First-in-human study of lomvastomig, a PD-1-TIM-3 bispecific antibody, in patients with advanced and/or metastatic solid tumors Authors: Kristoffer Staal Rohrberg, Ignacio Melero, Miguel F Sanmamed, Eva Muñoz-Couselo, Andrés Cervantes, Valentina Gambardella, Laurent Greillier, Pilar Garrido, Hye Ryun Kim, Dae Ho Lee, Antoine Italiano, Philippe Alexandre Cassier, Sandrine Hiret, Sanjeev Deva, Tae Min Kim, George R Blumenschein, Rikke Løvendahl Eefsen, Brian S Henick, Mario Perro, Pablo Umaña, Antonio Peixoto, Laura Lauener, Patrick Weber, Stefan Seeber, Christian Klein, Brian Higgins, Micol Guidi, Dominik Kraus, Merlind Mücke, Christian Heichinger, Vu-Long Tran, Daniela Schmid, Francesca Michielin, Ting Liu, Christine McIntyre, Henry Kao, Laura Codarri Deak, Christoph Markert, Victor Moreno Published: 2026 ABSTRACT AbstractBackgroundThis first-in-human clinical study explored lomvastomig, an immunoglobulin G1-based Fc-silenced bispecific antibody that simultaneously blocks the immune checkpoint receptors programmed cell death protein 1 (PD-1) and T-cell immunoglobulin domain and mucin domain-3.MethodsLomvastomig was characterized in cell cultures and preclinically in cancer mouse models. The phase 1, open-label, multicenter clinical study of lomvastomig included a dose-escalation part in patients with advanced and/or metastatic solid tumors and an expansion part with four tumor-specific cohorts, which enrolled checkpoint inhibitor (CPI)-experienced patients with melanoma and non-small-cell lung cancer (NSCLC) and CPI-naïve patients with SCLC and esophageal squamous cell carcinoma (ESCC). Primary and secondary objectives included safety/tolerability, maximum tolerated dose (MTD)/recommended dose for expansion (RDE), pharmacokinetics, drug receptor occupancy, and antitumor activity.Results39 and 95 patients were enrolled in the dose-escalation and expansion parts, respectively. Lomvastomig was well tolerated up to the highest tested dose of 2,100 mg every 2 weeks (Q2W). One dose-limiting toxicity was reported at 1,200 mg (grade 3 troponin T increase). No MTD was reached, and 2,100 mg Q2W was established as the RDE. Linear pharmacokinetics across the studied dose range suggested target saturation. Peripheral blood drug receptor occupancy on CD3+ and CD8+ was saturated at >90% throughout treatment for doses ≥70 mg. Objective responses were observed at 2,100 mg lomvastomig during dose-escalation (21%; n=19), and in the CPI-experienced melanoma (8%, n=38) and CPI-naïve ESCC (20%, n=15) expansion cohorts.ConclusionsLomvastomig had a tolerable and manageable safety profile at 2,100 mg Q2W. Clinical activity was limited in CPI-experienced patients with melanoma and NSCLC, while an encouraging signal was observed in CPI-naïve patients with ESCC.Trial registration numberNCT03708328 (registration date: 2018–10-09).Keywords: Immune Checkpoint Inhibitor, Immune modulatory, Immunotherapy, Skin Cancer, Lung Cancer WHAT IS ALREADY KNOWN ON THIS TOPIC Despite the success of therapies with immune checkpoint inhibitors (CPI), many patients still fail to respond or experience disease progression. Despite the success of therapies with immune checkpoint inhibitors (CPI), many patients still fail to respond or experience disease progression. We report nonclinical and initial clinical results for lomvastomig, an immunoglobulin G1-based, Fc-silenced bispecific antibody that blocks both programmed cell death protein 1 (PD-1) and T-cell immunoglobulin domain and mucin domain-3 (TIM-3). We report nonclinical and initial clinical results for lomvastomig, an immunoglobulin G1-based, Fc-silenced bispecific antibody that blocks both programmed cell death protein 1 (PD-1) and T-cell immunoglobulin domain and mucin domain-3 (TIM-3). WHAT THIS STUDY ADDS Lomvastomig had a tolerable and manageable safety profile in patients with advanced solid tumors. Lomvastomig had a tolerable and manageable safety profile in patients with advanced solid tumors. Consistent with results from clinical studies of other PD-1/TIM-3-based treatments, the overall clinical activity was low in the mostly heavily pretreated patient population included in this study. Consistent with results from clinical studies of other PD-1/TIM-3-based treatments, the overall clinical activity was low in the mostly heavily pretreated patient population included in this study. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY An encouraging signal in CPI-naïve patients with esophageal squamous cell carcinoma (ESCC) led to the initiation of a phase 2 study in patients with advanced or metastatic ESCC who are refractory or intolerant to fluoropyrimidine-based or taxane-based and platinum-based regimens (NCT04785820). An encouraging signal in CPI-naïve patients with esophageal squamous cell carcinoma (ESCC) led to the initiation of a phase 2 study in patients with advanced or metastatic ESCC who are refractory or intolerant to fluoropyrimidine-based or taxane-based and platinum-based regimens (NCT04785820). Introduction Immunotherapy with checkpoint inhibitors (CPIs), such as those targeting programmed death receptor-1 (PD-1)/programmed cell death ligand 1 (PD-L1), has emerged as a successful strategy for the treatment of various types of cancer, either as monotherapy or in various disease-specific combinations.1,3 However, treatment success is primarily limited to a subset of patients with inflamed tumor phenotypes, and primary or acquired resistance mechanisms often prevent clinically meaningful responses.4 T-cell immunoglobulin domain and mucin domain-3 (TIM-3), an inhibitory receptor involved in the regulation of innate and adaptive immune responses, is expressed on myeloid cells, natural killer cells, and dysfunctional T cells, often coexpressed with PD-1.5 6 Upregulated TIM-3 expression on T cells is associated with T-cell dysfunction, which potentially results in baseline or acquired resistance to anti-PD-1/PD-L1 immunotherapies.7,9 It is therefore conceivable that TIM-3 blockade could overcome or prevent resistance mechanisms to anti-PD-1/PD-L1 and help restore cytotoxic T-cell proliferation and enhance T-cell function. In animal models, dual blockade of TIM-3 and PD-1 restored T-cell function and suppressed tumor growth more effectively than targeting either pathway alone, thus supporting combination therapy.10 Phase 1 and 2 studies with various combinations of anti-PD-1 and anti-TIM-3 antibodies have shown a tolerable and manageable safety profile across several cancer indications.11 Based on data from early clinical investigations, no new safety concerns have been identified for bispecific antibodies targeting PD-1 and TIM-3, including AZD7789 in patients with relapsed or refractory classical Hodgkin lymphoma, AZD7789 in patients with stage IIIB–IV non-small-cell lung cancer (NSCLC), or LB1410 in patients with various advanced solid tumors.12,14 Here, we present results from the preclinical and clinical investigation of lomvastomig (RO7121661), a novel immunoglobulin G1 (IgG1)-based, Fc-silent bispecific antibody that simultaneously blocks two immune