The Dolaflexin-based Antibody–Drug Conjugate XMT-1536 Targets the Solid Tumor Lineage Antigen SLC34A2/NaPi2b

Antibody–drug conjugates (ADC) enable the tumor-targeted delivery of cytotoxic agents. This therapeutic modality reduces the systemic toxicity associated with potent chemotherapeutic agents while enhancing the antitumor activity (1, 2).

Of the seven FDA-approved ADCs, four target lineage-specific hematologic antigens that are not markedly differentially expressed in neoplastic versus nonneoplastic cells. CD30 is expressed in activated T and B cells (3, 4); brentuximab-vedotin, an ADC targeting CD30, is approved for the treatment of tumors arising from these cell populations, including Hodgkin lymphoma and systemic anaplastic large cell lymphoma (2, 5, 6). Inotuzumab ozogamicin targets CD22, a B-cell lineage antigen, and is approved for B-cell precursor acute lymphoblastic leukemia (7). Gentuzumab ozogamicin targets CD33, a myeloid lineage antigen, and is approved for acute myeloid leukemia (8). Polatuzumab vedotin targets CD79b, an antigen expressed in B cells and B-cell lymphoma (9).

In contrast, the strategy for solid tumor ADCs generally had been to focus on targets that are “overexpressed” in tumors versus normal tissue to avoid toxicity associated with antibody-mediated delivery of cytotoxin to normal tissues. For example, two approved ADCs for solid tumors, trastuzumab emtansine (2) and trastuzumab deruxtecan (10) target ERBB2 (HER2), which is significantly overexpressed in a subset of breast cancer versus normal tissue.

Recent clinical studies with other targets in solid tumors have yielded two approved entities, Nectin-4-targeted enfortumab vedotin (11) and Trop-2–targeted sacituzumab govitecan (12). Nectin-4 is expressed in normal skin, and while rash and alopecia were described in the clinical study, they were not the most prominent adverse effects. Trop-2 expression has also been described in normal tissues (13), but here the most prominent clinical adverse effects were similar to irinotecan, including neutropenia and diarrhea. Thus, it appears that targeting proteins expressed in normal epithelial and tumor tissue with ADCs is possible and that we could refine our strategies both for ADC target selection and technology to expand the target space (14, 15).

NaPi2b is a transmembrane sodium phosphate transporter protein encoded by the SLC34A2 gene and is expressed in both tumor and normal tissue. Expression by protein or RNA methods has been described in nonsquamous non–small cell lung carcinoma, nonmucinous ovarian carcinoma, and papillary thyroid carcinoma; normal tissue expression has been reported in lung, bronchus, and kidney (16). In normal lung tissue, NaPi2b plays a role in phosphate transport, and mutations in SLC34A2 are associated with pulmonary and testicular microlithiasis (17).

Lifastuzumab vedotin, an ADC that targets NaPi2b and utilizes the vcMMAE ADC linker-payload, has been evaluated in phase I and II clinical trials (18, 19). Lifastuzumab vedotin achieved a 34% overall response rate in the ITT population at the recommended phase II dose in patients with heavily pretreated platinum-resistant ovarian cancer, but it was not statistically superior to liposomal doxorubicin in a randomized phase II trial. In the biomarker analysis, using the anti-NaPi2b (10H1) mouse mAb IHC assay developed for this trial, most patients showed strong NaPi2b protein expression, precluding development of an expression/efficacy relationship. However, a qRT-PCR RNA expression method chosen to increase the analytic dynamic range showed a trend between NaPi2b expression and efficacy in the phase II subgroup analysis.

We hypothesized that NaPi2b would be a target well-suited to the Dolaflexin ADC technology (20). This polymer-based platform has several advantages over conventional ADC platforms. The Dolaflexin platform yields high drug–antibody ratio (DAR) ADCs with an average of 10–15 AF-HPA payload molecules per antibody, as opposed to 3–4 in many conventional ADCs. AF-HPA is cell permeable and can diffuse to adjacent cells to cause bystander cytotoxicity. AF-HPA is also metabolized intratumorally to an active metabolite, auristatin F (AF), that is essentially cell impermeable and is not a substrate for the PgP drug efflux pump. The resulting controlled bystander effect leads to efficacy in tumors with heterogeneous antigen expression and improved systemic tolerability (21). Furthermore, higher DAR ADCs might be efficacious in tumors types with a range of antigen expression.

