Abstract
Conventional antibody–drug conjugates (ADC) utilize native surface-exposed lysines or cysteines on the antibody of interest to conjugate cytotoxic payload. The nonspecific conjugation results in a mixture with variable drug-to-antibody ratios (DAR), conjugation sites, and ADCs that are often unstable in systemic circulation. ARX788 is an ADC consisting of a HER2-targeting antibody site-specifically conjugated with a potent antitubulin cytotoxic drug-linker, AS269. The site-specific conjugation is achieved by first incorporating the nonnatural amino acid, para-acetyl phenylalanine (pAF), into the antibody, followed by covalent conjugation of AS269 to the pAF to form a highly stable oxime bond resulting in a DAR 2 ADC. ARX788 exhibits significant, dose-dependent antitumor activity against HER2- expressing breast and gastric xenograft tumors. Pharmacokinetic (PK) studies in multiple species showed the highly stable nature of ARX788 with overlapping PK profiles for the intact ADC and total antibody. Metabolism studies demonstrated that pAF-AS269 was the sole major metabolite of ARX788, with no evidence for the release of free drug often observed in conventional ADCs and responsible for adverse side effects. Furthermore, ARX788 demonstrated a favorable safety profile in monkeys with a highest nonseverely toxic dose of 10 mg/kg, which was well above the efficacious dose level observed in preclinical tumor models, thus supporting clinical development of ARX788.
Introduction
Antibody–drug conjugates (ADC) are complex biomolecules made by conjugating a therapeutic chemical entity with a mAb. The specificity of the mAb to a tumor-targeted antigen and the ability to deliver potent cytotoxic drugs to target expressing cells has made ADCs an attractive class of molecules for cancer therapy (1, 2). Currently, there are over 70 ADCs in active clinical trials, with seven ADCs on the market: including two HER2-targeting ADCs, Ado-trastuzumab emtansine (anti-HER2 ADC for breast cancer, T-DM1, Kadcyla), and recently approved fam-trastuzumab deruxtecan-nxki (anti-HER2 ADC for adult patients with unresectable or metastatic HER2-positive metastatic breast cancer who received two or more prior anti-HER2 based regimens, DS-82011, Enhertu) (3–6). These marketed ADCs are made with conventional conjugation approaches that use exposed lysines or free cysteines formed via reduction of interchain disulfide bonds (7, 8) of an antibody. The availability of multiple lysine and cysteine residues for conjugation generally results in ADCs with a heterogeneous mixture of variable drug-to-antibody ratio (DAR) (7, 8). For example, gemtuzumab ozogamicin consists of calicheamicin attached to CD33 antibody with a DAR of 2 to 3, along with free antibody with no drug molecules. In the case of T-DM1, the average DAR is 3.5 with 0 to 8 payloads conjugated per antibody (3, 6).
The number of payloads on the antibody can significantly influence the stability, pharmacokinetics (PK), efficacy, and safety profile (7, 9). In theory, ADCs loaded with high DARs should exhibit higher efficacy, but ADC instability or nonspecific uptake can lead to safety concerns due to off-target toxicity. In addition, ADCs conjugated to hydrophobic payloads at high DARs have been shown to clear faster from the systemic circulation, which can result in suboptimal in vivo efficacy (10, 11). Most of the currently marketed ADC molecules are dosed at or near the MTD in clinic with a limited therapeutic window (12). For example, brentuximab vedotin as monotherapy is dosed at 1.8 mg/kg every 3 weeks (1.8 mg/kg MTD) with notable side effects including peripheral sensory neuropathy and neutropenia (13, 14), and T-DM1 is dosed at 3.6 mg/kg (3.6 mg/kg MTD) with side effects of thrombocytopenia and neutropenia (15, 16). While research on the cause of these toxicities in clinic is ongoing, the reported results thus far suggest the observed toxicity is likely target-independent (17). The therapeutic window or therapeutic index of an ADC is impacted by several factors, including but not limited to, the selection of target, drug, conjugation chemistry, linker design, patient selection, and dosing strategies. For example, the first approved ADC gemtuzumab ozogamicin was withdrawn from the U.S. market in 2010 due to no clear evidence of the drug's benefit to patient survival. However, gemtuzumab ozogamicin was reapproved by the FDA in 2017 after further studies with fractionated dosing, altered patient population, and different schedule in combination therapy or as a single agent (18) showed benefit. The gemtuzumab ozogamicin story demonstrates the complexities surrounding ADC treatments, including dose adjustments and patient selection, to obtain a favorable clinical benefit from the same therapeutic.
