Purpose:

We evaluated the activity of AZD8205, a B7-H4–directed antibody–drug conjugate (ADC) bearing a novel topoisomerase I inhibitor (TOP1i) payload, alone and in combination with the PARP1-selective inhibitor AZD5305, in preclinical models.

Experimental Design:

IHC and deep-learning–based image analysis algorithms were used to assess prevalence and intratumoral heterogeneity of B7-H4 expression in human tumors. Several TOP1i-ADCs, prepared with Val-Ala or Gly–Gly–Phe–Gly peptide linkers, with or without a PEG8 spacer, were compared in biophysical, in vivo efficacy, and rat toxicology studies. AZD8205 mechanism of action and efficacy studies were conducted in human cancer cell line and patient-derived xenograft (PDX) models.

Results:

Evaluation of IHC-staining density on a per-cell basis revealed a range of heterogeneous B7-H4 expression across patient tumors. This informed selection of bystander-capable Val-Ala–PEG8–TOP1i payload AZ14170133 and development of AZD8205, which demonstrated improved stability, efficacy, and safety compared with other linker–payload ADCs. In a study of 26 PDX tumors, single administration of 3.5 mg/kg AZD8205 provided a 69% overall response rate, according to modified RECIST criteria, which correlated with homologous recombination repair (HRR) deficiency (HRD) and elevated levels of B7-H4 in HRR-proficient models. Addition of AZD5305 sensitized very low B7-H4–expressing tumors to AZD8205 treatment, independent of HRD status and in models representing clinically relevant mechanisms of PARPi resistance.

Conclusions:

These data provide evidence for the potential utility of AZD8205 for treatment of B7-H4–expressing tumors and support the rationale for an ongoing phase 1 clinical study (NCT05123482).

See related commentary by Pommier and Thomas, p. 991

Translational Relevance

AZD8205 is a novel B7-H4–directed topoisomerase I inhibitor antibody–drug conjugate that is currently under evaluation in an ongoing phase 1 clinical study (NCT05123482). We conducted a comprehensive analysis of B7-H4 prevalence and an assessment of intratumoral heterogeneity in human tumors, which should aid in indication selection for AZD8205. A multiparametric assessment of well-annotated patient-derived xenograft (PDX) models demonstrated that the level of B7-H4 expression needed for AZD8205 response is context-dependent and incorporation of warhead sensitivity biomarkers could further refine approaches for patient selection. Furthermore, addition of the next-generation PARP1-selective inhibitor AZD5305 to AZD8205 treatment sensitized low B7-H4–expressing PDX tumors, including homologous recombination repair–proficient models that represent acquired or intrinsic mechanisms of resistance to PARP inhibition (PARPi). These findings support the implementation of this combination even in the post-PARPi or PARPi-resistant setting.

Chemotherapy has historically been the primary option for systemic treatment of patients with advanced or metastatic cancer. Although it rapidly kills tumor cells, chemotherapy also damages healthy cells and tissues, sometimes with irreversible consequences. Antibody–drug conjugates (ADC) are a rapidly growing class of therapeutic agents that are specifically designed to improve the delivery of chemotherapeutic agents by exploiting the target selectivity of mAbs to widen the therapeutic index (1, 2). To maximize their therapeutic benefit, ADCs require optimization of linker–payload and conjugation chemistry, alignment of the linker–payload mechanism of action with tumor biology, and identification of biomarkers of response and resistance.

The topoisomerase I (TOP1) enzyme controls the degree of DNA supercoiling and resolves the torsional stress that occurs during DNA transcription and replication by cutting and reannealing double-stranded DNA (dsDNA). TOP1–DNA cleavage complexes are specifically and reversibly trapped by TOP1 I inhibitors (TOP1i), which leads to replication- and transcription-mediated DNA damage, dsDNA breaks, and ultimately, cell death (3). DNA repair mechanisms are often impaired in human tumors compared with normal tissue (4), thus providing an attractive option for therapeutic intervention. Indeed, TOP1is (e.g., topotecan and irinotecan) have been used as chemotherapeutic agents for >20 years (5), and clear clinical benefits have been demonstrated with TOP1i ADCs in breast cancer with the approvals of sacituzumab govitecan-hziy (Trodelvy; ref. 6) for triple-negative disease and fam-trastuzumab deruxtecan-nxk (Enhertu; ref. 7) for HER2+ disease. Moreover, recent data from the DESTINY-Breast04 trial showed that Enhertu provided striking improvements over chemotherapy in both progression-free survival and overall survival in patients with HER2-low metastatic breast cancer (7), and additional TOP1i ADCs are currently under evaluation in several phase 1–3 studies in a variety of solid tumor types (8).

B7-H4 (VTCN1) is a transmembrane glycoprotein in the B7 superfamily that has been reported to negatively regulate T-cell function through interaction with an unknown co-receptor (9, 10). B7-H4 is nominally expressed in normal tissues yet is highly expressed in many types of cancer, including breast cancer, cholangiocarcinoma, endometrial carcinoma, non–small cell and squamous cell lung carcinoma, and ovarian carcinoma (11–15). B7-H4 overexpression in tumor tissues, together with its limited expression in normal tissue, makes it an attractive ADC target. This concept has been explored preclinically with the microtubule-disrupting auristatin linker–payload monomethyl auristatin E (16).

Here, we describe the development of AZD8205, a novel B7-H4–directed ADC currently being evaluated in a first-in-human, phase 1 study (NCT05123482) and the first ADC to make use of the novel TOP1i linker–payload AZ14170133 (AZ’0133). To advance our understanding of B7-H4 expression and intratumoral heterogeneity in human tumors, we used innovative digital pathology techniques, which, together with ADC biophysical assessments and in vivo efficacy and toxicology studies, inform AZD8205 design. We used multiparametric analyses to discover potential biomarkers associated with AZD8205 antitumor response in a collection of triple-negative breast cancer (TNBC) patient-derived xenograft (PDX) models. Finally, we investigated the activity of AZD8205 in combination with a next-generation PARP1-selective inhibitor, AZD5305; refs. 17, 18), in PDX models of BRCA wild-type (WT) and BRCA-mutant tumors representing proposed mechanisms of PARPi resistance.

Antibodies and reagents

Antibodies targeting human B7-H4 were generated in VelocImmune V2 mice by alternating immunization with SKBR3 cells and mouse B7-H4 extracellular domain chemically conjugated to keyhole limpet hemocyanin. Antibody INT016 was selected on the basis of specificity and cross-reactivity to human and cynomolgus monkey B7-H4, and antibody frameworks were mutated to the closest human germlines. The heavy- and light-chain amino acid sequences for INT016 are described previously in PCT Publication Number WO 2022/053650 A1.

Detailed information regarding surface plasmon resonance, in vitro cytotoxicity, internalization, and immunoblotting is provided in the Supplementary Data. Primary antibodies used for immunoblotting are listed in Supplementary Table S1.

Cell lines

HT29 and MDA-MB-468 cells were obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen, GmbH (Braunschweig, Germany) and the ATCC, respectively, and were cultured in RPMI-1640 medium containing 10% heat-inactivated FBS (Thermo Fisher Scientific). MX-1 cells were obtained from the National Cancer Institute (Bethesda, MD) and cultured in Dulbecco modified essential/F12 medium containing 10% FBS and 2 mmol/L L-glutamine (Thermo Fisher Scientific). The HT29 clonal cell line stably expressing human B7-H4 (HT29-huB7-H4) was generated by using lentivirus expression vectors prepared with the pPACK H1 HIV Lentivector Packaging Kit (System Biosciences), followed by transduction, geneticin selection, and isolation with FACS through direct staining with an anti–B7-H4–PE antibody (US Biological). All cell lines were cultured in a humidified atmosphere of 5% CO2 at a temperature of 37°C, authenticated by short tandem repeat DNA profiling (IDEXX BioResearch Laboratories), and found to be negative for Mycoplasma by PCR and the MycoSEQ Mycoplasma detection assay kit (Thermo Fisher Scientific).

Preparation of ADCs

The TOP1i AZ14170132 (AZ’0132; AstraZeneca) was covalently bound to native cysteines in antibody INT016 through a Val-Ala (VA) or Gly-Gly-Phe-Gly (GGFG) peptide linker with or without a PEG8 spacer, resulting in four distinct INT016-based ADCs with an approximate drug-to-antibody ratio (DAR) of 8. Briefly, A 100 mmol/L solution of TCEP [tris(2-carboxyethyl)phosphine] in PBS was added (12.5 mol/L equivalent/antibody, 0.3 mmol) to INT016 antibody (3.6 g, 24.0 μmol) in reduction buffer containing PBS and 1 mmol/L ethylenediaminetetraacetic acid and a final antibody concentration of 10 mg/mL. The reduction mixture was heated at 37°C for 2 hours (or until full reduction was observed by ultrahigh-performance liquid chromatography, UPLC) in an orbital shaker with gentle (60 rpm) shaking. Reduced antibody was removed from incubation and allowed to cool to room temperature. To this mixture, linker–payload was added as a dimethyl sulfoxide (DMSO) solution (11–14 mol/L equivalent/antibody, 26.4–33.6 μmol) for a 10% (v/v) final DMSO concentration. The solution was shaken for 1 hour at 25°C and then the conjugation was quenched with N-acetyl cysteine (5× excess compared with payload as a 100 mmol/L stock solution). Excess-free drug was removed via tangential flow filtration unit, using mPES, MidiKros 30-kDa fiber filter with a 375-cm2 surface area, into buffer containing 30 mmol/L histidine, 30 mmol/L arginine, pH 6.8. Extent of free drug removal was monitored by UPLC-RP with neat conjugate. After complete removal of free drug (11–16 DV), the ADC was formulated into 20 mmol/L histidine, 240 mmol/L sucrose, pH 6.0 (5 DV), and filtered with a 0.22-μm filter under a sterile atmosphere. DAR was calculated by UPLC-RP with reduced conjugate, and monomeric purity was measured by UPLC–size exclusion chromatography with neat conjugate. A non-targeting isotype control ADC (Iso-ADC) was synthesized in the same manner by using an isotype-matched antibody with a comparable DAR. Analytic characterization of the ADCs is summarized in Supplementary Table S2 and Supplementary Fig. S1.

