Cergutuzumab Amunaleukin in Combination with Atezolizumab in Patients with Carcinoembryonic Antigen–Positive Advanced/Metastatic Solid Tumors

Abstract

Purpose:

Cergutuzumab amunaleukin (CA) is an immunocytokine comprising a variant form of interleukin 2 (IL2) [constructed to avoid CD25 binding and regulatory T-cell (Treg) stimulation] fused to a carcinoembryonic antigen (CEA)–targeted antibody. This phase Ib open-label, multicenter dose-escalation and -expansion study (NCT02350673) evaluated the safety, activity, pharmacokinetics, and pharmacodynamics of CA plus atezolizumab in patients with advanced/metastatic CEA-positive solid tumors.

Patients and Methods:

Patients received escalating doses of CA (6–20/25 mg) with fixed dosages of atezolizumab (840 mg) every 2 weeks or escalating dosages of CA weekly (10–15/20 mg) with fixed dosages of atezolizumab (1,200 mg) every 3 weeks. Primary objectives include maximum tolerated dose (MTD), recommended dose for expansion (RDE), and safety.

Results:

Twenty-four patients were randomized to receive CA plus atezolizumab every 2 weeks and 45 patients to CA weekly plus atezolizumab every 3 weeks. A subgroup of patients (n = 5) received obinutuzumab before treatment to study the prevention of antidrug antibodies. The MTD was not determined; 15 mg weekly or 20 mg every 2 weeks of CA plus atezolizumab was the RDE. The safety profile was consistent with CA monotherapy and atezolizumab-based therapies. The addition of atezolizumab did not affect the pharmacokinetic profile of CA, and treatment induced the proliferation of T and NK cells in the blood without Treg expansion. Increases in pharmacodynamic markers (C-reactive protein, lymphocytes, sCD25, and cytokines) suggested immune activation despite limited antitumor activity (overall response rate: 13.5% with weekly/every-3-week regimen).

Conclusions:

The safety profile of this combination was manageable. Prominent pharmacodynamic effects were elucidated; antitumor activity was limited.

Translational Relevance.

Cergutuzumab amunaleukin (CA), a novel immunocytokine comprising a single moiety of a variant form of IL2 fused to a bivalent carcinoembryonic antigen (CEA)–targeted antibody, was developed to overcome the limitations of aldesleukin (high-dose IL2) by selectively promoting immune responses in the tumor microenvironment. IL2 was made defective in CD25 binding to avoid regulatory T-cell activation. CA monotherapy had a manageable safety profile and induced strong pharmacodynamic effects in phase I trials in patients with advanced/metastatic CEA-positive solid tumors; however, despite evident immune-modulatory effects in the tumor, clinical activity was modest. Upregulation of programmed-death ligand 1 (PD-L1) was observed in serial biopsies, suggesting adaptive resistance mechanisms, and preclinical studies have demonstrated superior activity of CA and anti–PD-L1 combination therapy. These findings provided the rationale for investigating the safety, pharmacodynamics, and efficacy of CA plus atezolizumab in patients with advanced/metastatic CEA-positive solid tumors reported in this publication.

Introduction

Clinical trials investigating the use of immune-based treatments have shown that immune checkpoint inhibitors (ICI), particularly PD-1/PD-L1 inhibitors, are effective in the treatment of cancer and consequently have emerged as a new standard of care for several solid tumors as they provide durable responses that extend overall survival (1). However, possibly because of immunosuppressive signals from the tumor microenvironment (TME), such as inactive CD8⁺ T cells and appearance of myeloid-derived suppressor cells and regulatory T cells (Treg), ICIs are only effective in a limited proportion of patients (1, 2). Efforts are needed to identify new drugs that can synergize with ICIs by modifying the immune microenvironment and ultimately enhance antitumor effects.

Immunocytokines can support immune activation by fusion of cytokines with tumor-specific antibodies (3). Cytokine therapy, such as IL2, may promote differentiation, proliferation, and homeostasis of immune effector cells, helping to overcome resistance to ICI therapy (4–6). High-dose IL2 (aldesleukin) is an approved immunotherapy for melanoma and renal cell carcinoma (7). Aldesleukin has demonstrated durable objective responses in some patients, but its use is limited by a challenging safety profile (8–10). Aldesleukin is most often associated with the development of capillary leak syndrome (CLS; caused by the stimulation of Tregs) and lethargy; consequently, use is restricted to a hospital setting in which patients can be closely monitored (7, 11). The effectiveness of aldesleukin is also potentially limited by an unwanted expansion of Tregs, which suppress antitumor immune responses (4, 12–16).

Carcinoembryonic antigen (CEA) is a protein that is often elevated in patients with cancer, with CEA expression levels in the serum or on the tumor cell surface used as biomarkers (17). Cergutuzumab amunaleukin (CA) is a novel immunocytokine comprising a single moiety of a variant form of IL2 (IL2v) fused to a bivalent CEA-targeted antibody (18–20). CA was designed to overcome the limitations of wild-type IL2 and enhance antitumor activity by favoring the expansion of cytotoxic T cells rather than Tregs (18) through lowered CD25 (IL2R-α) binding. A phase I single-agent dose-escalation study (NCT02004106) showed that CA monotherapy has a manageable safety profile and promising pharmacodynamics, such as the expansion of CD8⁺ T cells and NK cells but not of Tregs (Lassen; manuscript in preparation, Melero; manuscript in preparation).

Additionally, preclinical studies have shown that PD-L1 expression is upregulated by IL2 (21, 22), and upregulation of PD-L1 immune cells in the TME has been observed following treatment with CA monotherapy (Melero; manuscript in preparation). This increase in PD-L1 suggests an adaptive resistance mechanism to CA monotherapy. Mouse models have demonstrated synergistic effects between PD-L1 and IL2 therapies; Klein and colleagues (18) showed that the use of a PD-L1 inhibitor with a CA treatment regimen prolonged survival in mice, including rescue from exhaustion (23). This evidence provided a rationale to combine CA with anti–PD-L1 inhibitors such as atezolizumab.

In this study, we report the results of a first-in-human, phase Ib dose-escalation study investigating the safety, pharmacokinetics, pharmacodynamics, and antitumor activity of CA plus atezolizumab in patients with CEA-positive advanced/metastatic solid tumors.