checkpoint receptors, PD-1 and TIM-3, when co-expressed on activated T cells.15 Preclinical investigations aimed to evaluate the activity and mode of action of lomvastomig in cultures of immune cells and in tumor-bearing knock-in mouse models for the targeted human genes. The subsequent clinical investigations were conducted as a first-in-human, phase 1 dose-escalation and expansion study to evaluate the safety, pharmacokinetics (PK), pharmacodynamics, immunogenicity, and preliminary antitumor activity of lomvastomig as a single agent in patients with advanced/metastatic solid tumors. Materials and methods Preclinical methods Blood samples from healthy volunteers were recruited via the blood donation center, Zurich, Switzerland, with approval of the cantonal ethics committee, Zurich. Peripheral blood samples from cynomolgus monkeys were collected at the Roche Innovation Center Basel. Peripheral blood mononuclear cell (PBMC) samples from patients with melanoma were obtained in the context of a collaboration between Roche and the Department of Dermatology, Zurich University Hospital. The binding properties of lomvastomig to PD-1 and TIM-3 were assessed in a binding competition assay with bead-sorted monocytes (CD14 positive) and activated CD4 T cells isolated from the peripheral blood of healthy donors that were co-cultured at a 1:1 ratio before being exposed for 30 min at 4°C to increasing concentrations of AF-647-directly conjugated high-affinity or low-affinity anti-TIM-3 antibodies or lomvastomig. After a washing step, the cells were stained for CD3 (BD Biosciences), CD4 (BioLegend) and CD14 (BD Biosciences) with the antibody mix listed in online supplemental table S2. To assess the internalization of lomvastomig, activated CD4 T cells were exposed for up to 3 hours to directly conjugated parental anti-PD-1, parental anti-TIM-3, or lomvastomig, and monitored via confocal microscopy for internalization. The intensity of the fluorescent signal from the membrane of targeted cells was divided by the intensity of the fluorescent signal from the cytoplasm of the same cell, resulting in a ratio. Post image analysis was performed using Imaris software. For the drug-shaving assay, 3 days activated CD8 T cells were directly labeled with a membrane dye and pre-incubated with saturating concentrations of directly conjugated therapeutic antibodies (pembrolizumab, nivolumab, parental anti-PD-1, and lomvastomig; online supplemental table S4) before being co-cultured with macrophages. Time-lapse imaging was performed to visualize the drug-shaving action of the macrophages using confocal microscopy. Functional assays were conducted with PBMCs from patients containing detectable frequencies of tumor-specific CD8+T cells, which were re-stimulated overnight with antigen-spanning peptide pools derived from melanoma tumor-associated antigens (MAGEA1, MAGEA3, MAGEA4, Melan-A/MART-1, melanocyte protein Pmel 17 gp100, tyrosinase, and tyrosinase-related protein 2; JPT, PepMix Collection Melanoma) in the presence of lomvastomig vs the parental anti-PD-1 antibody alone or in combination with the parental anti-TIM-3 antibody. As readout, intracellular granzyme B and interferon (IFN)-γ were assessed via flow cytometry after staining the cells with the antibody mix listed in online supplemental table S5. For the preclinical assessment of efficacy, huPD-1 and huTIM-3 double-transgenic knock-in mice (Genoway) were challenged subcutaneously with the murine colon adenocarcinoma cell line MC38. On tumor engraftment, the mice received an intravenous induction dose of 10 mg/kg followed by intraperitoneal maintenance treatment with 5 mg/kg of either lomvastomig or the parental anti-PD-1 antibody, both given every 3 days. A group of mice was left untreated as a control group. A detailed description of the methods is provided in the online supplemental material and methods S1. Clinical study design and objectives The phase 1, open-label, multicenter, first-in-human clinical study of lomvastomig (NP40435; NCT03708328) included a dose-escalation part in patients with advanced and/or metastatic solid tumors and an expansion part with four tumor-specific cohorts, which enrolled patients with advanced melanoma, NSCLC, SCLC, or esophageal squamous cell carcinoma (ESCC) (for details see study population below). The primary objective of dose-escalation was to evaluate the safety, tolerability, and maximum tolerated dose (MTD) and/or the recommended dose for expansion (RDE) for lomvastomig; the secondary objectives included PK, drug receptor occupancy, anti-drug antibodies (ADAs), and preliminary clinical activity. The primary objective of the expansion was to evaluate the antitumor activity of lomvastomig; secondary objectives included safety, PK, ADAs, and biomarkers. Treatment and dose-escalation Patients were treated with an intravenous infusion of lomvastomig administered once every 2 weeks (Q2W) for up to 2 years or until disease progression, unacceptable toxicity, or withdrawal of consent. Patients could remain on treatment if they were still benefiting from the treatment at the end of the 2-year period. For dose-escalation, a flat dose of 70 mg lomvastomig Q2W was selected as the starting dose based on data from non-clinical toxicology and PK studies, and in vitro/ex vivo functional assays. The dose for the tumor-specific expansion part was derived based on the safety, PK, and/or drug receptor occupancy, and/or biomarker profile observed in the dose-escalation part. Dose-escalation was carried out according to a modified continual reassessment method with escalation with overdose control (EWOC) design and was based on the occurrence of dose-limiting toxicities (DLTs). The adverse events (AEs) that were defined as a DLT if assessed by the investigator as related to lomvastomig and occurring during the 21-day DLT period are listed in online supplemental material and methods S2. Study population Patients aged ≥18 years were eligible to enroll in the dose-escalation part if they had histologically or cytologically confirmed advanced and/or metastatic solid tumors who had progressed on a cancer therapy, for whom no effective standard therapy existed, or who declined treatment with approved therapies. The tumor-specific dose-expansion cohorts enrolled patients with histologically confirmed advanced metastatic melanoma previously treated with approved anti-PD-L1/anti-PD-1 agents with or without approved anti-cytotoxic T-lymphocyte associated protein 4 (CTLA-4) therapy and up to one additional treatment regimen; patients with histologically confirmed advanced NSCLC previously treated with approved PD-L1/PD-1 inhibitors and platinum-based chemotherapy, either sequentially or concurrently; patients with histologically confirmed SCLC who may have had prior chemotherapy, radiation therapy, or declined approved therapies for SCLC; and patients whose major lesion was histologically confirmed as squamous cell carcinoma or adenosquamous cell carcinoma of the esophagus and who had previously received no more than one prior line of treatment