XMT-1536 is an ADC targeting NaPi2b and is comprised of a unique humanized anti-NaPi2b antibody conjugated to Dolaflexin (22). Consistent with the higher DAR, XMT-1536 showed greater antitumor activity than lifastuzumab-vedotin in both ovarian cancer and non–small cell lung cancer (NSCLC) primary patient-derived xenograft models (PDX). Evaluation of XMT-1536 in toxicology studies in rat and cynomolgus monkey established a tolerability profile with a notable absence of neutropenia, the dose-limiting toxicity observed for lifastuzumab vedotin.

These data support clinical evaluation of XMT-1536 for the treatment of patients with NaPi2b-expressing tumors and the general concept of exploring lineage markers as targets for ADCs.

IHC was performed at QualTek Molecular Laboratories using an automated TechMate 500 or TechMate 1000 (BioTek Solutions/Ventana Medical Systems) platform.

Four-micron–thick sections were cut, dewaxed, and rehydrated through xylene and a series of alcohols. Slides were antigen retrieved in a standard steamer using BioGenex AR-10 retrieval solution. On the TechMate platform, further retrieval was performed with Proteinase K (DAKO). Following serum blocking, the primary antibody, comprised of a human-rabbit chimera of XMT-1535 (the antibody component of XMT-1536, described below), was applied for 30 minutes at room temperature, secondary antibody for 25 minutes at room temperature, then endogenous peroxidase block and nonbiotin polymer-based detection (rabbit Polink-2 Plus detection system, GBI) was used and finally a brief hematoxylin counterstaining

Tissues were either formalin-fixed, paraffin embedded (FFPE) xenograft tumors or banked human tumors obtained from Conversant BioSciences. Slides were scanned on an Aperio slide scanning system and images saved as 20× TIFF files. Representative fields were obtained from TIFF files using Windows Snipping Tool.

The targeting antibody in XMT-1536, XMT-1535, also known as Rebmab200, is a humanized version of the MX35 murine antibody that was generated from mice immunized with a mixture of four ovarian cancer cell lines (23) and that was later shown to recognize NaPi2b, the sodium-dependent phosphate transport protein 2b (NaPi2b) encoded by the SLC34A2 gene (24). The humanization strategy and characterization of XMT-1535 (Rebmab200) has been described previously (25, 26). Antibody material used in this work was material described in ref. 23 or a subsequent batch manufactured at a contract research organization. The antibody sequence is described in patent application US 2017/0266311A1 and referred to as XMT-1535 (27).

The structure and generalized process for production of Dolaflexin ADCs is shown in Supplementary Fig. S2. For the production of XMT-1536, the antibody XMT-1535 (350 mg, 9.97 mg/mL in sodium acetate, pH 5.5 buffer) was diluted with bioconjugation buffer (50 mmol/L triethyl ammonium acetate, 1 mmol/L EDTA, pH = 7.0) to provide a 5 mg/mL solution, to which was added 1.73 mg of TCEP (345 μL, 5 mg/mL, 2.5 equivalent) dropwise, and the resulting reaction mixture was stirred for 90 minutes at 37°C. In a separate flask, 315 mg (on a dry basis) of Dolaflexin (15.37 mL, 20.5 mg/mL, 0.9/1 wt/wt) was diluted with 16.13 mL of bioconjugation buffer to provide a 10 mg/mL solution, and the pH was adjusted to 6.5 with 1 N NaHCO₃. After 90 minutes, the reduced XMT-1535 solution was added by peristaltic pump to the vigorously stirred Dolaflexin solution over approximately 15–20 minutes. The reaction mixture was stirred at room temperature for an additional 45 minutes and excess maleimide was quenched by the addition of 14.2 mg of l-cysteine (dissolved in 2 mL conjugation buffer, 50 molar equiv.). After stirring for an additional 30 minutes, the reaction mixture was adjusted to pH 5.8 with 1 mol/L acetic acid and filtered through a 0.2-μm filter. The crude ADC was purified by SCX-HPLC (Thermo Fisher Life Technologies, POROS GoPure HS 50-μm resin prepacked column 1.2 cm D x 20 cm L, mobile phase A 50 mmol/L sodium acetate pH = 5.8, Mobile phase B 50 mmol/L sodium acetate 1 mol/L NaCl, with a step gradient 3% B and 9.5% B at 4 mL/minute in 100 minutes). Fractions corresponding to the desired product were pooled and concentrated by tangential flow filtration using a 10 kDa MWCO membrane, providing XMT-1535-Dolaflexin, now referred to as XMT-1536.

