Discovery and Characterization of Brigimadlin, a Novel and Highly Potent MDM2–p53 Antagonist Suitable for Intermittent Dose Schedules

The tumor protein p53 is a pivotal tumor suppressor protein and a mainstay of the body’s cellular anticancer defense mechanisms (1). p53 inactivation promotes the evasion of cell cycle arrest and apoptosis (2–4), and is essential for the initiation and the maintenance of tumors (5–7). In many cancers, the function of p53 is disrupted through loss-of-function mutations in the TP53 gene or increased levels of negative regulators, including the proto-oncogene mouse double minute 2 (MDM2). MDM2 inhibits p53 transcriptional activity, induces export of p53 from the nucleus, and acts as a p53 ubiquitin ligase that induces the proteasome-mediated degradation of p53 (4, 8, 9).

MDM2 amplification occurs in ∼3% of all cancers (10), with higher rates in certain tumor types (11). MDM2 amplification is frequently observed in tumors with wild-type p53 status, implying that amplification of MDM2 is the principal mechanism of p53 inactivation in these tumors. Analysis of patient tumor data derived from the American Association for Cancer Research (AACR) GENIE public database (v13.0; ref. 10) reveals that 94% of liposarcoma [LPS; including dedifferentiated liposarcoma (DDLPS) and well-differentiated liposarcoma (WDLPS)], 8% of bladder urothelial carcinoma, and 5% of biliary tract adenocarcinoma are both wild-type tumor protein p53 (TP53wt) and MDM2-amplified carcinomas (Methods and data in Supplementary Appendix 1).

Targeting MDM2 to restore p53 tumor suppressor activity is a promising therapeutic approach for patients with TP53wt tumors (12, 13). Initial clinical investigations indicated that MDM2–p53 antagonists showed preliminary signs of antitumor efficacy; however, treatment was associated with thrombocytopenia and neutropenia (14–16), likely resulting from inhibition of megakaryopoiesis and megakaryocyte differentiation upon p53 activation (17). Intermittent dosing has been proposed as an approach to increase the therapeutic window of MDM2–p53 antagonists by maintaining antitumor activity while mitigating target-related side effects (18). Clinically, this can only be achieved with an inhibitor that possesses excellent potency and pharmacokinetic (PK) properties. Spiro-oxindoles represent a promising class of MDM2–p53 antagonists (19–21) that historically have been limited by poor stability, leading to epimerization. Our goal was to discover and characterize a more stable alternative with improved potency and PK characteristics as well as good tissue penetration that would allow intermittent dosing schedules and thereby limit target-related adverse events in patients. Here, we present the discovery and properties of brigimadlin (BI 907828), a novel and highly potent investigational spiro-oxindole MDM2–p53 antagonist that disrupts the protein–protein interaction between MDM2 and p53. Brigimadlin has shown encouraging early clinical activity in patients with advanced TP53wt, MDM2-amplified solid tumors (22, 23) and LPS (24).

BI-0252 and BI-0282 were synthesized as previously described (21, 25). The synthesis of brigimadlin is described in detail in Supplementary Appendix 2. Alrizomadlin and milademetan were synthesized at WuXi AppTec, China.

Methods for MDM2 protein expression for crystallography are described in the Supplementary Methods. Crystallography data were collected at the Swiss Light Source (Paul Scherrer Institute) beamline PXII and processed using autoPROC (26) and STARANISO (27). Resolution cutoffs were determined using the default settings of STARANISO. The structure was solved using Protein Data Bank (PDB): 6I3S (28) as a search model. The structure was refined in alternating cycles of Crystallographic Object-Oriented Toolkit (29) and autoBUSTER (30). The refined structure was deposited at Research Collaboratory for Structural Bioinformatics (RCSB) PDB with ID: 8PWC.

Bidirectional permeability of test compounds across a Caco-2 cell monolayer was measured as previously described (31).

Cells (Supplementary Table S1) were cultured under standard conditions appropriate for the cell type [37°C, 5% CO₂ (or without CO₂ when cultured in Leibovitz’s L-15, e.g., SW982 cells), 95% humidity] in recommended growth medium. All cell lines were obtained from the American Type Culture Collection (ATCC), Centro Nacional de Investigaciones Oncológicas, Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (DSMZ), Japanese Collection of Research Bioresources Cell Bank (JCRB), the Korean Cell Line Bank (KCLD), and the National Institutes of Health (NIH; Supplementary Methods). All cell lines were authenticated by short tandem repeat analysis at Eurofins by Boehringer Ingelheim and were regularly tested for Mycoplasma (MycoAlert Mycoplasma Detection Kit, Lonza; Supplementary Methods).

AlphaScreen (PerkinElmer), a bead-based, nonradioactive amplified luminescent proximity homogeneous assay, was used to determine the half-maximal inhibitory concentration (IC₅₀) for the MDM2–p53 protein–protein interaction for BI-0252, BI-0282, and brigimadlin (Supplementary Methods). An AlphaScreen was also used to assess the MDMX–p53 protein–protein interaction IC₅₀ for BI-0282 and brigimadlin (Supplementary Methods).

Cells were plated in a 96-well format in recommended growth medium, and a serial dilution of brigimadlin was added 24 hours after cell seeding. At the same time, a time zero (T₀) untreated cell plate was measured. After 72- to 120-hour compound incubation, plates were assayed with alamarBlue (Invitrogen) or CellTiter-Glo (Promega) cell viability reagent according to the manufacturer’s instructions and measured using a multimode microplate reader (PerkinElmer). IC₅₀, GI₅₀, and GI₁₀₀ values were obtained by generating a concentration–response curve using a four-parameter log-logistic function. The IC₅₀ values were defined as 50% of maximal inhibitory concentration between 0 and the max response. GI₅₀ refers to half-maximal inhibitory concentration between the T₀ value and the max response. The complete inhibition of cell proliferation (GI₁₀₀) was calculated as the concentration at which the measurement of the treated cells was equal to the mean of T₀ measurements. Unbound IC₅₀ were calculated as 4.5% (unbound fraction in 10% fetal calf serum) of the total brigimadlin IC₅₀.

