A Phase 1 First-in-Human Study of FS118, a Tetravalent Bispecific Antibody Targeting LAG-3 and PD-L1 in Patients with Advanced Cancer and PD-L1 Resistance Free

Immune checkpoint blockade (ICB) can modulate immune cells within the tumor microenvironment, shifting the cells from an immunosuppressive or exhausted state to a proinflammatory state that favors an antitumor response (1).

Programmed death ligand 1 (PD-L1) is widely accepted as an immune checkpoint receptor, which delivers inhibitory signals and reduces T-cell–mediated effector responses through engagement with its receptors, programmed cell death protein 1 (PD-1) and CD80. PD-1/PD-L1 blockade restores functional T-cell responses in the tumor, leading to the approval of several PD-(L)1–targeting antibodies to treat patients with locally advanced or metastatic disease (2–5). Although these monotherapies have demonstrated meaningful responses and prolonged survival in a subset of patients, a large percentage of patients exhibit resistance to ICB therapy (6). Preclinical studies have indicated that compensatory upregulation of alternative checkpoint proteins such as lymphocyte-activation gene-3 (LAG-3) may be a mechanism of acquired resistance to anti–PD-(L)1 therapy (7, 8).

LAG-3 is an inhibitory checkpoint molecule that is predominantly expressed on the surface of T cells, natural killer (NK) cells, and plasmacytoid dendritic cells. LAG-3 is upregulated on the surface of activated and exhausted T cells and by a proportion of T-regulatory cells, which can confer suppressive activity against T effector cells by engaging peptide–MHC class II complexes (9–11). Although three splice variants of LAG-3 are proposed to exist (12), evidence is limited and soluble LAG-3 is most likely to arise from proteolytic cleavage of surface LAG-3 mediated by a disintegrin and metalloproteinase domain-containing protein 10 (ADAM10) and 17 (ADAM17; ref. 13). Cleavage of LAG-3 from conventional CD4⁺ T cells overcame anti–PD-1 resistance, and high LAG-3 and low ADAM-10 expression on CD4⁺ T cells in patients with squamous cell carcinoma of the head and neck (SCCHN) correlated with poor prognosis (14). Importantly, co-blockade of LAG-3 and the PD-1/L1 pathway modulates T-cell activity and enhanced antitumor T-cell responses in mouse models (8, 15, 16). Co-expression of LAG-3 and PD-1 on tumor-infiltrating lymphocytes (TIL) in clinical samples correlated with T-cell exhaustion and dysfunction (17, 18), and LAG-3 has been described previously as a poor prognostic marker in many cancers (19–21).

Several LAG-3–targeting molecules in clinical development have shown to be well tolerated with some responses as monotherapies reported (22, 23); however, combined targeting of LAG-3 and the PD-(L)1 axis may provide greater clinical benefit (24–26).

FS118 is a first-in-class tetravalent bispecific (mAb² format) mAb against LAG-3 and PD-L1 (27, 28). A LAG-3–binding site was introduced into the fragment crystallizable (Fc) region of a full-length human IgG₁ PD-L1 mAb (27), in addition to two mutations to abrogate Fcγ receptor binding (29). FS118-mediated blockade of LAG-3 and PD-L1 enhanced T-cell activity in vitro, and a mouse surrogate of FS118 significantly suppressed mouse tumor growth (28). Interestingly, a mouse surrogate of FS118 reduced cell-surface LAG-3 on T cells and increased soluble LAG-3 in the serum of mice, whereas the combination of mAbs targeting LAG-3 and PD-L1 increased LAG-3 expression in vivo. Given the compensatory upregulation of LAG-3 in response to anti–PD-(L)1 failures, and the mechanism of FS118-mediated shedding of cell-surface LAG-3 to overcome immune-mediated suppression, these data indicate that FS118 may extend clinical benefit beyond anti–PD-(L)1 monotherapy, providing rationale for this phase 1 clinical trial.

This open-label, multiple-dose, first-in-human phase 1 study aimed to characterize the safety, tolerability, pharmacokinetics (PK), and preliminary antitumor activity of FS118 in patients with advanced solid tumors and resistance to PD-(L)1 therapy (clinical trials identifier: NCT03440437). Pharmacodynamics of FS118 activity and immunogenicity measured by anti-drug antibodies (ADA) were also assessed.

