SGN-B6A: A New Vedotin Antibody–Drug Conjugate Directed to Integrin Beta-6 for Multiple Carcinoma Indications

This article is featured in Selected Articles from This Issue, p. 1363

The integrins are a large family of membrane-associated proteins with diverse roles in cellular adhesion, motility, and cytokinesis (1). Functional integrins exist as heterodimers consisting of single alpha and beta chains, with 18 known alpha chain isoforms and 8 known beta chain isoforms. Within this family, integrin beta-6 (which dimerizes exclusively with isoform alpha-v) is of particular interest for its role in cancer (2, 3). Integrin beta-6 is overexpressed in numerous solid tumors, and multiple investigators have noted its expression as a negative prognostic indicator in diverse cancers including colorectal, non–small cell lung carcinoma (NSCLC), gastric, and cervical (4–8). Previous investigators have shown by mRNA in situ hybridization that integrin beta-6 expression is restricted to epithelial tissues in nonhuman primates (9). This was subsequently confirmed by IHC using cryopreserved sections of both rhesus monkey and human tissues (10). We have verified these findings and observed expression at low levels in epithelial cells of various tissues including the gastrointestinal tract, skin, bronchiolar epithelium of the lung, tubular epithelium of the kidney, transitional epithelium of the renal pelvis and urinary bladder, and ductal epithelium of the pancreas, breast, and salivary gland.

In addition to the low constitutive expression noted above, integrin beta-6 expression is induced by tissue injury due to its role in tissue remodeling and wound repair (10), including its ability to activate TGFβ (11). This role in tissue remodeling is believed to be the function that malignant cells exploit through its overexpression, allowing them to become more invasive into surrounding healthy tissue (12). Integrin beta-6 has also been reported to be among the integrins that can promote the epithelial to mesenchymal transition as well as provide resistance to anoikis, thus increasing the metastatic potential of malignant cells expressing it (13–15).

Over the past decade, the long-promised potential of antibody–drug conjugates (ADC) in the field of cancer therapy has begun to be realized. Currently approved ADCs use a variety of different drug linker formats (16), which exert antitumor activity by releasing a cytotoxic payload that inhibits microtubule dynamics (17), blocks topoisomerase I (18), or directly binds to DNA (19). The most widely used clinically validated drug linker is the auristatin-based vedotin linker-payload system, in which monomethyl auristatin E (MMAE) is released, resulting in cell cycle arrest following microtubule inhibition (20, 21). Vedotin ADCs are designed to address unmet needs in oncology and expand treatment options for people living with cancer and have been approved in both hematologic and solid tumor indications. Recent studies have suggested that their clinical activity may be augmented when combined with checkpoint inhibitors (22, 23).

SGN-B6A is a novel ADC directed to integrin beta-6 to deliver the cytotoxic payload MMAE to tumor cells expressing the antigen. In this study, we demonstrate preclinically that SGN-B6A is well tolerated and capable of inhibiting tumor growth or regressing tumor volume in multiple integrin beta-6–positive carcinoma models. Further, our complete data package including pharmacokinetic (PK) and toxicology assessment of SGN-B6A in cynomolgus monkeys suggests that selective targeting of integrin beta-6 is preferrable to targeting pan alpha-v antigens due to the more limited normal tissue expression profile of integrin beta-6. Taken together, these studies indicate that integrin beta-6 is a viable therapeutic target and SGN-B6A has the potential to be a promising ADC for solid tumor malignancies.