The IgG-Dolaflexin, nonbinding ADC was constructed using the preceding protocol, with Rituxan (rituximab; Biogen & Genentech) as the antibody.

Lifastuzumab-vedotin was made according to the method described previously (16). DAR of these preparations was either 3.1 or 4.1.

Cyclic NaPi2b peptide was generated and used to coat 96-well high binding clear ELISA plates (Corning, 3369) at 1 μg/mL in sodium carbonate buffer, pH 9, to a total volume of 100 μL per well. After a 2-hour incubation, plates were washed four times with 100 μL per well TBS-Tween 20, blocked with 3% BSA in TBS-Tween 20 (100 μL per well) for 1 hour, and washed again. Test articles (run in duplicate in three assay runs) were added to the wells at a dose concentration range from 100 nmol/L to 0.002 nmol/L using an 8-point 3-fold serial dilution in TBS-Tween 20 buffer at 100 μL per well. Plates were incubated for 1 hour at room temperature with rocking then washed as above. Peroxidase-conjugated AffiniPure Goat Anti-Human IgG F(ab')2 fragment specific secondary antibody was used at 5,000x in TBS-Tween 20 buffer at 100 μL per well. Plates were again incubated for 1 hour with rocking followed by wash as above. TMB substrate was added at 50 μL per well and incubated until color appeared and then the reaction was quenched with 50 μL of 0.2N H₂SO₄. The optical density (OD) was measured at 450 nm with a SpectraMax M5 microplate reader (Molecular Devices). The K_d was calculated with GraphPad Prism software (GraphPad Software Inc.) by nonlinear regression using the one site-specific binding model.

OVCAR3 ovarian carcinoma cells (ATCC) were grown to approximately 90% confluency and released from the plate surface using Accutase. Cells were then washed once with ice-cold media, resuspended in ice-cold media containing 6% goat serum, seeded at a density of 50,000 cells per well and incubated on ice for 3 hours with a range of XMT-1536 or XMT-1535 concentrations (0.01–500 nmol/L, n = 5). The cells were washed three times with 100 μL ice-cold PBS, resuspended in 100 μL media with 2% goat serum and 5 μg/mL of Alexa Fluor 647-labeled goat anti-human IgG, and incubated in the dark on ice for 1 hour. The cells were washed three times with 100 μL ice-cold PBS and resuspended in 100 μL of ice-cold PBS with 1% paraformaldehyde. The median fluorescence value for each treatment was determined from 5,000 cells quantified on a MACSQuant flow cytometer (Miltenyi Biotec). The K_d was calculated with GraphPad Prism software (GraphPad Software Inc.) by nonlinear regression using the one site-specific binding model.

Cell identity and negative Mycoplasma status was confirmed by CellCheck 9 plus method at multiple passages (IDEXX BioResearch).

OVCAR3 cells were grown on chambered cover glass and incubated with ADC for 1 hour on ice. Cells were then washed with ice-cold PBS and incubated on ice with a goat anti-human IgG monovalent Fab fragment conjugated to Alexa Fluor -647 for 1 hour. After washing, the cells were transferred to a 37°C incubator to allow internalization for 0, 1, 3, 6, and 24 hours. The cells were then placed on ice, washed with ice-cold PBS, and incubated for 5 minutes with Cellmask Orange to visualize the plasma membranes and Lysosensor Green DND-189 to show colocalization with the lysosomes. After washing, the cells were imaged on a Zeiss Axio Vert.A1 FL inverted microscope.

OVCAR3 cells were grown in RPMI1640 media supplemented with 20% FBS and 1% penicillin/streptomycin, seeded at a density of 5,000 cells per well in 100 μL of growth media in a 96-well, white flat-bottom plate. Following overnight incubation, the media was replaced with 100 μL of fresh media containing the test compounds at a 3-fold titration up to 33 nmol/L. The treated cells were incubated for 96 hours at 37°C in the presence of 5% CO₂. Cell viability was determined using The CellTiter-Glo Luminescent Cell Viability Assay and luminescence was quantified with SpectraMax M5 microplate reader (Molecular Devices). Cell viability was normalized to untreated control and expressed as a percentage. IC₅₀ values were calculated with GraphPad Prism software (GraphPad Software Inc.) using 4-parameter, variable slope, dose–response curve fitting algorithm. Cell line studies were performed as n = 3 with each treatment.