for metastatic disease, which had to be a fluoropyrimidine and platinum-based regimen. All patients had to have radiologically measurable disease according to Response Evaluation Criteria in Solid Tumors (RECIST) V.1.1; Eastern Cooperative Oncology Group performance score 0 or 1; and a life expectancy of ≥12 weeks. Patients were excluded if they had known symptomatic central nervous system metastases; an active second malignancy; active or history of carcinomatous meningitis/leptomeningeal disease; evidence of significant, uncontrolled concomitant diseases, including diabetes mellitus, history of relevant pulmonary disorders, and known autoimmune diseases or immune deficiency, or other diseases with ongoing fibrosis; an active, or uncontrolled infection; a history of myocarditis; or if they were receiving immunomodulating agents or systemic immunosuppressive medications. A complete list of eligibility criteria is provided in online supplemental material and methods S2. Assessments Safety was monitored through physical examinations and assessments of vital signs, clinical laboratory parameters, and electrocardiograms. All AEs, regardless of their relationship to study treatment, were reported from the initiation of study treatment until 2 months (dose-escalation) or 3 months (dose-expansion) after the last dose of study treatment. The severity of AEs was graded according to National Cancer Institute Common Terminology Criteria for Adverse Events (NCI CTCAE) V.5.0. Biomarker assessments were performed using fresh or archival tumor biopsies. Further details are provided in online supplemental material and methods S2. PK analyses were carried out on serum samples using an ELISA. Receptor occupancy was assessed by flow cytometry through an anti-drug approach using a PE-labeled anti-PGLALA antibody that specifically binds to the P329G mutation present in lomvastomig as described in.16 Further details are provided in online supplemental material and methods S2. Measurable lesions were assessed using CT scans (or MRI) of the chest, abdomen, and pelvis up to 28 days before and every 8 weeks after the start of treatment during the first year, and every 12 weeks thereafter until disease progression or treatment discontinuation. Tumor response was evaluated according to RECIST V.1.1. Treatment beyond initial apparent progressive disease according to RECIST V.1.1 was allowed under defined criteria. Outcome measures included objective response rate (ORR) defined as confirmed complete response (CR) or partial response (PR), and disease control rate (defined as confirmed CR or PR or stable disease). Statistical analysis No formal sample size calculation was done for the dose-escalation and expansion parts of this study. Approximately 60 evaluable participants were planned to be enrolled for lomvastomig Q2W dose-escalation, with an additional 10–20 patients if confirmation of safety, PK, receptor occupancy, and/or biomarkers was needed prior to opening the expansion cohorts. Up to approximately 200 patients were planned to be enrolled in the expansion cohorts. The safety analysis included all patients who received at least one dose of study treatment. Patients were evaluable for antitumor activity if they received at least one dose of study treatment and had at least one post-baseline tumor assessment or had clinical progression before the first tumor assessment. Statistical analyses were performed using SAS V.9.4. PK parameters for lomvastomig were determined via non-compartmental analysis using Phoenix V.8.3.4. Results Preclinical characterization of lomvastomig Lomvastomig is an Fc-silenced IgG1 bispecific antibody with a 1+1 format that binds with a monovalent high-affinity arm to the human PD-1 receptor (250 pM) and with a monovalent relatively low-affinity arm to the human TIM-3 receptor (132 nM) (figure 1A). Knob-into-hole mutations and CrossMab technology prevent chain mispairing.17 On PD-1, it recognizes a unique glycopeptide epitope that largely overlaps with the binding sites of nivolumab and pembrolizumab, thereby interfering with PD-L1 and PD-L2 interactions. The roughly 500-fold difference in affinity between the PD-1 and TIM-3 binding arms is intended to achieve an avidity-driven selectivity binding of tumor-specific T cells that co-express both receptors. While a bivalent high-affinity TIM-3 antibody bound equivalently to monocytes and activated CD4 T cells (EC50: 104 vs 60 pM), lomvastomig showed a 13-fold reduction in binding to TIM-3+monocytes compared with activated CD4 T cells (EC50: 3023 vs 239 pM) (figure 1B). The bivalent parental anti-TIM-3 antibody bound to monocytes with a threefold lower affinity than to CD4 T cells (EC50: 17 664 vs 5368 pM) (figure 1B). Lomvastomig therefore preferentially binds to T cells that express either PD-1 alone or with PD-1 TIM-3, and has a reduced off-target binding to myeloid cells compared with high-affinity bivalent anti-TIM-3 antibodies. Preclinical in vitro and in vivo properties of the PD-1/TIM-3 bispecific antibody lomvastomig. (A) Schematic of lomvastomig. (B) Preferential binding of lomvastomig to T cells over monocytes/myeloid cells. (C) Lomvastomig was refractory to internalization in comparison to the bivalent anti-TIM-3 antibody as assessed by confocal microscopy. Parental anti-PD-1 was used as a negative control for internalization. (D) Lomvastomig was refractory to drug-shaving by macrophages, as opposed to PD-1-targeting IgG4 antibodies. (E) Lomvastomig enhanced ex vivo T-cell effector functions of PBMCs from patients with melanoma. (F) Lomvastomig significantly elicited antitumor activity against engrafted MC38-derived tumors in huPD-1xhuTIM-3 double-transgenic mice when compared with untreated (vehicle) and anti-PD-1 treated groups of mice. ANOVA, analysis of variance; BsAb, bispecific antibody; CR, complete response; IFN, interferon; PBMC, peripheral blood mononuclear cell; PD, progressive disease; PD-1, programmed death receptor-1; PR, partial response; SD, stable disease; TAA, tumor-associated antigen; TIM-3, T-cell immunoglobulin domain and mucin domain-3. Lomvastomig was also significantly less internalized by activated T cells on binding to TIM-3 than the corresponding conventional bivalent TIM-3 antibody (figure 1C). Furthermore, in contrast to nivolumab and pembrolizumab, lomvastomig was refractory to macrophage-mediated removal of the antibody from T cells (“drug-shaving”) (figure 1D), due to the introduction of P329G LALA mutations in the IgG1-based Fc region to prevent binding to Fc receptors.18 Lomvastomig showed in vitro functionality. Compared with PD-1 blockade alone, tumor antigen-specific T-cell effector functions (ie, granzyme B and IFN-γ secretion) in PBMCs from patients with melanoma were significantly higher when treated with lomvastomig (p<0.001), whereas the combination of anti-TIM-3 and anti-PD-1 parental antibodies did not reach significance (figure 1E). In the proof-of-concept efficacy studies in human PD-1 and TIM-3 double knock-in mice challenged with the murine colon adenocarcinoma cell line MC38, 