IHC detection of integrin beta-6 was developed and performed at Mosaic Laboratories (Lake Forest, CA). The procedure for IHC analysis of integrin beta-6 [mouse clone 6.2A1 E2 (24), expressed as hybridoma from ATCC] was performed on the Dako Autostainer Link 48 (Dako, Carpinteria, California). Four-micron tissue sections were mounted onto positively charged glass slides, dried, baked, deparaffinized with xylene, and rehydrated using sequential alcohol rinses (100%, 95%, 80%) followed by a final rinse in distilled water. Antigen retrieval was done by pretreating slides with FLEX Target Retrieval Solution, Low pH for 20 minutes at 97°C and cooling to 65°C in a PT Link. Slides were rinsed with FLEX Wash Buffer and transferred to the Autostainer Link 48. Slides were rinsed in FLEX Wash Buffer and incubated with FLEX Peroxidase-Blocking Reagent and primary antibody (0.1375 μg/mL) diluted in Dako Diluent for 30 minutes. Slides were incubated with FLEX+ Mouse Linker and FLEX horseradish peroxidase (HRP) for 20 minutes, and FLEX DAB+ Chromogen for 10 minutes, with rinses between incubations. After incubation with FLEX hematoxylin and additional buffer and water rinses, slides were removed from the autostainer and dehydrated in alcohol (95%, 100%), cleared in xylene, and mounted with a tape coverslipper (Sakura Fine-Tek, Torrance, California) or glass coverslipper (Leica Biosystems, Buffalo Grove, Illinois) in accordance with Mosaic Laboratories’ procedure.

The developed method was applied to tumor samples from the Mosaic Laboratories biorepository and evaluated by pathology review through a light microscope. The percentage of tumor cells staining at each of the following four levels was evaluated: 0 (no staining), 1+ (weak staining), 2+ (moderate staining) and 3+ (strong staining). The pathologist H-Score was calculated for membrane staining based on the summation of the product of percent of cells stained at each intensity using the following equation: (3 x % cells staining at 3+) + (2 × % cells staining at 2+) + (1 × % cells staining at 1+).

BALB/c mice were immunized three times with intraperitoneal injections of ∼5×10⁶ 3T3:huβ6 transfectants. Three days prior to fusion, mice received a final injection of purified recombinant human αvβ6 that was given intravenously (6 μg) and intraperitoneally (30 μg). Lymphocytes harvested from spleen and lymph nodes were fused to P3X63Ag8.653 myeloma cells using polyethylene glycol. Fused cells were recovered overnight in hybridoma growth media (IMDM containing 4 mmol/L glutamine, 10% fetal clone I, 10% cloning factor and penicillin/streptomycin). Following recovery, cells were spun down and then plated in semi-solid media consisting of CloneMatrix media supplemented with hybridoma growth media plus HAT for hybridoma selection and CloneDetect for IgG-production. Hybridomas were incubated for 10 days at 37°C. At day 10, IgG producing hybridoma clones were picked using a ClonePixFL (Molecular Devices) and transferred to 96-well plates containing IgG-depleted hybridoma growth media plus HT. Hybridoma culture supernatants were screened on 293F:huβ6 transfectants and positive clones identified using an Alexifluor-647 labeled secondary antibody. The plates were read in an FMAT 8200 (Applied Biosystems). Hybridomas that bound to 293F:huβ6 and 293F:cynoβ6 but not 293F:vector were expanded for conjugation as previously described (25). The resulting ADC panel was tested in binding and cytotoxicity assays. On the basis of its cytotoxic activity as an ADC on multiple αvβ6-positive tumor cell lines and its comparable affinity to human and cynomolgus forms of the antigen, 2A2 was selected as the lead antibody for humanization.

The murine 2A2 (m2A2) CDR domain sequences were engineered into suitable human framework domain sequences selected from the human germline repertoire based on framework homology, sequence usage, and conservation of canonical structure (amino acid sequence shown in Supplementary Table S1). Nucleotide sequences coding for the humanized 2A2 heavy and light chain variable regions were joined to coding regions for the human G1 and human kappa constant regions, respectively, to create the final antibody candidate designated h2A2.

SGN-B6A was generated by conjugation of the previously disclosed vedotin drug-linker to h2A2 native cysteine residues using standard methods to create an ADC with an average of 4 drugs per antibody (Supplementary Fig. S1; ref. 25).