OVCAR3 cells were treated with XMT-1536 ranging from 0 to 1 nmol/L and in some cases 30 nmol/L Bafilomycin A for 24 hours. The amount of nuclear DNA in each cell was quantified in 5,000 cells for each treatment by a MACSQuant flow cytometer (Miltenyi Biotec) using propidium iodide. The percentage of arrested cells was determined by dividing the number of cells in G₂–M phase by the total number of cells and multiplying by 100.

All procedures in animals were performed in compliance with the Animal Welfare Act, the Guide for the Care and Use of Laboratory Animals, and the Office of Laboratory Animal Welfare. Protocols were reviewed and approved by the Institutional Animal Care and Use Committees of the relevant facility.

Xenografts were initiated with OVCAR3 human ovarian carcinoma fragment (∼1 mm³) implanted subcutaneously in the right flank of 8-week-old female CB17-severe combined immunodeficiency (SCID) mice (Charles River Discovery Services North Carolina). Animals were randomized into treatment groups (n = 10) when the tumor target volume reached 100–150 mm³. Test articles were administered intravenously via tail vein injection. Mice received a single dose of either saline vehicle; XMT-1535 at 3 mg/kg; XMT-1536 (DAR 12.4) at 3 mg/kg; IgG1-Dolaflexin (DAR 18.1) at 3 mg/kg, or lifastuzumab vedotin (DAR 4.1) at 3 mg/kg. Tumors were measured twice per week. The study endpoint was tumor volume (TV) of 1000 mm³ or day 57, whichever came first. A partial response (PR) was defined as a tumor volume that was 50% or less of its day 1 volume for three consecutive measurements throughout the study, and equal to or greater than 13.5 mm³ for one or more of these three measurements. In a complete response (CR), the tumor volume was less than 13.5 mm³ for three consecutive measurements during the study. An animal with a CR at the termination of a study was additionally classified as a tumor-free survivor (TFS). The log rank (Mantel–Cox) and Gehan–Breslow–Wilcoxon tests were employed to assess the significance of the difference between the overall survival experiences of two groups. For statistical analyses, two-tailed tests were conducted at significance level P = 0.05.

Nu/Nu 7-week-old female mice were implanted with human NSCLC models CTG-0178 (moderately differentiated lung adenocarcinoma), CTG-0852 (poorly differentiated lung adenocarcinoma), and CTG-0860 (poorly differentiated lung adenocarcinoma), expressing NaPi2b mRNA (Champions Oncology). After the tumors reached 1–1.5 cm³, they were harvested and the tumor fragments of approximately 5 × 5 × 5 mm³ were implanted subcutaneously in study mice. Animals were randomized among the treatment groups (n = 8) when the tumor volumes reached approximately 100–300 mm³, and treated with saline vehicle, lifastuzumab vedotin (DAR 3.1) 3 mg/kg qwk × 3, nonspecific IgG1-Dolaflexin (DAR 16.2) 3 mg/kg qwk × 3, or XMT-1536 (DAR 13.1) 3 mg/kg qwk x 3. The study endpoint was tumor volume of 1,500 mm³. A PR was defined as a tumor volume ≤ 30% of tumor volume at day 0 for two consecutive measurements. A CR was defined as a mouse with no palpable tumor for two consecutive measurements over a 7-day period. A TFS was a CR that persisted until study completion. Statistical analyses were performed using one-way ANOVA followed by Dunnett multiple comparison tests.

Immunocompromised female mice (either Athymic Nude or CB-17 SCID, 6–12-week-old) were implanted with human primary xenograft tumors representing human ovarian or fallopian cancer (South Texas Accelerated Research). These models were selected to represent a diversity of treatment exposure (12 treatment-exposed and 8 treatment-naïve), but NaPi2b expression was unknown at the time of model selection. Mice were randomized at a mean tumor volume of 125–250 mm³. Three animals/model were dosed with XMT-1536 (DAR 12.0) at 3 mg/kg qwk × 3. The study endpoint was when tumors were 1 cm³ or at 45 days. Tumor from an untreated animal from each model was collected at the planned endpoint. Results were reported as %T/C and are shown as median best response, calculated as the median of the greatest difference from baseline for each tumor within each model at each examined timepoint.