In recent years, the ADC field has increased its focus and scrutiny on the conjugation site and chemistry to produce next-generation ADCs with site-specific conjugation technologies that offer high stability in the systemic circulation to limit off-target toxicity (11, 12, 19). Ambrx's pioneering work in a proprietary mammalian expression system to produce antibodies with nonnatural amino acid, para-acetyl phenylalanine (pAF), enables site-specific chemical conjugation via an oxime bond, resulting in ADCs with a homogeneous distribution of DAR 2 (20). Because of the high stability of the oxime bond and the use of noncleavable linkers, release of the cytotoxic payload occurs only after ADC is processed by endosomal/lysosomal proteases within target-expressing tumor cells. Ambrx has initiated several next-generation ADC programs, including ARX788, a novel HER2-targeting ADC with a potent antitubulin cytotoxic drug-linker (AS269) for the treatment of advanced cancers with HER2 expression. In this study, we present results from in vivo nonclinical evaluation of ARX788, a novel HER2-targeting ADC with a potent antitublin cytotoxic drug-linker (AS269) for the treatment of advanced cancers with HER2 expression. ARX788 is currently in phase I clinical trials in several countries including United States, China, Australia, and New Zealand.
Materials and Methods
All in vivo experiments were performed in accordance with the local guidelines of the Institutional Animal Care and Use Committee (IACUC).
In vivo efficacy studies
ARX788 was evaluated in subcutaneous xenografts of human tumor cell lines derived from breast and gastric cancer in BALB/c nude mice. Representative efficacy studies conducted in HCC1954 (HER2-expressing breast cancer cell line), MDA-MB-468 (non-HER2–expressing breast cancer cell line), and NCI-N87 (HER2-expressing gastric cancer cell line) are presented. The efficacy of ARX788 was evaluated at 0.55, 1.67, or 5 mg/kg compared against vehicle, T-DM1, and 15 mg/kg paclitaxel. Briefly, mice were inoculated subcutaneously in the right flank with respective tumor cells for tumor development. All groups received treatments via IV bolus administration except paclitaxel group which was via IP. Tumors were measured twice weekly using a caliper, and the tumor volume (TV) was expressed in mm3 using the formula: TV = 0.5ab2, where a and b are the long and short diameters of the tumor, respectively. The tumor volumes were used for calculation of treatment group/control group (T/C) values using the formula: T/C (%) = TRTV/CRTV × 100% (TRTV: relative tumor volume of the treatment group; CRTV: relative tumor volume of the control group). In addition, body weight was monitored as an indirect measure of toxicity approximately once every week in all xenograft studies. Statistical analysis was performed to assess the differences among treatment groups using one-way ANOVA.
DMPK
In vitro metabolism studies in HER2-expressing cell lines
Metabolism studies for ARX788 and pAF-AS269 were conducted in HER2-positive (HCC1954) and HER2-negative (MDA-MB-468) cells. Briefly, 3 × 105 cells in 6-well plates were incubated with 5 μg/mL ARX788 or 10 μmol/L pAF-AS269 or vehicle in triplicate and incubated for 0, 12, 16, and 24 hours. Following treatment, the plates were removed from the incubator and the supernatant was collected from the wells. The supernatant was further centrifuged to pellet the floating cells. The resulting supernatant was mixed with stop solution (acetonitrile containing 10 ng/mL diclofenac) and labeled as the medium sample. The cell pellet was combined with the remaining adherent cells in the plates, 450 μL of 10% formic acid in water was added, vortexed, incubated at −20°C for 5 minutes, and ultrasonicated in a bath for 1 minute. A total of 900 μL of the stop solution was then added to the mixture, vortexed, centrifuged, and the resulting supernatant was labeled as the cell sample. Both the medium sample and the cell sample were subjected to LC/MS analysis. The cell sample was also used for cell counting in a hemacytometer and protein content estimation by Bicinchoninic Acid Assay (BCA).
In vitro hepatocyte studies
The in vitro metabolism of ARX788 and pAF-AS269 were investigated in pooled human and monkey hepatocytes after a 120-minute incubation at 37°C. Briefly, the cells were thawed, counted, and diluted with prewarmed Williams Medium E. Hepatocytes were seeded at 1.25 × 106 cells with 40 μmol/L pAF-AS269 or 1 mg/mL ARX788 into a 24-well plate and incubated for 120 minutes. The assay included appropriate controls including blank (media without pAF-AS269 or ARX788) and 7-Ethoxycoumarin as the positive control. The reaction was stopped after 120 minutes by removing the plate from the incubator and adding 400-μL stop solution (0.1% formic acid in acetonitrile) to each well. Following gentle vortex and centrifugation (100 g for 10 minutes), the cell mixture from duplicate sample wells were transferred, combined into tubes, and centrifuged for 10 minutes at 15.7 × 1,000 g for 10 minutes. The supernatant was transferred into another tube and dried with nitrogen gas, and the residue was reconstituted, centrifuged and the resulting supernatant was transferred for LC/MS-MS analysis.