100% of lomvastomig-treated mice controlled tumor growth and 46% experienced complete tumor rejection (figure 1F). Only 50% of the mice receiving the parental anti-PD-1 controlled tumor growth, and 31% rejected their tumors. These data demonstrated the validity of co-targeting TIM-3 and PD-1 with a single bispecific antibody molecule, providing the preclinical rationale to initiate the first-in-human study with lomvastomig. Clinical investigation of lomvastomig For dose-escalation in the first-in-human study, 39 patients with advanced and/or metastatic solid tumors were enrolled in six cohorts at four sites in Spain and Denmark between October 2018 and October 2019. The tumor-specific dose-expansion part included 38 CPI-experienced patients with melanoma, 26 CPI-experienced and platinum-experienced patients with NSCLC, 15 CPI-naïve patients with SCLC, and 16 CPI-naïve patients with ESCC. The patients in these four tumor-specific cohorts were enrolled at 17 sites in six countries (Spain, Denmark, France, USA, New Zealand, and Republic of Korea) between November 2019 and May 2021. An overview of the dose-escalation and expansion cohorts is provided in online supplemental figure S1. Exposure All 134 enrolled patients received at least one dose of lomvastomig. The median duration of treatment was 71 days (range: 1–1,181 days) for patients enrolled in the dose-escalation and 64 days (range: 1–1,290 days), 43 days (range: 1–316 days), 44 days (range: 15–128 days), and 86 days (range: 1–710 days) for the melanoma-specific, NSCLC-specific, SCLC-specific, and ESCC-specific expansion cohort, respectively. As of the data cut-off (September 03, 2024), one patient had completed 24 months of treatment with lomvastomig and 133 patients had discontinued treatment with lomvastomig, mostly due to progressive disease (81%) (online supplemental figure S1). Baseline characteristics Overall, more men (63%) than women (37%) were enrolled in the study (table 1), and the patients were predominantly white, except for the ESCC expansion cohort which included six Asian and four white patients (the race was unknown in the remaining six patients). Table 1Patient baseline characteristicsCharacteristicDose-escalation* N=39Tumor-specific expansionMelanoma†‡ CPI-exp. N=38NSCLC§ CPI-exp. N=26SCLC CPI-naïve N=15ESCC CPI-naïve N=16Median age (range), years60.0 (32–74)62.5 (31–81)62.5 (41–78)64.0 (46–71)62.5 (41–80)Sex (male/female), n %23 (59)/16 (41)21 (55)/17 (45)15 (58)/11 (42)14 (93)/1 (7)12 (75)/4 (25)ECOG, n (%) 019 (49)19 (50)10 (39)5 (33)5 (31) 120 (51)19 (50)16 (62)10 (67)11 (69)Former/current smoker, n (%)27 (69)18 (47)23 (89)15 (100)14 (88)Disease status, n (%) Locally advanced1 (3)2 (5)2 (8)1 (7)2 (13) Metastatic38 (97)36 (95)24 (92)14 (93)14 (88)Brain metastasis, n (%)1 (2.6)4 (11)01 (7)0Prior treatment, n (%) CPI12 (31)38 (100)26 (100)00 Chemotherapy35 (90)1 (3)26 (100)15 (100)16 (100) Surgery¶25 (64)33 (87)16 (62)7 (47)6 (38) Radiotherapy25 (64)23 (61)17 (65)15 (100)11 (69)Prior metastatic treatment lines, median number (range)6 (1–9)2.5 (0–3)0.5 (0–3)1.5 (0–4)0 (0–3)PD-L1 expression in tissue at screening, n (%)** ≥50%3 (7.7)11 (29)10 (39)02 (13) ≥1% to <50%3 (7.7)12 (32)9 (35)1 (7)5 (31) <1%7 (18)10 (26)6 (23)12 (80)7 (44) Unknown26 (67)5 (13)1 (4)2 (13)2 (13)*The primary origins of the tumors of the patients were (n; %): SCLC (7; 18%), NSCLC (5; 13%), colon cancer (4; 10%), renal cell cancer (4; 10%), (gastro-)esophageal cancer (3; 8%), mesothelioma (3; 8%), head and neck cancer (2; 5%), pancreatic (2; 5%), and others in one patient each (3%) including bile duct, bladder, breast, cervical, endometrial, melanoma, rectal, thymoma, and thyroid cancers.†In the melanoma expansion cohort, 10 (26%) patients had received an anti-CTLA-4 antibody therapy and 7 patients (18%) a BRAF/MEK inhibition therapy. 13 patients (34%) had an LDH level of >1.5 ULN at screening.‡Seven patients (18.4%) had a BRAF V600E mutation. Includes all cancer-related surgeries and is not limited to whole tumor resections.§Of the 26 patients in this part of the study, 18 (69%) had non-squamous NSCLC, 7 (27%) had squamous carcinoma, and the histology of one patient (4%) was unknown.¶Includes all cancer-related surgeries and is not limited to whole tumor resections.**The data are from central PD-L1 testing. Patients were enrolled based on local PD-L1 testing, which was reported as positive by the investigators for the enrolled patients.CPI, checkpoint inhibitor; CTLA-4, cytotoxic T-lymphocyte associated protein 4; ECOG, Eastern Cooperative Oncology Group performance status; ESCC, esophageal squamous cell carcinoma; LDH, lactate dehydrogenase; NA, not applicable; NSCLC, non-small-cell lung cancer; PD-L1, programmed cell death ligand 1; SCLC, small-cell lung cancer; ULN, upper limit of normal. Patient baseline characteristics The primary origins of the tumors of the patients were (n; %): SCLC (7; 18%), NSCLC (5; 13%), colon cancer (4; 10%), renal cell cancer (4; 10%), (gastro-)esophageal cancer (3; 8%), mesothelioma (3; 8%), head and neck cancer (2; 5%), pancreatic (2; 5%), and others in one patient each (3%) including bile duct, bladder, breast, cervical, endometrial, melanoma, rectal, thymoma, and thyroid cancers. In the melanoma expansion cohort, 10 (26%) patients had received an anti-CTLA-4 antibody therapy and 7 patients (18%) a BRAF/MEK inhibition therapy. 13 patients (34%) had an LDH level of >1.5 ULN at screening. Seven patients (18.4%) had a BRAF V600E mutation. Includes all cancer-related surgeries and is not limited to whole tumor resections. Of the 26 patients in this part of the study, 18 (69%) had non-squamous NSCLC, 7 (27%) had squamous carcinoma, and the histology of one patient (4%) was unknown. Includes all cancer-related surgeries and is not limited to whole tumor resections. The data are from central PD-L1 testing. Patients were enrolled based on local PD-L1 testing, which was reported as positive by the investigators for the enrolled patients. CPI, checkpoint inhibitor; CTLA-4, cytotoxic T-lymphocyte associated protein 4; ECOG, Eastern Cooperative Oncology Group performance status; ESCC, esophageal squamous cell carcinoma; LDH, lactate dehydrogenase; NA, not applicable; NSCLC, non-small-cell lung cancer; PD-L1, programmed cell death ligand 1; SCLC, small-cell lung cancer; ULN, upper limit of normal. Of the 39 patients in the dose-escalation cohorts, 7 had SCLC, 5 had NSCLC, 4 had renal cancer, 4 had colon cancer, 3 each had mesothelioma and (gastro-)esophageal cancer, 2 each had head and neck or pancreatic cancer, and the remaining 9 had various other cancers. Their median age at baseline was 60 years. The median age at baseline was 62.5 years for patients enrolled in the melanoma, NSCLC, and ESCC expansion cohorts, and 64 years for patients in the SCLC expansion cohort. 