Patients and Methods

Study design and treatment

This was an open-label, multicenter, dose-escalation and -expansion phase Ib trial of CA in combination with atezolizumab (NCT02350673), which was carried out at 14 centers across six countries. This study was conducted in two parts (Supplementary Fig. S1A): part I was a dose-escalation assessment carried out according to a modified Continual Reassessment Method with Overdose Control and designed to evaluate the safety of CA and atezolizumab, and part II was an expansion of the MTD or recommended dose for expansion (RDE) established in part I. Patients in part I were further split to receive either CA (at doses of 6–20 mg or up-titration regimen 20/25 mg) plus a fixed, flat dosage of 840 mg of atezolizumab, both given every 2 weeks, hereafter referred to as the every-2-week/every-2-week regimen, or CA weekly (at 10 mg, 15 mg, and up-titration regimen 15/20 mg) plus atezolizumab 1,200 mg every 3 weeks, hereafter referred to as the weekly/every-3-week regimen (Supplementary Fig. S1B). Both CA and atezolizumab were administered intravenously. The first study treatment was administered on cycle 1 day 1, and the cycle length was 3 and 2 weeks for the weekly/every-3-week and every-2-week/every-2-week regimens, respectively. The dose-limiting toxicity (DLT) observation period was 21 days, counted from the first or higher dose administration of CA plus atezolizumab. At least three patients who completed the DLT observation period were enrolled into each cohort; the enrollment of the first and second patient in each cohort was staggered by 2 weeks. The starting dose of 6 mg and schedule for CA were informed by a CA monotherapy study (NCT02004106), for which the results have been published (Lassen; manuscript in preparation, Melero; manuscript in preparation). Patients received treatment until loss of clinical benefit, unacceptable toxicity, or withdrawal of consent. All patients had a chest X-ray at screening, before the start of the treatment, and during the study at any sign of pulmonary toxicity.

A dose-escalation substudy investigating the effect of obinutuzumab before treatment (2,000 mg) on the development of antidrug antibodies (ADA), overall safety and tolerability, pharmacodynamics, and antitumor activity of CA with atezolizumab was also conducted; the results have been published separately (24). In this substudy, patients received obinutuzumab before treatment, followed by 10 mg or 15 mg of CA weekly with 1,200 mg atezolizumab every 3 weeks (Supplementary Fig. S1).

Patients

Eligible patients were aged 18 years or older, had an Eastern Cooperative Oncology Group Performance Status (ECOG PS) of 0 to 1, had CEA-positive advanced/metastatic solid tumors, and whose disease had progressed on or who were intolerant to the standard-of-care therapy. Patients, except for those with non–small cell lung cancer (NSCLC), were required to have at least one tumor lesion accessible to biopsy. Ineligible patients were those with active or untreated central nervous system metastases and patients with an active second malignancy. Full eligibility criteria and study representation of underserved communities are included in Supplementary Data (Supplementary Methods; Supplementary Table S1, respectively).

CEA-positive tumors were defined as those with 20% or more of tumor cell membrane staining with at least moderate intensity by IHC. CEA assessment was performed locally or centrally, and retrospective central testing was done to confirm the local results (see "Study assessments" section for more details).

This study was conducted in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice guidelines, and the study protocol was approved by the local Ethics Committees/Institutional Review Boards. All patients provided written informed consent prior to receiving treatment.

Study objectives

The primary objectives of the study were to determine the safety and tolerability of CA plus atezolizumab and to identify the MTD for CA plus atezolizumab or an RDE. Secondary objectives were to describe the pharmacokinetic and pharmacodynamic effects and assess the antitumor activity of CA plus atezolizumab. Assessments of biomarkers were exploratory objectives.

Study assessments

Testing to confirm CEA positivity of tumors as part of the eligibility assessments (see “Patients" section within the "Patients and Methods” section) was performed locally or centrally by IHC with a CEA31 mouse monoclonal IgG1 anti-CD66/CEACAM5 antibody (RRID: AB_1158211) on a BenchMark ULTRA staining device with the ultraView DAB detection kit using an in-house validated procedure (20). The percentage of tumor cells with no CEA membrane expression or a CEA membrane staining intensity of level 1+ (weak), 2+ (moderate), or 3+ (strong) was visually assessed by board-certified pathologists. Pathologists were not blinded to the treatment arm when assessing tumor immune phenotype. To reduce intraobserver variability, interpretation training was provided to all readers. Retrospective central testing was done to confirm the local results with the same procedure (see below).

The MTD was defined as a dose with 20% to 30% probability of DLTs. The full definition for a DLT is in the Supplementary Methods. Safety was determined by adverse events (AE), laboratory tests, chest X-rays, vital signs, electrocardiograms, physical examinations, and ECOG PS, as well as by DLTs. AEs were evaluated using NCI Common Terminology Criteria for Adverse Events v4.03 and, regardless of relationship to study treatment, were reported until 30 days after the last dose of study drugs.

Efficacy outcomes included objective response rate [ORR; defined as a complete response (CR) or partial response (PR)], disease control rate [DCR; defined as CR, PR, or stable disease (SD)], and progression-free survival [PFS; defined as the time from randomization to the date of confirmed progressive disease or death, whichever occurred first per Response Evaluation Criteria In Solid Tumors (RECIST) criteria]. Antitumor activity and best overall response (BOR) were evaluated according to RECIST v1.1. Initial tumor assessments were completed upon screening; first computed tomography scans after treatment start were conducted at 8 and 12 weeks, then every 8 weeks for the first year, and every 12 weeks thereafter, until disease progression or treatment discontinuation.