PK studies
PKs of ARX788 were assessed in mice bearing tumors, rats, and monkeys. ARX788 was administered as an IV bolus for mouse and rat studies, and as an IV infusion (15–20 minutes) for cynomolgus monkey studies. Following ARX788 dosing, serum samples were collected at various time points for at least 14 days in all studies. Serum samples were evaluated in validated total antibody, intact ADC (detects DAR 2 ADCs only), and pAF-AS269 methods (Supplementary Methods, Section 1). PK analyses were performed using Phoenix WinNonlin software (version 6.3.1, Pharsight) using noncompartmental analysis methods.
Biodistribution studies in tumor-bearing mice
The tissue distribution of anti-HER2 antibody (mAb) and ARX788 in NCI-N87 tumor-bearing mice was investigated using radioisotope (125I)-labeling (label on the mAb) and tricarboxylic acid (TCA) precipitation. Forty-eight tumor-bearing mice were randomized into eight groups, each containing 6 mice: 3 males and 3 females. Four groups were injected intravenously with 5 mg/kg dose of 125I-ARX788, and the other four groups were injected with 125I-mAb. At 2, 24, 72, or 168 hours postdose, one group of mice injected with 125I-ARX788 or with 125I-mAb were euthanized following blood and urine sample collection. The thyroid, heart, lungs, liver, spleen, adrenals, kidneys, bladder, gonads (testis/ovary), mesenteric lymph nodes, small intestine, intestinal contents, jejunum, fat, muscle, bone marrow, femur, brain, and tumor tissues were collected. After weighing, the tissues were sliced and placed in 400-μL 20% TCA to precipitate protein, centrifuged, and the γ-radioactivity of the precipitate fraction and supernatant fraction were individually determined. Drug concentration was expressed as equivalents (μg Eq/mL).
Mass balance study in rats
A mass balance study was performed to determine the recovery of 14C-ARX788 (label on the AS269 drug-linker) in urine and feces following single intravenous administration at a dose of 20 mg/109 μCi/kg to male and female Sprague—Dawley (SD) Rats. Six SD rats were administered 14C-ARX788 and placed into metabolic cages, and the feces and urine samples were collected from the animals for up to 56 days postdose. The concentrations of 14C-ARX788 in urine and feces samples were determined using a liquid scintillation counter. All calculations were performed using general radiochemical calculation procedures.
Safety studies in rat and monkey
The safety of ARX788 was established using nonclinical single-dose and repeat-dose toxicology studies in SD rats and cynomolgus monkeys. Monkeys were selected because they are a preferred nonrodent species in which the PK disposition is very similar to that in humans, and similar target-binding affinity of ARX788 was observed (21, 22). The numbers of animals used in the toxicology studies were in accordance with the IACUC and necessary for meaningful interpretation of study results (single-dose rat study, 5/sex/group; single-dose monkey study, 2/sex/group; repeat-dose monkey study, 6 or 7/sex/group). In the rat single-dose toxicity study, ARX788 was administered as an IV bolus at dose levels ranging from 20 to 90 mg/kg. In monkey toxicity studies, an IV infusion (15–20 minutes) of ARX788 at dose levels ranging from 10 to 30 mg/kg, or 3.3 to 15 mg/kg, was evaluated in single- and repeat-dose studies, respectively. All studies generally included both males and females to evaluate difference between the sexes. The repeat-dose study in monkeys included a total of four repeat doses administered 21 days apart, and a subgroup of animals was designated for an additional 8-week recovery period to study the reversibility of observed toxicities. All toxicology studies included toxicokinetic assessments, clinical observations and pathology, necropsy and histopathology for complete safety assessments.
Results
ARX788 demonstrates higher antitumor activity than T-DM1 in HER2-expressing mouse xenograft models
The efficacy of ARX788 was evaluated in three breast or gastric cancer subcutaneous xenografts in BALB/c nude mice. Antitumor efficacy of ARX788 in HCC1954 (HER2-positive breast cancer cell line), NCI-N87 (HER2-positive gastric cancer cell line), and MDA-MB-468 (HER2-negative breast cancer cell line) are summarized in Fig. 1A–C.
The antitumor activity of ARX788 in HER2-positive breast and gastric cancer xenograft models. Tumor volumes (±SEM) are shown following a single dose of ARX788 or T-DM1 to female BALB/c nude mice bearing HCC1954 (HER2-positive breast cancer xenograft model, n = 10/group, A), NCI-N87 (HER2-positive gastric cancer xenograft model, n = 9/group, B), or MDA-MB-468 (HER2-negative breast cancer xenograft model, n = 10/group, C). T/C% dose response following ARX788 administration at 0.55, 1.67, and 5 mg/kg are presented in D.
In the HCC1954 xenograft model, ARX788 showed dose-dependent antitumor activity with tumor regression observed after a single dose of 1.67 mg/kg (T/C = 1.0%, P = 0.004) or 5 mg/kg (T/C = 0.2%, P = 0.003) ARX788 for up to 28 days. Treatment with ARX788 at 1.67 mg/kg resulted in similar antitumor activity as T-DM1 at 5 mg/kg (T/C = 7.1%, P = 0.005) suggesting ARX788 is more potent than T-DM1 (Fig. 1A).