XMT-1536 tissue distribution was evaluated in SCID mice bearing OVCAR3 subcutaneous xenografts. Animals received a single dose of XMT-1536 (DAR 12.0) at 3.0 mg/kg. Plasma and tissues, including tumor, lung, liver, kidney, skeletal muscle and spleen, from n = 4/timepoint at 5 minutes, 24 hours, 72 hours, 96 hours, 7 days, and 14 days, were harvested after mice were perfused with saline. Total AF-HPA, free AF-HPA, and free AF were analyzed in plasma and tissue samples via validated LC/MS-MS methods. For clarity, free AF-HPA and free AF refer to nonconjugated AF-HPA and AF. Total AF-HPA refers to total AF-HPA after hydrolysis to release all conjugated AF-HPA in the sample. Conjugated AF-HPA can be calculated by subtracting free AF-HPA from total AF-HPA.

Toxicology studies were performed under Good Laboratory Practice (GLP) for Nonclinical Laboratory Studies. XMT-1536 (DAR 11.6) was administered intravenously to 9-week-old Sprague-Dawley rats and 2.4 -3.5-year old cynomolgus monkeys on days 1 and 22, with terminal necropsy at day 29/30 and recovery necropsy at day 50. Doses of 1.5 mg/kg, 3 mg/kg, and 6 mg/kg were tested in the rat, in groups of 16 male and 16 female animals (10 main study and 6 recovery). Doses of 0.75 mg/kg, 1.5 mg/kg, and 3 mg/kg were tested in the monkey, in groups of 6 male and 6 female animals (4 main and 2 recovery studies). Both studies included in-life observations, clinical pathology, gross pathology, and histopathology on a full panel of tissues, and toxicokinetics. Four different analytes, including total XMT-1535, total AF-HPA, free AF-HPA, and free AF were analyzed in the animal plasma samples through validated LC/MS-MS methods. Exposure and pharmacokinetic (PK) parameters were assessed for each analyte. PK comparisons were also made to the prior mouse plasma PK in the tissue distribution study.

XMT-1536 was applied to cryosections of normal human, monkey, and rat tissues (n = 3 donors per tissue for human, n = 2 donors per tissue for monkey and rat), at two concentrations (5 and 0.25 μg/mL). XMT-1536 was detected with a secondary antibody raised in mouse against the toxin–linker conjugate and the DAKO EnVision+kit.

SLC34A2 expression was demonstrated at the protein level using an IHC assay for NaPi2b. NaPi2b IHC of a representative ovarian tumor confirmed detectable membrane expression of NaPi2b (Fig. 1A, and enlarged inset). In normal fallopian tube, NaPi2b expression was seen on the epithelial apical surface (Fig. 1B). A representative image of a lung adenocarcinoma confirmed membrane expression of NaPi2b in tumor cells (Fig. 1C, open arrow). In the enlarged inset, both cytoplasmic and membranous immunoreactivity are noted. The same tissue sample also demonstrated expression of NaPi2b protein on the apical surface of normal bronchial epithelium (Fig. 1C, solid arrow). NaPi2b expression was also detected in type 2 pneumocytes in lung parenchyma (Fig. 1D). The comparable expression of SLC34A2/NaPi2b in tumor and the matched normal tissue is consistent with expression being a lineage marker of the cell-of-origin, as opposed to part of the oncogenic process in these tumors.

Review of RNA-sequencing data from publicly available data sources showed levels of SLC34A2 RNA expression to be relatively higher in papillary thyroid carcinoma, lung adenocarcinoma, ovarian carcinoma and cholangiocarcinoma, in contrast to lung squamous cell carcinoma, where expression was relatively low (Supplementary Fig. S1). RNA expression also appeared relatively increased in uterine (endometrial) cancer and cholangiocarcinoma.

XMT-1536 and the unconjugated antibody XMT-1535 bound to a peptide corresponding to the epitope of human NaPi2b with equivalent affinity (K_d 0.82 ± 0.28 nmol/L and 0.73 ± 0.34 nmol/L for XMT-1536 and XMT-1535 antibody, respectively). XMT-1536 showed similar affinity for the NaPi2b epitope from human, cynomolgus monkey, rat, and mouse (Fig. 2A). Both XMT-1536 and the unconjugated XMT-1535 antibody bound to the NaPi2b-expressing OVCAR3 ovarian cancer cell line (∼66,000 NaPi2b antigens per cell) with nanomolar affinity (K_d 4.14 ± 3.08 nmol/L and 4.05 ± 2.62 nmol/L for XMT-1536 and unconjugated antibody, respectively; Fig. 2B). After binding, XMT-1536 internalized and trafficked to the lysosomal compartment within 3 hours (Fig. 2C).