In vitro ADC serum stability

Triplicate samples of each ADC were diluted in mouse, rat, or cynomolgus monkey serum and incubated with agitation at 37°C for 0, 1, 3, 7, or 15 days. Intact ADC content in samples was determined by immunoprecipitation followed by reduced reverse-phase high-performance liquid chromatography (HPLC)–mass spectrometry analysis, which was performed with a 1290 series HPLC instrument (Agilent) coupled to a 6520 Accurate-mass time-of-flight mass spectrometer (Agilent) with an electrospray ionization source. The percentage of ADC remaining at a given time point was calculated by using the formula:
formula
where ADCDx is the DAR of ADC at the specified incubation time (x) in days and ADCD0 is the DAR at day 0.

IHC

All samples were sourced through commercial vendors of human biological samples. All subjects from whom tissue samples were obtained provided written informed consent to allow samples to be used for research, in accordance with the Declaration of Helsinki and the International Council for Harmonization's Harmonized Tripartite Guideline E6(R1): Good Clinical Practice. This study was performed under the AstraZeneca Research Tissue Bank Ethical Review, Research Ethics Committee reference 16/EE/0334, as approved by the East of England National Research Ethics Service committee.

IHC detection of B7-H4 was performed on 4-μm formalin-fixed, paraffin-embedded tumor or tissue sections with an automated Leica Bond RX IHC staining platform (Leica Biosystems). Human normal and tumor tissue sections were stained with anti-human B7-H4 mouse immunoglobulin G1 (IgG1; clone ZY0EYT-D11; AstraZeneca), and tumor tissue sections derived from PDX were stained with anti-human B7-H4 rabbit IgG1 (CAL63; Abcam). Both antibodies were thoroughly validated for use in IHC with a range of known B7-H4–expressing cell lines, B7-H4–transfected cell lines, and human breast tumor tissue to ensure specificity and sensitivity (Supplementary Fig. S2). After being cover slipped and mounted, slides were scanned with a Leica Aperio Scanscope AT2 slide scanner and then reviewed and scored by a pathologist.

IHC detection of SLFN11 was performed with the Leica Bond Rx staining platform (Leica Biosystems) and rabbit anti-human SLFN11 mAb (cat. no. 34858; Cell Signaling Technology), and binding was demonstrated with Prolink anti-Rb horseradish peroxidase (HRP) detection reagent. IHC detection of gH2A.X and human IgG was performed with the Ventana Discovery Ultra staining platform (Roche Diagnostics) and rabbit anti-human phospho-histone H2A.X (Ser139; cat. no. 9718; Cell Signaling Technology) or rabbit anti-human IgG (H+L) antibodies (cat. no. 309–005–082; Jackson ImmunoResearch Laboratories). IHC binding was demonstrated with OmniMap anti-Rb HRP detection reagent. IHC detection of cleaved caspase-3 was performed with the Dako Autostainer Link48 staining platform (Agilent) with a rabbit anti-cleaved caspase-3 (Asp175) antibody (Cell Signaling Technology) and anti-rabbit Ig ImmPRESS-HRP reagent (Vector Laboratories), and DAB substrate buffer (Agilent).

Digital pathology analysis

Digitally scanned whole-slide images were analyzed with a three-step workflow executed in Python, version 3.8, and deep-learning algorithms were implemented in TensorFlow. (i), Whole-slide images were processed with a first deep-learning–based semantic segmentation algorithm [a DeepLabv3+ (ref. 19) decoder with an EfficientNetV2S as encoder (ref. 20)] to separate epithelial from non-epithelial regions. In the case of human tissue samples, the resulting region of interest was restricted to invasive tumor by pathologist-drawn tumor core annotations. (ii), Whole-slide images were processed with a second deep-learning–based semantic segmentation algorithm [A Unet (ref. 21) with ResNet50 encoder (ref. 22)] to classify pixels for epithelial cell membrane, cytoplasm, nucleus, and cell center of mass. (iii), The centers of epithelial tumor cells were detected with non-maxima suppression on the posterior for cell centers. Cell objects were then grown from those seeds by using a seeded watershed on the membrane posterior. Object classification and segmentation were saved in a custom .hdf5 file format (Definiens result container).

The signal for DAB-stained B7-H4 was quantified with color deconvolution into the hue–saturation–density (HSD) color mode (23). Image data in the red–green–blue color mode were converted into HSD, and stain contribution per pixel was estimated by calculating a weighted linear fit in the Euclidean chromaticity projection of the HSD space (axes referred to as cx, cy, density), where the reference points for brown (DAB) and blue (hematoxylin) were set to [–0.3, –0.3] and [0.4, 0.4] for cx and cy, respectively. The chromaticity coordinate of each pixel is projected onto the line connecting the two reference points. The signal contribution assigned to DAB stain is referred to as the optical density (OD) value, which provided a quantitative and continuous measure for the abundance of the target molecule B7-H4 within each cell. The readout is averaged per cell for its membrane and cytoplasm and saved to the Definiens result container for downstream data analysis.

Algorithm 1 was trained extensively on monoplex DAB-stained resections stained for various membrane markers with different assays, with associated pixel-precise annotations drawn by pathologists to achieve robust algorithm performance. Algorithm 2 was trained in a weakly supervised setting, where pathologists annotated epithelial cell centers. A color deconvolution to detect brown membrane as object boundaries combined with a distance map was then used to grow cell outlines from the centers as seeds. Cell outlines were corrected manually in a subsequent step in the program Developer XD.

Algorithm performance was assessed on an independent subset of the studied TNBC cohort (unseen and representative test data, approximately 30% of the dataset, chosen on the basis of tumor morphology and staining intensity by pathologists). Algorithm 1 for epithelium detection was found to segment tissue at a mean F1 score of 0.93 against three pathologists (inter-pathologist score, 0.934, 84 annotated regions of interest at 250 × 250 μm size, total of 34 whole-slide images). The combination of the deep-learning algorithm for cell pixel classification and its post-processing (algorithms 2 and 3) were found to detect cells very accurately (mean F1 score against three pathologists, 0.762; inter-pathologists score, 0.755; n = 9,051) and segment them reasonably well as assessed by average symmetric surface distance (ASSD; averaged against three pathologists 1.474 μm; inter-pathologist ASSD, 0.953 μm; n = 1,551).

Subsequent data analysis to aggregate readouts at case level and plot comparisons was executed in Python, relying on the libraries numpy, pandas, matplotlib, and seaborn.

In vivo efficacy studies

All animal experiments were conducted in a facility accredited by the Association for Assessment and Accreditation of Laboratory Animal Care under the guidelines of AstraZeneca's Institutional Animal Care and Use Committee and appropriate animal research approvals. The cell line xenograft models were established through subcutaneous implantation of cells in 50% Matrigel (Corning) in either the right mammary fat pad (MX-1, MDA-MB-468) or left flank (HT29, HT29-huB7-H4) of female CB-17 mice (Envigo). When tumor volume reached 150–200 mm3, mice were randomized and given a single intravenous injection of vehicle control (20 mmol/L histidine, pH 6; 240 mmol/L sucrose; 0.02% PS-80) or ADCs. Body weight and tumor measurements were determined once or twice weekly, and tumor volume was calculated with the formula
formula
PDX models of human TNBC were established through subcutaneous implantation of tumor fragments into the flank of athymic nude mice (Envigo) and tested at Xentech (Evry). Animals were matched by tumor volume, and vehicle control or ADCs were then administered as a single intravenous injection. When used, the PARP1-selective inhibitor AZD5305 (AstraZeneca) was formulated in water/HCl pH 3.5–4 and administered by oral gavage at a final dose volume of 10 mL/kg once daily for 28 days. Antitumor activity was determined on the last day, at which time all mice in the untreated control group remained in the study. If the end tumor volume (ETV) in the treatment group was less than the initial tumor volume (ITV), the antitumor response from baseline was calculated from the group means by using the formula:
formula
Otherwise, the antitumor response was expressed as a percentage of change in tumor volume in the treatment arm relative to that in the untreated control arm, calculated by the formula:
formula

For pharmacodynamic and pharmacokinetic studies, mice bearing either HT29 or HT29-huB7-H4 tumors (250–300 mm3) were administered a single intravenous injection of vehicle control or ADCs. Plasma and tumor samples were collected at indicated time points, when one portion of the tumor was fixed in formalin and embedded into paraffin blocks for IHC analysis, and the remaining portion was snap-frozen for analysis of total mAb, ADC, and free AZ’0132 warhead by LC/MS.

Bioanalysis of plasma and tumor samples

Mouse plasma was analyzed for total antibody and ADC with a qualified hybrid immunocapture LC/MS bioanalysis method with a quantification range of 0.1–15.0 μg/mL. Briefly, plasma samples, standards, and quality controls were incubated with magnetic beads conjugated with anti-HuFc polyclonal antibodies (Bethyl Laboratories) for approximately 2 hours at room temperature. The magnetic beads, together with captured target analyte, were washed multiple times with a pH-neutral buffer and then reconstituted with digestion buffer containing the stable isotope-labeled internal standards for the peptides and the warhead. The target analytes were then digested with trypsin directly at elevated temperature (∼70°C) for 1.5–2.5 hours. The supernatant from tryptic digestion was split into two parts; one was injected into an LC/MS instrument to monitor signature peptides (for quantification of total antibody concentration), and the other was further digested with papain to release the warhead before injecting into an LC/MS instrument to monitor the released AZ14170133 (AZ’0132) ADC (for quantification of ADC concentration). Mouse plasma was analyzed for total unconjugated AZ’0132 with a qualified plasma precipitation LC/MS bioanalysis method with a quantification range of 0.059–29.5 ng/mL. Briefly, the plasma sample, standards, and quality controls were mixed with stable isotope–labeled AZ’0132 as internal standards before protein precipitation with acetonitrile on a Waters Ostro plate (Waters). The eluate was then evaporated with nitrogen before reconstitution for LC/MS injection.

Tumor samples were collected with various weights and were homogenized with 300 μL of RIPA buffer. After centrifugation, the supernatant was quantified for total antibody, ADC, and unconjugated warhead concentration in a manner similar to that described above. The total warhead concentration and total protein concentration in the homogenate was also measured. The total concentration of AZ’0132 was determined after the homogenate was thoroughly digested with an excess of papain, following a plasma precipitation approach similar to that described above. The total protein concentration was measured with a commercially available BCA kit (Thermo Fisher Scientific).