The pharmacokinetic parameters were determined using noncompartmental analysis. The pharmacokinetic outcome measures of CA were AUC, drug concentration (trough concentration, C_min), and maximum drug concentration (peak concentration, C_max). Pharmacodynamic effects with respect to the immune cell count (CD4⁺, CD8⁺ T, and NK cells) in the peripheral blood were determined centrally using flow cytometry at IQVIA (formerly Quintiles and IMS Health). Soluble proteins, such as sCD25 and cytokines, were assessed centrally by Ella (R&D Systems) as markers of peripheral IL2v immune cell activation at Microcoat Biotechnologie. Archival tumor biopsies were requested from all patients. Fresh tumor biopsy samples were mandatory for all enrolled patients except for patients with NSCLC unless required for eligibility testing. On-treatment biopsies were collected on cycle 3 day 8 for the weekly/every-3-week regimen and on cycle 4 day 8 for the every-2-week/every-2-week regimen. Biopsies were centrally analyzed by IHC for CEA expression to determine the densities of different immune cell lineages (CD4⁺, CD8⁺, and NK cells) and to determine the expression of activation markers (Ki67 and PD-L1). IHC was done in single or duplex assays as follows: CEA (CEA31; RRID: AB_1158211) and PD-L1 (SP-142; Ventana Medical Systems) assays were performed and analyzed at Roche Tissue Diagnostics. Central testing of CEA was done with the same procedure as local testing, with cells assessed for membrane staining of any intensity. PD-L1 positivity was assessed as tumor cell score (percentage of PD-L1–positive tumor cells) and immune cell score (percentage of the tumor area covered by PD-L1–positive tumor-infiltrating immune cells). CD3 (2GV6; RRID: AB_2335978), perforin [5B10; recognizing perforin (RRID: AB_2042606)], Ki67 (30-9; RRID: AB_2631262), CD8 (SP239; RRID: AB_2756374), and FOXP3 (236A/E7; (RRID: AB_445284) assays were performed, digitized using a DP200 scanner (Ventana Medical Systems), and analyzed using in-house developed imaging algorithms at Roche Innovation Center Munich (24, 25). IHC positivity was assessed in the viable tumor area. C-reactive protein (CRP) levels and lymphocyte counts were locally assessed.

ADA analyses

Blood samples for the determination of ADAs against CEA-IL2v were analyzed by ELISA with a screening and confirmatory assay using CEA-IL2v as a capture ADA reagent. ADAs were determined at baseline and on day 1 of cycles 1 to 4 prior to the administration of CEA-IL2v. On-treatment ADA samples were taken before dose on cycle 1 day 1 (day 1) and before dose on cycle 4 day 1 (day 43). ADA incidence was defined as an ADA-positive titer at any time. The duration of ADA presence was defined as transient or persistent according to the definitions previously reported by Shankar and colleagues (26). Positive ADA titers were defined as low if they were ≤90.

Statistical analyses

All patients enrolled in this study who received at least one dose of CA plus atezolizumab were included in the safety population and assessment of the MTD, including the five patients who received obinutuzumab before treatment (described in Peters and colleagues; ref. 24). The five patients who received obinutuzumab before treatment were excluded from the efficacy population, which consisted of all other patients who received at least one dose each of CA plus atezolizumab and had at least one tumor assessment after starting treatment. These five patients were excluded from the efficacy population because they had a short follow-up (e.g., the decision to stop treatment occurred shortly after starting) or treatment was interrupted for these patients because of safety reasons.

ORR and DCR are summarized using relative frequencies and 90% confidence limits; PFS is summarized using time-to-event analyses and Kaplan–Meier curves. All patients in the safety population were included in the pharmacodynamic and pharmacokinetic analysis population.

Pharmacodynamic effects in the blood and TME were analyzed using a Wilcoxon matched-pairs signed-rank test.

Results

Patients

Between June 2015 and January 2018, 69 patients whose tumors were CEA positive were enrolled to receive CA plus atezolizumab. During the dose-escalation phase (part I), a total of 24 patients were enrolled and randomized to receive every-2-week/every-2-week dosing of CA plus atezolizumab (840 mg): 6, 10, and 15 mg, n = 5 each; 20 mg, n = 7; and 20/25 mg, n = 2, and 13 patients were enrolled to receive weekly/every-3-week dosing of CA plus atezolizumab (1,200 mg): 10 mg, n = 4; 15 mg, n = 1; and 15/20 mg, n = 8 (Supplementary Fig. S1). Five patients, all treated with the weekly/every-3-week regimen, also received obinutuzumab before treatment (24) during dose escalation of CA: 10 mg (n = 4) and 15 mg (n = 1; Supplementary Fig. S1). Following an early signal of clinical activity during the dose-escalation phase of the weekly/every-3-week regimen, 27 patients [12 patients with NSCLC and 15 patients with pancreatic ductal adenocarcinoma (PDAC)] were enrolled into the dose-expansion phase (part II) to receive 15 mg CA weekly and 1,200 mg atezolizumab every 3 weeks.

Patient baseline characteristics were comparable between the every-2-week/every-2-week and weekly/every-3-week CA plus atezolizumab regimens (Table 1); the median ages were 61 and 59.5 years, respectively, and 95.8% and 85.0% of patients were White. However, there were more male patients receiving the weekly/every-3-week regimen (37.5% vs. 57.5%). Of the 64 patients receiving CA plus atezolizumab, the most common primary tumor types were pancreatic cancer (n = 23, 64%), NSCLC (n = 22, 34%), and colorectal cancer (n = 14, 22%); overall, 50 (78.1%) patients were naïve to ICIs. Among the additional five patients pretreated with obinutuzumab who were included in the safety population, the median age was 59 years, all patients were White, and two patients (40%) were male. The most common tumor type was NSCLC (60.0%). Three patients (60%) receiving the pretreated regimen were naïve to ICIs.

Table 1.

Baseline characteristics.

Characteristic	Obinutuzumab before treatment (n = 5)	CA plus atezolizumab
		Every 2 weeks/every 2 weeks (n = 24)	Weekly/every 3 weeks (n = 40)
Median age, years (range)	59.0 (43–69)	61.0 (28–77)	59.5 (40–79)
≥65, n (%)	2 (40.0)	9 (37.5)	14 (35.0)
Male, n (%)	2 (40.0)	9 (37.5)	23 (57.5)
Ethnicity, n (%)	​
Latino or Hispanic	1 (20.0)	5 (20.8)	12 (30.0)
Not Latino or Hispanic	4 (80.0)	18 (75.0)	24 (60.0)
Unknown/not stated	0	1 (4.2)	4 (10.0)
Primary tumor type, n (%)	​
NSCLC	3 (60.0)	22 (34.4)
Colorectal	0	14 (21.9)
Pancreatic	2 (40.0)	23 (35.9)
Other	0	5 (7.8)
ICI naïve, n (%)	3 (60.0)	52 (81.3)

Abbreviations: CA, cergutuzumab amunaleukin; ICI, immune checkpoint inhibitor; NSCLC, non-small cell lung cancer.

All patients who were enrolled during the dose-escalation or -expansion phase discontinued treatment; the main reason for discontinuation was progressive disease (Supplementary Fig. S1).