Patients enrolled onto the study received FS118 intravenously once a week in three-week treatment cycles. The study followed an accelerated titration design; the first four cohorts (800 μg to 0.3 mg/kg) were enrolled sequentially as single-patient cohorts, and dose levels were progressed if no dose-limiting toxicities (DLT) or ≥ grade 2 adverse events were observed during the first cycle of treatment (intra-patient dose escalation was permitted). This was followed by a 3+3 ascending dose-escalation design (from 1 to 20 mg/kg) to determine the maximum tolerated dose (MTD). Dose-escalation cohorts were expanded at 3 mg/kg (n = 10), 10 mg/kg (n = 13), and 20 mg/kg (n = 13) to further explore PK, pharmacodynamics, and antitumor activity. Patients were considered to have completed treatment following 16 cycles (or 12 months) of FS118 treatment, or following confirmed progressive disease. Patients remaining on treatment at 12 months had the option to continue therapy.

Eligible patients had histologically confirmed locally advanced, unresectable or metastatic solid tumors and had exhausted available therapeutic options. Patients had received prior PD-(L)1 immune checkpoint inhibitors for a minimum of 12 weeks and showed subsequent progressive disease before enrolment. Prior treatment with LAG-3–targeting agents was not permitted.

Eligible patients were ≥18 years old, had an Eastern Cooperative Oncology Group performance status of 0–1, measurable disease (as per RECIST v1.1), and a life expectancy of at least three months. Patients were excluded if they had primary central nervous system tumors or metastases, a history of autoimmune disease, known infections, respiratory disease, encephalitis, seizures, or severe immune-related adverse events with prior anti–PD-(L)1–containing therapies. Patients with significant cardiac or laboratory abnormalities or a history of uncontrolled intercurrent illness were excluded (30).

The trial protocol and all amendments were approved by the appropriate institutional review board or independent ethics committee at each trial center. The trial was conducted according to the Declaration of Helsinki and International Conference on Harmonization Guidelines for Good Clinical Practice. All patients gave written informed consent before enrollment.

The primary objectives were to assess safety, define the MTD and/or recommended phase 2 dose (RP2D), and determine the PK parameters of FS118. The secondary objectives were to assess antitumor activity and characterize the immunogenicity (ADAs) of FS118. Exploratory objectives included characterizing the pharmacodynamic profile and correlating potential primary pharmacology with exposure. Radiographic assessments were conducted every eight weeks as per RECIST v1.1 and iRECIST. Toxicity was evaluated according to the National Cancer Institute (NCI) Common Terminology Criteria for Adverse Events (CTCAE, v4.03). The data cutoff value for the assessment of antitumor activity and safety was September 18, 2020.

Blood samples taken at each cycle were assessed for the development of ADAs using a validated, quasi-quantitative bridging electrochemiluminescence immunoassay. Samples were analyzed using a three-tier approach (screening, confirmatory, and titer determination). Affinity-purified rabbit anti-FS118 polyclonal antibodies (Envigo) acted as a positive control. The sensitivity of the screening assay was 10 ng/mL of positive control antibody, and the drug tolerance reported at 100 ng/mL of positive control antibody was 4.8 μg/mL.

Serum concentration of free FS118 was measured using a validated ligand-binding assay on the Gyrolab platform, based on capture and detection with the target molecules [biotin-labeled LAG-3 recombinant protein (BPS Bioscience) and Alexa Fluor 647–labeled PD-L1 (Bio-Techne)]. The assay sensitivity was 100 ng/mL. Standard serum PK parameters for FS118 were estimated using two-compartmental analysis.

Serum concentration of total soluble LAG-3 (sLAG-3) was quantified using an ELISA. Serum sLAG-3 was complexed with a saturating concentration of FS118, and the complex was captured with a non-competing anti–LAG-3 antibody clone 17B4 (Enzo Life Sciences) and detected with a non-competing anti-FS118 anti-idiotype antibody (Maine Biotechnology Services). Serum concentration data were used to calculate the AUC over the first week of treatment (AUC_0–7days) for each patient. The median (with 95% confidence interval) was calculated both for the serum concentration before each dosing cycle, and the AUC_0–7days, grouped by dose.

Tumor biopsies were collected before FS118 treatment for baseline targets expression, and samples were formalin-fixed, processed according to institutional best practices and paraffin-embedded. IHC was implemented on the BenchMark Ultra platform (Ventana/Roche), using anti–PD-L1 antibody clone SP263 (Roche/Ventana) and anti–LAG-3 antibody clone 17B4 (Abcam). The tumor proportion score (TPS) was calculated for PD-L1 expression, and PD-L1–positivity (PD-L1⁺) was defined as a TPS of ≥1%. For LAG-3 evaluation, five representative high-power fields (HPF, the area visible under the maximum magnification power) within the intra-epithelial tumor component and five HPF within the intra-tumoral stroma were selected and the average of these areas was used to assess the number of positive cells in each field. LAG-3–positivity (LAG-3⁺) was defined as one or more HPF with LAG-3⁺ cells.