Rat toxicity study

The four unique INT016-ADCs or vehicle control were administered intravenous to male Sprague–Dawley rats (n = 3 per group) once every 3 weeks for a total of two doses at 20 or 60 mg/kg. Animals were monitored throughout the 6-week study period for clinical signs, tolerability/mortality, body weight changes, and food consumption, and endpoints included clinical chemistry, hematology, and histopathology.

Statistical analysis

Data are presented as mean ± standard deviation or mean ± standard error of the mean. Statistically significant differences among INT016-ADCs in the MX-1 in vivo efficacy study were tested as described previously in the Supplementary Data. To evaluate associations with response versus nonresponse in the PDX models, the Wilcoxon test for group comparison was performed in the R statistical programming environment (version 4.1). A P value of <0.05 was considered statistically significant.

Data availability

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

B7-H4 expression and prevalence in tumor tissues

Previous reports have indicated that B7-H4 has limited expression in most normal tissues but is overexpressed in a variety of human carcinomas (11–15). However, due to the use of dissimilar methodologies and B7-H4 detection antibodies (10), as well as the generally subjective nature of IHC quantitation, it is difficult to accurately discern relative levels of B7-H4 across tumor types. To address this gap, we developed a validated IHC protocol (Supplementary Fig. S2) and used it to assess the expression of B7-H4 across a large panel of normal tissues and 2145 tumor samples across 15 indications.

B7-H4 was expressed in several tumor types and was most prevalent in endometrial carcinoma (94%), cholangiocarcinoma (89%), breast cancer (HER2+ 78%, TNBC 74%, ER+ 74%), and ovarian carcinoma (77%; Fig. 1A and B). In contrast, B7-H4 was detected in a limited number of normal human tissues, such as fallopian tube, lung, breast, and prostate, and was generally found in <10% of the total cells in the sample, restricted toward ductal or tubular epithelium, and primarily located on the apical luminal membrane (Supplementary Table S3).

Figure 1.

B7-H4 expression in human tumors. A, IHC was used to assess the expression of B7-H4 across multiple tumor types. The prevalence of B7-H4 expression in an indication is shown by the proportion of cells expressing B7-H4 positivity at any intensity. B, Representative images of B7-H4 IHC staining in endometrial cancer, cholangiocarcinoma, ER+ breast cancer or TNBC, and ovarian cancer. C, Quantitative image analysis showing the distribution of mean optical density (OD) of B7-H4 expression in each tumor cell membrane in primary tumor surgical resection, formalin-fixed, paraffin-embedded (FFPE) samples of TNBC (n = 196). Each truncated violin plot represents an individual donor sample. Lines represent medians.

Figure 1.

B7-H4 expression in human tumors. A, IHC was used to assess the expression of B7-H4 across multiple tumor types. The prevalence of B7-H4 expression in an indication is shown by the proportion of cells expressing B7-H4 positivity at any intensity. B, Representative images of B7-H4 IHC staining in endometrial cancer, cholangiocarcinoma, ER+ breast cancer or TNBC, and ovarian cancer. C, Quantitative image analysis showing the distribution of mean optical density (OD) of B7-H4 expression in each tumor cell membrane in primary tumor surgical resection, formalin-fixed, paraffin-embedded (FFPE) samples of TNBC (n = 196). Each truncated violin plot represents an individual donor sample. Lines represent medians.

Close modal

To understand B7-H4 intratumoral heterogeneity within individual tumors, a key factor influencing the activity of ADCs, a separate collection of histologically annotated samples of TNBC (n = 196) were stained and digitally analyzed with deep-learning–based image analysis algorithms to detect and segment the individual tumor epithelial cells and quantify the level of B7-H4 expression on each cell's membrane. This approach allowed us to study the distribution of B7-H4 expression on a per-cell basis, showing a range of heterogeneous B7-H4 expression within each patient sample across the cohort (Fig. 1C; Supplementary Fig. S3), which may become relevant for future studies of patient response to ADC treatment. B7-H4 expression and heterogeneity of expression were weakly correlated (r = –0.84, P < 0.01), suggesting that both variables may be important in identifying optimal patient populations.

Linker–warhead selection and development of anti–B7-H4 ADC AZD8205

Because of its heterogeneous expression profile, we hypothesized that B7-H4 may be best targeted by an ADC capable of “bystander killing,” whereby adjacent tumor cells could be eliminated through the transfer of released warhead from target-positive cells (24). First, we initiated a discovery campaign that yielded the anti–B7-H4 antibody INT016, which possesses several features desirable for use in an ADC: (i) Similar affinity for human and cynomolgus monkey B7-H4 (Supplementary Table S4; Supplementary Fig. S4), (ii) specific binding to B7-H4–expressing cells but not B7-H4–negative cells (Supplementary Fig. S5), and (iii) internalization and trafficking to the lysosome, a key feature for optimal release of the cytotoxic warhead (Supplementary Fig. S6). We next conjugated INT016 to the TOP1i warhead AZ’0132 and, to increase the likelihood of the bystander effect, used a cleavable VA or GGFG peptide linker, with or without a PEG8 spacer (Fig. 2A), resulting in four unique INT016-ADCs with an average DAR of 8. We then performed a head-to-head assessment of these four unique INT016-ADCs to enable selection of a lead candidate.

Figure 2.

Selection of AZD8205 linker–payload. A, The TOP1i warhead AZ’0132 was conjugated to the anti–B7-H4 antibody INT016, using either a cleavable VA or GGFG peptide linker, with or without a PEG8 spacer, to produce four unique ADCs with an approximate DAR of 8. B, Hydrophobic interaction chromatograms of INT016- and INT016-based ADCs. C,Ex vivo rat plasma stability studies, as assayed by immunoprecipitation followed by rapid LC/MS. D,In vivo antitumor activity resulting from a single intravenous bolus dose of 0.75 mg/kg INT016-ADCs in an MX-1 xenograft model studied in CB-17 SCID mice. Mean tumor volume ± standard error of the mean is shown. E, Toxicity was assessed in a repeat-dose study conducted in Sprague–Dawley rats with intravenous dosing at 20 or 60 mg/kg once every 3 weeks for a total of two doses. The table summarizes toxicities observed at either dose level, which occurred with increased severity in target organs over control samples. F, Schematic representation of AZD8205.

Figure 2.

Selection of AZD8205 linker–payload. A, The TOP1i warhead AZ’0132 was conjugated to the anti–B7-H4 antibody INT016, using either a cleavable VA or GGFG peptide linker, with or without a PEG8 spacer, to produce four unique ADCs with an approximate DAR of 8. B, Hydrophobic interaction chromatograms of INT016- and INT016-based ADCs. C,Ex vivo rat plasma stability studies, as assayed by immunoprecipitation followed by rapid LC/MS. D,In vivo antitumor activity resulting from a single intravenous bolus dose of 0.75 mg/kg INT016-ADCs in an MX-1 xenograft model studied in CB-17 SCID mice. Mean tumor volume ± standard error of the mean is shown. E, Toxicity was assessed in a repeat-dose study conducted in Sprague–Dawley rats with intravenous dosing at 20 or 60 mg/kg once every 3 weeks for a total of two doses. The table summarizes toxicities observed at either dose level, which occurred with increased severity in target organs over control samples. F, Schematic representation of AZD8205.

Close modal

Linker–payload hydrophobicity and stability contribute to the tolerability and efficacy of ADCs: hydrophobic ADCs tend to aggregate and have increased plasma clearance and nonspecific uptake, whereas unstable linkers result in premature release of the cytotoxic drug in circulation and reduced drug delivery to tumor cells. Together these characteristics can contribute to a narrow therapeutic window (25–27). We first examined the INT016-ADCs with hydrophobic interaction chromatography and observed a modest reduction in retention time (RT), indicating reduced hydrophobicity, for the VA linker–payload ADCs (INT016-AZ’0133, RT = 3.41; INT016-SG4010, RT = 3.37) compared with the GGFG linker–payload ADCs (INT016-SG4052, RT = 3.53; INT016-SG4057, RT = 3.76; Fig. 2B). Ex vivo serum stability experiments indicated that the INT016-ADCs containing the PEG8 spacer (INT016-AZ’0133 and INT016-SG4057) were strikingly more stable (23.5% and 29% drug loss, respectively, after 15 days) than those lacking PEG (INT016-SG4010 and INT016-SG4052; approximately 68% drug loss after 15 days; Fig. 2C; Supplementary Fig. S7). The ADCs had comparable potency in vitro (Supplementary Fig. S8). In vivo efficacy was evaluated in the MX-1 TNBC xenograft model with a multitudinous dose study design, which enabled comparison of each ADC across seven dose levels ranging from 0.125 to 2 mg/kg. With growth rate as a metric, all ADCs were active in vivo, showing similar efficacy at doses of >1 mg/kg (Supplementary Fig. S9). However, at the lower dose of 0.75 mg/kg, INT016-AZ’0133 provided the greatest efficacy (P < 0.001 against all other ADCs; Fig. 2D). In addition, growth rate analysis incorporating data from the entire dose range showed that INT016-AZ’0133 provided the most potent antitumor activity (P < 0.009 against all other ADCs; Supplementary Fig. S10).

In the comparative repeat-dose rat toxicity study, all ADCs were well tolerated and no deaths occurred after intravenous treatment once every 3 weeks for a total of two doses at either 20 or 60 mg/kg. The toxicities observed in this study were those anticipated due to TOP1 inhibition and included effects on the hemopoietic system, evidenced by decreases in some red cell parameters and/or minimal to mild increases in reticulocytes and/or platelets in all groups; changes in liver chemistry (increased triglycerides and decreased globulins); and mild hepatocellular necrosis (Fig. 2E). Some or all of these changes were noted in all ADC groups and at all dose levels except for INT016-AZ’0133, which produced changes only in the hematopoietic system. Kidney toxicity (unilateral chronic progressive nephropathy) was noted for all ADCs except INT016-AZ’0133, but a clear correlation with treatment was not determined. Together, these results indicate that INT016-AZ’0133 (henceforth referred to as AZD8205) is likely to provide the widest therapeutic window among the ADCs we evaluated. A schematic of AZD8205 is shown in Fig. 2F.