In the OVCAR3 cell line, XMT-1536 was cytotoxic in a 96-hour cellular cytotoxicity assay with subnanomolar potency (EC₅₀ 0.52 nmol/L expressed in AF-HPA payload equivalents), with higher potency compared with treatment with unconjugated payload (EC₅₀ 0.97 nmol/L, Fig. 2D). In contrast, a nonbinding control Dolaflexin ADC with similar DAR to XMT-1536 (IgG1-Dolaflexin) was 30-fold less potent than XMT-1536 in cytotoxicity assays (EC₅₀ 15.9 nmol/L).

Consistent with the antimitotic mechanism of action of the AF-HPA payload, treatment of OVCAR3 cells with XMT-1536 led to accumulation of cells in the G₂–M phase of the cell cycle. Cotreatment of OVCAR3 cells with Bafilomycin A (Baf.A) to block lysosomal acidification completely abrogated accumulation of cells in G₂–M (Table 1), indicating that internalization and lysosomal trafficking is necessary for the pharmacologic effect of XMT-1536.

XMT-1536 induced marked tumor regression of established tumors in the OVCAR3 ovarian carcinoma model when administered at a single 3 mg/kg dose (Fig. 3A). At this dose, there were 3 CR and 4 PR and T-C (difference in median time to endpoint for treated vs. control) was 30.9 days. XMT-1536 was well-tolerated with transient body weight loss of -6.6% at nadir (compared with −1.2% for vehicle control). Lifastuzumab-vedotin dosed once at 3 mg/kg resulted in no PRs or CRs and a T-C of 6.1 days. The nonbinding ADC IgG1-Dolaflexin and the unconjugated antibody XMT-1535 were inactive with 0/10 regressions and no significant tumor growth delay.

To confirm the results with OVCAR3 and extend the results to lung cancer, XMT-1536 was tested in a series of lung adenocarcinoma PDX models.

In model CTG-0178, XMT-1536 treatment did not result in tumor regressions when given at 3 mg/kg qwk ×3, but did cause significant delay of tumor growth, with a TGI of 67% (P≤0.01 vs. vehicle) at end of study, day 35. Treatment with the comparator, lifastuzumab-vedotin, or a nonbinding IgG1-Dolaflexin control ADC failed to cause tumor regression and did not lead to any significant growth delay (Fig. 3B).

In model CTG-0852, XMT-1536 achieved regressions at the 3 mg/kg qwk × 3 schedule, with 5 mice achieving PR, 1 CR, and 2 mice tumor free at the day 62 end of study. The nonbinding IgG1-Dolaflexin ADC did not induce tumor regression at the 3 mg/kg qwk × 3 schedule, but was associated with some tumor growth delay. Lifastuzumab-vedotin at 3 mg/kg was partially active, with 3/8 PR (Fig. 3C).

In model CTG -0860, XMT-1536 achieved PR in 2/7 mice at the 3 mg/kg qwk × 3 schedule, at day 60 end of study, with no responses seen in either the comparator or control group (Fig. 3D).

IHC was performed on representative vehicle blocks from each model and demonstrated the presence of NaPi2b target in each model.

Twenty patient-derived ovarian/fallopian tube adenocarcinoma xenografts were evaluated for response to XMT-1536. These models were chosen without knowledge of NaPi2b expression, to mimic an unselected patient population. XMT-1536 induced at least 50% reduction in tumor volume relative to baseline in 10/20 (50%) models when administered at a dose of 3 mg/kg once weekly for 3 weeks (Fig. 4A). Responses were seen in both treatment-naïve and treatment-exposed tumor models (5/8 and 5/12, respectively).

The NaPi2b IHC assay was applied to a control block for each model and scored using an H-score algorithm. Tumors representative of low, moderate and high NaPi2b staining are shown in Fig. 4B. Amongst tumors with H-score ≥70, 10/13 models (76%) achieved 50% or greater reduction in tumor volume after XMT-1536 treatment, versus 0/7 models (0%) with H-score <70 (Fig. 4C).

Twenty human ovarian tumors were stained and scored with the NaPi2b IHC assay (Fig. 4D; Supplementary Table S1). The distribution of H-scores was comparable in these human ovarian tumors and the tested patient-derived xenograft models. Twelve of 20 (60%) human ovarian tumors evaluated for NaPi2b expression had an H-score ≥70. The human tumors showed a trend toward a bimodal distribution, with most tumors having an H-score of 100 or above or 50 or below, suggesting that the H-score threshold of 70 identified in the xenograft panel may represent discrimination of two different tumor populations based upon NaPi2b expression and XMT-1536 sensitivity.