Profiling Relative Inhibition Simultaneously in Mixtures (PRISM) is a high-throughput multiplexed cancer cell line viability screening platform developed by the Broad Institute of the Massachusetts Institute of Technology (MIT) and Harvard (32, 33). Brigimadlin was tested in the PRISM assay on 931 cell lines at eight-point dose dilutions (threefold step dilutions) with a treatment duration of 5 days. The minimum and maximum doses were set at 0.00457 and 10 µmol/L, respectively. A total of 859 cancer cell lines passed technical quality control, but only 827 allowed for curve fitting and were not duplicated across PRISM cell line pools. Data processing from raw intensity values to computed area under the curve (AUC) values for each dose–response curve are described in the Supplementary Methods. The analysis code is publicly available on the code repository GitHub (https://github.com/cmap/dockerized_mts). For further details, see Supplementary Methods.

Cells were seeded 24 hours prior to drug treatment. After incubation under indicated conditions, cells were lysed on ice with buffer containing 20 mmol/L Tris pH 7.5, 150 mmol/L NaCl, 1 mmol/L EDTA, 1 mmol/L EGTA, and 1% Triton X-100, freshly supplemented with protease and phosphatase inhibitor (Thermo Fisher Scientific). Homogenates were centrifuged at 10,000 × g for 10 minutes at 4°C to clear cellular debris from the lysate.

Snap-frozen tumor tissue was disrupted and homogenized on ice with buffer containing 20 mmol/L Tris pH 7.5, 150 mmol/L NaCl, 1 mmol/L EDTA, 1 mmol/L EGTA, and 1% Triton X-100, supplemented with protease and phosphatase inhibitor (Thermo Fisher Scientific) using stainless steel beads and a TissueLyser (Qiagen). Homogenates were centrifuged at 10,000 × g for 10 minutes at 4°C.

Total protein concentration in lysates was measured using the Bio-Rad Protein Assay (Bio-Rad Laboratories) according to the manufacturer’s instructions. Equal amounts of total protein were separated by 4% to 12% Bis-Tris polyacrylamide gel electrophoresis (Bio-Rad Laboratories), transferred onto polyvinylidene difluoride (PVDF) membranes (Bio-Rad Laboratories), and then hybridized to specific primary antibodies (Supplementary Table S2) and horseradish peroxidase-conjugated secondary antibody (Agilent Technologies) for subsequent detection by enhanced chemiluminescence (Amersham GE Healthcare).

Sample DNA was isolated using DNeasy Blood & Tissue Kit (Qiagen). DNA quantity and quality were evaluated using a NanoDrop spectrophotometer (Supplementary Methods).

Cell or tumor lysates were prepared according to the Meso Scale Discovery (MSD) cell lysis protocol and analyzed in duplicate with MSD Apoptosis Panel Whole Cell Lysate Kit according to the manufacturer’s instructions. Plates were read on the SECTOR Imager 6000 plate reader. Duplicate readings were averaged and the average blank subtracted. Data were plotted as fold change relative to vehicle.

Cells were seeded in 96-well tissue culture plates 24 hours prior to drug treatment. After incubation with 10 and 100 nmol/L brigimadlin for 24 or 48 hours, cells were lysed using FastLane technology (Qiagen). FastLane lysates were directly used in real-time one-step reverse transcription PCR (RT-PCR) and run on a StepOnePlus Real-Time PCR instrument (Applied Biosystems). Conditions for one-step RT-PCR were 20 minutes at 50°C and 15 minutes at 95°C, followed by 45 cycles of 20 seconds at 95°C and 45 seconds at 60°C. TaqMan primer probe sets are listed in Supplementary Table S3. All were purchased from Applied Biosystems. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH, Thermo Fisher Scientific) was used as endogenous control.

Cells were seeded 24 hours prior to drug treatment. After incubation with 10 or 100 nmol/L brigimadlin for 48 hours, cells were lysed with RLT buffer (Qiagen). Total RNA from cell lysates was extracted using the QIAshredder kit (Qiagen) for cell homogenization and the RNeasy Mini Kit (Qiagen) for RNA isolation. Complementary DNA (cDNA) synthesis was performed using the SuperScript VILO cDNA Synthesis Kit (Life Technologies) according to the manufacturer’s instructions. qPCR using QuantiTect Multiplex PCR Kit (Qiagen) was run on a StepOnePlus Real-Time PCR instrument (Applied Biosystems). Conditions for two-step RT-PCR were 8 minutes at 95°C, followed by 45 cycles of 45 seconds at 94°C and 45 seconds at 60°C.

Snap-frozen tumor tissue was disrupted and homogenized on ice with QIAzol lysis reagent using stainless steel beads and a TissueLyser (Qiagen). Total RNA was obtained using RNeasy Plus Universal Mini Kit (Qiagen) according to the manufacturer’s protocol. RNA was converted to cDNA using SuperScript VILO cDNA Synthesis Kit (Life Technologies). qPCR with target genes and GAPDH as endogenous control was run on a StepOnePlus Real-Time PCR instrument (Applied Biosystems). Conditions for two-step RT-PCR were 8 minutes at 95°C, followed by 45 cycles of 45 seconds at 94°C and 45 seconds at 60°C.

Target messenger RNA (mRNA) levels were normalized against GAPDH levels. The ∆∆Ct method was used to compare the relative expression, and fold change in gene expression relative to vehicle control was calculated. RT-PCR data are represented as mean ± standard deviation for repeats.