Similarly, in the NCI-N87 xenograft model, ARX788 showed dose-dependent antitumor activity at 0.55, 1.67, and 5 mg/kg (Fig. 1B). A dose of 1.67 mg/kg ARX788 was relatively less effective (T/C = 56.6%, P = 0.066) when compared with the same dose in the HCC1954 breast cancer xenograft model (Fig. 1A). To note, antitumor activity with 5 mg/kg T-DM1 (T/C = 55.7%, P = 0.099) was similar to that of ARX788 at 1.67 mg/kg and inferior to ARX788 at 5 mg/kg, again indicating ARX788 is more potent than T-DM1.
Overall, ARX788 was well tolerated at all evaluated dose levels. Neither body weight loss nor clinical abnormality was observed in any ARX788 treatment groups. ARX788 as a single agent produced statistically significant, dose-dependent antitumor activity against established HER2-expressing HCC1954 (breast) and NCI-N87 (gastric) xenograft tumors (Fig. 1D). As expected, ARX788 was not effective in the MDA-MB-468 xenograft model, a HER2-negative breast cancer cell line (Fig. 1C). These results indicated that ARX788 efficacy is HER2-mediated and demonstrated that ARX788 has stronger antitumor activity than T-DM1 in HER2-expressing tumors. Notable ARX788 antitumor activity has also been observed in cell lines with lower HER2 expression levels; results have been submitted for publication in a separate manuscript.
DMPK
pAF-AS269 is the only metabolite of ARX788 found in HER2-expressing cell lines
Incubation of ARX788 for 24 hours with HCC1954 cells, which express high levels of HER2, resulted in reduced cellular proliferation, lower viable cell count, and lower protein content compared with HER2-negative MDA-MB-468 cells (Supplementary Fig. S1). LC/MS-MS analysis, with detection channel (ion-pairs) in multiple reaction monitoring method (MRM) set-up to identify 18 potential metabolites in a pilot study, showed the presence of a single metabolite of ARX788, pAF-AS269, which is derived from proteolytic degradation of ARX788 after internalization by HER2-expressing cells. The total level of pAF-AS269 released from ARX788 increased with time in HER2-expressing cells (HCC1954) but not in HER2-negative MDA-MB-468 cells (Fig. 2A). After 24-hour incubation of ARX788 with HCC1954 cells, the pAF-AS269 metabolite was primarily detected inside the cells, with lower levels observed in the medium fraction (Supplementary Fig. S1). Taken together, these results support ARX788 binding to HER2 antigen at the cell surface, followed by internalization and proteolytic degradation, and resulting in the formation of pAF-AS269 as the major and perhaps only active metabolite. Incubation of pAF-AS269 with HCC1954 and MDA-MB-468 cells for 24 hours showed no metabolites of pAF-AS269 in either cell line.
The metabolism of ARX788 in vitro. A, The amount of pAF-AS269 released after incubation with HER2-positive HCC1954 and HER2-negative MDA-MB-468 cell lines. B, Metabolites of pAF-AS269 following incubation with human and cynomolgus hepatocytes. Three pAF-AS269 metabolites, M1, M2, and M3, all in low abundance (< 2%), are shown along with the proposed reaction sites marked in red boxes.
Metabolites of pAF-AS269, but not ARX788, identified in hepatocytes
After incubation of ARX788 in human and cynomolgus monkey hepatocytes for 120 minutes at 37°C, no catabolites of ARX788 were found. However, incubation of pAF-AS269 in cynomolgus monkey hepatocytes for 120 minutes at 37°C led to identification of three metabolites (M1, M2, and M3), all in low abundance of less than 2% (Supplementary Table S1). The proposed pAF-AS269 metabolite structures are illustrated in Fig. 2B.
Metabolite M1: Electrospray ionization mass spectrometry (ESI-MS) gave the [M + 2H]2+ ion at m/z 564.8481, 7.9973 Da higher than that of parent, suggesting addition of an oxygen atom. Thus, the species was formed either through oxidation or hydroxylation. Fragmentation of this ion gave product ions at m/z 466.2906, m/z 205.1000, and m/z 186.1133, which suggested that oxidation occurred at the pyrrolidine containing portion. M1 is proposed to be the oxidized metabolite of pAF-AS269.
Metabolite M2: ESI-MS gave the [M + H]+ ion at m/z 732.4902, 380.1950 Da lower than that of parent, suggesting dealkylation. Fragmentation of this ion gave product ions at m/z 700.4639, m/z 619.3992, m/z 520.3268, m/z 335.1664, and m/z 170.1123, which suggests loss of a methoxy group, further fragmentation of the pentapeptide ion and final loss of tetraethylene glycol p-acetyl phenylalanine moiety (loss of C19H28N2O6). M2 is proposed as the dealkylation metabolite of pAF-AS269.
