A MET Targeting Antibody–Drug Conjugate Overcomes Gemcitabine Resistance in Pancreatic Cancer

Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive cancers with an overall 5-year survival rate of approximately 10%, and an incidence rate that continues to rise (1). PDAC lethality can be explained by its typical late stage at diagnosis, and by its resistance to chemotherapy. Most patients with PDAC are treated with regimens such as gemcitabine and nab-paclitaxel, FOLFIRINOX, or gemcitabine alone, depending on their health status. Unfortunately, while these regimens can improve survival, patients with inoperable disease cannot be cured and often rapidly develop treatment resistance and succumb to their cancers in less than 1 year. It is believed that the poor vascularization and desmoplastic stroma inherent to pancreatic cancer impair drug delivery and contribute to the poor responsiveness to standard cytotoxic therapies (2, 3). The dismal outcomes for patients with PDAC reflect an urgent need to develop more effective treatment approaches.

Antibody–drug conjugates (ADC) are an emerging class of molecules designed to specifically deliver cytotoxic agents to tumor cells. Typically, a highly cytotoxic agent, also called the “payload,” is covalently linked to a mAb specifically targeting a plasma membrane protein overexpressed at the surface of tumor epithelial cells (4). Upon binding to the protein, the ADC is internalized, and the cytotoxic agent is released following proteolytic cleavage of the linker in the lysosome. The therapeutic window of an ADC strategy is provided by the cell surface expression of the target protein, which must be high in tumor cells, while relatively low in normal tissue. Recently, ADC strategies for cancer therapy have regained interest because of the development of new linkers, potent cytotoxic payloads, and highly selective mAbs (4).

MET, also called HGFR, is a plasma membrane receptor tyrosine kinase overexpressed in several solid tumors (5), including in a subset of PDACs (6, 7). In PDAC, MET promotes cell proliferation and invasion, epithelial-to-mesenchymal transition, stem cell phenotype, and resistance to chemotherapy (8–12). Numerous inhibitors targeting MET or its downstream signaling pathway components have been tested over the last years with success limited to small subsets of patients with gene amplification (13, 14). To target all MET-overexpressing tumors, new targeting therapies such as ADCs are currently in development.

In this study, we report the quantification of MET expression and subcellular localization in human PDAC cell lines, tumors, and patient-derived xenografts (PDX). We then evaluated the efficacy of TR1801-ADC both in vitro and in vivo. TR1801-ADC is a newly developed and highly optimized humanized anti-MET antibody with limited agonist activity, conjugated to the pyrrolobenzodiazepine (PBD) dimer payload-linker, tesirine (SG3249), a highly efficient DNA cross-linking agent (15). Our work suggests that in MET-expressing tumors, a MET-targeted ADC can suppress PDAC growth and can synergize with chemotherapy to treat chemoresistant PDAC.

The human pancreatic cancer cell lines, BxPc3 and MiaPaca-2, were purchased from the ATCC, while the FG cell line was generously provided by David Cheresh (University of California, San Diego). All cell lines were cultured in DMEM or RPMI supplemented with 10% FBS at 37°C and 5% CO₂ and kept under 20 passages. Cell lines were tested for the presence of Mycoplasma DNA using the following primers (forward: GGCGAATGGGTGAGTAACACG and reverse: GGATAACGCTTGCGACCTAT). The 1334 cell line is an epithelial cell line derived from the human pancreatic cancer xenograft number 1334 (16).

Cell viability was determined by measuring the luminescence after adding the CellTiter-Glo 2.0 Reagent (Promega). Cancer cells were seeded overnight in growth media and incubated at 37°C, 5% CO₂, and 95% humidity. ADCs or the PBD warhead, SG3199, was added in serial dilutions starting with concentrations of 100 nmol/L for ADCs and 10 nmol/L for free drug. Cells were exposed to test articles for 5 days. IC₅₀ values were calculated by nonlinear regression using sigmoidal curve fitting in PRISM 7 (GraphPad).

For the evaluation of drug interaction, cells were exposed to test articles for 3 days prior to measuring cell viability as described above. Data were processed using Combenefit software under the classical HSA synergy model (17).