MaxiSorp 96 well plates (Nunc, Denmark) were coated overnight at 4°C with 1 μg/mL recombinant human integrin beta isoforms 1, 3, 5, 6, and 8 (R&D Systems, Minneapolis, MN) in 50 mmol/L carbonate buffer. Plates were washed with PBS + 0.05% Tween 20 (PBS-T) and blocked for 2 hours at room temperature in TBS blocking buffer (TBS, 0.05% Tween 20, 1% BSA). Plates were washed and then incubated for 2 hours with SGN-B6A or h2A2 in TBS binding buffer (TBS, 0.05% Tween 20, 1% BSA, 1 mmol/L MnCl₂). Plates were washed, incubated for 1 hour with 1:5,000 dilution of HRP labeled anti-human Fc (Jackson ImmunoResearch, West Grove, Pennsylvania), washed, and then incubated with tetramethyl benzidine substrate for 5 minutes. The reaction was stopped with 1 mol/L HCl. Absorbance at 450 nm was read using a Fusion HT plate reader (Perkin Elmer, Waltham, Massachusetts).

Saturation binding studies were conducted with AlexaFluor-647-labeled h2A2 using 293F:hu beta-6 and 293F:cyno beta-6 transfected cell lines. Cells (0.1×10⁶) were aliquoted into a 96-well plate and h2A2 added in buffer (TBS, 2% FBS, 0.5 mmol/L MnCl₂, 0.02% NaN₃). Cells were incubated for 1 hour then pelleted and washed 3 times with TBS, then resuspended in 120 μL of TBS. Fluorescent signal of binding was detected using LSR II flow cytometer (Becton Dickinson, San Jose, California). The binding data were fitted to a 4-parameter sigmoidal binding curve using GraphPad Prism (San Diego, California).

SGN-B6A was visualized on the plasma membrane of BxPC-3 cells and post-internalization by immunofluorescence by conjugating with AlexaFluor-555 NHS ester. Cells were dosed with 2 μg/mL SGN-B6A-AF555 and preincubated on ice for 30 minutes. Treated cells were washed with PBS to remove unbound ADC, then fixed and permeabilized or further incubated at 37°C for 4 hours. In addition to the signal of SGN-B6A-AF555, fixed cells were visualized with anti-lysosomal associated membrane protein 1 (LAMP-1) antibody (BD, catalog no. 562622) and Hoechst dye.

Cancer cell lines expressing alpha-v/beta-6 (BxPC-3, Detroit 562, HPAFII, and SW780) were purchased from ATCC (Manassas, Virginia) in 2013. BXPC3 and SW780 cells were maintained in RPMI1640 (Invitrogen) supplemented with 10% heat-inactivated FBS (HiFBS; Invitrogen). Detroit562 and HPAFII cells were maintained in Eagle Minimum Essential Medium (ATCC) supplemented with 10% HiFBS (Invitrogen). All cell lines were cultured at 37°C with 5% CO₂ in the relevant media mentioned above; cultures were maintained for less than 8 weeks. All cell line source vials were confirmed to be mycoplasma-negative using PCR protocols and authenticated by IDEXX Bio Analytics (Columbia, Missouri) using the CellCheck 16 Plus platform in July 2022.

For in vitro assays, cells were allowed to grow and recover until viability determined by Vi-CELL XR (Beckman Coulter, Indianapolis, Indiana) was above 90%. Cells were plated and allowed to adhere overnight. Serial dilutions of the ADCs were prepared in RPMI1640 supplemented with 20% FBS, added to each cell plate in triplicate, and incubated for 96 hours. CellTiter-Glo luminescent assay (Promega Corporation, Madison, Wisconsin) was prepared according to the manufacturer's protocol and luminescence determined using an EnVision plate reader (Perkin Elmer). Data were analyzed in GraphPad Prism using a nonlinear, 4-parameter curve fit model [Y = Bottom + (Top-Bottom)/(1+10^{((LogEC₅₀-X)*HillSlope))}].

All experiments involving animal handling and study designs were reviewed and approved by the Institutional Animal Care and Use Committees of the institutions where the work was performed, as described below.