DLTs and MTD

During dose escalation, three DLTs were reported in separate patients receiving CA plus atezolizumab: one grade 4 CLS with the 20 mg every-2-week/every-2-week regimen, one grade 3 blood bilirubin increase with the 15 mg weekly/every-3-week regimen, and one grade 3 gamma-glutamyltransferase increase with the 15 mg weekly/every-3-week regimen (Table 2).

Table 2.

Overview of AEs.

AE, n (%)	N = 69a
Total number of patients with at least one AE	69 (100)
Related AEb	69 (100)
Any-cause grade 3/4 AE	66 (95.7)
Related grade 3/4 AEb	43 (62.3)
Any-cause SAE	43 (62.3)
Related serious AEb	27 (39.1)
AE leading to withdrawal from treatment	4 (5.8)
AE leading to dose modification/interruption	29 (42.0)
AE with fatal outcomes	2 (2.9)
DLT	3 (4.3)
Most common (≥10% of patients) grade 3/4 AEs
Aspartate aminotransferase increased	10 (14.5)
Hypophosphatemia	9 (13.0)
Anemia	10 (14.5)
Fatigue	8 (11.6)
IRR	6 (8.7)

Abbreviations: AE, adverse event; DLT, dose-limiting toxicity; IRR, infusion-related reaction; SAE, serious adverse event.

^aIncludes five patients who received obinutuzumab before treatment.

^bAE related to at least one study drug (obinutuzumab, CA, or atezolizumab).

The MTD could not be determined, as it did not meet the statistical requirements (dose with 20%–30% probability of a DLT). However, based on the safety profile of the combination therapy, the DLTs reported, and integration of available clinical data, the RDEs were determined as 20 mg for the every-2-week/every-2-week regimen and 15 mg for the weekly/every-3-week regimen.

Safety

All patients enrolled in the study who received at least one dose of obinutuzumab, CA, or atezolizumab were included in the safety evaluation (n = 69). The median duration of treatment was 43 days for CA plus atezolizumab. AEs generally resolved and were manageable, with few severe AEs reported. There were no new or unexpected safety signals when CA was combined with atezolizumab (Table 2).

All 69 patients in the safety population experienced at least one AE (Table 2); the most frequently reported AEs (in more than 40% of patients) were pyrexia (n = 51, 73.9%), chills (n = 32, 46.4%), fatigue (n = 31, 44.9%), asthenia (n = 28, 40.6%), and infusion-related reactions (IRR; n = 28, 40.6%). All IRRs were grade 1 to 2 and associated with well-known symptoms of IRR such as chills, fever, flu-like symptoms, and pain; some patients additionally experienced nausea, rash or red/flushing face, dyspnea, tachycardia, and hypotension. Most symptoms associated with IRR resolved shortly after onset (minimum: 1 day; maximum: 9 days in one patient). There were no cases of cytokine release syndrome. In total, 100% of patients experienced any-grade drug-related AEs (related to at least one agent: CA, atezolizumab, or obinutuzumab) and 62.3% (n = 43) of patients experienced at least one grade 3/4 drug-related AE. Among those, 50.7% (n = 35) of patients experienced at least one grade 3/4 AE considered related to CA plus atezolizumab and 23.2% (n = 16) experienced at least one grade 3/4 AE considered related to CA only. The most common grade 3/4 AEs occurring in more than 10% of patients were aspartate aminotransferase increase (n = 10, 14.5%), hypophosphatemia (n = 9, 13%), fatigue (n = 8, 11.6%), and anemia (n = 10, 14.5%). No central nervous system–related AEs were reported. One patient experienced grade 4 drug-related CLS.

Overall, 62.3% (n = 43) of patients experienced one or more serious AE (SAE; Table 2), and 39.1% (n = 27) had a drug-related SAE. SAEs that were reported in four or more patients were IRRs (n = 11, 15.9%), pneumonia (n = 6, 8.7%), pyrexia (n = 5, 7.2%), and fatigue (n = 4, 5.8%).

AEs leading to dose modification/interruptions and withdrawal of treatment were seen in 42% (n = 29) and 5.8% (n = 4) of patients, respectively. Fifty-eight patients died; 48 deaths were due to progressive disease and two were due to AEs. Only one AE with a fatal outcome was deemed to be related to treatment; this was due to liver injury with the obinutuzumab pretreatment weekly/every-3-week regimen.

Antitumor activity

The efficacy analysis excluded the five patients who received obinutuzumab before treatment (described in Peters and colleagues; ref. 24). A total of 58 patients from parts I and II (every 2 weeks/every 2 weeks, n = 21; weekly/every 3 weeks, n = 37) were evaluable for antitumor activity; of these 58 patients, 13 were experienced with ICIs (Fig. 1A). The ORR was 4.8% for the every-2-week/every-2-week regimen (CR, n = 1; SD, n = 2; progressive disease, n = 17; and unevaluable, n = 1) and 13.5% for the weekly/every-3-week regimen (PR, n = 5; SD, n = 7; and progressive disease, n = 25). The DCR was 14.3% [90% confidence interval (CI), 5.86–30.87] and 32.4% (90% CI, 21.35–45.91), respectively. Seven patients had a reduction in tumor size for at least 6 months, including one patient with PDAC and one patient with NSCLC who had a reduction for more than 12 months (Supplementary Fig. S2).

Figure 1.

Waterfall plot of response-evaluable patients^a (A) and PFS according to tumor type (B). BOR, best overall response; CA, cergutuzumab amunaleukin; CEA, carcinoembryonic antigen; CI, confidence interval; CR, complete response; CRC, colorectal cancer; IC, immune cell; ICI, immune checkpoint inhibitor; NA, not available; NSCLC, non-small cell lung cancer; PD, progressive disease; PDAC, pancreatic ductal adenocarcinoma; PD-L1, programmed death-ligand 1; PFS, progression-free survival; PR, partial response; SD, stable disease; TC, tumor cell. ^aExcludes patient pretreated with obinutuzumab. CA dose is presented as the planned cycle 1 day 1 dose. CEA and PD-L1 status were determined centrally by IHC on fresh baseline tumor tissues (PD-L1 and CEA) or archival tumor tissues (if fresh baseline tumor tissue was not available; CEA only).