Mechanistic, pharmacodynamic, and pharmacokinetic studies of AZD8205

Despite similar sensitivity to the AZ’0132 warhead (EC50 = 1.04–1.48 nmol/L; Supplementary Fig. S11), AZD8205 was cytotoxic to B7-H4–expressing HT29-huB7-H4 cells but not to B7-H4–negative HT29 cells (Supplementary Fig. S8). This characteristic provided an opportunity to investigate bystander killing with an in vitro co-culture system (28). A lentivirus expression system was used to express GFP in HT29 cells, which were co-cultured at a 1:1 ratio with HT29-huB7-H4 cells in the presence of either 200 ng/mL AZD8205 or Iso-ADC before analysis by flow cytometry. AZD8205 was cytotoxic to the target-negative HT29-GFP cells only when in co-culture with the target-positive HT29-huB7-H4 cells (Fig. 3A), clearly demonstrating a bystander-killing effect.

Figure 3.

AZD8205 mechanism of action. A,In vitro bystander killing was evaluated by culturing B7-H4–positive HT29-huB7-H4 or B7-H4–negative HT29-GFP cells alone or together at a ratio of 1:1. The bar graph shows the relative cell counts determined by flow cytometry after treatment with either 200 ng/mL AZD8205 or Iso-ADC for 6 days. B,In vivo antitumor activity resulting from a single intravenous bolus dose of AZD8205 in HT29, HT29-huB7-H4, or heterogeneous xenograft models at HT29-huB7-H4:HT29 ratios of 10:90 or 40:60 in CB-17 SCID mice. Each value represents mean tumor volume ± standard error of the mean. Representative B7-H4 IHC staining in each model is shown above the graph. C, Image analysis and representative IHC images of pharmacodynamic changes in HT29-huB7-H4 xenograft tumors after treatment with AZD8205 or Iso-ADC indicate dose- and time-dependent accumulation of AZD8205, followed by an increase in γH2AX and CC-3 and an overall decrease in epithelial cell density. Solid lines indicate means; shaded areas denote 90% confidence intervals (n = 3). D, Plasma concentrations of AZD8205, Iso-ADC, total INT016 mAb (TAb), and total unconjugated warhead AZ’0132 or tumor concentrations of AZD8205, Iso-ADC, and total unconjugated AZ’0132 in HT29-huB7-H4 xenograft mice treated with 7 mg/kg AZD8205. Values indicate mean ± standard deviation (n = 3). Circles represent individual data points; dotted lines indicate geometric means.

Figure 3.

AZD8205 mechanism of action. A,In vitro bystander killing was evaluated by culturing B7-H4–positive HT29-huB7-H4 or B7-H4–negative HT29-GFP cells alone or together at a ratio of 1:1. The bar graph shows the relative cell counts determined by flow cytometry after treatment with either 200 ng/mL AZD8205 or Iso-ADC for 6 days. B,In vivo antitumor activity resulting from a single intravenous bolus dose of AZD8205 in HT29, HT29-huB7-H4, or heterogeneous xenograft models at HT29-huB7-H4:HT29 ratios of 10:90 or 40:60 in CB-17 SCID mice. Each value represents mean tumor volume ± standard error of the mean. Representative B7-H4 IHC staining in each model is shown above the graph. C, Image analysis and representative IHC images of pharmacodynamic changes in HT29-huB7-H4 xenograft tumors after treatment with AZD8205 or Iso-ADC indicate dose- and time-dependent accumulation of AZD8205, followed by an increase in γH2AX and CC-3 and an overall decrease in epithelial cell density. Solid lines indicate means; shaded areas denote 90% confidence intervals (n = 3). D, Plasma concentrations of AZD8205, Iso-ADC, total INT016 mAb (TAb), and total unconjugated warhead AZ’0132 or tumor concentrations of AZD8205, Iso-ADC, and total unconjugated AZ’0132 in HT29-huB7-H4 xenograft mice treated with 7 mg/kg AZD8205. Values indicate mean ± standard deviation (n = 3). Circles represent individual data points; dotted lines indicate geometric means.

Close modal

To explore bystander-killing potential in vivo, we combined HT29-huB7-H4 and HT29 cells at a ratio of either 10:90 or 40:60 to develop heterogeneous tumor models, as confirmed by IHC (Fig. 3B). In contrast with the B7-H4–negative HT29 model, AZD8205 significantly inhibited the growth of HT29-huB7-H4 xenograft tumors; regression was observed after a dose of either 5 or 2.5 mg/kg in the HT29-huB7-H4 homogeneous model or at 7 mg/kg in the 40:60 model. Treatment of the 10:90 model with 7 mg/kg AZD8205 resulted in 61% tumor growth inhibition (TGI) at day 29 but no regression from baseline levels, in contrast with the Iso-ADC control, in which no antitumor activity was observed. Furthermore, a dose of 3.5 mg/kg AZD8205 did not induce tumor regression in either heterogeneous model (10:90 or 40:60), in contrast with the tumor regression observed at the lower 2.5 mg/kg dose in the homogeneous HT29-huB7-H4 model (Fig. 3B). These results suggest that the bystander-killing capacity of AZD8205 enables antitumor activity in heterogeneous tumors and that a threshold of killing and/or level of B7-H4–positive cells may be needed to maximize this effect.

To further explore the mechanism of action of AZD8205, we used the HT29 and HT29-huB7-H4 xenograft models to evaluate pharmacodynamic and pharmacokinetic effects after a single intravenous injection of either AZD8205 or Iso-ADC (Supplementary Fig. S12). Dose- and time-dependent accumulation of AZD8205 was observed in the HT26-huB7-H4 tumors, as visualized by a human IgG IHC assay. This finding correlated with increased levels (approximately fourfold) of γH2AX foci, an indicator of DNA damage (29, 30), at 96–168 hours after treatment (Fig. 3C), which is consistent with the TOP1i mechanism of action and AZD8205 in vitro results (Supplementary Fig. S13). Elevated cleaved caspase 3 (CC-3) and an overall decrease in epithelial cell density as compared with Iso-ADC were also observed at 96 hours after treatment in AZD8205-treated tumors (Fig. 3C). In contrast, AZD8205 IgG staining was not detected in the HT29 (B7-H4–negative) tumors, and no decrease in epithelial cell density was observed after AZD8205 treatment (Supplementary Fig. S14). These results are in agreement with those of the HT29 in vivo xenograft study (Fig. 3B) and further support the target specificity of AZD8205.

Plasma concentrations of AZD8205 and total INT016 antibody were similar, and a low level of unconjugated AZ’0132 warhead was detected after a 7 mg/kg dose of the ADC (Fig. 3D); the observed peak plasma concentration was 0.4 ng/mL. In contrast, the estimated initial concentration of administered warhead was approximately 2,900 ng/mL, calculated from the observed ADC maximum concentration, the DAR, and the ADC and warhead molecular weights. The AZ’0132 plasma concentration was nearly 5-fold lower than the EC50 value observed for the in vitro cytotoxicity activity of AZ’0132 in HT29 and HT29-huB7-H4 cells (Supplementary Fig. S11). In contrast, after AZD8205 dosing, time-matched tumor concentrations of ADC and AZ’0132 warhead were up to 3- and 9-fold higher, respectively, than those of Iso-ADC, showing that the pharmacodynamic responses were driven by higher tumor exposure due to targeted warhead delivery by AZD8205.

Collectively, these data demonstrate that the primary mechanism of action of AZD8205 is intracellular delivery of the TOP1i warhead to B7-H4–positive cells, which leads to DNA damage and, ultimately, apoptotic cell death. After killing B7-H4–positive cells, the released warhead may also elicit antitumor activity against neighboring cells through the bystander effect.

AZD8205 antitumor activity in TNBC PDXs representing heterogeneous B7-H4 expression

The antitumor activity of AZD8205 was evaluated in a panel of 26 TNBC PDX models that represent a range of B7-H4 staining intensity and heterogeneity (Fig. 4A) as was observed in the TNBC patient samples (Figs. 1C and 4B). The level of B7-H4 expression in the PDX models measured by IHC and digital pathology was highly correlated with the level of gene expression measured using RNASeq (R = 0.86, P = 1.9e−06; Supplementary Fig. S15). Upon a single intravenous administration of 3.5 mg/kg AZD8205, the best response according to modified RECIST criteria was complete or partial response in 18 (69%), stable disease in 2 (8%), and progressive disease in 6 (23%; Fig. 4C). Examples of three PDXs that responded with progressive or stable disease or with complete or partial response are shown in Fig. 4D. Although a statistically significant association (P = 0.019) between B7-H4 expression and response was observed (Fig. 4E), we noticed that responses occurred in models representing a spectrum of B7-H4 expression, including those with a very low median membrane OD value. Overall, these results suggest that AZD8205 could be effective in tumors representing a range of B7-H4 expression and heterogeneity and that elevated levels of B7-H4 are generally associated with response.

Figure 4.

AZD8205 response in PDX models of TNBC and associations with B7-H4. A, Quantitative image analysis showing the distribution of mean optical density (OD) of B7-H4 expression in the tumor cell membrane in the TNBC PDX models (n = 26). B, Comparison of the median of the mean OD of B7-H4 expression in the tumor cell membrane measured in primary tumor surgical resection, formalin-fixed, paraffin-embedded samples of TNBC (n = 196; Fig. 1C) and the TNBC PDX models (n = 26; A). C, Waterfall plot representing responses after a single intravenous dose of 3.5 mg/kg AZD8205. Models were considered responsive to AZD8205 treatment if the percentage of tumor growth from baseline was –30% to –100%, inclusive. D, Response observed in HBCx-30, HBCx-8, and HBCx-10 models after a single intravenous dose of 3.5 mg/kg AZD8205. Data indicate mean tumor volume ± standard error of the mean (n = 3). PD, progressive disease; R, complete or partial response; SD, stable disease. E, Comparison of the median of the mean OD of B7-H4 expression in the tumor cell membrane measured in PDX responders and nonresponders. The line in the violin or box plot represents the median, and the standard deviation is represented in the box plots.

Figure 4.