Plasma protein binding of brigimadlin was assessed in vitro by equilibrium dialysis (DIANORM equilibrium dialysis device, DIANORM) using plasma from mouse, rat, minipig, and humans and protein solutions of human serum albumin and human alpha-1-acid-glycoprotein (hAGP; see Supplementary Methods for plasma collection). Intravenous and oral PK studies of brigimadlin were performed in mouse (BomTac:NMRI-Foxn1^nu, female), rat [Crl:WI(Han) male], dog (Chbb:Beagle, male), and mini pig (Göttingen, female). Quantification of brigimadlin from plasma or other media was performed by high-performance liquid chromatography–mass spectrometry (HPLC-MS; Sciex API 5000, Agilent 1200 dual pump, CTC autosampler) on an Xbridge BEH C18 2.1 × 50 mm 2.5 µm column with a 2-minute gradient method (A: 5 mmol/L NH₄Ac pH 4.0, B: ACN + 0.1% HCOOH; 0 minutes 95% A, 1 minute 5% A, 1.3 minutes 5% A, 1.4 minutes 95% A, 2 minutes 95% A). Quantification and PK interpretation were carried out at Boehringer Ingelheim RCV GmbH & Co. KG, Vienna. The animal experiments in mouse were carried out at Boehringer Ingelheim RCV GmbH & Co. KG, Vienna the experiments in rat at WUXI in Shanghai, and the experiments in dog and mini pig at the Laboratory of Pharmacology and Toxicology in Hamburg, in accordance with local law.

For pharmacodynamic (PD) biomarker studies in SJSA-1 tumor-bearing NMRI nude mice to assess GDF15 (growth/differentiation factor-15, or MIC-1; Supplementary Methods), serum GDF15 (protein) and tumor GDF15 (mRNA) were measured 24 hours after a single oral administration of brigimadlin (0.25, 0.75, 2, or 10 mg/kg) or vehicle (n = 3 mice for each treatment group). Serum GDF15 was quantified using an enzyme-linked immunosorbent assay [ELISA; human GDF15/MIC-1 Abcam #ab155432; measurements were taken at 450 nmol/L, in a Spectramax Plus (Molecular Devices); see Supplementary Methods]. Tumor GDF15 mRNA was assessed using a two-step RT-PCR assay (as above).

To evaluate the correlation of PK with PD biomarkers (p53 target genes and apoptosis), SJSA-1 tumor-bearing BALB/c nude mice received brigimadlin orally (p.o.; gavage; application volume 10 mL/kg) at 1 mg/kg twice daily (BID) × 1-day dosing schedule (sampled at 2, 4, 7, 9, 14, 24, and 31 hours after first dose) or at 2 mg/kg dose level as a single dose (sampled at 2, 4, 7, 24, and 31 hours after dosing; Supplementary Methods). Plasma concentrations of brigimadlin were measured by HPLC (Waters + API 6500 LC-MS/MS system; Supplementary Methods). PD biomarker analysis was carried out at Boehringer Ingelheim RCV GmbH & Co. KG, Vienna, using a two-step RT-PCR and the MSD apoptosis assay.

SJSA-1 tumor-bearing BALB/c nude mice (n = 60) were randomized into five treatment groups (n = 10 per group) when mean tumor size reached ∼150 mm³. Mice received brigimadlin or vehicle (0.5% Natrosol; gavage; application volume 10 mL/kg p.o.), either BID × 1 day per week (QW; brigimadlin 1 mg/kg) or as a single dose QW (brigimadlin 1, 1.5, or 2 mg/kg). Randomization was performed based on a “matched distribution” method (Study Director software, version 3.1.399.19).

Patient-derived tumor-bearing BALB/c nude mice (SA3283, LU0861, LU6903) were randomized when mean tumor size reached ∼200 mm³. Twenty mice per model were randomized to brigimadlin 2 mg/kg p.o. QW [for 2 weeks (SA3283), 3 weeks (LU0861), or 5 weeks (LU6903)] or to vehicle (n = 10 per group). Randomization was performed as for the SJSA-1 xenograft model.

Mice were monitored daily for morbidity and mortality; tumor volumes were measured three times per week (see Supplementary Methods for details of measurement of tumor volume and calculation of tumor growth inhibition). All statistical analyses were performed in R v3.3.1. All tests are two sided unless otherwise specified, and P-values of <0.05 were regarded as significant.

Animal experiments were approved by the Institutional Animal Care and Use Committees of Boehringer Ingelheim RCV GmbH & Co. KG (Vienna, Austria) or Crown Bioscience Inc. (Taicang, China).

Exploratory PK parameters of brigimadlin were determined from plasma analysis of a sub-cohort of patients in the expansion cohort of the phase Ia/Ib dose escalation trial (Clinicaltrials.gov identifier: NCT03449381) after a single oral dose of 45 mg on Day 1 of Cycle 1 using non-compartmental analysis and planned sampling times. For PK analysis, 3-mL blood samples were taken using EDTA as anticoagulant, and plasma was obtained subsequently. Brigimadlin was quantitated by validated liquid chromatography–mass spectroscopy (LC-MS/MS) methods using a calibration curve in the range of 20 to 20,000 nmol/L. PK parameters were calculated using SAS. NCT03449381 was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines, it was approved by all relevant institutional review boards, and all patients provided written informed consent.

EDTA plasma samples were taken from patients in the expansion cohort of NCT03449381 after a single oral dose of 45 mg on Day 1 of Cycle 1. Planned sampling times were 5 minutes before and 4, 8, 24, 48, and 72 hours after brigimadlin administration. Circulating GDF15 was measured using the Quantikine ELISA Human GDF15 Immunoassay (R&D Systems; Supplementary Methods).

Brigimadlin plasma exposure was simulated based on a population PK model. Simulations were performed for a dose range of 30 to 60 mg brigimadlin and for standard body weight of 70 kg. Calculations of unbound concentrations in human and mice were based on f_u (unbound fraction) of 0.22% for human and 0.34% for mice.