Efficacy-evaluable patients with NSCLC (n = 18) had an ORR of 16.7%. One patient had a CR (20 mg every 2 weeks/every 2 weeks), two patients had a PR (15 mg weekly/every 3 weeks), and three patients achieved SD (20 mg every 2 weeks/every 2 weeks, n = 1; 15 mg weekly/every 2 weeks, n = 2); all other patients with NSCLC (n = 12) had progressive disease (15/20 mg weekly/every 3 weeks, n = 3; 15 mg weekly/every 3 weeks, n = 8; and 10 mg every 2 weeks/every 2 weeks, n = 1; Fig. 1A). The median PFS was 1.9 months [95% CI, 1.6–not available (NA); Fig. 1B]. Thirteen efficacy-evaluable patients with NSCLC had previously received an ICI, including one patient with a PR.

Efficacy-evaluable patients with pancreatic cancer (n = 21) had an ORR of 9.5%; two patients had a PR (10 mg weekly/every 3 weeks and 15 mg weekly/every 3 weeks, n = 1 each), five had SD (10 mg weekly/every 3 weeks, n = 1; 15 mg weekly/every 3 weeks, n = 4), and 14 had progressive disease (20/25 mg every 2 weeks/every 2 weeks, n = 1; 15/20 mg weekly/every 3 weeks, n = 3; and 15 mg weekly/every 3 weeks, n = 10; Fig. 1A). The median PFS was 1.7 months (95% CI, 1.5–3.3; Fig. 1B). All patients with PDAC were naïve to ICI therapy.

The combination of atezolizumab with CA did not seem to provide marked benefit for patients with colorectal cancer (n = 14; median PFS: 2.4 months; 95% CI, 1.9–5.8) or other cancer types (n = 5; median PFS: 1.7 months; 95% CI, 1.6–NA; Fig. 1; Supplementary Fig. S2).

Pharmacokinetics

Pharmacokinetic analyses were conducted in all efficacy-evaluable patients, and characteristics were followed longitudinally within each dose regimen. CA concentrations exhibited nonlinear pharmacokinetics by time and dose (Supplementary Fig. S3). Upon dose 1 of cycle 1, there was an effect of reduced exposure, possibly because of target-mediated drug disposition because of a higher expression of the IL2 receptor.

Overall, more than 80% of patients developed ADAs, with a complete loss of exposure (no measurable concentration) observed in more than 30% of patients during subsequent cycles (Fig. 2). Some patients who experienced a response also developed ADAs, but these cases generally seemed to have lower ADA titers than nonresponders; a direct correlation between ADAs and response could not be determined because of the insufficient numbers. Most ADAs were directed against the CEA-binding domain of the chimeric CA molecule rather than the IL2v domain. The addition of atezolizumab did not alter the pharmacokinetic profile previously observed with monotherapy (Lassen; manuscript in preparation, Melero; manuscript in preparation).

Figure 2.

Antidrug antibodies (ADA) impact on exposure. In patients with the lowest ADA titers, ADA had no impact on exposure [area under the curve (AUC)] or C_max (green boxes). In patients with intermediate ADA titers, ADA had minimal to no impact on exposure or C_max (yellow boxes). For patients with the highest ADA titers, nearly all had a strong reduction in AUC and C_max (red boxes). C1, cycle 1; C4/5, cycle 4/5; C_max, maximum serum concentration.

Pharmacodynamics

In peripheral blood, treatment with CA plus atezolizumab induced significant increases in NK and CD8⁺ T-cell numbers, whereas there were no significant changes in CD4⁺ T-cell numbers, including Tregs (Fig. 3A). Increased levels of NK, CD8⁺, and CD4⁺ T cells were accompanied by a marked increase in their proliferation status as determined by Ki67 (Fig. 3B). Notably, larger increases in NK, CD8⁺, CD4⁺, and Treg numbers in peripheral blood with treatment correlated with a longer PFS (Supplementary Fig. S4). Less prominent changes were observed in the TME, in which the median density of NK and T cells only just doubled following treatment (Supplementary Fig. S5A).

Figure 3.

Box and line plots showing absolute counts of immune cells (NK cells, CD8⁺ T cells, CD4⁺ T cells, and CD4⁺ Tregs; A) and relative counts of proliferating immune cells (NK cells, CD8⁺ T cells, and CD4⁺ T cells; B) in peripheral blood determined by flow cytometry at baseline and following treatment with CA plus atezolizumab per the indicated regimens. Samples were collected prior to dosing on dosing days. Medians are given for all samples plotted; fold changes and significance of the difference between the timepoints indicated by the orthogonal line (black lines and fonts) were calculated from paired samples only. The number of pairs is given underneath the respective P values, which were determined using a Wilcoxon matched-pairs signed-rank test. Color-coding indicates the level (low/medium/high) of the first CA dose received. The significance of the difference between dosing schedules (weekly/every 3 weeks on cycle 3 day 4 vs. every 2 weeks/every 2 weeks on cycle 4 day 4) was determined using a Wilcoxon rank-sum test (blue lines and fonts). Bsl, baseline; CxDy, study cycle (x) day (y); NK, natural killer; ns, not significant; Treg, regulatory T cells.

There was a trend for higher increases in CD8⁺ and CD4⁺ T-cell numbers in peripheral blood in later cycles with the less frequent dosing regimens (every 2 weeks/every 2 weeks), particularly for intermediate and high doses (data not shown). Although the expansion of proliferating NK and T cells in the periphery was similar in cycle 1 for both the weekly/every-3-week and every 2-week/every 2-week regimens, this expansion was more sustained with the every-2-week/every-2-week regimen (Fig. 3B). Conversely, the intratumor inflammatory index [i.e., the ratio of tumor-infiltrating lymphocytes (NK and CD8⁺ T cells) to Tregs (FOXP3+)] was to some extent higher with the weekly/every-3-week regimen than every 2 weeks/every 2 weeks (Supplementary Fig. S5B). Consistently, as per pathologist assessment of paired on-treatment and baseline tumor biopsies, more patients displayed a notable enrichment in their tumor immune phenotype (immune cells moving into the intraepithelial tumor as opposed to being restricted to the tumor stroma). This was more frequently observed with CA dosing weekly/every 3 weeks (n = 5/7 patients compared with n = 0/8 for every 2 weeks/every 2 weeks). Note that of the 15 evaluable paired biopsies, 14 were collected from the same location and tissue longitudinally. Representative microscopy images are shown in Supplementary Fig. S6.