Studies with cell line-derived xenografts were performed at Seagen (Bothell, Washington). Seagen is fully accredited by the Association and Accreditation of Laboratory Animal Care. Cells were injected subcutaneously into 5 to 8 female nude mice per group (Envigo, Indianapolis, Indiana) and randomly divided into study groups once the tumors reached approximately 100 mm³. SGN-B6A was dosed at 1 or 3 mg/kg weekly for 3 doses via the IP route, and nonbinding control ADC at 3 mg/kg weekly for 3 doses. Untreated control mice were included in each study. Animals were euthanized when tumor volumes reached 500 to 1,000 mm³. Tumor sizes were measured regularly, and tumor volume was calculated with the formula (volume = ½ × length × width × width).

Studies with patient-derived xenograft (PDX) models of NSCLC were conducted at Champions Oncology (Hackensack, New Jersey). Models were grown subcutaneously in nude mice (Envigo). When sufficient stock animals reached 1,000 to 1,500 mm³ tumor volume, tumors were harvested for re-implantation into pre-study animals. Pre-study animals were implanted with tumor fragments harvested from stock animals. When tumors reached an average tumor volume of 150 to 300 mm³, animals were matched by tumor volume into treatment or control groups (N = 3) to be used for dosing with SGN-B6A or nonbinding control ADC at 3 mg/kg weekly for 3 doses. Untreated control mice were included in each study. Animals were euthanized when tumor volumes reached 1,500 mm³ or at day 60 post-dose. Tumor sizes were measured twice weekly, and tumor volume was calculated with the formula (volume = ½ × length × width × width).

A Good Laboratory Practice study of SGN-B6A PK and toxicity was conducted in cynomolgus monkeys at Charles River Laboratories (Mattawan, Michigan). SGN-B6Awas intravenously administered to 3 male and 3 female monkeys (per dose group) by slow bolus injection at 1-week intervals over a 4-week period at 3, 4, and 5 mg/kg and at a 3-week interval at 6 mg/kg. The following parameters and endpoints were evaluated in this study: mortality, clinical observations, body weight, blood pressure, respiration rate, body temperature, opthalmoscopic, physical, neurologic, and electrocardiographic examinations, clinical pathology parameters (hematology, coagulation, clinical chemistry, and urinalysis), PK parameters, gross necropsy findings, and histopathologic examinations.

For assessment of PK, plasma was collected at various time points with di-potassium ethylenediaminetetraacetic acid (K2EDTA), and total antibody (TAb) concentrations were determined by a validated ELISA with an anti-idiotypic monoclonal antibody (mAb) directed to h2A2. A sandwich ELISA configuration was used in microtiter plates coated with an anti-idiotypic mAb directed to the h2A2 antibody component of SGN-B6A. Bound SGN-B6A was detected with a biotinylated anti-idiotypic mAb directed to the h2A2 antibody component of SGN-B6A. The lower and upper limit of quantitation for this method were 32 and 512 ng/mL, respectively.

Raw data for this study were generated at Seagen (Bothell, Washington) and under contract at Mosaic Laboratories (IHC; Lake Forest, California), Champions Oncology (PDX models; Hackensack, New Jersey), and Charles River Laboratories (Mattawan, Michigan). Data generated in this study are available within the article and its Supplementary Data files or from the corresponding author upon reasonable request.

All animal studies were conducted in accordance with protocols reviewed and approved by the Institutional Animal Care and Use Committee at Seagen or the external testing facility that conducted the studies.

The RNA expression of integrin beta-6 (ITGB6 gene) in cancer and normal tissues was assessed using The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression datasets (Supplementary Fig. S2). Pancreas, lung, esophageal, head and neck squamous cell carcinoma (HNSCC), and bladder carcinomas were among the many tumor types that showed relatively high integrin beta-6 RNA levels, along with lower levels in normal tissues including lung, bladder, and kidney, consistent with previous reports (9). IHC staining of formalin-fixed, paraffin-embedded tissues (FFPE) confirmed the expression of integrin beta-6 protein in tumor sections from ovarian, breast, cutaneous skin squamous cell carcinoma, HNSCC, esophageal, and lung (adenocarcinoma and SCC) cancers. Fig. 1A–E shows representative images of integrin beta-6 staining along with the corresponding H-scores based on staining intensity and percentage of positive tumor cells in the section. Scatter plots of the H-scores for individual tumor samples illustrate the expression profile of integrin beta-6 within these tumor types (Supplementary Fig. S3) and show good concordance with the RNA data in relative expression across indications.