Data, scientific, and medical researchers can request access to clinical study data when it becomes available on https://vivli.org/ and earliest after publication of the primary manuscript in a peer-reviewed journal, regulatory activities are complete, and other criteria are met. Please visit https://www.mystudywindow.com/msw/datasharing for further information.

The p53-binding pocket of MDM2 represents a challenge for inhibitor design due to the unique orientation of a hydrophobic triad of three sub-pockets, named after the p53 amino acids which bind to them (the Leu26, Trp23, and Phe19 pockets). Optimal filling of the entire hydrophobic triad is required for a compound to display sufficient potency. The spiro-indolinone scaffold has emerged as an ideal complement for the MDM2-binding pocket hydrophobic triad. The first generation of spiro[3H-indole-3,3′-pyrrolidin]-2(1H)-one MDM2–p53 antagonists was prone to epimerization at the carbons at positions 2′and 3′ by a retro-Mannich/Mannich reaction in solution (20, 21). We addressed the stability issue of the first generation of spiro-oxindole MDM2–p53 antagonists by designing a spiro[3H-indole-3,2′-pyrrolidin]-2(1H)-one scaffold, in which the nitrogen of the pyrrolidine ring was moved by one position, with the hypothesis that this should avoid epimerization. Other groups addressed this problem by attaching a spirocycle to the pyrrolidine ring as seen for the clinical compounds milademetan and alrizomadlin (34). Compound 2 (Fig. 1A), which only occupies two of the three lipophilic sub-pockets of MDM2 (Leu26 and Trp23 pockets), showed comparable in vitro potency to similar spiro[3H-indole-3,3′-pyrrolidin]-2(1H)-one inhibitors (21). Addressing the Phe19 pocket by introduction of a cyclopropylmethyl group to the pyrrolidine ring and further structure-based optimization by rigidification led to BI-0252 (Fig. 1A). The synthesis and characterization of this molecule have previously been described (21). To further improve binding affinity, we designed compounds for which the side chain of His96_MDM2 had an H-bond acceptor instead of an H-bond donor (NH in BI-0252). In the first series of compounds, we replaced the secondary amine of BI-0252 with the carbonyl group of an amide. This yielded more potent compounds in cellular assays that were limited due to unfavorable PK properties. The findings within this series have previously been described (28).

A second series of compounds aimed to address the His96_MDM2 interaction with an H-bond acceptor nitrogen of a five-membered aromatic heterocycle. This led to the structurally more rigid 2H-indazole derivative, BI-0282 (Fig. 1A), which showed improved potency and PK properties compared with BI-0252 [IC₅₀ in SJSA-1 (TP53wt) cells in vitro: BI-0252, 471 nmol/L; BI-0282, 163 nmol/L; Supplementary Table S4]. The synthesis and characterization of BI-0282 have previously been described (25). An X-ray structure of BI-0282 suggested that the interaction with His96_MDM2 was likely still not optimal. Further optimization resulted in the highly rigidified tetra annulated scaffold of brigimadlin which showed low efflux but high permeability and solubility in simulated gastrointestinal fluids (Supplementary Table S5). Finally, placing a methyl group in the ortho position to the carboxylic acid further improved the PK properties, resulting in brigimadlin (Fig. 1A and B). We obtained an X-ray co-crystal structure of brigimadlin that confirmed that brigimadlin binds to the p53 pocket of MDM2. The indolinone group binds in the Trp pocket, and the indolinone -NH proton forms a hydrogen bond to the backbone carbonyl of Leu54. The Leu pocket is filled with a halogenated phenyl group, which also is positioned for a pi–pi interaction with His96, while the Phe pocket is addressed with a cyclopropyl methyl group. The indazole group forms a hydrogen bond to the side chain of His96.

The challenging synthetic sequence (<3% overall yield, 10-step longest linear sequence) was optimized to a new synthetic route, featuring a global reduction-reductive amination sequence and an oxidative indazole closure leading directly to the desired scaffold in significantly improved overall yields.

Brigimadlin demonstrated higher potency than BI-0252 and BI-0282 in the AlphaScreen assay and the MDM2-amplified SJSA-1 cancer cell line in in vitro viability assays but retained selectivity versus MDMX, a p53-interacting protein with a high homology to MDM2 (Supplementary Table S4). Brigimadlin demonstrated lower IC₅₀ values in cell lines with TP53wt than in TP53mut cell lines (Supplementary Tables S4 and S6). A cell panel viability screen (PRISM) was carried out to investigate the therapeutic potential of brigimadlin across multiple tumor types and explore additional biomarkers of response (32, 33). In the 827 cell lines that passed technical quality control, allowed for curve fitting, and were not duplicated across PRISM cell line pools, the assays revealed a wide range of sensitivities (IC₅₀ range: 0.0046–9.8 µmol/L; Supplementary Appendix 3). Based on plotting the sensitivity profile (1-AUC) and curve fitting, three distinct response classes were defined: 221 cell lines were highly sensitive to brigimadlin (sensitivity >0.5, IC₅₀ range 0.0046 to 0.2035 µmol/L, median IC₅₀: 0.1074 µmol/L), 40 cell lines showed an intermediate response to brigimadlin (0.25≥ sensitivity ≤0.5, IC₅₀ range 0.0662 to 7.2098 µmol/L, median IC₅₀: 1.2399 µmol/L), and 566 cell lines showed resistance to brigimadlin (sensitivity <0.25, IC₅₀ range 2.1880 to 9.7898 µmol/L, median IC₅₀: 7.2740 µmol/L).

Whole-exome mutations, somatic copy number alterations, and gene expression profiles were matched to the response profile of the cell line panel to explore biomarkers of sensitivity to brigimadlin. The strongest predictor of sensitivity to brigimadlin was TP53 mutation status. In cell lines with available TP53 mutation data (822/827), 90% (198/221) of the sensitive cell lines had TP53wt status, and 83% (469/563) of the resistant cell lines had TP53mut status (Fig. 2A).