Metabolite M3: ESI-MS gave the [M + 2H]2+ ion at m/z 549.84, 7.01 Da lower than that of parent, suggesting demethylation. Fragmentation of this ion gave product ions at m/z 205.10 and m/z 170.12, suggesting the loss of a methyl group from the tripeptide moiety. M3 is proposed as the demethylation metabolite of pAF-AS269.
On the basis of the mass spectrometric response, the metabolites M1, M2, and M3 were of low abundance (0.11%, 0.67%, and 1.46% of starting material, respectively) when compared with parent pAF-AS269. With human hepatocytes, only two metabolites of pAF-AS269, M2 and M3, were detected by ESI-MS following incubation of pAF-AS269, accounting for 0.21% and 0.62% of total mass peak area, respectively.
ARX788 was cleared preferentially by fecal excretion in rats
Following a single intravenous administration of [14C]-ARX788 at a dose of 20 mg/109 μCi/kg to SD rats, a total of 92.4% (including cage wash 0.877%) of the administered radioactivity was recovered in urine and feces 56 days postdose, while 9.54% of the administered radioactivity remained within the body (Fig. 3A). Fecal excretion (70.4%) was the dominant elimination route followed by urinary excretion (11.4%).
Mass balance and body distribution of ARX788 in rodents. A, Cumulative percentage dose recovery of 14C-ARX788 in urine, feces, and cage wash in a rat mass balance study. B, Biodistribution of 125I-ARX788 in tumor-bearing mice.
Biodistribution studies show HER2-targeted exposure of ARX788 in tumor tissues
The tissue distribution of 125I-ARX788 at various time points following intravenous injection in tumor-bearing mice is presented in Fig. 3B. Tissues sorted in descending order (from highest to lowest) by total radioactivity AUC were: serum, tumor tissue, adrenals, urine, lungs, kidneys, bladder, femurs, small intestine, heart, spleen, gonads, mesenteric lymph nodes, jejunum, muscle, liver, bone marrow, fat, intestinal contents, and brain. ARX788 concentrations in tumor tissue remained relatively stable at approximately 10 μg Eq/mL from 2 hours through 72 hours postdose, indicating HER2-targeted distribution into the tumor environment. In other tissues, ARX788 peaked 2 hours postdose followed by a gradual decrease in concentrations through 168 hours. Ratios of tumor to blood ARX788 concentrations were 0.25, 0.65, 1.46, and 1.64 at 2, 24, 72, and 168 hours postdose, respectively. The tissue biodistribution of 125I-anti-HER2 mAb in tumor-bearing mice was similar to that of 125I-ARX788 (Supplementary Fig. S2).
Prolonged serum stability of ARX788 in rodents and monkeys
The PKs of ARX788 was studied in tumor-bearing mice, rats, and monkeys (Table 1). Bioanalytical assays were developed to only detect intact ARX788 with DAR 2 (intact ADC) or all unconjugated and conjugated species (total antibody). Overall, the PK of ARX788 demonstrated slow clearance and a prolonged serum half-life in all species, similar to other IgG mAb therapies. PKs were similar between sexes in all species evaluated.
PK parameters following a single ARX788 dose in monkeys, rats, and tumor-bearing mice.
In mice bearing HER2-positive HCC1954 and NCI-N87 tumors, at dose levels of 1.67 and 5 mg/kg, ARX788 exposure was dose-proportional (Fig. 4A) with clearance and volume of distribution ranging from 0.0118 to 0.012 mL/minute/kg and 0.144 to 0.159 L/kg, respectively; and half-life (T1/2) ranging from 152 to 160 hours (Table 1). In rats, exposure to ARX788 increased with dose in a less than dose-proportional manner at dose levels ranging from 20 to 90 mg/kg (Fig. 4B). Clearance and volume of distribution ranged from 0.314 to 0.826 mL/h/kg and 125 to 71.6 mL/kg, respectively (Table 1). The resulting T1/2 ranged from 290 to 57.2 hours with an apparent trend for increase in clearance with increase in dose. It is possible at the high dose levels, PK may have been compromised either due to impact of toxicity or due to ADA formation. However, the exact reason for this trend is not understood.
ARX788 intact ADC serum concentrations over time in rodents and cynomolgus monkeys. A, The intact ADC concentrations in tumor-bearing mice at doses of 1.67 and 5 mg/kg. B, The intact ADC concentrations in rats at 20, 60, and 90 mg/kg doses. C, The intact ADC concentrations in cynomolgus monkeys at doses ranging from 0.3 to 30 mg/kg. D, The observed linearity of AUClast (Table 1) with corresponding doses from 0.33 to 30 mg/kg in cynomolgus monkeys.