12 patients (31%) in the dose-escalation part had previously received a CPI. In the melanoma and NSCLC expansion cohorts, all patients had prior CPI exposure. In the melanoma expansion cohort, 10 (26%) patients had also previously received an anti-CTLA-4 antibody therapy and 7 (18%) patients a BRAF/MEK inhibition therapy. All patients enrolled in the SCLC and ESCC expansion cohorts were CPI-naïve. All patients with SCLC received prior platinum-based regimens. Safety During the dose-escalation part of the study, a patient with metastatic thymoma in the 1,200 mg Q2W lomvastomig cohort had a grade 3 AE of troponin T increase (see details below) which was considered a DLT. No other DLT was noted up to the maximum administered dose of 2,100 mg Q2W which was generally well tolerated by the 19 patients treated with this dose (ie, 49% of the 39 patients enrolled into dose-escalation) (table 2, online supplemental table S6). Therefore, no MTD was defined, and 2,100 mg Q2W was established as the RDE. There was no obvious difference between the AE profiles of the dose-escalation and the subsequent dose-expansion cohorts (table 2, online supplemental table S6). Table 2Most common adverse eventsMedDRA preferred term, n (%)Dose-escalation N=39Dose-expansion N=95Total N=134Most common (in ≥10% total patients) AEsGradeGradeGradeAny3Any3Any3Patients with at least one adverse event Asthenia8 (20.5)1 (2.6)30 (31.6)2 (2.1)38 (28.4)3 (2.2) Dyspnea14 (35.9)1 (2.6)16 (16.8)2 (2.1)30 (22.4)3 (2.2) Fatigue7 (17.9)022 (23.2)1 (1.1)29 (21.6)1 (0.7) Nausea9 (23.1)018 (18.9)027 (20.1)0 Anemia8 (20.5)1 (2.6)19 (20.0)9 (9.5)27 (20.1)10 (7.5) Decreased appetite6 (15.4)021 (22.1)1 (1.1)27 (20.1)1 (0.7) Constipation4 (10.3)022 (23.2)026 (19.4)0 Diarrhea8 (20.5)1 (2.6)15 (15.8)023 (17.2)1 (0.7) Abdominal pain7 (17.9)014 (14.7)2 (2.1)21 (15.7)2 (1.5) Vomiting6 (15.4)015 (15.8)021 (15.7)0 Cough8 (20.5)012 (12.6)020 (14.9)0 Back pain7 (17.9)1 (2.6)11 (11.6)018 (13.4)1 (0.7) Pruritus5 (12.8)013 (13.7)018 (13.4)0 Arthralgia7 (17.9)010 (10.5)1 (1.1)17 (12.7)1 (0.7) Aspartate aminotransferase increased7 (17.9)1 (2.6)10 (10.5)017 (12.7)1 (0.7) Headache10 (25.6)06 (6.3)016 (11.9)0 Fever4 (10.3)012 (12.6)016 (11.9)0 Alanine aminotransferase increased5 (12.8)010 (10.5)015 (11.2)0 Weight decreased4 (10.3)011 (11.6)015 (11.2)0AE, adverse event; MedDRA, Medical Dictionary for Regulatory Activities. Most common adverse events AE, adverse event; MedDRA, Medical Dictionary for Regulatory Activities. Almost all patients (127 of 134; 95%) in the dose-escalation and expansion cohorts reported at least one AE (online supplemental table S6). Overall, the most common (≥20% of patients) AEs of any causality were asthenia (28%), dyspnea and fatigue (22% each), and nausea, anemia, and decreased appetite (20% each) (table 2). The most common AEs with a potential immune-mediated etiology (imAEs), regardless of causality, were laboratory test alterations in the form of increased aspartate aminotransferase (13%), alanine aminotransferase (12%), and gamma-glutamyltransferase (10%) (online supplemental table S7). Among imAEs commonly reported with CPI therapies, rash was reported in 7% of patients, colitis and hypothyroidism in 3% patients each, and adrenal insufficiency, immune-mediated myositis, and hyperthyroidism in 2% of patients each. Such imAEs tended to occur more than 90 days after the first dose, including rash (4%), adrenal insufficiency (2%), and colitis and hypothyroidism (1% each). No delayed imAEs (ie, occurring 90 days after the last dose19) were reported. Most AEs in the dose-escalation and expansion cohorts had a maximum severity of grade 1 or 2. The most common grade 3 AEs were anemia in 10 patients (8%) and asthenia and dyspnea in 2 patients each (2%). 11 patients (8%) had at least one grade 3 AE considered related to lomvastomig, the most common of which were anemia, asthenia, increased gamma-glutamyltransferase, and lymphopenia (2 patients each). Three patients had a grade 4 AE, one of which was considered related to lomvastomig (respiratory failure, see below). AEs with an outcome of death were reported for one patient in the NSCLC expansion cohort (internal hemorrhage) and one patient in the ESCC expansion cohort (sepsis); both events were considered not related to lomvastomig. In the dose-escalation part, two patients (5%) experienced AEs leading to withdrawal of lomvastomig treatment. One was a patient in their 70s with mesothelioma who experienced a grade 2 AE of acute kidney injury that was considered treatment-related and occurred on study day 210, 20 days after the last 2,100 mg dose of lomvastomig. The other was a patient in their 40s with metastatic thymoma, type B3 without known paraneoplasias, who experienced a grade 3 immune-mediated myositis 1 day after the second 1,200 mg lomvastomig dose on study day 16, after which lomvastomig was discontinued. 3 days later, the patient had a grade 3 troponin T increase, which was considered a DLT and related to study treatment. Echocardiography showed normal heart function and cardiac MRI did not show signs of myocarditis. Muscle biopsy confirmed necrotizing inflammatory myopathy. The patient was treated with corticosteroids. On study day 38, the patient experienced a grade 4 event of respiratory failure and was hospitalized in an intensive treatment unit (ITU). The patient was treated with intravenous immunoglobulin, the event of respiratory failure was resolved on the same day. At the time of respiratory failure, the respirator showed weakened respiratory muscle strength suspected due to neuromuscular disease (myositis). On study day 144, the event of immune-mediated myositis was considered resolved. Circulating troponin T levels remained elevated at the time of study discontinuation. The investigator considered immune-mediated myositis and respiratory failure to be related to lomvastomig and the disease under study. In the dose-expansion part, three patients (3%) had AEs leading to withdrawal of lomvastomig treatment, which included alcohol abuse, internal hemorrhage, and immune-mediated enterocolitis. Of these, the AE of immune-mediated enterocolitis was considered related to study drug; it occurred on study day 33 with a grade 2 severity in a patient in the ESCC expansion cohort. PK, drug receptor occupancy, and immunogenicity PK and peripheral blood drug receptor occupancy analyses were conducted in all patients in the dose-escalation cohort, and characteristics were followed longitudinally within each dose cohort. Lomvastomig concentrations exhibited linear elimination after a single administration across the dose range studied (figure 2A,B). The data available for multiple administration of the 210 mg and 2,100 mg doses showed similar concentration-time profiles, but with higher concentrations than for the single administration, indicating drug accumulation. Pharmacokinetic profile, exposure, and drug receptor occupancy of lomvastomig. (A) Lomvastomig plasma concentration in cycles 1 (left panel) and 5 (right panel). (B) Dose-normalized area under the curve, across dose levels in cycle 1. (C) Drug receptor occupancy in peripheral CD3+T cells (left panel) and CD8+T cells (right panel); mean±SD, mean reported when n≥3. Following a single dose, systemic exposure increased with dose escalation from 70 mg to 2,100 mg, with the