For in vitro purposes, gemcitabine (TSZCHEM) was prepared fresh by solubilizing into the appropriate tissue culture medium. For in vivo purposes, gemcitabine was solubilized into 0.9% NaCl solution at 20 mg/mL. TR1801-ADC and secukinumab-ADC were obtained from Tanabe Research Laboratories Inc., and prepared to a final concentration of 1 mg/mL, as described previously (15).

Human tumors were collected and utilized in accordance with institutional review board–approved and Institutional Animal Care and Use Committee (IACUC)-approved protocols at the University of California, San Diego (UCSD, La Jolla, CA). Resected tumor tissue was implanted fresh into the pancreas of NOD SCID gamma (NSG) mice, as described previously (18, 19).

Six-week-old male NSG mice were obtained from The Jackson Laboratory (stock #005557) and orthotopically grafted with human PDX tumor. Briefly, 2 × 2 mm pieces of tumor were sutured in between the tail and the head of the pancreas as described previously (20). Mice were randomized when tumor volumes exceeded 65 mm³. Secukinumab-ADC was administered intravenously at 1 mg/kg and TR1801-ADC was administered intravenously at either 0.5 or 1 mg/kg. Both ADCs were freshly diluted at 20 mg/mL in saline solution prior to injection. Secukinumab-ADC, used here as a negative control, is an ADC targeting IL17A, secreted by T cells, mast cells, and neutrophils and not expressed by pancreatic cancer cells. Secukinumab has been approved by the FDA for the treatment of psoriasis, psoriatic arthritis, and ankylosing spondylitis (21).

Gemcitabine treatment was used as a standard of care and provided intraperitoneally at 50 mg/kg on a weekly basis. The corresponding volume of saline solution was provided to the control group. Mice were euthanized when tumor volumes exceeded 2,000 mm³ or if any sign of ulceration or distress was observed according to our IACUC-approved protocol (protocol S09158).

Tumor growth was evaluated at regular intervals by ultrasound imaging, typically on a weekly basis. Tumor volumes were calculated using the formula V = 4/3 × π × tumor radius³. Tumor volumes relative to dosing time ± SEM were plotted using GraphPad Prism.

Tumor growth inhibition (TGI) index was defined as [1 – (mean relative tumor volume of treated tumors)/(mean relative tumor volume of control tumors)] × 100%. Average TGI ± SEM was plotted for all treatment groups at different timepoints. A complete response was defined as a tumor volume <25 mm³ for 2 weeks in a row. Statistical evaluation of tumor growth was conducted pairwise against the corresponding control group with P values obtained from type II ANOVA using the TumGrowth software (22). Statistical evaluation of dynamic TGI index differences between groups was conducted on GraphPad Prism using repeated measures ANOVA, with Bonferroni multiple comparison test for comparisons between two groups. Overall survival times were estimated by Kaplan–Meier methods. Mice alive at the end of the follow-up period were censored. Statistical evaluation of mouse survival was performed using the log-rank (Mantel–Cox) test on GraphPad PRISM.

For all experiments, differences were considered significant for P < 0.05 and reported in the figures as follows: *, 0.01 ≤ P < 0.05; **, 0.001 ≤ P < 0.01; ***, P < 0.001.