AZD8205 response in PDX models of TNBC and associations with B7-H4. A, Quantitative image analysis showing the distribution of mean optical density (OD) of B7-H4 expression in the tumor cell membrane in the TNBC PDX models (n = 26). B, Comparison of the median of the mean OD of B7-H4 expression in the tumor cell membrane measured in primary tumor surgical resection, formalin-fixed, paraffin-embedded samples of TNBC (n = 196; Fig. 1C) and the TNBC PDX models (n = 26; A). C, Waterfall plot representing responses after a single intravenous dose of 3.5 mg/kg AZD8205. Models were considered responsive to AZD8205 treatment if the percentage of tumor growth from baseline was –30% to –100%, inclusive. D, Response observed in HBCx-30, HBCx-8, and HBCx-10 models after a single intravenous dose of 3.5 mg/kg AZD8205. Data indicate mean tumor volume ± standard error of the mean (n = 3). PD, progressive disease; R, complete or partial response; SD, stable disease. E, Comparison of the median of the mean OD of B7-H4 expression in the tumor cell membrane measured in PDX responders and nonresponders. The line in the violin or box plot represents the median, and the standard deviation is represented in the box plots.

Close modal

Assessment of additional biomarkers associated with AZD8205 response in TNBC PDXs

Twenty-five of the 26 PDX models examined in this study have been previously characterized for cytogenetic and other biological markers by a variety of procedures, including testing for RAD51 foci, a proposed dynamic biomarker of homologous recombination repair (HRR) deficiency (HRD; ref. 31); whole-exome sequencing; RNA sequencing; and an assay for response to PARPi (32). We leveraged this rich dataset to interrogate biomarkers for AZD8205 response beyond B7-H4 expression alone.

Alterations in genes involved in DNA repair and HRD have been associated with sensitivity to TOP1i chemotherapy (5). Therefore, we evaluated the RAD51 score, where a negative result (RAD51-negative; ≤10% RAD51 cutoff value) is considered to be indicative of HRD and has been shown to be an accurate and dynamic biomarker in these PDX models for predicting response to PARPi and platinum-based treatments (32, 33). Seven of the seven RAD51-negative PDX models (100%) responded to 3.5 mg/kg AZD8205 treatment, and five of seven (71%) achieved complete response (Fig. 5A). In the remaining 18 models, which are HRR proficient (HRP; RAD51 score, >10%), the best response was complete or partial response in 10 (67%), stable disease in 2 (13%), and progressive disease in 6 (40%). Although responders in the HRD group represented a wide range of B7-H4 expression, there was a statistically significant association between elevated levels of B7-H4 and response in the HRP group (P = 0.027; Fig. 5B). These results indicate that AZD8205 is active in both HRP and HRD tumors and illustrate that biomarkers of AZD8205 response are multifactorial, involving an interplay of warhead sensitivity and B7-H4 expression. In the HRP setting, AZD8205 response is positively associated with elevated expression of B7-H4, whereas HRD tumors, which are expected to have increased sensitivity to AZ’0132 (Supplementary Fig. S16), may have a lower threshold for response.

Figure 5.

AZD8205 response in PDX models of TNBC and associations with defective DNA repair. A, Waterfall plot representing responses after a single intravenous dose of 3.5 mg/kg AZD8205 in TNBC PDX models categorized as HRD or HRP based on a RAD51 foci test. B, Comparison of the median of the mean optical density (OD) of B7-H4 expression in the tumor cell membrane measured in PDX responders and nonresponders in TNBC PDX models, subdivided by RAD51 foci status. The solid line in the box plot represents the median, and the standard deviation is represented in the box plots. C, Antitumor activity resulting from a single intravenous dose of 3.5 mg/kg AZD8205 in models previously shown to be resistant to PARPi treatment (32). Data indicate tumor volume mean ± standard error of the mean (n = 3).

Figure 5.

AZD8205 response in PDX models of TNBC and associations with defective DNA repair. A, Waterfall plot representing responses after a single intravenous dose of 3.5 mg/kg AZD8205 in TNBC PDX models categorized as HRD or HRP based on a RAD51 foci test. B, Comparison of the median of the mean optical density (OD) of B7-H4 expression in the tumor cell membrane measured in PDX responders and nonresponders in TNBC PDX models, subdivided by RAD51 foci status. The solid line in the box plot represents the median, and the standard deviation is represented in the box plots. C, Antitumor activity resulting from a single intravenous dose of 3.5 mg/kg AZD8205 in models previously shown to be resistant to PARPi treatment (32). Data indicate tumor volume mean ± standard error of the mean (n = 3).

Close modal

AZD8205 treatment provided robust antitumor activity in a number of HRP models (Fig. 5C), including those representing unknown mechanisms of resistance to PARPi (HBCx-12B, HBCx-23, HBCx-8, and HBCx-1). Additional examples include a BRCA1 hypomorphic model (HBCx-11), models with a loss of BRCA1 promoter hypermethylation (HBCx-9 and HBCx-24), and a BRCA1-mutant model with alterations in the 53BP1-shieldin pathway (HBCx-28), all of which have been shown to confer PARPi resistance (34).

Finally, we investigated a role for Schlafen 11 (SLFN11) in AZD8205 response, because SLFN11 expression has been associated with TOP1i as well as PARPi sensitivity in preclinical studies (35, 36). We measured SLFN11 levels in these PDXs by IHC and found a range of expression that correlated well with SLFN11 gene expression determined by RNA sequencing (R = 0.7, P = 9.02e−05; Supplementary Fig. S17). In contrast with HRD, we did not detect a statistically significant association between levels of SLFN11 protein (P = 1) or gene expression (P = 0.45) and response to AZD8205 in this study (Supplementary Fig. S18).

Combination of AZD8205 with next-generation PARP1-selective inhibitor AZD5305

PARP1 limits the cytotoxicity of TOP1i by enhancing the excision and repair of TOP1 cleavage complexes (37). Therefore, we tested the combination of AZD8205 with a next-generation PARP1-selective inhibitor, AZD5305 (18), in a collection of HRP PDX models that would not be expected to be sensitive to PARPi alone, including BRCA WT tumors and models representing post-PARP resistance mechanisms (34). In the high B7-H4–expressing, BRCA WT HBCx-39 model, AZD5305 sensitized the tumor to a single, suboptimal dose of 1.25 mg/kg AZD8205, resulting in tumor regression in five of five mice, in contrast with 24.5% TGI observed in the AZD8205 monotherapy arm (Fig. 6A). Greater activity of combination treatment was also observed with a higher dose (3.5 mg/kg) of AZD8205, extending the duration of response over that of AZD8205 monotherapy (Fig. 6B). Similar results were obtained in other high B7-H4–expressing models, including the BRCA1-mutant HBCx-24 model (Fig. 6C), where PARPi resistance is probably driven through BRCA1 promotor methylation, and the HBCx-11 model (Fig. 6D), which is BRCA1 hypomorphic and resistant to PARPi (32). The combination benefit observed in HBCx-39 and HBCx-11 is intriguing, as these models not only represent a BRCA WT (HBCx-39) or PARPi-resistant setting (HBCx-11) but are also SLFN11 negative (Supplementary Fig. S17). Because SLFN11 loss is a proposed resistance biomarker for both TOP1i and PARPi (35, 36, 38–41), the robust combination activity observed in these models was not anticipated. We were also surprised to find that combination benefit extended to HBCx-8, which has very low B7-H4 expression and is a BRCA1-mutant HRP model resistant to PARPi (Fig. 6E). In this model, a single low dose of 1.25 mg/kg AZD8205 provided modest antitumor activity (32.9% TGI), whereas combination with AZD5305 provided tumor regression in five of five mice. Similar combination benefit was observed in a low B7-H4–expressing BRCA WT model, HBCx-2 (Fig. 6F), as improved activity was observed with increasing doses of the ADC.

Figure 6.

AZD8205 antitumor efficacy in combination with the PARP1-selective inhibitor AZD5305 in TNBC PDX models. Tumors were established as described previously in the Supplementary Data and the Materials and Methods, and treatments were administered at the doses indicated in each panel. ADCs were delivered as a single bolus intravenous injection, and AZD5305 was delivered by oral gavage once daily for 28 days. A and B, Antitumor efficacy resulting from (A) 1.25 mg/kg or (B) 3.5 mg/kg ADC treatment alone or in combination with AZD5305 in the high B7-H4–expressing BRCA WT HBCx-39 model. C and D, Activity resulting from treatment of (C) high B7-H4–expressing BRCA1-mutant RAD51-positive HRP HBCx-24 and (D) BRCA1-hypomorphic RAD51-positive HRP HBCx-11 models. E and F, Efficacy resulting from treatment of (E) HBCx-8, a RAD51-positive HRP BRCA1-mutant model, and (F) HBCx-2, a BRCA WT model with low B7-H4 expression. Data indicate tumor volume mean ± standard error of the mean (n = 4 or 5 animals per group).

Figure 6.

AZD8205 antitumor efficacy in combination with the PARP1-selective inhibitor AZD5305 in TNBC PDX models. Tumors were established as described previously in the Supplementary Data and the Materials and Methods, and treatments were administered at the doses indicated in each panel. ADCs were delivered as a single bolus intravenous injection, and AZD5305 was delivered by oral gavage once daily for 28 days. A and B, Antitumor efficacy resulting from (A) 1.25 mg/kg or (B) 3.5 mg/kg ADC treatment alone or in combination with AZD5305 in the high B7-H4–expressing BRCA WT HBCx-39 model. C and D, Activity resulting from treatment of (C) high B7-H4–expressing BRCA1-mutant RAD51-positive HRP HBCx-24 and (D) BRCA1-hypomorphic RAD51-positive HRP HBCx-11 models. E and F, Efficacy resulting from treatment of (E) HBCx-8, a RAD51-positive HRP BRCA1-mutant model, and (F) HBCx-2, a BRCA WT model with low B7-H4 expression. Data indicate tumor volume mean ± standard error of the mean (n = 4 or 5 animals per group).

Close modal

To maximize the patient benefits of ADCs, careful consideration must be given to several design features, including target selection, linker–payload and conjugation chemistry, alignment of the payload mechanism of action to tumor biology, and biomarkers of response and resistance. In this study, we applied digital pathology to comprehensively understand B7-H4 expression with an innovative approach that revealed high intratumoral heterogeneity in human tumors. This finding, together with those of empirical studies assessing ADC biophysical properties and in vivo efficacy and safety, informed the design of AZD8205, a novel B7-H4–directed TOP1i-ADC. Using a collection of PDX models that closely resemble the B7-H4 expression pattern observed in human tumors, we evaluated the antitumor activity of AZD8205 alone and in combination with a next-generation PARP1-selective inhibitor, AZD5305, and applied multiparametric analyses to understand biomarkers associated with response. The clinically relevant insights provided by these studies include the finding that the underlying sensitivity of tumor cells to the TOP1i warhead may influence the level of target expression needed for response, and surprisingly, that the addition of AZD5305 can sensitize even very low B7-H4–expressing tumors to AZD8205 treatment independently of HRD status or SLFN11 positivity. The results of this study also support inclusion of γH2AX as a candidate pharmacodynamic marker to demonstrate AZD8205 target engagement during phase 1 clinical trials. Alternatively, measuring the level of ADC or unconjugated warhead in tumor biopsies after treatment could more directly demonstrate targeted delivery and release of payload to tumors.

