional biomarker analyses and data have been reported separately.¹⁸

Antitumor activity of single-agent CA was evaluated according to RECIST v1.1²¹ for objective response rate (ORR; defined as a complete response or partial response), disease control rate [DCR; defined as complete response, partial response, or stable disease (SD)], progression-free survival (PFS), and overall survival. ORR, DCR, and PFS were also reported according to immune-related modified RECIST (mRECIST).²²

Statistical analyses

Patients were followed until PD (according to RECIST v1.1 or mRECIST, whichever occurred later) or until treatment discontinuation due to study withdrawal. All patients who received ≥1 dose of CA were included in the safety, pharmacokinetic, and pharmacodynamic populations. Preliminary antitumor activity of single-agent CA was evaluated in all patients within the safety population who had more than one postbaseline tumor assessment. Descriptive statistics and exploratory hypothesis testing are presented. No formal statistical hypothesis testing was conducted. The data cutoff date was December 06, 2016.

Study approval

The study protocol was approved by the local Ethics Committees/Institutional Review Boards at each center (see Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2026.107697). At Dr Lassen’s center (Rigshospitalet, Copenhagen, Denmark), the protocol was approved by the Scientific Ethics Committees for the Capital Region of Denmark (vek@regionh.dk); no approval ID was provided. The study was conducted in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice guidelines. All patients provided written informed consent prior to any study-related procedures.

Results

Patients

Between 30 December 2013, and 21 January 2016, 60 patients were enrolled. Of these, five patients received single-dose treatment with CA in part I, 31 received q2w CA in part II, and 24 received qw CA in part II (Supplementary Figure S1, available at https://doi.org/10.1016/j.esmoop.2026.107697). All 60 patients were included in the safety and pharmacokinetic analysis populations. Only part II patients were included in the pharmacodynamic analysis population. Overall, 53 patients were evaluable for antitumor activity: 5/5, 27/31, and 21/24 patients in part I, part II q2w, and part II qw, respectively.

Patient baseline demographics and clinical characteristics were generally similar across treatment groups (Table 1). The median age was 60 years in part I and part II q2w, and 62.5 years in part II qw. The most common primary tumor in part I, part II q2w, and part II qw was colorectal cancer (CRC; 40%, 44%, and 57%, respectively) followed by rectal cancer (40%, 19%, and 14%, respectively). Patients had received a median of three prior lines of therapy for metastatic disease.

Table 1.

Baseline demographic and clinical characteristics

Characteristic	Part I (n = 5)	Part II q2w (n = 31)	Part II qw (n = 24)
Median age, years (range)	60 (52-63)	60 (36-79)	62.5 (46-78)
Age, n (%)
<65	5 (100)	21 (68)	14 (58)
≥65	0	10 (32)	10 (42)
Male, n (%)	5 (100)	18 (58)	18 (75)
ECOG PS, n (%)
0	3 (60)	12 (39)	17 (71)
1	2 (40)	19 (61)	7 (29)
Ethnicity, n (%)		(n = 30)	(n = 23)
Hispanic or Latino	2 (40)	8 (27)	4 (17)
Not Hispanic or Latino	3 (60)	22 (73)	17 (74)
Not stated	0	0	2 (9)
Median weight, kg (range)	77.2 (67.0-87.6)	75.5 (47.4-126.0)	78.6 (49.1-117.0)
Disease stage, n (%)
IV	5 (100)	25 (81)	20 (83)
IVB	0	1 (3)	0
IVC	0	0	1 (4)
Unknown	0	5 (16)	3 (13)
Primary tumor site, n (%)a	(n = 5)	(n = 27)	(n = 21)
Colon	2 (40)	12 (44)	12 (57)
Rectum	2 (40)	5 (19)	3 (14)
Pancreas	0	4 (15)	2 (10)
Lung	1 (20)	2 (7)	0
Small intestine	0	1 (4)	1 (5)
Breast	0	0	1 (5)
Stomach	0	1 (4)	0
Cervix	0	1 (4)	0
Bile duct	0	0	1 (5)
Mixed adeno-neuroendocrine	0	1 (4)	0
Esophagus	0	0	1 (5)

ECOG PS, Eastern Cooperative Oncology Group performance status; qw, weekly; q2w, every 2 weeks.

^aPatients evaluable for antitumor activity.