Immunophenotyping was performed using flow cytometry on live fresh-blood cells. The panel included antibodies to CD45 (HI30, BD Biosciences, 563204), CD3 (SK7, BD Biosciences, 564001), CD19 (HIB19, BioLegend, 302240), CD4 (SK3, BD Biosciences, 563028), CD8 (SK1, BD Biosciences, 564629), CD56 (HCD56, BioLegend, 318334), and Ki67 (20Raj1, eBioscience, 46-5699-42). Immune cell populations were defined as: CD4 T-cells (CD45⁺/CD3⁺/CD4⁺/CD8⁻/CD19⁻), CD8 T-cells (CD45⁺/CD3⁺/CD8⁺/CD4⁻/CD19⁻) and NK cells (CD45⁺/CD3⁻/CD19⁻/CD56⁺/CD16^+/−). All values were reported as a percentage of total CD45⁺ leucocytes, and only cell subsets with more than 50 cells were considered.

A Mann–Whitney two-tailed test was used to assess correlations between time on FS118 treatment and prior ICB therapy in the context of resistance to PD-(L)1 therapy, as well as to test the statistical significance between dose levels with respect to sLAG-3 in patient serum and the difference in FS118 PK profile between ADA-negative and confirmed positive patients. A paired Wilcoxon signed-rank test was applied to analyze changes in leukocytes in patient blood, between baseline and on treatment values. Multiple P values were corrected with the Benjamani–Hochberg approach. Statistical tests were performed using GraphPad Prism and R (R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.r-project.org/).

The data generated in this study are available upon reasonable request from the corresponding author (M. Morrow).

Forty-three patients with advanced solid tumors were treated with FS118 across eight dose-escalation cohorts. Single-patient cohorts were dosed with FS118 at fixed doses of 800 μg (Cohort 1), 2,400 μg (Cohort 2), 0.1 mg/kg (Cohort 3) or 0.3 mg/kg (Cohort 4; Supplementary Fig. S1). Dose-escalation cohorts were then expanded following a 3+3 design to further explore PK, pharmacodynamics, and preliminary antitumor activity, resulting in a further 39 patients treated at 1 mg/kg (n = 3, Cohort 5), 3 mg/kg (n = 10, Cohort 6), 10 mg/kg (n = 13, Cohort 7) or 20 mg/kg (n = 13, Cohort 8).

The representativeness of this study of PD-L1–refractory solid tumors is discussed in Supplementary Table S1. The age of patients across all cohorts ranged from 30 to 85 years with a median age of 59 years (Table 1). The majority of patients were male (60.5%) and white (67.4%). Patients with a wide range of tumor types were enrolled (Supplementary Table S1). Advanced non–small cell lung cancer (NSCLC) and SCCHN were the most common tumor types, comprising 20.9% and 14.0% of all patients, respectively. Patients enrolled on study were heavily pre-treated, having received a median of three prior regimens (range, 1–11) for locally advanced/metastatic disease; 37.2% received four or more prior regimens (Table 1). The median number of prior ICB regimens was one (range, 1–3), with 21.0% of patients receiving two or more prior ICB regimens, and the median total time on prior ICB regimens was 38.9 weeks.

A total of 469 treatment-emergent adverse events were observed across all patients enrolled on study, which were present in 42 of the 43 patients enrolled (97.7%; Table 2).

Infusion of FS118 was well tolerated, where grade 1–2 infusion-related reactions were observed in five of 43 patients (11.6%; Table 2). In total, five patients discontinued treatment with FS118, which was not drug related in four of these patients (treatment withdrawn due to myocardial infarction, disease progression, abdominal pain, and tracheal obstruction). One patient withdrew due to treatment-related elevation of alkaline phosphatase (ALP) levels (grade 3) following treatment of FS118 at 10 mg/kg. No DLTs were observed, and the MTD was not reached.

Of the 43 patients enrolled on the trial, 25 (58.1%) had treatment-related adverse events (TRAE), of which two patients (4.7%), both of whom were dosed at 10 mg/kg, experienced grade 3 TRAEs (Table 2); one patient presented with elevated alanine aminotransferase, aspartate aminotransferase and bilirubin that resolved without the need to reduce the dose of FS118, and another patient demonstrated elevated ALP that resulted in the discontinuation of treatment. No treatment-related grade 4, 5 or serious adverse events relating to FS118 were reported.