In line with the systemic inflammatory response, CRP levels were strongly increased following treatment and remained elevated for about 1 week after treatment (Supplementary Fig. S7A). Similarly, strong increases in sCD25 and cytokines in plasma during treatment indicated systemic activation of the immune system (Fig. 4A). After 2 weeks of treatment on the every-2-week/every-2-week regimen, sCD25 levels returned to baseline, although they remained elevated with the weekly/every-3-week regimen (Fig. 4A). Subsequent predosing data for both schedules showed the cytokine levels returning to baseline within 1 week. Weekly/every-3-week dosing resulted in accumulated subsequent increases in cytokine levels compared with every-2-week/every-2-week dosing (Fig. 4). Additionally, systemic pharmacodynamic effects as exemplified by CRP levels and lymphocyte count increases were sustained despite high levels of ADAs (Supplementary Fig. S7).

Figure 4.

Levels of soluble proteins in the peripheral blood (plasma) determined by Ella at baseline and following treatment with CA plus atezolizumab per the indicated regimens. A–E, respectively, represent sCD25, TNFα, IFNγ, IL6, and IL10. Samples were collected prior to dosing on dosing days. Medians are given for all samples plotted; the significance of the difference between time points indicated by the orthogonal line was calculated from paired samples only (the number of pairs is given underneath the orthogonal line) using a Wilcoxon matched-pairs signed-rank test (black lines and fonts). Color-coding indicates the level (low/medium/high) of the first CA dose received. The significance of the difference between dosing schedules (weekly/every 3 weeks on cycle 3 day 2 vs. every 2 weeks/every 2 weeks on cycle 4 day 2) was determined using a Wilcoxon rank-sum test (blue lines and fonts). Bsl, baseline; CA, cergutuzumab amunaleukin; CxDy, study cycle[x] day[y]; ns, not significant.

Among all patient samples tested in this study, only samples from two patients (one with PR and one with progressive disease as BOR) were high in microsatellite instability (MSI; data not shown). Notably, of the patients with an evaluable PD-L1 status, none of the patients who responded (CR or PR as BOR) had PD-L1 scores equal to or greater than 1 at baseline. Two of five (40%) patients with SD and seven of 29 (24%) patients with progressive disease had PD-L1 scores equal to or greater than 1 at baseline (Fig. 1A). There was an increase in tumor-infiltrating PD-L1–positive immune cell levels following treatment, but this was not significant (Supplementary Fig. S8). CEA expression did not change in response to treatment (Supplementary Fig. S9), and there was a trend for poorer PFS outcomes in patients with higher CEA expression at baseline (Fig. 1A CEA heat map).

Discussion

CA was designed to overcome the limitations of wild-type IL2 and enhance antitumor activity by favoring the expansion of cytotoxic T cells rather than Tregs (18). CEA and its wider family of CEA-related cell adhesion molecules are highly expressed across multiple solid tumor types (25), making them ideal candidates for linking with cytotoxic agents to direct their antitumor potency. Several CEA-targeting approaches have been tested in clinical trials (Melero; manuscript in preparation; refs. 27–30), with more in preclinical development (31, 32). For example, anti-CEA chimeric antigen receptor T-cell (CAR-T) therapy has been investigated in patients with solid tumors, namely, colorectal cancer; however, efficacy was limited (33), and CEA-directed CAR-T therapy may carry risks of severe AEs such as refractory colitis (34). Thus far, no treatments targeting CEA have been approved, suggesting that combination with additional agents such as ICIs may be needed to improve success. Indeed, previous studies have suggested that inhibiting PD-L1, such as with atezolizumab, may work synergistically with CA to overcome an adaptive resistance mechanism and prolong survival for patients (Melero; manuscript in preparation; ref. 18). A similar approach combining the CEA-CD3 epsilon chain bispecific cibisatamab with atezolizumab has been evaluated in a phase I study, with the combination warranting further examination in patients with microsatellite-stable colorectal cancer (35).

In this study, the MTD of CA plus atezolizumab could not be determined. Considering our experiences with CA monotherapy (Lassen; manuscript in preparation, Melero; manuscript in preparation) and the similarities of the safety profiles between monotherapy and combination therapy with atezolizumab, the dose was not escalated until stopping criteria were met to minimize the risk to patients. Nevertheless, the RDE was determined at 20 mg for the every-2-week/every-2-week regimen and 15 mg for the weekly/every-3-week regimen.

In this phase Ib dose-escalation study in patients with CEA-positive advanced/metastatic solid tumors, CA plus atezolizumab had a manageable safety profile at the tested doses; no central nervous system–related AEs were reported, and there was only one report of treatment-related CLS. This suggests that CA plus atezolizumab may have a safety profile that is improved and favorable, in terms of skin, cardiovascular, and respiratory AEs, compared with aldesleukin (7). The overall safety profile of CA plus atezolizumab in this study is similar to that observed in the CA monotherapy study, in which 100% of patients experienced at least one treatment-related AE, and 75% of patients experienced at least one grade 3/4 AE (Lassen; manuscript in preparation). The atezolizumab-mediated AEs reported in our study were also consistent with the known safety profile of atezolizumab-based regimens (36–38).

The treatment and management of advanced solid tumors remains challenging despite the development of new therapeutic agents. IL2 has been used alone or in combination with other cancer therapies and has been shown to induce durable responses (11, 16, 39). Limited clinical activity was seen in patients in this study, including in patients with colorectal cancer. This was similar to a CA monotherapy study in which the ORR was 0% in the intention-to-treat population (with a population primarily consisting of patients with colorectal cancer; Lassen; manuscript in preparation). The activity of ICIs is commonly associated with PD-L1 expression in many tumor types, but in this study, we observed responses only in patients with PD-L1–negative disease: no patients achieving a CR or PR had PD-L1–positive tumors. Previously, no obvious relationship between antitumor activity and PD-L1 status had been observed with the novel immunocytokine FAP-IL2v as monotherapy (40) or in combination with atezolizumab (38). One possible conclusion from our data is that CA may be capable of enabling or enhancing an immune response to ICI in some patients with noninflamed tumors. However, the overall efficacy profile of atezolizumab plus CA was similar to that observed with CA monotherapy previously (Lassen; manuscript in preparation), suggesting that the addition of atezolizumab to CA was not sufficient to overcome any immunosuppressive mechanisms in most patients as had been hypothesized based on the observed upregulation of PD-L1 in the monotherapy study (Lassen; manuscript in preparation, Melero; manuscript in preparation).