In monkeys, PK was studied in a wide dose range from 0.3 to 30 mg/kg (Fig. 4C). Clearance ranged from 1.22 to 0.262 mL/hour/kg, with an apparent trend for decrease in clearance with increase in dose. A general linearity in ARX788 exposure was observed from 0.3 to 30 mg/kg dose in monkeys (Fig. 4D). The resulting T1/2 values ranged from 99.9 to 219 hours. The PK results from monkeys at high dose levels of 15 mg/kg and above must be interpreted with caution due to observed safety findings, and possible impact of ADA that was not studied within the single-dose studies.
The stable, site-specific conjugation of AS269 drug-linker with the anti-HER2 antibody results in high serum stability of ARX788 in the systemic circulation. The circulating levels of intact ADC and total antibody remained at similar levels throughout the duration of the monkey PKs and toxicity studies, demonstrating the AS269 drug-linker remains attached to the antibody and is not released prematurely into circulation (Fig. 5A and B). pAF-AS269, the sole metabolite of ARX788 identified in the in vitro metabolism studies (Fig. 2A), was only measurable at later time points at very low quantities near the lower limit of quantitation (LLOQ) of the assay (Fig. 5A and B). This data confirms the stability of ARX788, which should maximize targeted delivery of the payload to HER2-expressing cells and limit off-target toxicity. In addition, no marked drug accumulation was observed for ARX788 following administration of four doses of ARX788 given 21 days apart in cynomolgus monkeys (Fig. 5B).
ARX788 stability in cynomolgus PK studies. A, Serum concentrations of intact ADC, total antibody, and pAF-AS269 in monkeys after a single dose of ARX788 at 1.25, 5, and 30 mg/kg doses. B, Serum concentrations of intact ADC, total antibody, and pAF-AS269 after the first and the fourth doses at 3.3 and 10 mg/kg dose levels in a ARX788 repeat-dose study in monkeys. The pAF-AS269 at 3.3 mg/kg is detected in only one animal; the other two animals show no detectable levels of pAF-AS269.
ARX788 is tolerated with the highest nonseverely toxic dose of 10 mg/kg in monkeys
A single-dose administration of ARX788 was evaluated in SD rats and cynomolgus monkeys. A repeat-dose study was also performed in monkeys with four repeat doses at every 3 weeks dosing regimen, followed by an 8-week recovery period for the observation of reversibility, persistence, and/or delayed occurrence of toxic effects. Following a single dose, ARX788 was tolerated in rats and monkeys up to 90 and 20 mg/kg, respectively, indicating that monkey was the more sensitive species. The observed difference in sensitivity is likely due to cross-reactivity of ARX788 with monkey HER2, but not with rat HER2 (22). In the repeat-dose study, monkeys tolerated four repeat doses of ARX788 up to 10 mg/kg, which was designated as the highest nonseverely toxic dose (HNSTD) based on the overall toxicity profile and reversible nature of the observed toxicity findings.
Detailed ARX788 toxicity profiles and comparisons between rat and monkey are summarized in Table 2. Clinical observations including decreased body weight, decreased activity, hunched posture, inappetence, changes in serum chemistry parameters, and histopathology findings in kidney and gastrointestinal tract at doses ≥10 mg/kg/dose were considered adverse due to the severity. Body weight decreases persisted to the recovery phase for males at all dose levels; however, there were no significant body weight changes for the females at any dose level by the end of recovery phase. ARX788-related pathologic findings in the kidney and rectum for animals in the 10 mg/kg/dose group were considered adverse because of the potential for detrimental effects on organ function and/or overall health of the animals. All other findings were considered nonadverse because of their minimal or mild severity and/or the lack of probability for affecting the overall health of the animals. Changes in hematology were mostly resolved by the end of the recovery phase and were considered ARX788-related, but nonadverse due to the transient nature, minimal effect and/or reversibility.
ARX788-related observed toxicity in rat (single dose only) and monkey (single and repeat dose) safety studies.
Clinical chemistry changes observed in levels of AST, GGT, TCHO, TG, TP, A/G, UREA, and CRE were correlated with microscopic findings in kidney and liver, and they were mostly recovered by the end of the recovery phase except for increased TCHO, TG, and decreased A/G noted at 10 and/or 15 mg/kg/dose. Changes in body weight and serum chemistry (AST and ALT) in the repeat-dose monkey study are presented in Supplementary Figs. S3 and S4, respectively. Because of the reversibility of most of the histopathologic and clinical observations, and all animals tolerated four doses of ARX788 at 10 mg/kg through in-life and recovery phases, the dose of 10 mg/kg was determined as the HNSTD for both males and females, and 3.3 mg/kg was the no observed adverse effect level dose. Repeat-dose studies were not conducted in rats.