CEA expression testing

Overall, CEA expression was similar for the 48 patients with both local and central results; the median symmetric percentage difference was 17.2%. Differences were mostly driven by cases where different samples were tested locally and centrally. All patients had ≥20% of tumor cell membranes with moderate or intense staining for CEA by IHC on archival or freshly obtained tumor samples. In patients with CEA+ tumors, no correlation between CEA expression levels and response was observed (data not shown). For one patient with adenocarcinoma of the cervix and one patient with mixed adenoneuroendocrine carcinomas, screening biopsies were negative for CEA when tested centrally. In both cases, archival material was used for prospective testing to determine eligibility locally. For the patient with adenocarcinoma of the cervix, the archival sample was also submitted for central analysis and was positive. Further results on centrally tested results have been published previously.¹⁸

MTD

A summary of DLTs is provided in Table 2. No DLTs were observed in part I. During the part II q2w dosing schedule, one patient experienced two DLTs (grade 4 hypophosphatemia and thrombocytopenia, onset in cycle 1, day 5, and day 6, respectively) after the first administration of 40 mg q2w, leading to a reduction in the CA dose to 30 mg. This occurred in the first patient of the 40 mg cohort; no additional patients were enrolled, and dose escalation was halted. Two of 11 patients treated with 30 mg q2w experienced DLTs [grade 2 CLS and grade 3 fatigue (n = 1 each), both after the first administration of CA]. Consequently, the MTD was declared as 30 mg for the q2w schedule.

Table 2.

Overview of DLTs

Dose schedule	Patients treated (n)	Patients with DLT (n)	Adverse event	Most extreme NCI CTCAE grade	Action taken with study drug
DLTs in part II q2w
10 mg q2w	11	0	–	–	–
20 mg q2w	6	0	–	–	–
30 mg q2w	11	1	CLS	2	Withdrawn
		1	Fatigue	3	Withdrawn
40 mg q2w	3	1	Hypophosphatemiaa	4	Reduced
			Thrombocytopeniaa	4
DLTs in part II qw
6 mg qw	4	0	–	–	–
10 mg qw	3	0	–	–	–
20 mg qw	6	1	Oliguria	3	None
		1	LFT elevation	3	Reduced
		1	Dyspnea	3	Withdrawn
20/25 mg qw	6	1	Hypotension	3	Reduced
		1	Thrombocytopenia	3	Reduced
20/20/30 mg qw	5	0	–	–	–

CLS, capillary leak syndrome; DLT, dose-limiting toxicity; LFT, liver function test; NCI CTCAE, National Cancer Institute Common Terminology Criteria for Adverse Events; qw, weekly; q2w, every 2 weeks.

^aOccurring within the same patient.

In the part II qw schedule, DLTs occurred in three patients at 20 mg qw [grade 3 oliguria and abnormal liver function tests (LFT; n = 1 each) after the first infusion, and grade 3 dyspnea (n = 1) after the second infusion] and in two patients at 25 mg qw during the intrapatient 20/25 mg dose uptitration [grade 3 hypotension and thrombocytopenia (n = 1 each)]. The patient with oliguria recovered, and the CA dose was unchanged. For the DLTs of abnormal LFT, hypotension, and thrombocytopenia, the patients recovered, and the CA dose was reduced. Due to the toxicity observed at 25 mg, and safety data available from the q2w schedule, dose escalation was halted, and the MTD was not defined for the qw regimen.

Safety

The most frequent any-grade AEs were pyrexia [part I: n = 4 (80%); part II q2w: n = 18 (58%); part II qw: n = 18 (75%)] and IRR [part II q2w: n = 18 (58%); part II qw: n = 13 (54%)]. All patients in parts I and II experienced ≥1 treatment-related AE. The majority of AEs occurred within the first week after the first CA administration and resolved prior to the next infusion.

Overall, 45 (75%) patients experienced ≥1 grade 3/4 AE, of which the most frequent were IRR [n = 11 (18%) in parts I and II], ƴ-glutamyl transferase increased [n = 8 (13%)], hypophosphatemia [n = 7 (12%)], and fatigue [n = 6 (10%)]. Grade 3/4 pyrexia occurred in three patients (5%). CLS occurred in four patients (7%) in part II of the study (q2w CA: n = 3; qw CA: n = 1). AEs associated with CLS in these four patients included grade 1 edema (n = 3), grade 1 weight increase (n = 2), grade 1 and 2 pleural effusion (n = 2), grade 2 hypotension (n = 2), and grade 2 hypoalbuminemia (n = 1). All CLS events were classed as serious, grade 2 in severity, and related to the study drug. All four patients recovered after receiving appropriate treatment. Edema AEs (edema, peripheral edema, and pulmonary edema) were reported by 10 patients in part II q2w and eight patients in part II qw at a range of doses from 6 to 30 mg, with no pattern of dose dependency. All edema AEs were grades 1 or 2, resolved, and did not impact study treatment adherence, except for one grade 3 event with 20 mg CA leading to permanent study discontinuation.