geometric mean for maximum concentration (Cmax) ranging from 21.5 µg/mL to 732 µg/mL and the area under the curve over 14 days (AUC14d) ranging from 140 to 4,640 day×µg/mL (online supplemental table S8). After multiple administrations, drug exposure was substantially higher, indicating accumulation. At steady state, a comparison reveals significant drug accumulation; for the 2,100 mg dose, the geometric mean Cmax approximately doubled from 732 µg/mL in cycle 1 to 1,360 µg/mL in cycle 5. The total exposure (AUC14d) for this same dose showed an accumulation ratio of approximately 2.4, increasing from 4,640 day×µg/mL to 11,200 day×µg/mL between cycle 1 and cycle 5. Peripheral blood drug receptor occupancy in CD3+ (figure 2C, left panel) and CD8+T cells (figure 2C, right panel) was observed with lomvastomig across all doses and persisted throughout the treatment period. ADA analysis was conducted in all patients and a total of 28 of 127 post baseline evaluable participants (22%) presented positive ADA titers, of which none were associated with AEs suggestive of immunogenicity. Clinical activity In the dose-escalation part, one patient had a CR, and three patients had a PR as best overall response (BOR) (figure 3); all these patients were treated in the 2,100 mg lomvastomig cohort (n=19), which had an ORR of 21% (95% CI 7.5% to 41.9%) (online supplemental table S9). The patient with the CR had a history of microsatellite instability-high (MSI-high) metastatic colon cancer and had previously received the combination of 5-fluorouracil, leucovorin, and oxaliplatin; the reduction in the sum of target lesion diameters was 50% on study day 56, 95% on study day 99 (which was 2 weeks after the seventh dose of lomvastomig, after which the treatment was interrupted due to a serious AE of pneumonia (considered not related to treatment) and then not resumed), and 100% approximately 6 months later, with a CR duration of 25 months at the data cut-off. The PRs occurred in one patient with CPI-experienced NSCLC, one patient with CPI-naïve ESCC, and one patient with CPI-naïve SCLC, with durations of response of 19, 8.5, and 5.6 months, respectively. Antitumor activity results based on RECIST V.1.1. Best per cent change in sum of diameters of target lesion from baseline (column color indicates the lomvastomig dose) and confirmed response in patients enrolled in the dose-escalation cohorts (N=39). CPI, checkpoint inhibitor (including anti-PD-(L)1, anti-CTLA-4 and anti-LAG-3); CR, complete response; CTLA-4, cytotoxic T-lymphocyte associated protein 4; ESCC, esophageal squamous cell carcinoma; L of Tx, lines of metastatic therapy; LAG-3, lymphocyte-activation gene 3; N, no; NA, not available; NSCLC; non-small-cell lung cancer; PD, progressive disease; PD-L1, programmed cell death ligand 1; PR, partial response; RECIST, Response Evaluation Criteria in Solid Tumors; SCLC; small-cell lung cancer; SD, stable disease; TAP, tumor area positive; TC, tumor cells; Y, yes. In the expansion cohorts, six patients achieved PR as BOR. Of these, three were in the CPI-experienced melanoma cohort, which had an ORR of 8% (95% CI 2.2% to 19.2%) and median progression-free survival (mPFS) of 1.8 months (90% CI 1.7 to 2.0) for the 38 response-evaluable patients. All three responders were among the 30 patients with non-uveal melanoma and had not received prior anti-CTLA-4 treatment (figure 4A). Two of the patients had a durable PR (17.7 and 18.5 months), one of whom had liver and brain metastases at baseline. In addition, 3 of the 15 response-evaluable patients in the CPI-naïve ESCC expansion cohort had PR as BOR, resulting in an ORR of 20% (95% CI 5.7% to 44.0%) and mPFS of 3.6 months (90% CI 1.5 to 5.7). The PR was durable in two patients (10.6 and 22.0 months), both of whom had received definitive concurrent chemo-radiotherapy but no prior systemic therapy in the metastatic setting (figure 5A). Response over time and association with select target associated biomarker levels measured in the tumor microenvironment in CPI-experienced patients enrolled in the melanoma cohort (N=38). (A) Swimlane plot. The left panel provides a detailed heatmap of key biomarkers, including PD-L1 (SP263) TC% and TAP%. Additional variables measured include immune cell markers (CD8+, CD8+PD-1+TIM3+, PD-1+, TIM3+), and clinical characteristics. Biomarker intensity is represented using color gradients, as explained in the legend. Biomarker heatmap intensity is represented by color scales: orange for PD-L1 markers and blue for immune cell markers. The right panel depicts patient-specific outcomes using a swimmer plot, showing response types (CR, PR, SD, PD) and the duration of clinical benefit over time, alongside other notable events such as discontinuation or death (X). (B) Baseline composition of CD8+ and CD8– cell phenotypes determined by mIF, based on co-expression of median densities of immune checkpoint markers PD-1, LAG-3, TIM-3 and tumor cell markers SOX10/KRT. Stacked bar plots display the relative fractions of distinct cell phenotypes within the CD8+ and CD8− cell populations. Colors correspond to specific phenotypes as defined in the legend. (C) Association of baseline IHC and mIF biomarker levels with clinical response. Box plots comparing PD-L1 (TC% and TAP%), CD8+, CD8+PD-1+TIM3+, PD-1+, and TIM3+cell populations between groups of patients with CR/PR and SD/PD responses in the CPI-experienced melanoma cohort. Statistical significance p values are indicated for comparisons between groups. An example image of a patient with the IF markers is shown in online supplemental figure S3. BOR, best overall response; CPI, checkpoint inhibitor (including anti-PD-(L)1, anti-CTLA-4 and anti-LAG-3); CR, complete response; CTLA-4, cytotoxic T-lymphocyte associated protein 4; IF, immunofluorescence; IHC, immunohistochemistry; L of Tx, lines of metastatic therapy; LAG-3, lymphocyte-activation gene 3; mIF, multiplex immunofluorescence; N, no; NE, not evaluable; PD, progressive disease; PD-1, programmed death receptor-1; PD-L1, programmed cell death ligand 1; PR, partial response; Rad Tx, radiotherapy; SD, stable disease; TAP, tumor area positive; TAP%, tumor area positivity fraction; TC, tumor cells; TC%, tumor cell fraction; TIM-3, T-cell immunoglobulin domain and mucin domain-3; Trt disc/compl, treatment discontinued/completed; UKN, unknown; Y, yes. Response over time and association with select target associated biomarker levels measured in the tumor microenvironment in CPI-naïve patients enrolled in the ESCC cohort (N=15). (A) Swimlane plot. The left panel provides a detailed heatmap of key biomarkers, including PD-L1 (SP263) TC% and TAP%. Additional variables measured include immune cell markers (CD8+, CD8+PD-1+TIM3+, PD-1+, TIM3+), and clinical characteristics. Biomarker intensity is represented using color gradients, as explained in the legend. Biomarker heatmap intensity is represented by color scales: orange for PD-L1 markers and blue for immune cell markers. The right panel depicts patient-specific outcomes using a swimmer plot, showing response