More than 80% of patients developed ADAs, with a loss of exposure seen in more than 30% of patients during later cycles. This may be due to ADAs having a higher impact on inflammation in the TME than in the peripheral blood, resulting in reduced intratumoral CA exposure and dampened pharmacodynamic effects (26, 41). Peters and colleagues (24) examined B-cell depletion as a strategy to mitigate CA formation of ADAs and demonstrated that obinutuzumab pretreatment before CA administration in patients with CEA-positive solid tumors is a feasible, safe, and potent ADA mitigation strategy (24), but the triplet therapy was not further pursued. However, this ADA risk mitigation approach could not be transferred to the CA plus atezolizumab combination. In the subgroup of obinutuzumab pretreated patients receiving the combination treatment, one of five patients experienced hepatotoxicity with fatal outcomes, which was assessed as related to the triplet therapy (24). Similar to results in our FAP-IL2v plus atezolizumab combination study (38), the present study (CA plus atezolizumab) did not lead to a significant correlation between ADA positivity and clinical response.

We report that treatment with CA plus atezolizumab induced strong expansion and proliferation of NK and CD8⁺ T cells, with no significant changes in increasing Treg counts, which highlights the benefits of IL2v over the properties of aldesleukin (7). More frequent dosing of CA (weekly) led to a significant, but apparently less sustainable, expansion of proliferating NK and T cells. This could be due to either the differentiation of more exhausted phenotypes or more frequent dosing leading to increased NK- and T-cell infiltration into the tumor and thereby reducing the frequency of proliferating cells in the periphery. This observation is in line with the improved immune phenotypes observed for the weekly/every-3-week regimen and with the better ORR for the weekly/every-3-week regimen compared with the every-2-week/every-2-week regimen (13.5% vs. 4.8%, respectively). Key cytokines showed an initial increase following treatment, and steady increases in sCD25 also suggested systemic activation.

The observed changes in immune profile were less prominent in the TME than in the periphery. The IL2v component in CA is constitutively active and does not require CA binding to CEA for binding to and activation of its receptor. The immune infiltrate profile seems to be less affected by the relative advantage of CA accumulation in the tumor via CEA binding than by the immunosuppressive TME. Conversely, CA in the bloodstream can activate immune cells in the absence of such immunosuppression.

Higher baseline CEA positivity in the tumor continued to show a trend toward a shorter PFS, indicating that treatment with CA plus atezolizumab could not overcome the worse prognosis associated with higher CEA levels (42–47).

Data collected from patients in this study showed limited clinical activity based on RECIST criteria, at least in this heavily pretreated series of patients; therefore, the study sponsor decided to discontinue the development of CA. The decision to discontinue development was not related to safety concerns. Limitations of this study include the relatively small sample size and the fact that the therapeutic activity may have been influenced by cancer type and prior therapies. With respect to mismatch repair, there were too few MSI-high patients to draw any meaningful conclusions from this study. Further investigation in patients with defined tumor types (e.g., MSI-high colorectal cancer and other tumors) may be warranted. Similarly, resistance to ICIs is a complex process; the addition of IL2 may not have been sufficient to overcome multiple immunotherapy-related resistance mechanisms.

Conclusions

CA in combination with atezolizumab was manageable, and the safety profile was considered acceptable and consistent with IL2v and atezolizumab monotherapy in patients with locally advanced and/or metastatic CEA-positive solid tumors. However, CA plus atezolizumab showed limited antitumor activity in this population. Further research may be warranted to identify other therapies that synergize with ICIs or other immunotherapies and elicit antitumor activity in patients with CEA-positive advanced/metastatic solid tumors.

Supplementary Material

Supplementary Figure 1

Fig S1

Supplementary Figure 2

Fig S2

Supplementary Figure 3

Fig S3

Supplementary Figure 4

Fig S4

Supplementary Figure 5

Fig S5

Supplementary Figure 6

Fig S6

Supplementary Figure 7

Fig S7

Figure S8

Supplementary Figure 8

Figure S9

Supplementary Figure 9

Supplementary Methods 1

Supplementary Methods

Supplementary Table 1

Table S1

Data Availability

Qualified researchers may request access to individual patient-level data through the clinical study data request platform (https://vivli.org/ourmember/roche/). Further details on F. Hoffmann-La Roche Ltd’s criteria for eligible studies are available at https://vivli.org/members/ourmembers/. For further details on F. Hoffmann-La Roche Ltd’s Global Policy on the Sharing of Clinical Information and how to request access to related clinical study documents, see https://www.roche.com/innovation/process/clinical-trials/data-sharing.