Serious AEs occurred in approximately two-thirds of patients (Table 3), the most frequent being IRR [part II q2w: n = 5 (16%); part II qw: n = 6 (25%)]. There were no grade 5 (fatal) AEs.

Table 3.

Summary of AEs

Patients, n (%)	Part I (n = 5)	Part II q2w (n = 31)	Part II qw (n = 24)
AE, any cause	5 (100)	31 (100)	24 (100)
Treatment-related AE	5 (100)	31 (100)	24 (100)
Grade 3 AE, any cause	4 (80)	23 (74)	17 (71)
Grade 4 AE, any cause	1 (20)	8 (26)	3 (13)
Grade 5 AE, any cause	0	0	0
Grade 3 treatment-related AE	0	16 (52)	13 (54)
Grade 4 treatment-related AE	0	5 (16)	2 (8)
Serious AE	3 (60)	21 (68)	15 (63)
Treatment-related serious AE	0	19 (61)	12 (50)
AEs leading to treatment withdrawal	1 (20)	1 (3)	1 (4)
AEs leading to dose modification/interruption	1 (20)	4 (13)	7 (29)
Dose-limiting toxicity	0	3 (10)	5 (21)
Capillary leak syndrome AEs	0	3 (10)	1 (4)
Edema AEsa	0	10 (32)	8 (33)
AEs reported in ≥20% of patients
Pyrexia	4 (80)	18 (58)	19 (79)
Infusion-related reaction	0	18 (58)	13 (54)
Fatigue	1 (20)	15 (48)	11 (46)
Nausea	1 (20)	10 (32)	8 (33)
Decreased appetite	1 (20)	13 (42)	6 (25)
Vomiting	2 (40)	9 (29)	8 (33)
Diarrhea	2 (40)	10 (32)	5 (21)
Chills	0	4 (13)	9 (38)
Constipation	2 (40)	10 (32)	7 (29)
Weight decreased	2 (40)	8 (26)	6 (25)
Anemia	2 (40)	7 (23)	5 (21)
Aspartate aminotransferase increased	0	6 (19)	6 (25)
Dyspnea	0	11 (36)	5 (21)
Headache	1 (20)	8 (26)	3 (13)
Hypophosphatemia	0	7 (23)	5 (21)
Peripheral edema	0	6 (19)	7 (29)
Asthenia	2 (40)	8 (26)	2 (8)

AE, adverse event; MedDRA, Medical Dictionary for Regulatory Activities; qw, weekly; q2w, every 2 weeks.

^aEdema AEs (MedDRA preferred term: edema, peripheral edema, and pulmonary edema).

Three patients withdrew due to AEs, one each in part I (grade 1 nausea and vomiting considered unrelated to study drug), part II q2w (grade 4 IRR considered treatment related), and part II qw (grade 3 dyspnea considered treatment related). Both of the related AEs resolved without sequelae; the patient with grade 4 IRR permanently discontinued study treatment as per protocol.

Transient lymphopenia was observed in patients shortly after treatment initiation. This effect may be a cytokine-related phenomenon, because the lymphopenia resolved swiftly into transient lymphocytosis, which could be a consequence of CA-induced immune cell expansion. Similar reductions in thrombocyte count were reported in approximately half of the patients; two events were considered DLTs (described above).

Overall, 52/60 (87%) patients died after study discontinuation: four patients (80%) in part I (all due to PD), 25 (81%) in part II q2w [21 (68%) due to PD], and 23 (96%) in part II qw [21 (88%) due to PD]. In part II q2w, the cause of death was unknown for three patients (10%). A patient in part II q2w died from gastrointestinal bleeding 42 days after the last dose of study drug, which was considered due to PD. In part II qw, one patient died from unknown causes, and another patient died from bacterial peritonitis due to cholangitis 82 days after the last dose of the study drug.