types (CR, PR, SD, PD) and the duration of clinical benefit over time, alongside other notable events such as discontinuation or death (X). (B) Baseline composition of CD8+ and CD8− cell phenotypes determined by multiplex mIF, based on co-expression of median densities of immune checkpoint markers PD-1, LAG-3, TIM-3, and tumor cell markers SOX10/KRT. Stacked bar plots display the relative fractions of distinct cell phenotypes within the CD8+and CD8- cell populations. Colors correspond to specific phenotypes as defined in the legend. (C) Association of baseline IHC and mIF biomarker levels with clinical response. Box plots comparing PD-L1 (TC% and TAP%), CD8+, CD8+PD-1+TIM3+, PD-1+, and TIM3+cell populations between groups of patients with CR/PR and SD/PD responses in the CPI-naïve ESCC cohort. P values are indicated for comparisons between groups. BOR, best overall response; CPI, checkpoint inhibitor (including anti-PD-(L)1, anti-CTLA-4 and anti-LAG-3); CR, complete response; CTLA-4, cytotoxic T-lymphocyte associated protein 4; ESCC, esophageal squamous cell carcinoma; IF, immunofluorescence; IHC, immunohistochemistry; L of Tx, lines of metastatic therapy; LAG-3, lymphocyte-activation gene 3; mIF, multiplex immunofluorescence; N, no; NE, not evaluable; PD, progressive disease; PD-1, programmed death receptor-1; PD-L1, programmed cell death ligand 1; PR, partial response; Rad Tx, radiotherapy; SD, stable disease; SCR, screening; TAP, tumor area positive; TAP%, tumor area positivity fraction; TC, tumor cells; TC%, tumor cell fraction; TIM-3, T-cell immunoglobulin domain and mucin domain-3; Trt disc/compl, treatment discontinued/completed; UKN, unknown; Y, yes. All 15 response-evaluable patients in the CPI-naïve SCLC expansion cohort had progressive disease as BOR, the mPFS was 1.7 months (90% CI 1.6 to 1.8). 9 of the 25 response-evaluable patients (36%) in the CPI-experienced NSCLC expansion cohort had SD as BOR, the mPFS was 1.7 months (90% CI 1.6 to 1.9). Online supplemental table S8 provides a summary of the antitumor activity per RECIST V.1.1 for all cohorts. Biomarker analyses Individual biomarker data obtained from baseline tumor samples in the dose-expansion cohorts are shown in CPI-experienced patients with melanoma (figure 4A,B), CPI-naïve patients with ESCC (figure 5A,B), and CPI-experienced patients with NSCLC and CPI-naïve patients with SCLC (online supplemental figure 2A,C). To characterize the spatial architecture of the tumor microenvironment, we used a quantitative 5-plex immunofluorescence (mIF) assay (CD8, PD-1, LAG-3, TIM-3, and SOX10/KRT) to assess the median baseline prevalence of 20 cell phenotypes in the tumor region across tumor samples in the different cohorts (figures 4B and 5B and online supplemental figures 2A,C and 3). Marker-negative cell phenotypes (PD-1/lymphocyte-activation gene 3 [LAG-3]/TIM-3 negative) were numerically the most prevalent in both the CD8+ and CD8− subgroups, the median per cent density of CD8+T cells expressing TIM-3 or co-expressing PD-1 and TIM-3 was highest in CPI-experienced patients with melanoma (49% and 11%), followed by CPI-experienced patients with NSCLC (31% and 4%), and CPI-naïve patients with ESCC (27% and 3%) and patients with SCLC (4% and 0.2%). TIM-3 prevalence in CD8− immune cells and SOX10+/KRT+cell subsets was low (<1% to 5%). No clear association between PD-L1 expression (tumor area positivity or tumor cell score) and BOR was observed in the CPI-experienced melanoma and CPI-naïve ESCC cohorts (figures 4C and 5C). Numerically higher baseline expression of several target-associated immune cell biomarkers were observed in responders compared with non-responders, including CD8, PD-1, TIM-3, and triple-positive CD8/PD-1/TIM-3 (mIF assay). However, the small sample size and number of responders limited the statistical power and thus the detection of any meaningful differences between responders and non-responders. Discussion Lomvastomig has been designed to target dysfunctional tumor antigen-specific T lymphocytes expressing PD-1 and TIM-3, with the aim of establishing or restoring an effective antitumor immune response in patients with cancer who do not respond to or become resistant to current PD-1/PD-L1-directed therapies. The preclinical characterization presented here has shown that the design features of lomvastomig may prevent several factors from limiting its clinical activity. These include reduced off-target binding to TIM-3+ (myeloid) only cells through an avidity-driven selectivity binding to T cells in the tumor microenvironment expressing both PD-1 and TIM-3 or PD-1 alone, a low internalization rate by T cells compared with that of classical bivalent anti-TIM-3 antibodies on binding to the receptor and macrophage-mediated clearance of the antibody from T cells surface (“drug-shaving”) which is observed with IgG4 and IgG1 antibodies targeting PD-1.20 Conversely, LY3415244, a PD-L1/TIM-3 bispecific antibody and sabatolimab, a high affinity bivalent anti-TIM-3 antibody, which mainly bind to myeloid cells, have shown high immunogenicity and TIM-3 internalization,21 respectively, in addition to being subject to macrophage-mediated clearance. The IgG1 bispecific PD-1/TIM-3 antibody sabestomig has a similar design to lomvastomig; however, unlike lomvastomig, its TIM-3 binder does not block the interaction between TIM-3 and phosphatidylserine. This may result in reduced antitumor activity, since phosphatidylserine is known to create an immunosuppressive environment in the tumor.22 Lomvastomig’s design features, together with the findings that it improved the tumor-reactive effector functions of T cells from PBMCs of patients with melanoma as well as tumor growth control and eradication in a mouse model for the targeted human genes, provided the rationale for the initiation of the first-in-human study. In the dose-escalation part of the first-in-human study, lomvastomig was well tolerated up to the highest dose of 2,100 mg Q2W tested in heavily pretreated patients with advanced solid tumors. This dose was also well tolerated by patients enrolled in the tumor-specific expansion cohorts. No significant safety concerns were identified, and the safety profile was similar to that of anti-TIM-3 and anti-PD-l therapies administered alone or in combination and other bispecific antibodies targeting PD-1 and TIM-3.11,1423 One DLT was noted at a dose of 1,200 mg Q2W (grade 3 AE of troponin T increase) during dose-escalation; however, the MTD was not reached. The PK data were linear throughout the dose range studied suggesting that the target was saturated at 2,100 mg. In the absence of safety concerns and given that responses were only observed at 2,100 mg Q2W, this dose was established as the RDE. The patient with the DLT event also experienced grade 4 respiratory failure; however, this event was readily resolved with intensive management in the ITU. It is well known that patients with thymoma are predisposed to autoimmunity, and the occurrence of myocarditis, myositis, and myasthenia can overlap in patients with thymoma treated