The subset of highly sensitive TP53wt cell lines was enriched for tumor types such as liposarcoma, renal cell carcinoma, rhabdomyosarcoma, and melanoma (Fig. 2B). Some tumor types (e.g., non–small cell lung cancer) showed a bimodal distribution with respect to sensitivity, whereas exocrine tumors (pancreas cancer) or small-cell lung cancer cell lines (SCLC) were generally more resistant. However, these cancer types were represented by relatively few cell lines (SCLC: n = 4; pancreatic cancer: n = 11), making it difficult to draw firm conclusions about potential resistance mechanisms.

Sensitivity to brigimadlin was further examined across selected TP53wt MDM2-amplified and non-amplified cancer cell lines (Supplementary Table S7). MDM2 mRNA expression and protein levels were observed to be higher in MDM2-amplified versus non-amplified cell lines (Fig. 3; Supplementary Fig. S1; Supplementary Table S7). The IC₅₀ values for brigimadlin were similar in MDM2-amplified cell lines (range 5 to 240 nmol/L) versus non-amplified cell lines (range 7 to >2,000 nmol/L; Supplementary Table S7). However, GI₁₀₀ values were generally lower in MDM2-amplified cell lines (GI₁₀₀ values for the three most sensitive lines: RH-18, 11 nmol/L; 94T778, 12 nmol/L; SJSA-1, 20 nmol/L) than in MDM2 non-amplified cell lines (GI₁₀₀ values for the three most sensitive lines: SNU-719, 31 nmol/L; LS 174T, 49 nmol/L; LS 180, 54 nmol/L). In SJSA-1 and 94T778 cells, brigimadlin exhibited lower IC₅₀, GI₅₀, and GI₁₀₀ relative to other MDM2–p53 antagonists (alrizomadlin, milademetan; Supplementary Table S8). All three compounds induced similar upregulation of p53 target genes and apoptosis although higher concentrations were required for alrizomadlin and milademetan (Supplementary Fig. S2).

Brigimadlin treatment led to greater increases in mRNA levels of p53 target genes in the MDM2-amplified cell lines, SJSA-1 and 94T778, than in the non-amplified U-2 OS and SW982 cell lines [Fig. 3B (SJSA-1, U-2 OS); Supplementary Fig. S1A (94T778, SW982)]. Dose-dependent increases in PUMA (BBC3), CDKN1A (P21), MDM2, CD80, BAX, and BIM mRNA expression levels were observed in MDM2-amplified and non-amplified cell lines (Fig. 3B; Supplementary Fig. S1A). No consistent changes were observed in BCL-XL expression (antiapoptotic marker). MDM2, p53, and cleaved caspase-3/7 protein levels (Fig. 3C; Supplementary Fig. S1B) showed dose-dependent upregulation in both MDM2-amplified and non-amplified cell lines, with greater upregulation observed in MDM2-amplified cells. In the MDM2-amplified SJSA-1 cell line, elevated p53 protein levels decreased again at 72 hours of exposure to brigimadlin. This can be explained by the p53–MDM2 negative feedback loop, in which p53 induces high levels of MDM2 that target p53 for degradation and hence limit the levels of p53 protein. Dose-dependent increases in cleaved caspase-3 and cleaved PARP levels were observed in the two MDM2-amplified cell lines SJSA-1 and 94T778 (Fig. 3D; Supplementary Fig. S1C). Following incubation with brigimadlin (100 nmol/L, 4 hours), increased p53 immunoreactivity was observed in SJSA-1 cell nuclei and to a lesser degree in U-2 OS cells, versus vehicle-treated control samples (Supplementary Fig. S3), suggesting stabilization and nuclear translocation of p53.

Brigimadlin on-target toxicity was examined using an in vitro human megakaryocyte hematopoietic progenitor colony-forming assay (3D CFU-MK). Brigimadlin inhibited colony formation of CFU-MK with IC₅₀ and IC₉₀ values of 25 and 119 nmol/L, respectively. The morphology (size and/or density) of colonies was compromised at all concentrations tested. We then studied the effect of increasing brigimadlin concentrations (up to 100 nmol/L) on normal cell lines, fibroblasts derived from human foreskin (BJ) and cells derived from human retinal pigment epithelium (hTert RPE-1). Both cell lines displayed growth inhibition upon treatment with brigimadlin (IC₅₀ of 12 and 16 nmol/L, respectively), as expected for cells with TP53wt (Supplementary Table S9). In addition, we measured caspase-3/7-mediated apoptosis (IncuCyte Live-Cell Analysis). Brigimadlin did not induce apoptosis in these nonmalignant cells, whereas SJSA-1 tumor cells displayed strong caspase-3/7 activation starting at 24 hours and peaking at 52 hours upon exposure to brigimadlin (Supplementary Fig. S4). The apparent differential sensitivity between tumor cells and normal cells suggests that there is likely a therapeutic window for brigimadlin for the treatment of MDM2-amplified, TP53wt tumors.

High oral bioavailability and dose linearity with low variability in AUC and maximum plasma concentration (C_max) were observed across different preclinical species for brigimadlin (Table 1). The plasma protein binding of brigimadlin was very high (>99%) in all species tested (Supplementary Table S10) and did not display dependency on concentration or sex. Human serum albumin was found to be the major binding protein for brigimadlin in human plasma, whereas hAGP binding was of negligible importance.

A dose-dependent increase of GDF15 protein levels, a biomarker of p53 target engagement (35), could be detected in serum samples of SJSA-1 tumor-bearing BALB/c mice 24 hours after a single oral dose of brigimadlin (0.25, 0.75, 2, or 10 mg/kg; Fig. 4A). The increase of GDF15 in serum was positively correlated to the degree of GDF15 mRNA induction observed in tumor samples (Fig. 4B and C). The observed correlation supports the use of plasma GDF15 levels as a biomarker for target engagement in patients (22).