Antidrug antibodies
Antidrug antibodies (ADA) were detected in the repeat-dose toxicology studies in cynomolgus monkeys (Supplementary Fig. S5). The overall incidence of the presence of ADAs was 3 of 12 animals (33.3%) in the 3.3 mg/kg dose group, 2 of 12 (16.7%) animals in the 10 mg/kg dose group, and 4 of 14 (21.4%) animals in the 15 mg/kg dose group; the presence of ADAs was observed on days 64, 99, or 120. Consistent with an ADA response, decreased serum levels of ARX788 were observed in all animals with positive ADA results.
Discussion
HER2 is one of a family of four EGFR tyrosine kinases and is overexpressed in a subset of breast and gastric cancer tumors (23). ARX788 is a novel ADC with a nonnatural amino acid, pAF, site-specifically incorporated into an anti-HER2 antibody, then chemically conjugated to an antitubulin cytotoxic drug-linker, AS269, via an oxime bond. The optimization of conjugation site, linker design, and DAR homogeneity to render high stability to an ADC molecule in the systemic circulation was found to be of paramount importance to limit off-target toxicity in the clinic (19, 20). Thus, ARX788 may exhibit less off-target toxicity and a favorable therapeutic window in the clinic. Here, we report the key findings from nonclinical studies which support the clinical development of ARX788.
ARX788 showed dose-dependent antitumor activity, and superior efficacy when compared with equivalent doses of T-DM1, in the HER2-positive HCC1954 breast and NCI-N87 gastric cancer xenograft models (Fig. 1). However, no antitumor activity was observed with ARX788 in the HER2-negative MDA-MB-468 xenograft model, indicating the selectivity of ARX788. Previous studies have also shown treatment with an isotope control ADC did not result in any antitumor activity (22), thus demonstrating ARX788 activity is mediated by HER2 expression in tumor cells. Given the antibody used for both ARX788 and T-DM1 are similar, the superior efficacy of ARX788 versus T-DM1 could potentially be attributed to the difference in released payload, as well as the site-specific conjugation technology which results in a highly stable ADC as demonstrated in PK and metabolism studies.
For conventional ADCs, such as T-DM1, the half-life of ADC is shorter than that of total antibody and the released free payload concentrations are proportional to ADC concentrations (24–27). However, in the case of ARX788 in rat and monkey tox studies, the serum concentration profiles for intact ADC were overlaid with that of total antibody, and the half-life values were similar for intact ADC and total antibody (Fig. 5). The ARX788 PK profile in monkeys followed a typical mAb PK profile with slow clearance and low volume of distribution, and a trend for target-mediated drug disposition resulting in a decrease in clearance with increase in dose levels followed by dose-linearity.
The only detectable metabolite, pAF-AS269, appeared slowly in the systemic circulation at low concentrations near the assay LLOQ, suggesting ADC internalization and catabolism led to the release of pAF-AS269. In a repeat-dose cynomolgus monkey toxicity study of ARX788 (Q3W × 4 doses), low systemic levels of pAF-AS269 were observed at or just above the LLOQ of 0.2 ng/mL for the 10 mg/kg ARX788 dose group. After the final and fourth dose of 10 mg/kg ARX788 on day 64, a Cmax of approximately 0.5 ng/mL was observed indicating very low levels of metabolite were detected in the systemic circulation.
Furthermore, the stability of ARX788 and the metabolite pAF-AS269 were studied in human and monkey hepatocytes, and in cell lines with and without HER2 expression (Fig. 2A). Incubation of ARX788 in HER2-positive cells for 24 hours resulted in a single metabolite, pAF-AS269, with no evidence for the formation of free AS269 drug-linker. Similarly, incubation of ARX788 in hepatocytes for 120 minutes did not result in any detectable metabolites. Overall, the metabolism studies demonstrated that AS269 drug-linker was not released following ARX788 incubation in hepatocytes or HER2-positive cell lines, confirming the high stability of the noncleavable linker and the oxime bond formed after site-specific conjugation.
In studies to evaluate pAF-AS269 metabolism, incubation of pAF-AS269 with human and monkey hepatocytes resulted in the identification of three metabolites in low abundance (<2%) as products of oxidation, dealkylation, or demethylation (monkey only). Notably, no peaks that correlated to AS269 were observed. Additional metabolism studies for pAF-AS269 such as reaction phenotyping and Cytochrome P450 proteins (CYP) inhibition and induction studies are planned to understand the potential for drug–drug interactions.
A mass balance study of 14C-ARX788 in rats demonstrated fecal excretion to be the dominant route with >70% of the dose recovered in feces, followed by urinary excretion at 11.4% of the dose. Therefore, it is possible that bile plays a key role in the excretion of ARX788.