Overall, any observed immunogenicity, as defined by the presence of confirmed positive ADAs, was typically transient in nature (i.e., did not persist for longer than three consecutive cycles) and had no significant impact on FS118 levels. Immunogenicity was observed in 21 of 43 patients (48.8%) treated with FS118. Four patients (9.3%) had ADAs present at baseline (Supplementary Table S2). It was noted that a smaller proportion of patients were ADA-positive at higher dose levels compared with patients treated with lower dose levels of FS118 [three out of 13 patients in the 20 mg/kg (23.1%) cohort, compared with seven out of 10 patients in the 3 mg/kg cohort (70%)]. All ADA confirmed positive patients at 10 and 20 mg/kg demonstrated ADAs that were transient, whereas in the 1 and 3 mg/kg cohorts (total of 13 patients), two patients (66.7%) and one patient (10.0%), respectively, demonstrated ADAs for three or more consecutive cycles. In addition, in the higher dose levels of 10 and 20 mg/kg, no patients and two patients (15.4%), respectively, presented with ADAs at the last treatment cycle, compared with two patients (66.7%) and five patients (50.0%) in the 1 and 3 mg/kg cohorts.

The median FS118 treatment duration was eight weeks (a range of 0.1–79 weeks, as of the data cutoff value, September 18, 2020). Across all patients, a disease control rate (DCR) of 46.5% was observed (Table 3). In both the 10 and 20 mg/kg cohorts, seven out of 13 patients (53.8%) achieved stable disease (SD) per RECIST v1.1 (hereafter referred to as SD). A DCR of 30% (three out of 10 patients) was observed in the 3 mg/kg cohort. No complete or partial responses (PR) per RECIST v1.1 were observed across all dose levels. Overall, nine of 43 patients (20.9%) surpassed 20 weeks of treatment with FS118 (Fig. 1), which included patients with advanced and heavily pre-treated anaplastic thyroid carcinoma (ATC), leiomyosarcoma, NSCLC, malignant peritoneal mesothelioma, cervical cancer, ovarian cancer, and SCCHN. At the data cutoff value, 41 patients had completed treatment and two patients remained on FS118 treatment at 79 weeks (ATC) and 56.1 weeks (leiomyosarcoma). As of March 2022, the ATC patient remained on FS118 treatment for over three years with disease control, including a late confirmed PR.

We hypothesized that a patients’ outcome on their previous anti–PD-(L)1 therapy, and whether they had primary or acquired resistance (6), could influence antitumor activity of FS118. Post hoc analysis explored correlations between FS118 treatment duration and resistance status to prior anti–PD-(L)1 therapy (acquired or primary resistance). It was observed that most patients who received FS118 treatment for greater than 20 weeks had previously achieved a PR, complete response (CR), or SD for more than three months as their best overall response to prior anti–PD-(L)1 therapy (Supplementary Fig. S2A; see Supplementary Data for more details). These patients were considered to have an acquired resistance phenotype (herein referred to as “acquired resistance” patients). Patients with progressive disease per RECIST v1.1 (hereafter referred to as PD) or whose best overall response to prior anti–PD-(L)1 therapy was SD for three months or less were considered to have a primary resistance phenotype (herein referred to as “primary resistance” patients). None of the patients with primary resistance remained on FS118 for more than 20 weeks, whereas it was noted that eight acquired resistance patients remained on study for greater than 20 weeks [seven of these patients for as long as six months (26 weeks); Supplementary Fig. S2A]. We also sought to assess whether the classification of resistance as defined by Kluger and colleagues (31) had any correlation with FS118 treatment duration. Using this classification, when acquired resistance was defined as a best overall response of CR, PR or SD for greater than six months and primary resistance defined as PD or SD for less than six months, a difference in FS118 treatment duration was not observed between acquired resistance versus primary resistance patients (Supplementary Fig. S2B).