In order to assess p53 activation and antitumor activity induced by a single high weekly dose of brigimadlin versus the same accumulative dose administered as BID dosing once a week, SJSA-1 tumor-bearing BALB/c nude mice received either a single oral dose of brigimadlin at 2 mg/kg or two oral doses (7-hour interval) of brigimadlin at 1 mg/kg. Compared with BID dosing, a single oral dose of brigimadlin (2 mg/kg) was associated with a trend toward higher AUC and C_max (Fig. 4D; PK parameters in Supplementary Table S11) and a stronger induction of p53 target gene expression (CDKN1A, GDF15, PUMA, MDM2, CD80, BAX, BIM; Fig. 4E) and markers of apoptosis (cleaved PARP, cleaved caspase-3; Fig. 4F).

Brigimadlin efficacy with weekly dosing was assessed (Fig. 5A) according to two dose schedules. In single-dose groups, mice received one dose of brigimadlin (1, 1.5, or 2 mg/kg) on Day 1 and on Day 8. In the split-dose group, mice received brigimadlin 1 mg/kg BID (7-hour interval) on Day 1 and Day 8. A control group received vehicle (split dose). The single-dose regimens showed a clear dose–response relationship. The split-dose group (1 + 1 mg/kg) had a less pronounced and shorter effect than the 2 mg/kg single dose. Efficacy was similar in the split-dose group (1 + 1 mg/kg) and the 1.5 mg/kg-treated groups. These experiments demonstrated that brigimadlin treatment leads to durable antitumor activity with single doses administered with a 1-week interval, supporting the use of a high-dose intermittent schedule in clinical trials.

Extending our analysis, brigimadlin demonstrated antitumor activity in patient-derived xenograft (PDX) models derived from several MDM2-amplified, TP53wt tumor types. Prolonged tumor regression was observed following two doses of brigimadlin (2 mg/kg, Day 1 and Day 8; Fig. 5B) in a PDX model of MDM2-amplified undifferentiated pleomorphic sarcoma (SA3283). Tumor stasis and prolonged tumor growth control were observed in a PDX model of squamous cell lung carcinoma (LU0861; Fig. 5C). In another PDX model of squamous cell lung carcinoma, five doses of brigimadlin at 2 mg/kg QW induced complete and prolonged tumor regression (LU6903; Fig. 5D). Ex vivo analysis of the LU0861 PDX tumors 30 hours after treatment with either vehicle or a single oral dose of brigimadlin (2 mg/kg) showed increased levels of MDM2 and p21 mRNA in the brigimadlin-treated group compared with the vehicle control (Supplementary Fig. S5). In contrast to the SJSA-1 tumors (Fig. 4D), no apoptosis (measured by cleaved caspase-3 and cleaved PARP) was detected, in line with tumor stasis as a best response to brigimadlin in this PDX model compared with regression observed in the SJSA-1 CDX model. However, a limitation of this dataset remains as ex vivo biomarker data are only available for one of the PDX models tested (LU0861) and are not available for the other two PDX models.

In our in vivo studies, no toxicity (as evidenced by changes in body weight) was observed in mice with either of the dosing schedules used.

Brigimadlin showed high systemic exposure in an exploratory PK analysis of 27 patients who received 45 mg every 3 weeks (Q3W), the recommended dose for expansion in the phase Ib expansion cohort of NCT03449381 (geometric mean C_max: 2,880 nmol/L; AUC_0–∞: 173,000 nmol/L × hours/L; Supplementary Table S12). Median time to peak concentration (T_max) was 5.0 hours. Brigimadlin demonstrated a low gMean clearance/F (7.34 mL/minutes) and a long plasma elimination half-life (42 hours), with a low volume of distribution/F (Vz/F; 26.7 L). Inter-patient variability in exposure was moderate. These data are consistent with the PK data observed for patients enrolled in the phase Ia dose escalation part of the study (22). GDF15 levels were measured in plasma from the same 27 patients enrolled in the phase Ib study. Treatment with a single oral dose of brigimadlin (45 mg) induced a strong absolute change from baseline in GDF15, lasting at least 24 to 72 hours after dosing (Supplementary Fig. S6).

The clinical exposure was back translated to mouse studies in order to enhance the specificity and relevance of preclinical results. We attempted to match the observed clinical unbound exposure to the preclinical unbound exposure (Supplementary Table S13). Based on population PK modeling, unbound C_max in humans after the first 45 mg dose was calculated to be 5.4 nmol/L, with an unbound AUC_0–inf of 349.8 nmol/L × hours. Unbound C_max over unbound IC₅₀ was calculated to be approximately 10-fold. In mice, it was calculated that an oral dose of 1.5 mg/kg would confer an unbound C_max of 7.5 nmol/L and an unbound AUC_0–inf of 127.5 nmol/L × hours. The unbound C_max over unbound IC₅₀ was calculated to be 14-fold in mice at this dose. In conclusion, oral doses of 1.5 to 2 mg/kg in mice were estimated to result in unbound plasma exposure in mice close to the unbound plasma exposure reached with the 45 mg dose in patients, supporting the doses chosen in the preclinical studies.

In summary, brigimadlin demonstrated improved drug-like properties compared with previous related compounds (BI-0252 and BI-0282), displaying high potency and favorable PK characteristics.

We describe the discovery and characterization of brigimadlin, a novel MDM2–p53 antagonist with high potency and favorable PK properties. Its highly optimized properties permit intermittent dosing, a strategy proposed to mitigate on-target toxicity (thrombocytopenia) while maintaining strong efficacy.

Brigimadlin demonstrated improved IC₅₀ (2 nmol/L) for the inhibition of the MDM2–p53 protein–protein interaction in an AlphaScreen assay compared with previous molecules from the same series [BI-0252 (4 nmol/L), BI-0282 (5 nmol/L)]. An X-ray co-crystal structure of MDM2 and brigimadlin confirmed that brigimadlin binds to the p53-binding pocket of MDM2 in a way that would disrupt interaction with p53. Previously published data using immunoprecipitation and in situ proximity ligation assays confirmed the strong interference of brigimadlin with MDM2–p53 protein–protein interaction (36). Across a broad cancer cell line panel, 221 out of 827 lines were highly sensitive to brigimadlin (IC₅₀ range 4.6 to 203.5 nmol/L, median 107 nmol/L).