The same IHC assay that was applied to staining human ovarian tumors was used on a panel of 20 human lung adenocarcinomas. Sixteen of 20 (80%) of tumors showed strong staining for NaPi2b (Supplementary Fig. S3).

A tumor-bearing mouse biodistribution study demonstrated the specific targeting of XMT-1536 to NaPi2b expressing tumors. Maximum concentration of total AF-HPA in tumor tissue was observed at 96 hours after XMT-1536 administration, consistent with delayed penetration and target-dependent accumulation of the ADC in the tumor tissue. Exposure to total AF-HPA, free AF-HPA, and free AF in other tissues (i.e., lung, liver, kidney, skeletal muscle, and spleen) was at least an order of magnitude less than that seen in tumor tissue. All analytes continued to be detectable in tumor tissue at 14 days (Fig. 5; Supplementary Fig. S4). NaPi2b target expression is conserved in mouse and mouse lung target tissue expression is shown in Supplementary Fig. S5.

XMT-1536 was evaluated in repeat dose toxicity studies in rat and cynomolgus monkey. The findings allowed us to establish the highest nonseverely toxic dose, used in determining the initial dose level in a phase I clinical study. Toxicokinetics were evaluated in rat and monkey. Total antibody, total AF-HPA, free AF-HPA, and free AF, were evaluated to characterize PK of XMT-1536 in the plasma. Exposure to each analyte was approximately dose-proportional in both species (Fig. 6A). Furthermore, there was no gender impact on PK in either species. In rats, the 3 mg/kg dose resulted in total AF-HPA payload AUC_0-∞ of 251,000 hour/ng/mL in females and 278,000 hour/ng/mL in males. The exposure for total AF-HPA in monkeys after a 3 mg/kg dose was similar, with AUC_0-∞ of 299,000 hour/ng/mL in females and 321,000 hour/ng/mL in males. In both species, the systemic maximum concentration of free AF-HPA and free AF after the 3 mg/kg XMT-1536 dose was less than 2 ng/mL with a T_max generally later for AF compared with AF-HPA (>72 vs. <24 hours, respectively; Fig. 6A). The exposures for rat, monkey, and tumor-bearing mouse are shown in Fig. 6B.

XMT-1536 used as a primary IHC reagent produced membrane and cytoplasmic staining in a variety of cell types. Human tissues with immunoreactivity included, but were not limited to, fallopian tube, kidney, bile duct epithelium, pulmonary pneumocytes and bronchial epithelium, ovarian surface epithelium; pancreatic ductal epithelium; mesothelium; thyroid follicles; endocervix; and endometrium. Overall, there was concordance among the human, monkey, and rat tissue staining. One subtle difference in staining was the lower level of expression on the bronchial epithelium in rats compared with monkeys and humans although overall there was expression of NaPi2b in lung tissue for rats, monkeys, and humans. In the eye there was plasma membrane staining in human but not in rat or monkey. In the monkey, occasional weak immunoreactivity was seen in the eye epithelial cytoplasm.

ADCs have shown promise in the treatment of cancer, but the identification of appropriate targets (especially for solid tumor ADCs) and the limitations of first-generation linker–payload platforms have remained challenges to achieving broader clinical success.

We, and others, (16) have demonstrated NaPi2b expression in NSCLC adenocarcinoma and epithelial ovarian cancer and in the presumed normal tissues of origin, type II pneumocytes and fallopian tube epithelium, respectively (28, 29). Our study uses a novel IHC reagent and the data is generally comparable to protein and RNA data expression data described by others. However, our IHC method was calibrated with outcome in a large series of primary xenograft models and this may allow us to predict levels of NaPi2b expression that will correspond to responsiveness to XMT-1536 in human tumors. XMT-1536 targets NaPi2b using the Dolaflexin platform. Consistent with the known mechanism of action of the auristatin class (30), XMT-1536 induced cell-cycle arrest in G₂–M that was dependent upon antigen-mediated ADC internalization and lysosomal processing. XMT-1536 was capable of inducing regressions in multiple PDX models of NSCLC adenocarcinoma and epithelial ovarian cancer, including durable complete regressions in several models.

Biodistribution studies in mice showed no antigen-dependent ADC accumulation in normal lung, a tissue with target expression, and substantially lower exposure to AF-HPA and AF in the lung compared with tumor. This difference between tumor and lung could be due to target accessibility, rate of antibody internalization, or other factors.