Using our three-month resistance definition, no patients with primary resistance were on study beyond 20 weeks of FS118 treatment, compared with a subset of patients with acquired resistance (Supplementary Fig. S2A and Fig. 1A and B). Moreover, the DCR in patients with an acquired resistance phenotype was 54.8% (17 out of 31 patients) in patients receiving dose levels of 1 mg/kg or above, compared with 25.0% (two out of eight patients) in patients with primary resistance. In addition, patients who were treated with ICB therapy as their last prior treatment before treatment with FS118 (this included anti–TIM-3 or anti–CTLA-4 in addition to anti–PD-(L)1 therapy) achieved a significantly longer treatment duration (P = 0.029) on FS118 (median = 11.1 weeks) compared with patients who did not have ICB therapy as their last prior treatment (median = 7.3 weeks; Fig. 1C). Of those nine patients who surpassed 20 weeks treatment, eight patients were characterized by both acquired resistance to anti–PD-(L)1 therapy and receipt of any ICB as their last prior therapy, suggesting that this population of patients may have the greatest benefit from FS118 therapy. No clear correlation between time on prior ICB regimens and FS118 treatment was observed (Supplementary Fig. S3).

Suitable pre-treatment tumor core biopsies were obtained from 28 of 43 patients and were evaluated for PD-L1 and LAG-3 tumor expression. Of these, 17 patients were double positive for both PD-L1 and LAG-3 (60.7%), six were LAG-3 single positive (21.4%) and three were single positive for PD-L1 (10.7%; Fig. 2). Two patients were positive for LAG-3 with unknown PD-L1 expression. No patients were negative for both PD-L1 and LAG-3. The patient with ATC, whose PD-L1 and LAG-3 status were unknown when starting on this study, was later confirmed to have tumor-expressing PD-L1 and LAG-3 from an archival biopsy collected 15 months earlier. Although none of the LAG-3 single positive patients remained on study for more than 20 weeks, of the nine evaluable patients that surpassed 20 weeks of FS118 therapy, all with acquired resistance to prior anti–PD-(L)1 therapy, five demonstrated expression of both LAG-3 and PD-L1 in the tumor microenvironment (Fig. 2).

FS118 exhibited dose-linear PK within the evaluated dose range of 1–20 mg/kg (Fig. 3A; Supplementary Table S3) with a rapid phase half-life of 23.3 hours (one day) and a terminal half-life of 92.4 hours (3.9 days), as determined by PK data fit to a biexponential decay, characteristic of a two-compartment model.

A dose-dependent increase in total sLAG-3 was observed (Fig. 3B), indicating target engagement and suggesting that FS118 mediates LAG-3 shedding. Levels of sLAG-3 were higher in the 10 and 20 mg/kg cohorts compared with 1 and 3 mg/kg across the first four cycles of treatment (Fig. 3C). Moreover, no statistical differences in the levels of soluble LAG-3 were observed between 10 and 20 mg/kg (Fig. 3B). Peak levels of sLAG-3 were observed two to four days after administration of FS118 and elevation was maintained throughout the dosing interval in the 10 and 20 mg/kg cohorts (Supplementary Fig. S4A and S4B). Trough sLAG-3 concentrations at the end of each cycle were elevated compared with baseline levels (Fig. 3C), demonstrating the prolonged pharmacodynamic effect of FS118. There was no significant difference in baseline levels of sLAG-3 in those patients with acquired resistance compared with those with primary resistance to prior PD-(L)1 therapy (Supplementary Fig. S4C).

Analysis of immune cell subsets in the periphery revealed an increase in the percentage of circulating lymphocytes in patients who demonstrated SD, compared with those with PD, as their best response on FS118 treatment (Fig. 3D–F). The proportions of CD4⁺ and CD8⁺ T cells in patients with SD were significantly higher (P = 0.015 and P = 0.021, respectively) six weeks after initiating FS118 therapy compared with their baseline level (Fig. 3D and E). In comparison, patients with PD demonstrated significantly reduced circulating CD8⁺ T cells (P = 0.014) and NK cells (P = 0.014), and a trend toward reduced circulating CD4⁺ cells following FS118 treatment. The percentage of NK cells also showed a trend toward an increase in patients with SD after six weeks of treatment, but this was not found to be statistically significant (P=0.165; Fig. 3F). In support of the increased peripheral lymphocytes observed in patients with SD, a significant increase in the percentage of proliferating (Ki67⁺) CD4⁺ and CD8⁺ T cells and NK cells (P = 0.039, P = 0.039, and P = 0.0053, respectively) was observed seven days after the first dose of FS118. A similar increase in the percentage of proliferating (Ki67⁺) CD4⁺ and CD8⁺ T cells was also observed in patients with PD (P = 0.010 and P = 0.047, respectively) though as stated above, this did not correlate with an increase in peripheral lymphocytes in these patients (Supplementary Fig. S5A–S5C). These effects on peripheral lymphocytes were shown not to be dose-dependent (Supplementary Fig. S5D–S5F).