Consistent with its mechanism of action as an MDM2–p53 antagonist, sensitivity to brigimadlin correlated with TP53wt status and MDM2 amplification. Brigimadlin promoted the accumulation of p53 in the nucleus of treated cells (36) and evoked dose-dependent p53 activation in TP53wt MDM2-amplified cell lines and to a lesser extent in TP53wt MDM2 non-amplified cell lines, associated with the induction of p53 target gene expression, e.g., PUMA, P21, BAX, BIM, and CD80. CD80 is one of the most potent costimulatory molecules by which immune cells limit cancer progression. p53 activity stimulates the expression of CD80 in human tumor cells (37); enhanced expression of CD80 on tumor cells has been previously observed in vitro and in vivo in response to treatment with siremadlin (38) and is associated with increased antitumor T-cell activity. Our experiments demonstrated that brigimadlin also increases the expression of CD80 on tumor cells, in vitro and in vivo, which may contribute to its antitumor activity via immunomodulation. MDM2 itself is another p53 target gene and mediates an autoregulatory feedback loop (39); consistently, treatment with brigimadlin was associated with increased MDM2 expression.

Brigimadlin demonstrated high bioavailability, high exposure, and dose-linear pharmacokinetics in several species. Preclinical in vivo experiments demonstrated that treatment with weekly doses of brigimadlin led to durable antitumor activity, supporting the use of a high-dose intermittent treatment schedule in clinical trials. Previous reports indicate that increased serum GDF15 levels may be used as a biomarker of p53 activation (22, 40). Brigimadlin dose-dependent induction of tumor GDF15 mRNA and serum GDF15 protein levels was observed in vivo. Increased serum GDF15 was also observed in patients treated with brigimadlin, indicating p53 activation.

p53 activation by MDM2–p53 antagonists leads to cell cycle arrest in both normal cells and neoplastic cells. p53 activation inhibits hematopoietic progenitors and megakaryopoiesis; accordingly, neutropenia and thrombocytopenia are established on-target toxicities associated with MDM2–p53 antagonists (17, 41, 42). The tolerability profile of MDM2–p53 antagonists is of enduring interest. Our data and previous investigations indicate that normal cells are more resistant to apoptosis following treatment with MDM2–p53 antagonists than neoplastic cells (43, 44). In clinical experience with brigimadlin to date, hematotoxicity consistent with MDM2–p53 antagonism has been observed. In the phase I study, the most common grade ≥3 treatment-related adverse events were thrombocytopenia (25.9%) and neutropenia (24.1%); however, these were manageable, and no patients discontinued treatment with brigimadlin due to a hematologic treatment-related adverse event (22). In addition to the convincing preclinical and early clinical PK and PD data, brigimadlin showed acceptable tolerability (thrombocytopenia was observed but was manageable) and encouraging efficacy in early clinical trials, particularly in patients with MDM2-amplified, TP53wt liposarcoma or biliary tract carcinoma (23, 24).

In the exploratory PK analysis described here, patients who received 45 mg brigimadlin Q3W, the recommended dose for expansion, showed high systemic exposure and a long elimination half-life (42 hours). Serum GDF15 peaked 48 hours after dose, indicating durable p53 activation and adding further support to the strategy of intermittent dosing. In studies of other MDM2–p53 antagonists, in patients who received siremadlin (1 to 400 mg; multiple dosing schedules with either a 21- or 28-day cycle), median half-life (Day 1 of Cycle 1) varied from 6.4 to 31.1 hours (45); in patients who received milademetan (Day 1 of Cycle 1; 260 mg once daily on Days 1–3 and 15–17/28 days), the half-life was 10.0 hours (46). A phase III trial of milademetan in patients with DDLPS did not meet its primary endpoint (47). In light of the published clinical data and the fact that mice are not generally considered to be accurate models to assess potential on-target side effects of drugs, we have not evaluated the effects of intermittent doses of brigimadlin on the blood parameters in preclinical in vivo tumor models. These models were exclusively used to demonstrate the efficacy of brigimadlin when it was administered using intermittent dosing, which resulted in an exposure similar to that reached in patients. The high potency and long elimination half-life of brigimadlin permit intermittent dosing in patients with a single dose once every 3 weeks, as a means of mitigating the adverse effects while maintaining efficacy.

Clinical development of brigimadlin is ongoing. Brigimadlin monotherapy is under investigation in a phase I trial (NCT03449381), in the randomized phase II/III Brightline-1 trial (NCT05218499) assessing brigimadlin versus doxorubicin as first-line treatment in patients with advanced DDLPS [for which the FDA has granted fast track designation status (48)], and in the phase II Brightline-2 study (NCT05512377) assessing brigimadlin as second-line treatment in patients with MDM2-amplified solid tumors, including biliary tract and pancreatic cancers. The FDA has granted orphan drug designation for brigimadlin for the potential treatment of biliary tract cancer (49).

The findings presented here support the continued clinical evaluation of brigimadlin using intermittent (Q3W) dosing in patients with aberrant MDM2 activity-driven TP53wt tumors and underscore the rationale for investigating the broad utility of brigimadlin in MDM2-amplified cancers, such as DDLPS.