The pharmacokinetics of XMT-1536 were assessed in mice, rats, and monkeys after intravenous administration. PK analyses showed approximately dose proportional exposure of XMT-1536 in both rat and monkey. Systemic free AF-HPA and AF concentrations were several orders of magnitude lower than total AF-HPA in all animal species. In addition, specific tumor targeting of auristatin payload by XMT-1536 was illustrated in the mouse tissue distribution study. Overall, the PK characteristics of XMT-1536 are consistent with those of other ADCs in the clinic, despite the much higher DAR value of 10–15.

We postulated that a feature of a lineage-associated antigen target would be broad expression in tumors arising from a given cell lineage. We detected NaPi2b expression in many NSCLC adenocarcinoma and epithelial ovarian cancer cases using IHC, with an estimated 60% of ovarian cancer cases having NaPi2b-expression at a level associated with regressions in the PDX models. Notably, in both the NSCLC adenocarcinoma and ovarian cancer tumor panels evaluated here, there was an apparent bimodal distribution of NaPi2b H-scores, with a minority of tumors showing very low expression. These data suggest that NaPi2b IHC might be used to enrich for patient populations that respond to XMT-1536 therapy.

Despite recent advances in the treatment of non–small cell lung cancer and ovarian cancer, there remains a significant need for new therapies for patients with heavily pretreated disease, immunologically cold tumors (in NSCLC), or tumors with intact DNA repair mechanisms. On the basis of the data presented here, clinical investigation of XMT-1536 in these diseases and other NaPi2b-expressing tumors is warranted. XMT-1536 currently is being explored in early clinical studies of ovarian and lung adenocarcinoma. Clinical activity has been seen at doses of 20 mg/m² q4w and higher. At the MTD of 43 mg/m² q4w, severe toxicities such as neutropenia and neuropathy seen in other ADC studies have not been observed following treatment with XMT-1536 (31). Finally, the data presented here suggest that other solid tumor lineage antigens might be suitable as therapeutic targets for next-generation ADCs.

N.D. Bodyak reported other from Mersana Therapeutics outside the submitted work. R. Mosher reported personal fees and other from Mersana Therapeutics during the conduct of the study; in addition, R. Mosher had a patent for US 2019/0160181 pending. C. Bu reported other from Mersana outside the submitted work. D.R. Demady reported other from Mersana Therapeutics during the conduct of the study. M.J. DeVit reported personal fees from Mersana Therapeutics outside the submitted work. D.R. Gumerov reported personal fees from Mersana Therapeutics during the conduct of the study. W. Lee reported other from Mersana Therapeutics outside the submitted work. P.U. Park reported personal fees from Mersana Therapeutics during the conduct of the study; personal fees from Mersana Therapeutics outside the submitted work; in addition, P.U. Park had a patent for Mersana Therapeutics pending and issued. D.A. Bergstrom reported a patent for US2017/0266311 pending to Mersana Therapeutics and a patent for US2019/0160181 pending to Mersana Therapeutics. No disclosures were reported by the other authors.

N.D. Bodyak: Conceptualization, supervision, investigation, methodology, writing–original draft. R. Mosher: Formal analysis, supervision, investigation, methodology, writing–original draft, writing–review and editing. A.V. Yurkovetskiy: Conceptualization, formal analysis, supervision, investigation, methodology, writing–original draft. M. Yin: Investigation, methodology. C. Bu: Investigation. P.R. Conlon: Investigation. D.R. Demady: Supervision, investigation. M.J. DeVit: Investigation. D.R. Gumerov: Investigation, methodology. D. Xiao: Investigation. V.R. Gurijala: Investigation. W. Lee: Investigation. D. McGillicuddy: Investigation. P.U. Park: Conceptualization, supervision, investigation. L.L. Poling: Investigation. M. Protopopova: Investigation. L. Qin: Investigation. C.A. Stevenson: Investigation. E. Ter-Ovanesyan: Investigation. A. Uttard: Investigation. J. Xu: Investigation. L. Xu: Investigation, methodology. D.A. Bergstrom: Conceptualization, formal analysis, supervision, writing–original draft. T.B. Lowinger: Conceptualization, formal analysis, supervision, validation, investigation, methodology, writing–original draft, writing–review and editing.

The authors wish to thank Charles River Laboratories, Champions Oncology, and START for their contributions to the in vivo work; QualTek Molecular Laboratories for their performance and interpretation of IHC assays; Frontage for sample analysis; Drs. Marc Damelin, Dorin Toader, and Radha Iyengar for contributions in preparation of this manuscript; and Jeffrey Zurita and Scott Collins for preparation of figures.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.