Pharmacokinetic results

Pharmacokinetic parameters of CA after single-dose administration of 0.1-6 mg and after single- and multiple-dose administration of 6-40 mg are summarized in Supplementary Tables S2 and 3, respectively (available at https://doi.org/10.1016/j.esmoop.2026.107697). Following a 2-hour IV infusion, CA serum concentrations increased rapidly and reached C_max around 2-4 hours after starting the infusion (Supplementary Figure S2, available at https://doi.org/10.1016/j.esmoop.2026.107697). The pharmacokinetic profiles were consistent with a saturable and linear clearance mechanism, typical of target-mediated drug disposition. However, at 6-40 mg, approximate dose-proportional exposures (AUC and C_max) were observed within a cycle.

No serum accumulation of CA was observed following multiple treatment cycles; instead, serum exposure of CA decreased (Supplementary Table S3, available at https://doi.org/10.1016/j.esmoop.2026.107697). There was a 40%−50% reduction in AUC between cycles 1 and 4 (q2w) and cycles 1 and 5 (qw) for all except one patient. Because clearance showed time- and dose-dependent behavior, no estimates of CA half-life are provided. However, as an example, the observed half-life following a 20-mg qw dose was estimated to be 18 and 7 hours on cycles 1 and 5, respectively. In general, the observed reduction in exposure/increased clearance, mainly due to target-mediated drug disposition, was more pronounced following a qw regimen than a q2w regimen. Examination of dose-normalized exposure as a function of predose IL-2R-expressing cell counts (i.e. CD8+ and CD4+ T cells and NK cells in the periphery) at cycles 1 and 4/5 showed that, overall, IL-2R-expressing cell counts were inversely related to exposure (Supplementary Figure S3, available at https://doi.org/10.1016/j.esmoop.2026.107697). Overall, 43/51 (84%) evaluable patients showed postdose positive ADA titers at any time following CA administration. Results from a study examining the effects of obinutuzumab-pretreatment on the formation of ADAs have been reported previously.²³

Pharmacodynamic results

As expected from the hypothesized mechanism of action, patients receiving CA demonstrated preferential increases in the blood that were highest for NK cells (6.8-fold expansion, P < 0.001) followed by CD8+ T cells (2.6-fold expansion, P < 0.001) and to a lesser extent CD4+ T cells (1.4-fold expansion, P = 0.003), without preferential expansion of T_reg (P = 0.53; Figure 1). No significant changes were observed in the absolute count of macrophages/monocytes and B cells (data not shown). Evaluable pretreatment tumor samples were generally low in PD-L1 expression, with a negative score (IC = 0) in 13/51 (25%) of the evaluated patients’ samples (Figure 2B), and no patient had >10% of the tumor area covered by PD-L1-positive immune cells.¹⁸ More pharmacodynamic results, including PD-L1 regulation, have been reported separately.¹⁸

Figure 1.

Cergutuzumab amunaleukin significantly expanded NK (CD3-CD56+/CD16+) cells, CD8+ T cells, and to a lesser extent CD4+ T cells without preferentially expanding T_reg (CD3+CD4+CD25+CD127-/loFOXP3+). ABS numbers and percentages (%) of the indicated lymphocyte subpopulation were counted by flow cytometry. Data are shown per indicated dose for values from baseline and C4D1 predose measurements combined from qw and q2w schedules. ∗P < 0.1; ∗∗P < 0.05, and ∗∗∗P < 0.01.

ABS, absolute numbers; C4D1, cycle 4 day 1; NK, natural killer; NS, not significant; T_reg, regulatory T cells; qw, weekly; q2w, every 2 weeks.

Figure 2.

Response to cergutuzumab amunaleukin for patients with postbaseline tumor assessments. (A) Sum of largest tumor diameter (expressed as % of baseline sum of largest diameter) over time. Tumor types are indicated by different colors. All tumors, apart from RCC, expressed CEA above the required cutoff (including mixed histologies e.g. adenosquamous differentiation). (B) The smallest sum of the largest tumor diameter postbaseline (expressed as % of baseline sum of largest diameter). Patients with a reduction in target lesions did not meet objective response criteria. Patients with shrinking target lesions indicated as PD had new lesions or unequivocal progression in nontarget lesions. Color-coding indicates the PD-L1 status of an archival or a fresh baseline tumor sample. Patients who progressed prior to postbaseline tumor assessment are excluded.