with immune checkpoint inhibitors, although myocarditis has been ruled out by cardiac MRI for the patient with thymoma in this study. The pattern of imAEs reported in this study appears to be consistent with the known safety profiles of anti-PD-1/L1 immunotherapies, although the distinct criteria used to define imAEs make direct cross-trial comparison not feasible. Lomvastomig exhibited linear PK across the studied dose range, suggesting target saturation. The peripheral drug receptor occupancy of both CD3+ and CD8+ was also saturated at >90% for lomvastomig doses of ≥70 mg throughout the treatment period. This is likely driven by the higher affinity binding to PD-1 over TIM-3, allowing avidity-mediated selectivity. During dose-escalation, antitumor activity was only observed at the highest dose of 2,100 mg tested and included a CR in a patient with MSI-high metastatic colon cancer and PRs in a CPI-experienced patient with NSCLC and two CPI-naïve patients, one with ESCC, and one with SCLC. During the expansion part, responses were limited to PRs observed in 3 of 38 (8%) CPI-experienced patients with melanoma and 3 of 15 (20%) CPI-naïve patients with ESCC. None of the patients in the CPI-naïve SCLC and CPI-experienced NSCLC expansion cohorts achieved a response. Most patients had no or low PD-L1 expression at enrollment, which may have limited the potential for responses in this study given that response to PD-1-based therapies has been associated with PD-L1 positivity in tumors.24 In addition, owing to the lack of predictive biomarkers for the effective use of TIM-3 blockade, it was not possible to select optimal candidates for anti-TIM-3 therapy. Consistently, biomarker analyses from this study showed limited association between select T-cell markers and clinical outcome; however, a trend was observed towards a higher baseline expression level of TIM-3 and the co-expression of CD8/PD-1/TIM3 on immune cells in responders in comparison to non-responders in the CPI-experienced melanoma and CPI-naïve ESCC cohorts. Consistent with the primary treatment rationale, it is plausible that dual checkpoint inhibition by lomvastomig could overcome resistance mechanisms to previous anti-PD-1/PD-L1 therapies by restoring cytotoxic T-cell proliferation in responding patients who, on average, expressed higher levels of TIM3 on exhausted CD8 T cells. In other studies, dual blockade of TIM-3 and PD-L1 generally resulted in better outcomes than anti-TIM-3 blockade alone, although the response rates remained low.25 26 The ORR was 6% in patients with advanced solid tumors treated with the combination of the anti-TIM-3 antibody sabatolimab and anti-PD-1 antibody spartalizumab,26 and 10% at the highest tested dose of the anti-TIM-3 antibody cobolimab combined with the PD-1 inhibitor dostarlimab in patients with advanced/metastatic NSCLC previously treated with anti-PD-1 therapy.27 Interim data for the monovalent, bispecific anti-TIM-3 and anti-PD-1 antibody AZD7789 indicated limited clinical activity in patients with NSCLC previously treated with anti-PD-1 therapy, with 7 of 19 patients achieving stable disease, including 2 patients with unconfirmed PRs.13 Results of the phase 1 study of the bivalent, anti-PD-1 and anti-TIM3 bispecific antibody LB1410 in patients with advanced solid tumors or lymphoma also showed only limited clinical activity.28 None of these studies have reported conclusive evidence of an association between potential biomarkers and treatment response. However, a trend towards higher TIM-3 levels was observed in responders,26 consistent with our findings. Further investigation is warranted to understand the roles of PD-1 and TIM-3 in predicting responses to dual checkpoint inhibitor treatments, thereby enabling a biomarker-based patient selection. TIM-3 is a complex molecule with multiple ligands, and the intracellular signaling is poorly understood.29 The blockade of TIM-3 is postulated to reverse exhaustion of T cells10 and may also have an effect on myeloid cells,30 although the latter is unlikely for lomvastomig given its low affinity TIM-3 binder. The contribution of the TIM-3 binding part of lomvastomig to the overall efficacy remains unclear in the clinic. Even the CR in the CPI-naïve patient with MSI-high metastatic colon cancer and encouraging results in CPI-naïve patients with ESCC could still be explained by the high-affinity PD-1 blocking part of the bispecific molecule. Ongoing clinical trials with other anti-TIM-3 agents will collectively shed light on the potential or lack of potential of this target. Furthermore, larger studies with more strictly defined populations will overcome the limitations inherent in this early-phase dose escalation study, in which the inclusion of different tumor types and a heterogeneous population have limited the generalizability of the clinical results. In conclusion, 2,100 mg lomvastomig administered Q2W had a tolerable and manageable safety profile across various advanced solid tumor indications. Consistent with results from studies of other PD-1/TIM-3-based treatments, the overall clinical activity was low in the mostly heavily pretreated patient population included in this study. An encouraging signal was observed in CPI-naïve patients with ESCC. This has led to the initiation of a randomized, blinded phase 2 study in patients with advanced or metastatic CPI-naïve ESCC who are refractory or intolerant to fluoropyrimidine-based or taxane-based and platinum-based regimens (NCT04785820). This study uses nivolumab as an active comparator and will therefore generate important and meaningful data to help decipher the contributions of TIM-3 to the overall antitumor activity of lomvastomig. Supplementary material Acknowledgements The authors would like to thank the patients, their families, and the participating study centers. We also thank Franclim Ribeiro, Annika Blank, Tobias Rutishauser, Florian Heil and Fabian Junker. Medical writing assistance was provided by Christian Seitz, PhD, and was funded by F. Hoffmann-La Roche Ltd. Footnotes Funding: This study was funded by F. Hoffmann-La Roche Ltd. No grant numbers apply. Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: The protocol of the animal study was approved by the Cantonal Ethics Commission of Zurich (license ZH193/2014, ZH225-17). The clinical study was conducted at multiple sites. The trial protocol and amendments were approved by the respective institutional review boards and ethic committees. Participants gave informed consent to participate in the study before taking part. The study was conducted in accordance with the Declaration of Helsinki and ICH Good Clinical Practice (GCP) Guidelines. Data availability free text: Data are available on reasonable request. 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 here: 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 re-identification. Data availability statement Data are available upon reasonable request. References Associated Data Supplementary Materials Data Availability Statement Data are available upon reasonable request.