Tissue distribution studies of 125I- ARX788 in tumor-bearing mice showed targeted distribution into the tumor environment as soon as 2 hours postdose, through 72 hours postdose. It should be noted that 125I radiolabels are nonresidualizing in nature. After ARX788 is catabolized inside the cells, the nonresidualizing free 125I can be effluxed out, limiting the ability to measure sustained accumulation or tumor uptake. Therefore, the observed distribution into tumor tissues does not reflect sustained accumulation; instead, it indicates relative targeted distribution of ARX788 into the tumor microenvironment at measured time points. To note, the mechanism of ADCs eventually leads to cell lysis after release of payload from ADC catabolism, which could pose inherent challenges to studying sustained accumulation into tumor tissues.
In vitro binding studies confirmed ARX788 bound to both human and cynomolgus HER2 with similar affinity, but not to rat HER2 (22). Therefore, ARX788 safety studies were conducted in rat and monkey to study antigen-independent and antigen-dependent toxicities. The safety profiles in both rats and monkeys were generally similar with major safety findings attributed to liver and kidney-related toxicities, which correlated to changes in serum chemistry parameters and microscopic findings of liver necrosis and nephropathy (Table 2). On the basis of the tolerated dose levels, monkey was the more sensitive species.
In tissue cross-reactivity studies in monkeys (Supplementary Table S2), specific biotin-ARX788 staining was observed in the bladder, breast, cerebellum, cerebral cortex, colon, eye, fallopian tube, small intestine (duodenum), kidney (glomerulus, tubule), lung, pancreas, prostate, skin, spinal cord (thoracic), thyroid, tonsil, ureter, and uterus (cervix and endometrium). Although there was no major toxicity observation specific to monkeys, HER2 expression in several tissues suggests the observed toxicity in monkeys could be a combination of on-target activity of ARX788 as well as off-target distribution of ARX788. Notably, neutropenia and thrombocytopenia are generally the major safety concerns in the clinic for first-generation ADCs, including T-DM1. However, neutropenia and thrombocytopenia were not predominant in the ARX788 rat or cynomolgus studies, suggesting the benefit of site-specific conjugation and the highly stable nature of ARX788 (28).
Overall, ARX788 has shown consistent antitumor activity across several HER2-expressing breast and gastric cancer tumor models. In a breast cancer xenograft model, significant antitumor activity was observed at 1.67 mg/kg with a corresponding AUC of approximately 1,200 μg*hour/mL. The observed exposure in monkeys at the HNSTD of 10 mg/kg was 23,000 μg*hour/mL, indicating the therapeutic exposure was achieved with a favorable therapeutic window. Allometric scaling based on monkey PK parameters projected the human efficacious dose could range between 0.33 and 1 mg/kg. A starting dose of 0.33 mg/kg, based on 1/10 of the HNSTD in monkey and 1/3 human to monkey body surface area conversion factor, was thus recommended for the first-in-human phase I study (21). The starting dose of ARX788 in clinic is similar to the starting dose for T-DM1 as per published phase I PK results (27).
In conclusion, ARX788 with Ambrx novel site-specific conjugation offers unique advantages with superior efficacy, stability in systemic circulation, and a favorable safety profile. Thus, ARX788 is a promising next-generation ADC for the treatment of breast and gastric cancers with HER2 overexpression and is currently being evaluated in phase I clinical trials (NCT03255070).
Disclosure of Potential Conflicts of Interest
At the time of publication, Prathap Nagaraja Shastri is an Associate Scientific Director at Janssen Research and Development, LLC. L. Skidmore is an Associate Director at Ambrx. Y. Gu is a Head of Research at Transcenta Holdings. S. Yao is a Senior Director at Ambrx. G. Xia is a CSO at NovoCodex Biopharmaceuticals. No potential conflicts of interest were disclosed by the authors.
Authors' Contributions
Conception and design: Y. Gu, F. Tian, S. Yao, G. Xia
Development of methodology: P.N. Shastri, J. Zhu, L. Skidmore, Y. Ji, G. Xia
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.N. Shastri, J. Zhu, Y. Ji, S. Yao, G. Xia
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P.N. Shastri, J. Zhu, L. Skidmore, Y. Ji, G. Xia
Writing, review, and/or revision of the manuscript: P.N. Shastri, J. Zhu, L. Skidmore, Y. Gu, F. Tian, S. Yao, G. Xia
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P.N. Shastri, J. Zhu, L. Skidmore, X. Liang, Y. Ji, F. Tian, S. Yao, G. Xia
Study supervision: X. Liang, F. Tian, S. Yao, G. Xia
Acknowledgments
We would like to thank Sukumar Sakamuri from Ambrx for helpful discussion in the preparation of manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Footnotes
Note: Supplementary data for this article are available at Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/).
Mol Cancer Ther 2020;19:1822–32
- Received July 12, 2019.
- Revision received December 1, 2019.
- Accepted June 2, 2020.
- Published first June 4, 2020.
- ©2020 American Association for Cancer Research.
References
Volume 19, Issue 9