On the basis of the evident good safety profile of FS118, lower frequency of immunogenicity at dose levels ≥10 mg/kg, PK and observed pharmacodynamic activity (increased immune cell proliferation and sustained increased sLAG-3 over the dosing period), 10-mg/kg dose level was selected for further evaluation in the clinic.

In this first-in-human phase 1 dose-escalation and expansion study, FS118 was well tolerated with no DLTs observed up to the maximum administered dose of 20 mg/kg, and the MTD was not reached. The RP2D was established at 10 mg/kg QW. PK showed a terminal half-life of 3.9 days. Pharmacodynamic changes consistent with the bispecific mechanism of action of FS118 were observed. SD was observed, particularly in patients with acquired resistance to PD-(L)1 blockade, ICB as their most recent therapy and/or co-expression of LAG-3 and PD-L1 in the tumor.

FS118 demonstrated a good safety profile following weekly intravenous administration of FS118 at all dose levels assessed up to and including 20 mg/kg, characterized by the occurrence of mainly grade 1 and 2 TRAEs. In this study, TRAEs were observed in 58.1% of patients, with grade ≥3 TRAEs observed in only two patients (4.7%), and no adverse events were dose-limiting. The TRAE rate of FS118 compares favorably with PD-L1 inhibitors, where TRAE rates of 63.9%–81.1% (grade ≥3 rates of 4.9-17.0%) have been observed in a phase 1 ICB-naïve clinical setting (32, 33), and to the combination of LAG-3 and PD-1 mAbs, where TRAE rates of 81.1% (grade ≥3 rates of 18.9%) were observed with relatlimab and nivolumab combination (34) and TRAE rates of 69.4% (grade ≥3 rates of 9.1%) were observed with ieramilimab and spartlizumab combination (25). These data indicate that FS118 may have the potential to be better tolerated than PD-L1–targeting therapies alone or a combination of PD-1 and LAG-3–targeting therapies.

The patients enrolled on this study were heavily pre-treated, suggesting that the low rate of grade ≥3 TRAEs observed with FS118 therapy may have resulted from the pre-selection of patients with better tolerance to ICB therapies. We also speculate that the favorable safety profile of FS118 may also be attributed to its unique mechanism of action and localized activity in the tumor microenvironment. Tumor-localized FS118 activity may also explain the relatively rapid clearance of FS118 from the serum when compared with other PD-L1 mAbs (35).

SD was observed in 20 of the heavily pre-treated 43 patients (46.5%), with nine patients (20.9%) demonstrating long-lasting disease control for over 20 weeks. Of these nine patients, two patients had SCCHN and two patients had NSCLC. A patient with advanced ATC and a patient with advanced leiomyosarcoma who both presented with acquired resistance to recent PD-(L)1 therapy, demonstrated the longest time on FS118 treatment of 79 weeks and 56.1 weeks, respectively (at data cutoff value). Moreover, as of March 2022, the ATC patient remained on FS118 treatment for over three years with disease control, including a late confirmed PR.

On the basis of our observations, we speculate that upregulation of LAG-3 and PD-L1 may be markers of acquired resistance. Upregulation of LAG-3 on TILs in anti–PD-1–resistant tumors has been documented preclinically (7, 8, 36, 37) and clinically (36, 38). In this study, 60.7% of evaluable patients’ tumors were positive for both LAG-3 and PD-L1 before the start of FS118 treatment. Studies in melanoma targeting LAG-3 in patients with resistance to anti–PD-(L)1 therapies have demonstrated preliminary clinical benefit (39) and more recently, a study assessing the combination of fianlimab (anti–LAG-3) and cemiplimab (anti–PD-1) reported a much lower overall response rate in patients with melanoma previously exposed to prior PD-(L)1 inhibitors compared with ICB-naïve patients (13.3% and 63.6% respectively; ref. 40). Furthermore, in a phase 1/2 study assessing anti–PD-1 and LAG-3, all responding patients had not received prior checkpoint therapy (25).

In this study, a DCR of 54.8% (17 out of 31 patients) was observed in patients with acquired resistance to prior PD-(L)1 ICB therapy who were treated with at least 1 mg/kg FS118 (Fig. 1A). Given that blockade of the PD-(L)1 axis may result in upregulation of LAG-3 to drive acquired resistance (6, 7, 36), modulation of the LAG-3 pathway in addition to PD-L1 blockade by FS118 may be an approach to address acquired resistance. However, there is a possibility that the disease control observed in this study could be due to re-challenge with PD-L1 blockade, as it has been demonstrated that re-challenge with immune checkpoint inhibitors can result in re-induction of responses in ICB-sensitive tumor types (41). Further clinical studies in acquired resistance patients will be required to assess the potential for FS118 to provide patient benefit in this setting.