The anti-integrin beta-6 antibody 2A2 was discovered using a mouse hybridoma campaign from a panel of ∼1100 clonal using criteria that included binding specificity, ADC potency, and affinity to human and cynomolgus monkey homologues. Although several antibodies met selectivity and affinity criteria, m2A2 consistently delivered the most favorable potency in cytotoxicity assays using the vedotin payload and was selected for humanization. Following humanization, the reformatted h2A2 was conjugated with the vedotin drug-linker to create SGN-B6A and its binding shown to be exclusive to the integrin dimer alpha-v/beta-6 with no detectable binding to other alpha-v integrins by ELISA (Fig. 2A). In a cell-based assay, saturation binding of h2A2 to human and cynomolgus monkey integrin beta-6 were highly concordant, with EC₅₀ values of 0.3 and 0.2 nmol/L, respectively (Supplementary Fig. S4A). Similarly, saturation binding curves of h2A2 and SGN-B6A to human integrin beta-6 expressing cells were similar, indicating that conjugation does not impact binding (Supplementary Fig. S4B).

The cell surface binding and intracellular trafficking of SGN-B6A were evaluated in BxPC-3 cells by fluorescence microscopy. At t₀, SGN-B6A could be observed as a halo of red staining on the cell membrane (Fig. 2B). By 4 hours, the red staining was no longer localized to the plasma membrane, and the appearance of orange color in intracellular vesicles indicated colocalization with the green-stained lysosomes. These data are consistent with receptor-mediated internalization of SGN-B6A followed by intracellular drug delivery through a lysosomal pathway. The in vitro cytotoxicity of SGN-B6A was evaluated using the four integrin beta-6-expressing carcinoma cell lines described in Supplementary Table S2, which includes the observed integrin beta-6 expression determined by QFACS. SGN-B6A exhibited antigen-dependent cytotoxic activity in all four models with EC₅₀ values ranging from 10 to 200 ng/mL (Fig. 2C). The observed differences in apparent potency are likely to be related to biological factors such as mitotic index or efflux pump expression rather than arising from the narrow range of integrin beta-6 expression. SGN-B6A was inactive against two MMAE-sensitive cell lines that do not express integrin beta-6 (Supplementary Fig. S5).

Though the primary mechanism of action of SGN-B6A is predicted to be delivery of MMAE as a cytotoxic payload, we explored any potential contributions that may occur through immune-mediated effector function. SGN-B6A effector function activity was evaluated in vitro using SW780 bladder cancer cells at doses ranging from 0.001 μg/mL up to 50 μg/mL depending on the assay. The results revealed that SGN-B6A displays antibody-dependent cellular phagocytosis activity (EC₅₀ = 0.7 μg/mL) but not antibody-dependent cellular cytotoxicity or complement-dependent cytotoxicity (Supplementary Fig. S6).

The in vivo activity of SGN-B6A was determined in subcutaneous xenograft models using tumor cell lines (N = 3) and PDX (N = 21) models. SGN-B6A demonstrated dose-dependent antitumor activity in all three cell line models, although there was variability in the depth and durability of response (Fig. 3A–C). In the Detroit 562 model, 3 mg/kg SGN-B6A induced durable tumor regressions out to day 40, with 2 of 8 animals maintaining complete tumor regressions until the end of study (Fig. 3A). With BxPC-3, all 8 animals dosed at 3 mg/kg had reductions in tumor volume, and one maintained complete regression (Fig. 3B). HPAFII was the least responsive model with only a slight decline in tumor volume during the dosing period, but the tumor resumed growth upon dosing cessation (Fig. 3C). Supplementary Figure S7 provides tumor volumes for individual animals in each of these studies at the last timepoint prior to animals being sacrificed from the untreated group, including statistical comparisons between groups. As has been observed for other models when treated with vedotin ADCs, the nonbinding ADC exhibited activity ranging from slight tumor growth delay to tumor stasis during the dosing interval (26). IHC staining of integrin beta-6 is shown alongside the growth curves for each tumor xenograft in Fig. 3, and the results revealed a distinct pattern of expression. Unlike HPAFII, the Detroit 562 and BxPC-3 tumors exhibited substantially greater expression of integrin beta-6 at the tumor periphery that diminished toward the center, increasing the potential relevance of MMAE-mediated bystander activity in these models (27).