Authors’ Disclosures

I. Melero reports grants and personal fees from Roche during the conduct of the study, as well as grants and personal fees from Bristol Myers Squibb, AstraZeneca, Genmab, and PharmaMar and personal fees from Curon Biopharmaceutical, F-star Therapeutics, Catalym, Boehringer Ingelheim, HotSpot Therapeutics, Highlight therapeutics, Bright Peak Therapeutics, Pioneer Pharmaceutical, and BioNTech outside the submitted work. J. Tabernero reports other support from Accent Therapeutics, Alentis Therapeutics, AstraZeneca, Boehringer Ingelheim, Bristol Myers Squibb, Carina Biotech, Cartography Biosciences, Chugai Pharmaceutical, Daiichi Sankyo, F. Hoffmann-La Roche, Genentech, Johnson & Johnson/Janssen Pharmaceuticals, Eli Lilly and Company, Marengo Therapeutics, Menarini Group, Merus, MSD, Novartis, ONO PHARMA USA, Peptomyc, Pfizer, Pierre Fabre, QUANTRO Therapeutics, Scandion Oncology, Scorpion Therapeutics, Servier, SOTIO Biotech, syntelios AG, Taiho Pharmaceutical, Takeda Oncology, TOLREMO Therapeutics, Oniria Therapeutics, 1 TRIALSP, and Pangaea Oncology outside the submitted work. N. Steeghs reports nonfinancial support from multiple entities during the conduct of the study, as well as grants, personal fees, and nonfinancial support from multiple entities outside the submitted work. D.G.J. Robbrecht reports other support from Johnson & Johnson, MSD, Merck AG, and AstraZeneca outside the submitted work. S. Peters reports personal fees from AbbVie, Amgen, Arcus Biosciences, AstraZeneca, Bayer, BeiGene, BioNTech, BerGenBio, Bicycle Therapeutics, Biocartis, BioInvent, Blueprint Medicines, Boehringer Ingelheim, Bristol Myers Squibb, Clovis Oncology, Daiichi Sankyo, Debiopharm, Eli Lilly and company, F-star Therapeutics, Foundation Medicine, Genmab, Genzyme, Gilead Sciences, GSK, HUTCHMES, Illumina, Incyte, Ipsen, iTeos Therapeutics, Janssen Pharmaceuticals, Qlucore, Merck Sharp and Dohme, Merck Serono, Nuvation Bio, Nuvalent, Nykode Therapeutics, Novartis, Novocure, PharmaMar, Promontory Therapeutics, Pfizer, Regeneron Pharmaceuticals, Roche/Genentech, Sanofi, Takeda, Zymeworks, and Mirati Therapeutics outside the submitted work. E.M. O’Reilly reports other support (research funding to institution) from Genentech/Roche, BioNTech, AstraZeneca, Arcus Biosciences, Elicio Therapeutics, Parker Institute, NIH/NCI, Digestive Care, Break Through Cancer, Agenus, Amgen, and Revolution Medicines; other support (consultant/data safety monitoring board, uncompensated) from Arcus, Amgen, AstraZeneca, Ability Pharma, Alligator BioSciences , Pfizer, Agenus, BioNTech, Ipsen, Ikena Oncology, Merck, Immuneering Corporation, MOMA Therapeutics, Novartis, Astellas Pharma, Bristol Myers Squibb, Revolution Medicines, Regeneron Pharmaceuticals, and Tango Therapeutics; other support (travel) from BioNTech and Arcus; and other support from American Association for Cancer Research, American Society of Clinical Oncology, iMedx, Research To Practice, Stand Up To Cancer, and NIH/NCI Cancer Center Support Grant/Core Grant P30 CA008748 NCI/NIH P50 CA257881-01A1. E. Calvo reports employment with START, HM Hospitales; other support (consulting or advisory role) from Adcendo, Amunix Pharmaceuticals, ANAVEON, AstraZeneca/MedImmune, Bristol Myers Squibb, Chugai Pharmaceutical, Diaccurate, Elevation Oncology, Ellipses Pharma, Genmab, Janssen Pharmaceuticals/Cilag, MonTa Biosciences, MSD Oncology, Nanobiotix, Nouscom, Novartis, OncoDNA, PharmaMar, Roche/Genentech, Servier, TargImmune Therapeutics, T-knife Therapeutics, and Syneos Health; other support (leadership role) from BeiGene, European Organisation for Research and Treatment of Cancer, Merus NV, Novartis, PharmaMar, Sanofi, START, CRIS Cancer Foundation, Foundation PharmaMar, and Investigational Therapeutics in Oncological Sciences; and ownership interests and ownership of stocks in HM Hospitales, Oncoart Associated, and START. A. Cervantes reports grants from Genentech, Merck Serono, MSD, Bayer, Servier, Elly Lilly and Company, Natera, Novartis, Takeda, and Átelas Pharma and grants and personal fees from Roche and AbbVie outside the submitted work. N. Hafez reports grants from Genentech outside the submitted work. C. Habigt reports nonfinancial support from Ashfield MedComms GmbH during the conduct of the study, as well as personal fees from Roche Diagnostics GmbH outside the submitted work. D. Dejardin reports personal fees from Roche and other support from Roche outside the submitted work. I. Martinez Quetglas reports other support from F. Hoffmann-La Roche Ltd outside the submitted work. J. Duarte reports personal fees from Roche outside the submitted work, as well as ownership of Roche shares. C. Adessi reports personal fees from F. Hoffmann-La Roche during the conduct of the study, as well as personal fees from F. Hoffmann-La Roche outside the submitted work. C. Boetsch reports personal fees from F. Hoffmann-La Roche during the conduct of the study, personal fees from F. Hoffmann-La Roche outside the submitted work, and employment with F. Hoffmann-La Roche and ownership of F. Hoffmann-La Roche stock. S. Evers reports personal fees, nonfinancial support, and other support from F. Hoffmann-La Roche AG outside the submitted work. J. Charo reports personal fees from Roche during the conduct of the study, personal fees from Roche outside the submitted work, a patent for EPO. 22207100.3 pending, and ownership of Roche stock. V. Teichgräber reports other support from Roche during the conduct of the study, as well as other support from Roche outside the submitted work. U. Lassen reports grants from Roche during the conduct of the study, as well as grants from Bristol Myers Squibb, Eli Lilly and Company, Pfizer, Incyte, Janssen Pharmaceuticals, GSK, and Novartis outside the submitted work. No disclosures were reported by the other authors.

Authors’ Contributions

I. Melero: Investigation, writing–review and editing. J. Tabernero: Investigation, writing–review and editing. N. Steeghs: Investigation, writing–review and editing. D.G.J. Robbrecht: Investigation, writing–review and editing. S. Peters: Investigation, writing–review and editing. N.A. Rizvi: Investigation, writing–review and editing. E.M. O’Reilly: Investigation, writing–review and editing. E. Calvo: Investigation, writing–review and editing. R.L. Eefsen: Investigation, writing–review and editing. N. Leighl: Investigation, writing–review and editing. A. Cervantes: Investigation, writing–review and editing. N. Hafez: Investigation, writing–review and editing. C. Habigt: Conceptualization, formal analysis, writing–review and editing. E. Andersson: Conceptualization, formal analysis, writing–review and editing. D. Dejardin: Conceptualization, formal analysis, writing–review and editing. E. Rossmann: Conceptualization, formal analysis, writing–review and editing. I. Martinez Quetglas: Conceptualization, formal analysis, writing–review and editing. G. Babitzki: Conceptualization, formal analysis, writing–review and editing. J. Duarte: Conceptualization, formal analysis, writing–review and editing. C. Adessi: Conceptualization, formal analysis, writing–review and editing. C. Boetsch: Conceptualization, formal analysis, writing–review and editing. S. Evers: Conceptualization, formal analysis, writing–review and editing. J. Charo: Conceptualization, formal analysis, writing–review and editing. V. Teichgräber: Conceptualization, formal analysis, writing–review and editing. U. Lassen: Formal analysis, investigation, writing–review and editing.