CEA, carcinoembryonic antigen; IC, tumor-infiltrating immune cell; PD, progressive disease; PD-L1, programmed death-ligand 1; qw, weekly; q2w, every 2 weeks; RCC, renal cell carcinoma; SD, stable disease.

Antitumor activity

Overall, 53 patients were evaluable for antitumor activity. No patient achieved an objective response (RECIST v1.1). SD was evident in six patients (part I: n = 1; part II q2w: n = 2; part II qw: n = 3), showing a DCR of 20% [90% confidence interval (CI) 4.60-56.47], 7% (90% CI, 2.16-17.74), and 13% (90% CI, 5.11-27.49), respectively (Figure 2). Median duration of SD was 4.5 months (95% CI, 4.5-not reached). Median investigator-assessed PFS (RECIST v1.1) was 84 (95% CI, 17-141), 78 (95% CI, 58-83), and 77 days (95% CI, 59-82) for part I, part II q2w, and part II qw, respectively. Analyses based on mRECIST yielded comparable results.

Discussion

CA was designed to exploit the immune stimulatory effects of IL-2, by avoiding T_reg activation and skewing the effects toward immune effector cells, and to increase the therapeutic window through tumor targeting and abrogation of CD25 binding.¹⁵

A 21-day DLT period for both schedules was chosen with the safety profile of CEA-IL2v in mind and to capture well-known class effects of proleukin (IL-2), which typically manifest within the first week. In this study, the MTD was determined as 30 mg for the q2w schedule. This was based on the occurrence of well-known IL-2 class-effect AEs (grade 4 hypophosphatemia and thrombocytopenia) occurring at 40 mg CA and resulting in a reduction to 30 mg. At high doses, CA displayed an AE profile similar to other IL-2 compounds, i.e. CLS, edema, hypotension, pyrexia, flu-like illness, and fatigue. However, typical aldesleukin-related AEs such as CNS or skin toxicity²⁴ were not observed in this small sample size. Reductions in thrombocyte counts, another well-known IL-2-mediated AE, were observed, but these were transient with recovery within 24 hours of onset. The toxicity profile appears to be more related to the IL-2v moiety and secondary cytokine effects, but some gastrointestinal toxicity was seen. Patients receiving CA were able to be managed in an outpatient setting, especially after they had adapted to their first or second treatment cycle. AEs were manageable, allowing patients to remain on treatment for a long period of time, as opposed to only 1 or 2 possible treatment cycles with high-dose aldesleukin.²⁵

Owing to the immunoglobulin-cytokine fusion protein design, CA has a favorable pharmacokinetic profile over aldesleukin. The prolonged half-life allows dosing of patients qw or q2w, yet still delivering IL-2 exposures higher than those achieved with high-dose aldesleukin schedules (600 000 IU/kg every 8 hours, days 1-5).²⁴ It is hypothesized that CA mediates its own clearance; the IL-2v moiety of CA induces expansion of total circulating IL-2R-expressing cells, resulting in an increased number of IL-2R subunits that enhance saturable target-mediated clearance. The higher the baseline IL-2R-expressing cell count, the greater the reduction in dose-normalized exposure. More frequent dosing did not deliver higher cumulative exposure levels, because target induction and the number of IL-2R-expressing cells increased drug clearance (i.e. CA has time- and dose-dependent pharmacokinetics). Modeling efforts suggested that retaining the compound in the TME through CEA binding may help increase exposure levels locally while apparent systemic exposure is decreased with time, higher doses, and more frequent dosing.^17,20

CA promoted the expansion of NK cells and T cells (CD8+ and CD4+) in peripheral blood. Steady increases in soluble CD25 (sCD25) also suggested a systemic activation effect of CA.^18,20 T_reg are unable to utilize IL-2v effectively due to missing CD25 engagement. Nonetheless, T_reg can derive IL-2R signaling through IL-2Rβγ in the same way as conventional CD25-negative T cells.¹⁵ Further, T_reg will react to endogenous wild-type IL-2 from activated and cycling immune cells. Hence, despite the IL-2v design, a very low and insignificant expansion of T_reg in the periphery was documented in response to CA dosing.