Next, we broadened the evaluation of SGN-B6A to include PDX models of human NSCLC. Twenty-one engrafted models representing adenocarcinoma (10 models) and squamous cell carcinoma (11 models) histologies were dosed with 3 mg/kg SGN-B6A or nonbinding ADC weekly for 3 doses. The observed antitumor activity of SGN-B6A across these 21 PDX models is shown as a percent change from baseline at best response (Fig. 4A). The observed responses to the nonbinding ADC are shown in Fig. 4B; because the models are arranged in the same left-right order in these two panels, the contribution of antigen-dependent activity can be seen by comparison. To assess any potential relationship between integrin beta-6 activity and response, tumors from the untreated control groups from each model were stained by IHC and the response of those models to SGN-B6A treatment is shown plotted against the resulting H-scores (Fig. 4C). While tumor regressions were observed across the expression continuum, among the nine models with an H-score above 100 there were seven that resulted in tumor volume regression. Figure 4C also illustrates that although H-scores did skew higher for squamous cell models over adenocarcinoma models, regressions were observed for both tumor histologies. The observed H-scores for the PDX models also correlated well with expression of ITGB6 by RNA in the database provided by Champions Oncology (Supplementary Fig. S8). Complete growth curves for all models are presented in Supplementary Fig. S9.

After establishing the preclinical antitumor activity of SGN-B6A using multiple in vivo models, we evaluated the PK and tolerability of SGN-B6A in cynomolgus monkeys. Cynomolgus monkeys (3 male, 3 female per dose group) received 4 weekly doses of SGN-B6A at 3, 4, and 5 mg/kg or 2 doses of SGN-B6A at 6 mg/kg every 3 weeks. All doses were tolerated, and the resulting antigen-independent dose-limiting hematologic toxicities were consistent with those of other vedotin ADCs (28, 29). Briefly, the hematology findings included moderate to marked reversible reductions in neutrophils and reticulocytes, corresponding with minimal to mild decreases in sternal bone marrow cellularity observed by histopathology. Minimal to mild epithelial toxicity in the urothelial tract was also observed, along with single-incidence findings of cell injury in the esophagus, trachea, and cecum at doses of 5 mg/kg weekly or 6 mg/kg every 3 weeks. These epithelial findings coincide with tissues known to express integrin beta-6, and therefore were considered to be possibly related to antigen expression. No life-threatening toxicities were observed at any dose; therefore, the highest non-severely toxic dose was considered 5 mg/kg/dose when administered weekly for 4 consecutive weeks. Plasma exposure (total antibody) of SGN-B6A was approximately linear with dose and did not exhibit a PK profile suggestive of target-mediated drug disposition (Fig. 5; Supplementary Table S3). Antibody-conjugated MMAE exhibited slightly faster clearance than the total antibody, consistent with the known properties of the vedotin platform (refs. 30–33; Supplementary Fig. S10).

The aim of using ADCs in cancer therapy is to improve the therapeutic index of cytotoxic agents and minimize drug exposure in normal tissue. SGN-B6A uses the same vedotin drug-linker system (maleimidocaproyl-valine-citrulline-MMAE) that has been clinically validated with ADCs including brentuximab vedotin, polatuzumab vedotin, and enfortumab vedotin (34–36). The proven efficacy of these ADCs in a variety of malignancies paired with their generally predictable and manageable adverse event profile provides a strong rationale for the further development of vedotin ADCs directed against additional tumor-associated antigens.