Our results demonstrate that B7-H4 is an attractive ADC target because of its minimal expression in normal tissues and overexpression in several solid tumor types. Beyond our prevalence data, which agree with those of previous reports (11–15), our use of digital pathology advanced our understanding of B7-H4 expression by quantitatively assessing B7-H4 in hundreds of thousands of cells within an individual tumor and revealing for the first time the level of heterogeneity in human tumors. These data informed our strategy to therapeutically target B7-H4 by using an ADC with a cleavable linker-warhead capable of bystander killing. In this study, the target-dependent activity of AZD8205 in the HT29 and mixed tumor models, together with the lack of antitumor activity observed with the Iso-AZ’0133 ADC, suggest that the mechanism of AZD8205 bystander activity is intracellular release of the AZ’0132 warhead into B7-H4–positive cells and subsequent diffusion to surrounding cells. In preclinical studies, bystander-capable ADCs provide improved antitumor activity and have been shown to overcome resistance to ADCs that are not capable of this effect (42, 43). Of the five ADCs currently approved by the FDA for the treatment of solid tumors, four use bystander-capable linker–warheads: Enhertu, Trodelvy, enfortumab vedotin-ejfv (Padcev), and tisotumab vedotin-tftv (Tivdak). Although ADC activity in heterogeneous tumors can be attributed at least in part to the bystander-killing effect, intratumoral heterogeneity can still influence the activity of bystander-capable ADCs, as evidenced in our isogenic mixed tumor models by the decrease in AZD8205 activity with decreasing proportions of B7-H4–positive tumor cells. To address this complexity, we evaluated the antitumor activity of AZD8205 primarily in PDX models, which recapitulate the B7-H4 expression pattern observed in primary patient samples more faithfully than cell line–derived models.

ADCs are complex biotherapeutics comprising a mAb conjugated to a small-molecule drug. Thus, we hypothesized that AZD8205 activity could be influenced not only by B7-H4 expression but also by the inherent sensitivity of tumor cells to the released warhead. We therefore sought to understand the contribution of tumor cell biology to observed activity. Several reports have shown that genomic alterations in DNA repair genes and defects in cell-cycle checkpoints can sensitize tumor cells to TOP1i (44–47), a finding that aligns well with our PDX study results: HRD tumors responded to AZD8205 treatment even with a very low level of B7-H4 or a low proportion of B7-H4–positive cells, whereas in HRP tumors AZD8205 activity was more clearly associated with elevated levels of B7-H4. These results indicate that the level of target expression needed for ADC activity is context dependent and that incorporation of warhead sensitivity biomarkers could further refine approaches for patient selection.

Despite preclinical evidence, the clinical relevance of HRD status and TOP1i-ADC sensitivity is unclear. The association between tumor TROP-2 expression and germline BRCA1/2 mutation status on sacituzumab govitecan (SG) efficacy was investigated in the phase 3 ASCENT study (48), in which SG was found to improve survival outcomes over standard-of-care chemotherapy regardless of TROP-2 expression level or germline BRCA1/2 mutation status. Because of the small number of patients with low TROP-2–expressing tumors or positive BRCA1/2 germline mutations, the investigators were unable to determine the predictive value of germline BRCA1/2 mutations for SG response. Further clinical study in this area is warranted for SG and other TOP1i-ADCs, as identification of TOP1i sensitivity biomarkers beyond target expression may further widen the therapeutic window and best identify those patients who are most likely to respond.

Pharmacologic inhibition of the DNA damage response pathway is another attractive approach to maximize the response to TOP1i-ADCs and provide greater antitumor activity. The potential for synergy between TOP1i drugs and PARPi has already been demonstrated in preclinical models (49–52), but clinical development of PARPi in combination with TOP1i chemotherapy (e.g., irinotecan, topotecan) has been challenging due to overlapping hematological toxicity (5, 53). In this work, we evaluated the combination of AZD8205 and AZD5305, a potent PARP1-selective inhibitor demonstrating reduced hematological toxicity in preclinical models in comparison with non-selective PARPi (17, 18). In our PDX studies, the combination of AZD8205 and AZD5305 provided greater antitumor activity than monotherapy, resulting in greater tumor regression or improved duration of response even in HRP models with low B7-H4 expression. Remarkably, this combination benefit was observed with low, suboptimal doses of AZD8205 as well as in HRP BRCA1 WT or BRCA1-mutant models representing mechanisms of PARPi resistance. These findings suggest that this combination may be effective in a PARPi-resistant setting and in tumors with low B7-H4 expression, extending the potential utility and patient benefit of this approach, and that maximum ADC activity need not be achieved to capitalize on this mechanism.

Interestingly, AZD8205 monotherapy and AZD8205-AZD5305 combination benefit was observed in models that were negative for SLFN11, for which low/negative expression has been associated with reduced potency for both TOP1i and PARPi in preclinical models (35, 38, 40, 46). A limitation of our study is that we focused solely on TNBC, so more work is needed in other B7-H4–expressing tumor types to more comprehensively evaluate associations between SLFN11 expression and response. The literature suggests that associations between SLFN11 and TOP1i or PARPi sensitivity may vary by tumor type; for example, an association between SLFN11 expression and sensitivity to the PARPi olaparib or talazoparib was observed preclinically in small-cell lung cancer (38, 40) but not in TNBC (54). In a separate TNBC preclinical PDX study, SLFN11 expression was associated with response to the TOP1i irinotecan but was most predictive when combined with markers of BRCAness and RB1 loss (44). Further investigations are necessary to understand how the biology of different tumor types may influence the impact of SLFN11 and to fully exploit its potential as a sensitivity marker for both TOP1i and PARPi.

Targeted delivery of a TOP1i is expected to provide an advantage over systemic chemotherapy, yet a recent case report (55) from the phase 1b SEASTAR study demonstrates that although concurrent treatment of the PARPi rucaparib and the TOP1i-ADC SG provided promising antitumor activity, neutropenia remained as a dose-limiting toxicity (DLT). SG uses a linker that allows the TOP1i SN-38 to be released in the tumor microenvironment without prerequisite tumor cell internalization and cleavage from the anti–TROP-2 antibody. In mouse xenograft studies, treatment with 1 mg/kg SG resulted in release of unconjugated SN38 in mouse plasma at levels of >100 ng/mL, and decreasing levels of SN-38 were detected up to 48 hours after treatment (56). In contrast, we conducted a similar mouse xenograft pharmacokinetic study and found that AZD8205 was stable in circulation, as evidenced by close concordance between ADC and total antibody analytes at all time points and the detection of unconjugated AZ’0132 warhead at levels of 0.4 ng/mL after treatment with 7 mg/kg ADC (Fig. 3D). Although we hypothesize that delivery of a TOP1i warhead by the more stable AZD8205 could improve upon this toxicity profile, use of the PARP1-selective inhibitor AZD5305 may further widen the therapeutic window. Data from the ongoing phase 1/2a PETRA trial in 46 patients with advanced cancer showed that compared with first-generation PARPi, AZD5305 provided superior target engagement, promising clinical activity, and favorable tolerability with no DLT after treatment with 10–90 mg once daily (55). In addition, a “gapped schedule” could be used to widen the therapeutic index for TOP1i-ADC and PARPi combinations (5, 57). The potential for this approach was recently demonstrated in the clinic: Whereas a continuous schedule of SG in combination with the PARPi talazoparib was associated with multiple DLTs due to severe myelosuppression (including four of seven patients with febrile neutropenia), a staggered schedule was generally well tolerated (58).

Collectively, the data shared in this report provide evidence for the potential utility of AZD8205 for the treatment of B7-H4–expressing tumors and suggest potential biomarkers that could be used for patient selection and combination approaches to further improve antitumor activity. A first-in-human phase 1 study in patients with advanced solid tumors is currently underway (NCT05123482).

K. Kinneer reports a patent for WO 2022/053650 A1 pending, employment at AstraZeneca, and stock ownership and/or stock options or interests in AstraZeneca. P. Wortmann reports personal fees from AstraZeneca during the conduct of the study as well as a patent for WO2022053685A2 (a scoring method for an anti-B7H4 ADC therapy) issued. Z.A. Cooper reports personal fees from AstraZeneca during the conduct of the study. N.J. Dickinson reports a patent for WO2020200880A1 issued and WO2021148501A1 issued. L. Masterson reports a patent for US11446292B2 issued and EP3946464B1 issued. M. Davies reports employment with and holds shares in AstraZeneca. A. Lewis reports other support from AstraZeneca during the conduct of the study as well as other support from AstraZeneca outside the submitted work. Y. Huang reports grants and personal fees from AstraZeneca during the conduct of the study as well as personal fees from AstraZeneca outside the submitted work. A.I. Rosenbaum reports employment with and hold shares in AstraZeneca. J. Yuan reports grants and personal fees from AstraZeneca during the conduct of the study as well as personal fees from AstraZeneca outside the submitted work. N. Monks reports employment with and holds shares in AstraZeneca. R.J. Christie reports employment with and holds stock/financial interests in AstraZeneca. F.A. Tosto reports employment with and stock ownership/stock interests in AstraZeneca. Y. Wallez reports employment with and holds shares in AstraZeneca. E. Leo reports employment with and holds shares in AstraZeneca. M.R. Albertella reports other support from AstraZeneca outside the submitted work. D.A. Tice reports personal fees from Arcellx, Inc. outside the submitted work. P.W. Howard reports other support from AstraZeneca outside the submitted work as well as a patent for US11,446,292 issued. N. Luheshi reports personal fees and other support from AstraZeneca during the conduct of the study as well as personal fees and other support from AstraZeneca outside the submitted work. P. Sapra reports personal fees from AstraZeneca during the conduct of the study. No disclosures were reported by the other authors.