Pharmacological activity associated with FS118 was observed as increases in circulating and proliferating lymphocytes. Increases in CD4⁺ and CD8⁺ T cells following treatment with FS118 observed in patients with stable disease, but not in patients with progressive disease, may indicate that a T-cell–mediated immune response is a good prognostic indicator. A sustained elevation of soluble LAG-3 was achieved in the serum as a dose-dependent marker of FS118 activity, indicating prolonged pharmacodynamic activity and a saturation of LAG-3 engagement is reached at 10 mg/kg. On the basis of our preclinical data, we hypothesize that LAG-3 shedding drives this increase (42), although it is possible that alternative splicing of LAG-3 contributes to this effect (12). As LAG-3 expression has been associated with an exhaustive T-cell phenotype, and with LAG-3⁺ NK cells with impaired cytotoxic functions (43), it could be hypothesized that shedding of LAG-3 from the cell surface of effector cells by FS118 may reverse these phenotypes, as supported by data herein demonstrating increased peripheral effector cells following FS118 treatment.

The highest DCR was observed at 10 and 20 mg/kg (53.8% of patients in both cohorts), indicating that optimal pharmacology and antitumor activity were achieved at dose levels of 10 mg/kg and above. This is further supported by the pharmacology data showing that soluble LAG-3 was not markedly increased in serum samples of patients dosed at 20 mg/kg compared with serum samples of patients dosed at 10 mg/kg, suggesting that a higher concentration of FS118 does not necessarily translate to increased activity and a threshold for activity may have already been reached at 10 mg/kg. In further support of this, no pharmacodynamic differences in circulating lymphocytes were observed between dose levels, which suggest saturation of pharmacological effects at lower dose levels. In addition, although the majority of ADAs observed were shown to be transient, two patients from 1 to 3 mg/kg cohorts (from a total of n=3 and n=10 patients, respectively) presented with ADAs for more than three consecutive cycles, consistent with other observations that lower doses of biotherapeutics can be more immunogenic than higher doses (44). Taken together, these data were used to select a RP2D of 10 mg/kg weekly for future FS118 studies.

In this first-in-human study, FS118 was well tolerated without DLT, and the safety profile compared favorably to other ICB therapies. Stabilization of disease was observed in a subset of patients and a RP2D was identified for future studies. Further assessment is warranted to assess the clinical benefit FS118 may bring to patients who have progressed on prior anti–PD-(L)1 therapy. A phase 2 clinical trial exploring the antitumor activity of FS118 in patients with PD-L1- and LAG-3–positive SCCHN with acquired resistance to PD-(L)1 inhibitor therapy is ongoing (ClinicalTrials.gov: NCT03440327).