Integrin beta-6, the target of SGN-B6A, is a membrane protein with low levels of constitutive expression on normal epithelial tissue but with higher levels of expression on a wide variety of tumors. IHC analyses of human tumor samples reported in the literature and replicated at Seagen, Inc., have revealed that integrin beta-6 is expressed at a high frequency in multiple carcinomas including NSCLC (squamous and adenocarcinoma), HNSCC, esophageal, and skin. The IHC results reflected RNA expression data from TCGA. The expression pattern of integrin beta-6 in malignancies with an unmet medical need suggests that it could be a promising target for ADC therapy.

The h2A2 antibody used for SGN-B6A is specific to the beta-6 subunit of the alpha-v/beta-6 integrin dimer and does not bind to other alpha-v dimers. This fact differentiates SGN-B6A from other integrin-directed antibody-based therapeutics that have been tested in oncology indications such as intetumumab (37) and abituzumab (38), both of which bind to the alpha-v subunit. Because of the more widespread expression pattern of alpha-v integrin, both of these antibodies exhibited substantial target-mediated disposition and rapid clearance at the lower tested doses (39). Interestingly, the abituzumab study included IHC analysis of alpha-v as well as the individual beta-3, beta-5, beta-6, and beta-8 subunits. Of the beta subunits, only expression of beta-6 had any prognostic value, with high expression in the abituzumab arm leading to the greatest treatment benefit and high expression in the control arm resulting in poorer survival (6). These results provide further rationale for targeting integrin beta-6 specifically rather than through a pan-alpha-v integrin antibody.

Selection of h2A2 antibody as the lead antibody for SGN-B6A was driven largely by its ability to effectively deliver drug as opposed to acting functionally through ligand blockade of extracellular matrix protein interactions or inhibition of latency-associated peptide (LAP) activation of TGFβ. In vitro evaluation of m2A2 showed only weak blockade of LAP binding to integrin beta-6 (Supplementary Fig. S11) and is therefore unlikely to directly inhibit LAP activation of TGFβ. Though functional blockade of integrin receptors has been a viable therapeutic strategy for certain autoimmune and cardiovascular disorders (40), antibody-based targeting of integrins has yet to be validated within the sphere of oncology. Reported clinical data on safety and efficacy have generally shown that integrin-based antibody therapies are well tolerated but lack sufficient efficacy (41–43). Functional redundancy of integrins and the potential for target-mediated drug disposition may contribute to the limited therapeutic efficacy seen in oncology thus far. Our strategy, in contrast, is based on the clinically validated approach of antibody-mediated drug delivery. As such, the anticancer activity of SGN-B6A is expected to be mediated primarily through cellular internalization and release of the cytotoxic microtubule-disrupting agent MMAE. This mechanism has been described extensively for other vedotin ADCs and includes the possibility of immunogenic cell death induction (44–46).

The in vivo antitumor activity of SGN-B6A was examined in murine xenograft models derived from three of the cell lines used for the in vitro studies. Effects of SGN-B6A ranged from tumor growth delay in the least responsive model to reductions in tumor volume and complete eradication of tumor in up to 25% of the treated mice in the more responsive models. Varying degrees of antigen-independent activity were observed from the nonbinding ADC in these studies. This is consistent with prior vedotin ADC experience and may arise from uptake of the ADC by tumor associated macrophages or other infiltrating murine immune cells (26). Given the highly membrane permeable characteristics of the MMAE payload, such localized delivery through antigen-independent mechanisms may allow distribution within the tumor and observable growth inhibition. Differences in activity between xenograft models (from both SGN-B6A and the nonbinding ADC) may arise from such differences in the tumor microenvironment or from variation in tumor growth rate, degree of vascularization, or other uncharacterized properties.