K. Kinneer: Conceptualization, resources, formal analysis, supervision, investigation, visualization, methodology, writing–original draft, writing–review and editing. P. Wortmann: Data curation, software, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. Z.A. Cooper: Resources, investigation, visualization, writing–original draft, writing–review and editing. N.J. Dickinson: Investigation, methodology, writing–review and editing. L. Masterson: Resources, investigation, methodology, writing–review and editing. T. Cailleau: Investigation, methodology, writing–review and editing. I. Hutchinson: Investigation, methodology, writing–review and editing. B. Vijayakrishnan: Resources, investigation, methodology, writing–review and editing. M. McFarlane: Resources, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. K. Ball: Formal analysis, validation, visualization, writing–review and editing. M. Davies: Formal analysis, investigation, visualization, writing–review and editing. A. Lewis: Resources, validation, investigation, visualization, methodology, writing–review and editing. Y. Huang: Resources, formal analysis, visualization, methodology, writing–review and editing. A.I. Rosenbaum: Resources, supervision, methodology, writing–review and editing. J. Yuan: Formal analysis, investigation, methodology, writing–review and editing. J. Chesebrough: Resources, formal analysis, investigation, visualization, methodology, writing–review and editing. J. Anderton: Formal analysis, visualization, methodology, writing–review and editing. N. Monks: Resources, supervision, investigation, methodology, writing–review and editing. S. Novick: Software, formal analysis, visualization, methodology, writing–original draft, writing–review and editing. J. Wang: Data curation, formal analysis, visualization, writing–review and editing. N. Dimasi: Resources, formal analysis, supervision, investigation, methodology, writing–review and editing. R.J. Christie: Supervision, investigation, methodology, writing–review and editing. D. Sabol: Investigation, methodology, writing–review and editing. F.A. Tosto: Investigation, methodology, writing–review and editing. Y. Wallez: Investigation, visualization, methodology, writing–review and editing. E. Leo: Conceptualization, investigation, methodology, writing–review and editing. M.R. Albertella: Methodology, writing–review and editing. A.D. Staniszewska: Methodology, writing–review and editing. D.A. Tice: Resources, supervision, writing–review and editing. P.W. Howard: Resources, supervision, methodology, writing–review and editing. N. Luheshi: Supervision, visualization, writing–review and editing. P. Sapra: Resources, supervision, writing–review and editing.

We gratefully acknowledge Mahammad Ahmed for the preparation of the ADCs used in this study; Ben Ruddle for conducting ex vivo serum stability studies; Si Mou for PK assay development; Minjong Park and Inna Vainshtein for image acquisition and internalization analysis; Martin Korade, Rosa Carrasco, and Laura Choi for in vitro assay support; and Deborah Shuman for editing the article and formatting the figures (all from AstraZeneca). We also thank PPD Laboratory Services for sample analysis. This study was funded by AstraZeneca.

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.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