Following standard response assessment using RECIST v1.1, an objective response to treatment was not documented and the DCR was low. Heavily pretreated patients with CEA+ solid tumors that are not usually responsive to cytokines, especially IL-2, were enrolled. The most common diagnosis was CRC, which is considered immunologically “cold” with low intrinsic immunogenicity and is not known to be sensitive to immune modulators, with the exception of programmed cell death protein-1 (PD-1) inhibition in the small subset of microsatellite high (MSI-H) cancers.^26,27 Although the MSI/MMR status of patients with CRC was not determined in this study, the general immunological characteristics of CRC make it less likely that tumor control would be seen in those patients, given the immune-modulatory mechanism of action of CA that requires the presence of a responsive immune compartment. Only if such a compartment is sufficiently engaged by treatment can tumor control be expected. IL-2v is unlikely to exert direct antitumor or tumor cell toxic effects. Rather, effects need to be established indirectly through activation/expansion of tumor-reactive immune cells. Responses to treatment are likely to form and develop slowly over time, and immediate tumor shrinkage is unlikely. If at all, proinflammatory processes may rather result in an influx of cells and swelling of the tumor, which could be misinterpreted as pseudoprogression. No pseudoprogression could be verified in the current patient population. Preferential expansion of immune cells in the peripheral blood and increases in sCD25²⁰ with CA suggest that systemic effects might also occur in the TME if such cells were present, or if these cells migrated from the blood to the TME. The immune stimulatory mechanism of action of CA suggests synergistic combination with other immunomodulatory agents may help unlock potential antitumor activity, and the manageable safety profile of CA would support a combinatorial approach.

Limitations of this study include the small sample size and the preponderance of CRC in the study population. Although some severe IL-2-like toxicities were reported in the highest dose cohorts, definitive conclusions cannot be reached regarding the dose dependency of AEs due to the small sample sizes in the different dose cohorts.

Following completion of this study, extensive biomarker analyses were carried out to interrogate the mechanism of action of CA as a selective immune modulator¹⁸ and a subsequent study of CA in combination with PD-L1 inhibitor atezolizumab in CEA-positive solid tumors was completed,²⁸ to enable a more comprehensive assessment of CA. The study of CA plus atezolizumab showed similar findings to the current study, in showing systemic activation but limited clinical activity, suggesting that inhibition of the PD-L1:PD-1 immune checkpoint is not sufficient to overcome any immunosuppressive mechanisms. Because CA caused systemic activation as monotherapy and in combination with atezolizumab, further investigation may be warranted in patients with tumor types that are more susceptible to immunotherapy such as MSI-H CRC. However, there were too few patients with MSI-H tumors in the combination study to assess activity,²⁸ and in the current study, MSI status was not determined.

ADAs against CA developed in more than 80% of patients in the current study, which is consistent with the proportion observed in the subsequent combination study with atezolizumab.²⁸ The absence of clinical responses in the current study prevents any analysis of the impact of ADAs, and measures to mitigate/reduce the occurrence of ADAs to CA are warranted if this treatment is developed further. However, it is worth noting that in the combination study and also in a study of another novel IL2-based immunocytokine, FAP-IL2v,²⁹ there was no correlation between ADA positivity and clinical response.

Conclusions

Despite the abolishment of CD25 binding of IL-2v, high doses of CA retain IL-2-like safety features. However, at doses below MTD, CA enables management in an outpatient setting and has a manageable safety profile even in at-risk patients. Mechanistically, the design concept works in terms of immune pharmacodynamic effects with no activation of T_reg, but induction of systemic inflammation (sCD25) and promotion of effector cells (NK, CD8+, and CD4+ T cells). Clinical activity of CA was modest and indicated a trend toward certain subgroups; in tumors with high intrinsic immune cell infiltration and an innate susceptibility to immunotherapy such as MSI-H CRC, the combination of cytokine stimulus and checkpoint inhibition blockade may tip the scales of tumor control and be more potent at inducing and maintaining a meaningful antitumor response.

Consent for publication

Not applicable.

Data availability

Data are available on reasonable request. For eligible studies, qualified researchers may request access to individual patient-level clinical data through a data request platform. At the time of writing, this request platform is Vivli (https://vivli.org/ourmember/roche/). For up-to-date details on Roche’s Global Policy on the Sharing of Clinical Information and how to request access to related clinical study documents, see https://go.roche.com/data_sharing. Anonymized records for individual patients across more than one data source external to Roche cannot, and should not, be linked due to a potential increase in risk of patient reidentification.