In a PDX study of NSCLC, SGN-B6A induced tumor reductions in 12 of 21 models, including 9 models with reductions greater than 30%. As with the CDX models, the nonbinding vedotin ADC demonstrated some growth delay in a subset of these models, but in no case was able to decrease tumor volume. Tumor volume reductions with SGN-B6A treatment were observed in models representing both squamous cell and adenocarcinoma histologies, and across a range of integrin beta-6 expression assessed by IHC. However, of the 9 models with an H-score above 100, 7 (78%) were shrunk by SGN-B6A treatment, versus 5 of 12 models (42%) with H-scores below this value. While not conclusive, the results of this small signal-finding study provide some early suggestion that SGN-B6A activity may be proportional to integrin beta-6 expression in a larger sample set. Annotation of patient pretreatment status (naive vs. pretreated) is available for most of these PDX models and has been noted in Supplementary Fig. S9. As one might expect, many of the best-responding models were treatment naive; however, tumor reductions were observed in models derived from pretreated patients as well. Additional details regarding the individual PDX models and characteristics of the patients from which they were derived are provided in Supplementary Table S4.

These in vivo studies demonstrate the efficacy of SGN-B6A in multiple models at relevant doses. Moreover, the doses of SGN-B6A tolerated in cynomolgus monkeys are similar or greater than those of other vedotin ADCs that have been approved across multiple oncology indications. To illustrate, Table 1 provides data on two such vedotin ADCs approved in solid tumors and allows for benchmarking of the preclinical data presented here for SGN-B6A with corresponding preclinical data for those ADCs. Although dosing schedules vary across the published preclinical data packages, SGN-B6A compares favorably against these other ADCs with antitumor activity being observed in multiple models at similar dose intensity and demonstrated tolerability in cynomolgus monkeys at higher dose intensity. By leveraging this comparison against historical data from now-approved ADCs that also use the vedotin platform, we believe that the results reported here support the potential of SGN-B6A as an ADC for the treatment of multiple carcinoma indications. Accordingly, it is currently being evaluated in a phase I clinical study at multiple sites in North America and Europe (NCT04389632).

R.P. Lyon reports a patent for anti-avb6 antibodies and ADCs pending; and employment by and stock ownership in Seagen, Inc. L. Westendorf reports as an employee, I own Seagen stock. C.J. Hale reports I own Seagen stock. J.L. Stilwell reports other support from Seagen outside the submitted work. N. Yeddula reports employee of Seagen and holds stock in the company. K.M. Snead reports she owns stock in Seagen Inc. V. Kumar reports I own Seagen stocks. G.I. Patilea-Vrana reports employment with and stock ownership of Seagen Inc. K. Klussman reports I have stock ownership in Seagen Inc. M.C. Ryan reports other support from Seagen, Inc. during the conduct of the study; other support from Seagen, Inc; and other support from Seagen, Inc. outside the submitted work; in addition, M.C. Ryan has a patent for US Patent 9,493,566 issued to Seagen. No disclosures were reported by the other authors.

R.P. Lyon: Conceptualization, supervision, writing–original draft, writing–review and editing. M. Jonas: Investigation. C. Frantz: Investigation, writing–original draft. E.S. Trueblood: Investigation. R. Yumul: Investigation. L. Westendorf: Investigation. C.J. Hale: Formal analysis, visualization. J.L. Stilwell: Supervision. N. Yeddula: Investigation. K.M. Snead: Investigation. V. Kumar: Formal analysis. G.I. Patilea-Vrana: Formal analysis. K. Klussman: Investigation. M.C. Ryan: Conceptualization, supervision, investigation, writing–original draft, writing–review and editing.

The authors would like to acknowledge the following individuals for their contribution to this work: Dennis Benjamin, for years of mentorship as we developed this project; Brad Meyer and Heather Kostner for screening anti-integrin beta-6 antibodies to identify the 2A2 clone; Rob Lawrence for assistance in visualizing ITGB6 expression by RNASeq; Janelle Taylor for maintenance of our research bell bank; Iliyana Mikell for assistance in preparing this manuscript. The experiments described in this manuscript were funded by Seagen Inc.

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.