1.
Drago
JZ
,
Modi
S
,
Chandarlapaty
S
.
Unlocking the potential of antibody–drug conjugates for cancer therapy
.
Nat Rev Clin Oncol
2021
;
18
:
327
44
.
2.
Tong
JTW
,
Harris
PWR
,
Brimble
MA
,
Kavianinia
I
.
An insight into FDA approved antibody–drug conjugates for cancer therapy
.
Molecules
2021
;
26
:
5847
.
3.
Pommier
Y
.
Drugging topoisomerases: lessons and challenges
.
ACS Chem Biol
2013
;
8
:
82
95
.
4.
O'Connor
MJ
.
Targeting the DNA damage response in cancer
.
Mol Cell
2015
;
60
:
547
60
.
5.
Thomas
A
,
Pommier
Y
.
Targeting topoisomerase I in the era of precision medicine
.
Clin Cancer Res
2019
;
25
:
6581
9
.
6.
Bardia
A
,
Tolaney
SM
,
Loirat
D
,
Punie
K
,
Oliveira
M
,
Rugo
HS
, et al
.
ASCENT: a randomized phase III study of sacituzumab govitecan vs. treatment of physician's choice in patients with previously treated metastatic triple-negative breast cancer [abstract]
.
In:
Proceedings of the ESMO Virtual Conference
;
2020
Sep 19–Oct 18;
Switzerland
:
ESMO
;
2020
.
Abstract nr LBA17
.
7.
Modi
S
,
Saura
C
,
Yamashita
T
,
Park
YH
,
Kim
SB
,
Tamura
K
, et al
.
Trastuzumab deruxtecan in previously treated HER2-positive breast cancer
.
N Engl J Med
2020
;
382
:
610
21
.
8.
Hafeez
U
,
Parakh
S
,
Gan
HK
,
Scott
AM
.
Antibody–drug conjugates for cancer therapy
.
Molecules
2020
;
25
:
4764
.
9.
Podojil
JR
,
Miller
SD
.
Potential targeting of B7-H4 for the treatment of cancer
.
Immunol Rev
2017
;
276
:
40
51
.
10.
Smith
JB
,
Stashwick
C
,
Powell
DJ
.
B7-H4 as a potential target for immunotherapy for gynecologic cancers: a closer look
.
Gynecol Oncol
2014
;
134
:
181
9
.
11.
Bregar
A
,
Deshpande
A
,
Grange
C
,
Zi
T
,
Stall
J
,
Hirsch
H
, et al
.
Characterization of immune regulatory molecules B7-H4 and PD-L1 in low and high grade endometrial tumors
.
Gynecol Oncol
2017
;
145
:
446
52
.
12.
Liang
L
,
Jiang
Y
,
Chen
JS
,
Niu
N
,
Piao
J
,
Ning
J
, et al
.
B7-H4 expression in ovarian serous carcinoma: a study of 306 cases
.
Hum Pathol
2016
;
57
:
1
6
.
13.
Parra
ER
,
Villalobos
P
,
Zhang
J
,
Behrens
C
,
Mino
B
,
Swisher
S
, et al
.
Immunohistochemical and image analysis-based study shows that several immune checkpoints are co-expressed in non–small cell lung carcinoma tumors
.
J Thorac Oncol
2018
;
13
:
779
91
.
14.
Tringler
B
,
Zhuo
S
,
Pilkington
G
,
Torkko
KC
,
Singh
M
,
Lucia
MS
, et al
.
B7-h4 is highly expressed in ductal and lobular breast cancer
.
Clin Cancer Res
2005
;
11
:
1842
8
.
15.
Zhao
X
,
Guo
F
,
Li
Z
,
Jiang
P
,
Deng
X
,
Tian
F
, et al
.
Aberrant expression of B7-H4 correlates with poor prognosis and suppresses tumor-infiltration of CD8+ T lymphocytes in human cholangiocarcinoma
.
Oncol Rep
2016
;
36
:
419
27
.
16.
Leong
SR
,
Liang
WC
,
Wu
Y
,
Crocker
L
,
Cheng
E
,
Sampath
D
, et al
.
An anti-B7-H4 antibody–drug conjugate for the treatment of breast cancer
.
Mol Pharm
2015
;
12
:
1717
29
.
17.
Illuzzi
G
,
Staniszewska
AD
,
Gill
SJ
,
Pike
A
,
McWilliams
L
,
Critchlow
SE
, et al
.
Preclinical characterization of AZD5305, a next-generation, highly selective PARP1 inhibitor and trapper
.
Clin Cancer Res
2022
;
28
:
4724
36
.
18.
Johannes
JW
,
Balazs
A
,
Barratt
D
,
Bista
M
,
Chuba
MD
,
Cosulich
S
, et al
.
Discovery of 5-{4-[(7-ethyl-6-oxo-5,6-dihydro-1,5-naphthyridin-3-yl)methyl]piperazin-1-yl}-N-m ethylpyridine-2-carboxamide (AZD5305): a PARP1- DNA trapper with high selectivity for PARP1 over PARP2 and other PARPs
.
J Med Chem
2021
;
64
:
14498
512
.
19.
Chen
L-C
,
Zhu
Y
,
Papandreou
G
,
Schroff
F
,
Adam
H
.
Encoder-decoder with atrous separable convolution for semantic image segmentation
.
In:
Ferrari
V
,
Hebert
M
,
Sminchisescu
C
,
Weiss
Y
,
editors
.
Lecture Notes in Computer Science
,
September 8–14
,
2018
;
Munich, Germany. New York
:
Springer
. p.
833
51
.
20.
Tan
M
,
Le
QV
.
EfficientNetV2: smaller models and faster training [abstract]
.
In
:
Proceedings of the 38th International Conference on Machine Learning, PMLR
2021
:
139
;
10096
10106
.
Available at
http://proceedings.mlr.press/v139/tan21a/tan21a.pdf.
Accessed November 18, 2022
.
21.
Ronneberger
O
,
Fischer
P
,
Brox
T
.
U-Net: convolutional networks for biomedical image segmentation
.
In:
Navab
N
,
Hornegger
J
,
Wells
WM
,
Frangi
AF
,
editors
.
Lecture Notes in Computer Science
,
October 5–9
,
2015
;
Munich, Germany. New York
:
Springer
. p.
234
41
.
22.
He
K
,
Zhang
X
,
Ren
S
,
Sun
J
.
Deep residual learning for image recognition
.
arXiv
.
Cornell University
;
2015
.
Available at
https://arxiv.org/abs/1512.03385.
Accessed July 27, 2022
.
23.
van Der Laak
JA
,
Pahlplatz
MM
,
Hanselaar
AG
,
de Wilde
PC
.
Hue-saturation-density (HSD) model for stain recognition in digital images from transmitted light microscopy
.
Cytometry
2000
;
39
:
275
84
.
24.
Kovtun
YV
,
Goldmacher
VS
.
Cell killing by antibody–drug conjugates
.
Cancer Lett
2007
;
255
:
232
40
.
25.
Buecheler
JW
,
Winzer
M
,
Tonillo
J
,
Weber
C
,
Gieseler
H
.
Impact of payload hydrophobicity on the stability of antibody–drug conjugates
.
Mol Pharm
2018
;
15
:
2656
64
.
26.
Lyon
RP
,
Bovee
TD
,
Doronina
SO
,
Burke
PJ
,
Hunter
JH
,
Neff-LaFord
HD
, et al
.
Reducing hydrophobicity of homogeneous antibody–drug conjugates improves pharmacokinetics and therapeutic index
.
Nat Biotechnol
2015
;
33
:
733
5
.
27.
Simmons
JK
,
Burke
PJ
,
Cochran
JH
,
Pittman
PG
,
Lyon
RP
.
Reducing the antigen-independent toxicity of antibody–drug conjugates by minimizing their non-specific clearance through PEGylation
.
Toxicol Appl Pharmacol
2020
;
392
:
114932
.
28.
Li
JY
,
Perry
SR
,
Muniz-Medina
V
,
Wang
X
,
Wetzel
LK
,
Rebelatto
MC
, et al
.
A biparatopic HER2-targeting antibody–drug conjugate induces tumor regression in primary models refractory to or ineligible for HER2-targeted therapy
.
Cancer Cell
2016
;
29
:
117
29
.
29.
Redon
CE
,
Weyemi
U
,
Parekh
PR
,
Huang
D
,
Burrell
AS
,
Bonner
WM
.
Gamma-H2AX and other histone post-translational modifications in the clinic
.
Biochim Biophys Acta
2012
;
1819
:
743
56
.
30.
Rogakou
EP
,
Pilch
DR
,
Orr
AH
,
Ivanova
VS
,
Bonner
WM
.
DNA double-stranded breaks induce histone H2AX phosphorylation on serine 139
.
J Biol Chem
1998
;
273
:
5858
68
.
31.
Castroviejo-Bermejo
M
,
Cruz
C
,
Llop-Guevara
A
,
Gutierrez-Enriquez
S
,
Ducy
M
,
Ibrahim
YH
, et al
.
A RAD51 assay feasible in routine tumor samples calls PARP inhibitor response beyond BRCA mutation
.
EMBO Mol Med
2018
;
10
:
e9172
.
32.
Pellegrino
B
,
Herencia-Ropero
A
,
Llop-Guevara
A
,
Pedretti
F
,
Moles-Fernandez
A
,
Viaplana
C
, et al
.
Preclinical in vivo validation of the RAD51 test for identification of homologous recombination-deficient tumors and patient stratification
.
Cancer Res
2022
;
82
:
1646
57
.
33.
Graeser
M
,
McCarthy
A
,
Lord
CJ
,
Savage
K
,
Hills
M
,
Salter
J
, et al
.
A marker of homologous recombination predicts pathologic complete response to neoadjuvant chemotherapy in primary breast cancer
.
Clin Cancer Res
2010
;
16
:
6159
68
.
34.
Prados-Carvajal
R
,
Irving
E
,
Lukashchuk
N
,
Forment
JV
.
Preventing and overcoming resistance to PARP inhibitors: a focus on the clinical landscape
.
Cancers
2021
;
14
:
44
.
35.
Murai
J
,
Feng
Y
,
Yu
GK
,
Ru
Y
,
Tang
SW
,
Shen
Y
, et al
.
Resistance to PARP inhibitors by SLFN11 inactivation can be overcome by ATR inhibition
.
Oncotarget
2016
;
7
:
76534
50
.
36.
Zoppoli
G
,
Regairaz
M
,
Leo
E
,
Reinhold
WC
,
Varma
S
,
Ballestrero
A
, et al
.
Putative DNA/RNA helicase Schlafen-11 (SLFN11) sensitizes cancer cells to DNA-damaging agents
.
Proc Natl Acad Sci U S A
2012
;
109
:
15030
5
.
37.
Pommier
Y
,
O'Connor
MJ
,
de Bono
J
.
Laying a trap to kill cancer cells: PARP inhibitors and their mechanisms of action
.
Sci Transl Med
2016
;
8
:
362ps17
.
38.
Lok
BH
,
Gardner
EE
,
Schneeberger
VE
,
Ni
A
,
Desmeules
P
,
Rekhtman
N
, et al
.
PARP inhibitor activity correlates with SLFN11 expression and demonstrates synergy with temozolomide in small-cell lung cancer
.
Clin Cancer Res
2017
;
23
:
523
35
.
39.
Nogales
V
,
Reinhold
WC
,
Varma
S
,
Martinez-Cardus
A
,
Moutinho
C
,
Moran
S
, et al
.
Epigenetic inactivation of the putative DNA/RNA helicase SLFN11 in human cancer confers resistance to platinum drugs
.
Oncotarget
2016
;
7
:
3084
97
.
40.
Stewart
CA
,
Tong
P
,
Cardnell
RJ
,
Sen
T
,
Li
L
,
Gay
CM
, et al
.
Dynamic variations in epithelial-to-mesenchymal transition (EMT), ATM, and SLFN11 govern response to PARP inhibitors and cisplatin in small-cell lung cancer
.
Oncotarget
2017
;
8
:
28575
87
.
41.
Ballestrero
A
,
Bedognetti
D
,
Ferraioli
D
,
Franceschelli
P
,
Labidi-Galy
SI
,
Leo
E
, et al
.
Report on the first SLFN11 monothematic workshop: from function to role as a biomarker in cancer
.
J Transl Med
2017
;
15
:
199
.
42.
Kovtun
YV
,
Audette
CA
,
Ye
Y
,
Xie
H
,
Ruberti
MF
,
Phinney
SJ
, et al
.
Antibody–drug conjugates designed to eradicate tumors with homogeneous and heterogeneous expression of the target antigen
.
Cancer Res
2006
;
66
:
3214
21
.
43.
Ogitani
Y
,
Hagihara
K
,
Oitate
M
,
Naito
H
,
Agatsuma
T
.
Bystander killing effect of DS-8201a, a novel anti-human epidermal growth factor receptor 2 antibody–drug conjugate, in tumors with human epidermal growth factor receptor 2 heterogeneity
.
Cancer Sci
2016
;
107
:
1039
46
.
44.
Coussy
F
,
El-Botty
R
,
Chateau-Joubert
S
,
Dahmani
A
,
Montaudon
E
,
Leboucher
S
, et al
.
BRCAness, SLFN11, and RB1 loss predict response to topoisomerase I inhibitors in triple-negative breast cancers
.
Sci Transl Med
2020
;
12
:
eaax2625
.
45.
Maede
Y
,
Shimizu
H
,
Fukushima
T
,
Kogame
T
,
Nakamura
T
,
Miki
T
, et al
.
Differential and common DNA repair pathways for topoisomerase I- and II-targeted drugs in a genetic DT40 repair cell screen panel
.
Mol Cancer Ther
2014
;
13
:
214
20
.
46.
Pommier
Y
.
Topoisomerase I inhibitors: camptothecins and beyond
.
Nat Rev Cancer
2006
;
6
:
789
802
.
47.
Jo
U
,
Murai
Y
,
Agama
KK
,
Sun
Y
,
Saha
LK
,
Yang
X
, et al
.
TOP1-DNA trapping by exatecan and combination therapy with ATR inhibitor
.
Mol Cancer Ther
2022
;
21
:
1090
102
.
48.
Bardia
A
,
Tolaney
SM
,
Punie
K
,
Loirat
D
,
Oliveira
M
,
Kalinsky
K
, et al
.
Biomarker analyses in the phase III ASCENT study of sacituzumab govitecan versus chemotherapy in patients with metastatic triple-negative breast cancer
.
Ann Oncol
2021
;
32
:
1148
56
.
49.
Cardillo
TM
,
Sharkey
RM
,
Rossi
DL
,
Arrojo
R
,
Mostafa
AA
,
Goldenberg
DM
.
Synthetic lethality exploitation by an anti-Trop-2-SN-38 antibody–drug conjugate, IMMU-132, plus PARP inhibitors in BRCA1/2–wild-type triple-negative breast cancer
.
Clin Cancer Res
2017
;
23
:
3405
15
.
50.
Chowdhuri
SP
,
Das
BB
.
Top1-PARP1 association and beyond: from DNA topology to break repair
.
NAR Cancer
2021
;
3
:
zcab003
.
51.
Murai
J
,
Pommier
Y
.
PARP trapping beyond homologous recombination and platinum sensitivity in cancers
.
Annu Rev Cancer Biol
2019
;
3
:
131
50
.
52.
Murai
J
,
Zhang
Y
,
Morris
J
,
Ji
J
,
Takeda
S
,
Doroshow
JH
, et al
.
Rationale for poly(ADP-ribose) polymerase (PARP) inhibitors in combination therapy with camptothecins or temozolomide based on PARP trapping versus catalytic inhibition
.
J Pharmacol Exp Ther
2014
;
349
:
408
16
.
53.
Chen
EX
,
Jonker
DJ
,
Siu
LL
,
McKeever
K
,
Keller
D
,
Wells
J
, et al
.
A Phase I study of olaparib and irinotecan in patients with colorectal cancer: Canadian Cancer Trials Group IND 187
.
Invest New Drugs
2016
;
34
:
450
7
.
54.
Winkler
C
,
Armenia
J
,
Jones
GN
,
Tobalina
L
,
Sale
MJ
,
Petreus
T
, et al
.
SLFN11 informs on standard of care and novel treatments in a wide range of cancer models
.
Br J Cancer
2021
;
124
:
951
62
.
55.
Yap
TA
,
Im
SA
,
Schram
AM
,
Sharp
A
,
Balmana
J
,
Baird
RD
, et al
.
PETRA: first in class, first in human trial of the next-generation PARP1-selective inhibitor AZD5305 in patients with BRCA1/2, PALB2, or RAD51C/D mutations
.
Abstract CT007
.
Cancer Res
2022
;
82
(
12 Suppl
):
CT007
.
56.
Goldenberg
DM
,
Cardillo
TM
,
Govindan
SV
,
Rossi
EA
,
Sharkey
RM
.
Trop-2 is a novel target for solid cancer therapy with sacituzumab govitecan (IMMU-132), an antibody–drug conjugate (ADC)
.
Oncotarget
2015
;
6
:
22496
512
.
57.
Dhawan
MS
,
Rahimi
R
,
Karipineni
S
,
Wilch
S
,
Zigman
E
,
Aggarwal
RR
, et al
.
Phase I study of rucaparib and irinotecan in advanced solid tumors with homologous recombination deficiency mutations
.
J Clin Oncol
2020
;
38
:
3513
.
58.
Bardia
A
,
Coates
JT
,
Spring
L
,
Sun
S
,
Juric
D
,
Thimmiah
N
, et al
.
Sacituzumab govitecan combination with PARP inhibitor talazoparib in metastatic triple-negative breast cancer: translational investigation
.
Abstract 2638
.
Cancer Res
2022
;
82
(
12 Suppl
):
2638
.

Supplementary data