T.A. Yap reports other support from MD Anderson Cancer Center and Seagen; grants and personal fees from Acrivon, Artios, AstraZeneca, Bayer, BeiGene, Clovis, EMD Serono, F-star, Merck, Pfizer, and Repare; grants from BioNTech, Blueprint, BMS, Constellation, Cyteir, Eli Lilly, Forbius, GlaxoSmithKline, Genentech, Haihe, ImmuneSensor, Ionis, Ipsen, Jounce, Karyophar, KSQ, Kyowa, Mirati, Novartis, Ribon Therapeutics, Regeneron, Rubius, Sanofi, Scholar Rock, Seattle Genetics, Tesaro, Vivace, and Zenith; and personal fees from AbbVie, Acrivon, Adagene, Almac, Aduro, Amphista, Athena, Atrin, Avoro, Axiom, Baptist Health Systems, Boxer, Bristol Myers Squibb, C4 Therapeutics, Calithera, Cancer Research UK, Cybrexa, Diffusion, Genmab, Glenmark, GLG, Globe Life Sciences, GSK, Guidepoint, Idience, Ignyta, I-Mab, ImmuneSensor, Institut Gustave Roussy, Intellisphere, Janssen, Kyn, MEI Pharma, Mereo, Natera, Nexys, Novocure, OHSU, OncoSec, Ono Pharma, Pegascy, PER, Piper-Sandler, Prolynx, ResTORbio, Roche, Schrodinger, Theragnostics, Varian, Versant, Vibliome, Xinthera, Zai Labs, and ZielBio during the conduct of the study. P.M. LoRusso reports other support from AbbVie, Agios, Five Prime, GenMab, Halozyme, Roche-Genentech, Genentech, CytomX, Takeda, SOTIO, Cybrexa, Agenus, Tyme, IQVIA, TRIGR, Pfizer, ImmunoMet, Black Diamond, GlaxoSmithKline, QED Therapeutics, AstraZeneca, EMD Serono, Shattuck, Astellas, Salarius, Silverback, MacroGenics, Kyowa Kirin, Kineta, Zentalis, Molecular Templates, ABL Bio, SK Life Science, STCube, Bayer, I-MAB, Seagen, imCheck, Relay Therapeutics, Stemline, Compass BADX, Mekanist, Mersana, BAKX Therapeutics, Scenic Biotech, Qualigen, Roivant, and NeurotTrials outside the submitted work. D.J. Wong reports grants from F-star during the conduct of the study as well as grants from Bristol-Myers Squibb, Genentech/Roche, Sanofi, Blueprints Medicines, Lilly, Merck Sharp and Dohme, AstraZeneca, Kura Oncology, Enzychem Lifesciences, Elevar Therapeutics, Top Alliance BioSciences, Checkmate Pharmaceuticals, Bicara Therapeutics, and Gilead Sciences and grants and personal fees from Regeneron outside the submitted work. K.P. Papadopoulos reports other support from F-star during the conduct of the study as well as personal fees from Basilia, Bicycle, and Turning Point Therapeutics and other support from 3D Medicines, AbbVie, ADC Therapeutics, Amgen, Anheart, Bayer, Daiichi Sankyo, Incyte, Jounce Therapeutics, Lilly, Linnaeus, Merck, Mersana, Mirati, Pfizer, Regeneron, Revolution Medicines, Syros Pharmaceuticals, Tempest Therapeutics, and Treadwell Therapeutics outside the submitted work. J.-B. Holz reports personal fees from F-star during the conduct of the study as well as personal fees from F-star outside the submitted work. R.C.A. Sainson reports other support from F-star during the conduct of the study. C.J. Shepherd reports holding shares in F-star during the conduct of the study. F. Germaschewski reports service as an independent biomarker consultant supporting Biotech/Pharma and CRUK; however, there is no conflict of interest with F-star with bispecifics or the specific targets; biotechs supported are NovalGen, Evgen Pharma, UCB, Ellipses Pharma, Molecular Templates, and Sqz Biotech; current biotechs supported are Apollo Therapeutics, Tribune Therapeutics, Orchard Therapeutics, Dynamicure, and CRUK. D. Gliddon reports a patent for WO2020229626A1 pending. L. Kayitalire reports other support from F-star during the conduct of the study as well as other support from F-star outside the submitted work. No disclosures were reported by the other authors.

T.A. Yap: Resources, investigation, writing–review and editing. P.M. LoRusso: Resources, investigation, writing–review and editing. D.J. Wong: Resources, investigation, writing–review and editing. S. Hu-Lieskovan: Resources, investigation, writing–review and editing. K.P. Papadopoulos: Resources, investigation, writing–review and editing. J.-B. Holz: Conceptualization, data curation, methodology, project administration, writing–review and editing. U. Grabowska: Validation, project administration, writing–review and editing. C. Gradinaru: Data curation, formal analysis, methodology, writing–review and editing. K.-M. Leung: Supervision, validation, investigation, visualization, writing–review and editing. S. Marshall: Supervision, validation, investigation, visualization, writing–review and editing. C.S. Reader: Data curation, validation, visualization, writing–original draft, writing–review and editing. R. Russell: Data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. R.C.A. Sainson: Supervision, validation, writing–review and editing. C.J. Seal: Data curation, validation, visualization, writing–original draft, writing–review and editing. C.J. Shepherd: Data curation, supervision, validation, investigation, writing–original draft, writing–review and editing. F. Germaschewski: Data curation, investigation, methodology, writing–review and editing. D. Gliddon: Conceptualization, supervision, investigation, methodology, writing–review and editing. O. Stern: Investigation, visualization, methodology, writing–original draft, writing–review and editing. L. Young: Conceptualization, supervision, writing–review and editing. N. Brewis: Conceptualization, supervision, writing–review and editing. L. Kayitalire: Supervision, writing–review and editing. M. Morrow: Supervision, writing–original draft, project administration, writing–review and editing.

This study was funded and supported by F-star Therapeutics Ltd., Cambridge, UK. We thank the patients, their families, investigators, and research staff at all study-sites. The authors would also like to thank the F-star team for support with study operations, data analysis and QC, and critical review of this article.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.