A. Gollner reports grants from the Austrian Research Promotion Agency during the conduct of the study, as well as patent for WO2016001376A1, WO2017060431A1, WO2016026937A1, and WO2015155332A1 issued. D. Rudolph reports grants from Boehringer Ingelheim during the conduct of the study and full-time employment with Boehringer Ingelheim. U. Weyer-Czernilofsky reports grants from Austrian Research Promotion Agency FFG during the conduct of the study. R. Baumgartinger reports grants from the Austrian Research Promotion Agency FFG during the conduct of the study and full-time employment with Boehringer Ingelheim. P. Jung reports grants from the Austrian Research Promotion Agency FFG and personal fees from Boehringer Ingelheim RCV during the conduct of the study, as well as personal fees from Boehringer Ingelheim RCV outside the submitted work. H. Weinstabl reports grants from the Austrian Research Promotion Agency FFG during the conduct of the study, as well as a patent for WO2017060431 issued. J. Ramharter reports grants from the Austrian Research Promotion Agency FFG during the conduct of the study, as well as a patent for WO 2017/060431 issued. R. Grempler reports grants from the Austrian Research Promotion Agency FFG outside the submitted work, a patent for Boehringer Ingelheim pending and issued, employment with Boehringer Ingelheim, as well as additional fund from the Austrian Research Promotion Agency FFG more than 3 years from 04/2011 to 03/2014 (grants 832260, 837815, 842856). J. Quant reports grants from the Austrian Research Promotion Agency FFG during the conduct of the study. J. Rinnenthal reports grants from the Austrian Research Promotion Agency outside the submitted work, as well as full-time employment with Boehringer Ingelheim RCV GmbH & Co KG from 2012 until 2020. A.P. Pitarch reports a patent for CA3226022A1 pending. B. Golubovic reports grants from the Austrian Research Promotion Agency FFG (grants 832260, 837815, 842856) during the conduct of the study, as well as full-time employment with Boehringer Ingelheim. D. Gerlach reports grants from the Austrian Research Promotion Agency FFG during the conduct of the study, as well as full-time employment with Boehringer Ingelheim RCV. G. Bader reports personal fees from Boehringer Ingelheim and grants from the Austrian Research Promotion Agency FFG during the conduct of the study, as well as personal fees from Boehringer Ingelheim outside the submitted work. K. Wetzel reports grants from the Austrian Research Promotion Agency FFG during the conduct of the study, no other relationships/conditions/circumstances that present a potential conflict of interest, as well as employment with Boehringer Ingelheim. S. Otto reports grants from the Austrian Research Promotion Agency during the conduct of the study. C. Mandl reports grants from the Austrian Research Promotion Agency FFG during the conduct of the study, as well as full-time employment with Boehringer Ingelheim RCV GmbH & Co KG. G. Boehmelt reports grants from the Austrian Research Promotion Agency FFG during the conduct of the study. D.B. McConnell reports financial support by the Austrian Research Promotion Agency FFG more than 3 years from 04/2011 to 03/2014 (grants 832260, 837815, 842856). N. Kraut reports grants from the Austrian Research Promotion Agency FFG during the conduct of the study, as well as full-time employment with Boehringer Ingelheim. P. Sini reports grants from the Austrian Research Promotion Agency FFG during the conduct of the study, as well as personal fees from Boehringer Ingelheim outside the submitted work.

A. Gollner: Conceptualization, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. D. Rudolph: Conceptualization, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. U. Weyer-Czernilofsky: Conceptualization, resources, data curation, investigation, methodology, writing–review and editing. R. Baumgartinger: Conceptualization, resources, data curation, investigation, visualization, methodology, writing–original draft. P. Jung: Investigation, visualization, writing–review and editing. H. Weinstabl: Conceptualization, data curation, investigation, visualization, methodology, writing–original draft, writing–review and editing. J. Ramharter: Conceptualization, writing–original draft, writing–review and editing, structural design and synthesis of brigimadlin. R. Grempler: Conceptualization, investigation, visualization, writing–original draft, writing–review and editing, final manuscript approval, clinical pharmacokinetic and pharmacodynamic data (GDF-15). J. Quant: Conceptualization, data curation, supervision, investigation, visualization, methodology, writing–original draft, writing–review and editing. J. Rinnenthal: Data curation, supervision, writing–review and editing. A.P. Pitarch: Conceptualization, formal analysis, methodology, writing–original draft, writing–review and editing. B. Golubovic: Conceptualization, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. D. Gerlach: Conceptualization, data curation, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. G. Bader: Data curation, formal analysis, investigation, visualization, methodology, writing–review and editing, protein crystallography. K. Wetzel: Investigation, visualization, methodology, writing–original draft, writing–review and editing. S. Otto: Methodology, writing–original draft, writing–review and editing. C. Mandl: Data curation, methodology. G. Boehmelt: Conceptualization, data curation, methodology, writing–review and editing. D.B. McConnell: Conceptualization, supervision, methodology, writing–review and editing. N. Kraut: Conceptualization, data curation, supervision, investigation, visualization, methodology, writing–original draft, writing–review and editing. P. Sini: Conceptualization, data curation, supervision, investigation, visualization, methodology, writing–original draft, writing–review and editing.

We thank Mark Pearson and Flavio Solca for their insightful discussions and comments. We thank Reinhard Sailer, Marion Schmid, and Sven Schmidt for support about clinical PK and PD data analysis and quality control. We would also like to thank Andrew S. Boghossian, Matthew G. Rees, Melissa M. Ronan, and Jennifer A. Roth of the Broad Institute of MIT and Harvard (Cambridge 02142 MA, USA) for their contributions to the PRISM screening experiments and analysis. All authors meet criteria for authorship as recommended by the International Committee of Medical Journal Editors (ICMJE). Medical writing support for the development of this manuscript, under the direction of the authors, was provided by Jim Sinclair, PhD, of Ashfield MedComms, an Inizio Company, and funded by Boehringer Ingelheim. Boehringer Ingelheim was given the opportunity to review the manuscript for medical and scientific accuracy as well as intellectual property considerations. These investigations were funded by Boehringer Ingelheim. Boehringer Ingelheim is grateful for financial support by the Australian Research Promotion Agency FFG (grants 832260, 837815, and 842856).