Staining on PDX tumors was performed on formalin-fixed, paraffin-embedded tumor samples. Tissue sections were fixed overnight in 10% NBF and transferred to 70% ethanol. Paraffin embedding and cutting of tissue sections were performed by the UCSD Histology Resource Center (La Jolla, CA). Slides were baked overnight at 60°C and were deparaffinized with two rinses in xylene, and hydrated using an alcohol gradient. Antigen retrieval was conducted using 1 × citrate buffer, pH 6, Antigen Retrieval Solution pH low (eBioscience) at 95°C–100°C for 20 minutes. After cooling, blocking was realized using Bloxall (Vector Laboratories) for 10 minutes and then incubated in blocking buffer (10% donkey serum + 5% BSA in 1 × TBS + 0.1% Tween20) for 1 hour at room temperature. Slides were incubated overnight at 4°C with primary antibody CONFIRM anti-total MET SP44 (1:200, Roche) or rabbit isotype control (1:200, Millipore) in blocking buffer. Sides were then washed three times in TBS + 0.1% Tween20 and incubated with anti-rabbit horseradish peroxidase (HRP)-conjugated secondary antibody for 30 minutes at room temperature. Slides were then counterstained using hematoxylin and bluing reagents. Imaging was done on a Nikon Eclipse E600 Upright Fluorescent Microscope from UCSD Microscopy Core (La Jolla, CA). For IHC on tissue microarrays (US Biomax), slides were treated at 100°C in EDTA buffer, pH 9, for 20 minutes for antigen unmasking. Primary rabbit mAb, MET SP44 (1:200, Abcam, #ab227637), or rabbit IgG isotype control (1:200, Abcam, #ab27478) was incubated with tissue sections for 60 minutes at room temperature. Goat anti-rabbit IgG-HRP conjugate was used as a detection antibody (Leica Biosystems, #DS9800) at 25 μg/mL for 60 minutes at room temperature. Slides were scanned with the Nano-Zoomer Image System (Hamamatsu), and IHC staining intensity (H-score) was calculated according to the following formula: total score = (% at 0) × 0 + (% at 1) × 1 + (% at 2) × 2 + (% at 3) × 3, with 0, no staining; 1, weak staining; 2, medium staining; and 3, strong staining.

For immunofluorescence, cells were grown subconfluent on 4-well Chamber Slides (Nunc Lab-Tek, Thermo Fisher Scientific). Cells were then fixed on ice with 4% paraformaldehyde for 15 minutes, then washed in cold PBS three times, and permeabilized with PBS Triton X100 0.5% for 5 minutes on ice. Cells were washed in PBS glycine 0.1 mol/L for 15 minutes. After 1-hour blocking in PBS BSA 1.5%, cells were incubated overnight with anti-MET primary antibody (#8198) at 4°C. Cells were then washed three times with PBS Tween20 0.1% and incubated for 2 hours with secondary antibody conjugated to Alexa Fluor 594. Finally, slides were washed and mounted with ProLong Medium (Themo Fisher Scientific).

Precise quantification of plasma membrane MET receptors was performed as described previously (23). Briefly, pancreatic cancer cells were stained with an anti-MET conjugated with Alexa Fluor 488 and then analyzed on a BD Accuri Flow Cytometer (BD Biosciences). Fluorescence values for each cell were compared with a standard established using the Quantum Simply Cellular Microbeads Kit (Bangs Laboratories) to estimate the number of MET receptors detected.

Initially we sought to characterize the expression of MET in PDAC cells to define it as a potential ADC target. According to The Cancer Genome Atlas, PDAC is among the tumor types expressing the highest level of MET (Supplementary Fig. S1). We also observed that MET is overexpressed in PDAC versus normal pancreatic tissue in Affymetrix datasets publicly available from the GENT2 platform (ref. 24; Fig. 1A). These findings confirm previous reports of MET overexpression in PDAC (6, 7, 25, 26). In addition, MET protein expression was evaluated by IHC on a tissue microarray containing 91 cores of human pancreatic tumors and five normal pancreatic tissues. Staining revealed considerable variability of expression in tumor cells, while stromal cells remained unstained (Fig. 1B). H-scores attributed to MET staining showed that 59% of PDACs expressed MET, 9% with a very strong expression level (H-score > 150; Fig. 1C). MET staining was not found to be associated with tumor stage or tumor subtype.

To select PDAC cell lines suitable for in vitro evaluation, a panel of cell lines was analyzed for MET expression. Western blot evaluation of MET protein expression revealed a variable level of expression, with the MIA PaCa-2 (MP2) cell line negative and high expressors including the Kras-mutant PDX-derived cell lines, 1334 and BxPC3 (Fig. 1D).

Efficient drug delivery by an ADC requires the targeted protein to be localized at the plasma membrane of tumor cells. Because activated receptor tyrosine kinase can be partially retained within the cytoplasm without full plasma membrane expression, we conducted plasma membrane protein biotinylation followed by MET immunoprecipitation (27–29). Streptavidin immunoblotting revealed that a fraction of MET proteins was biotinylated and, therefore, located at the plasma membrane in both FG and BxPC3 cells (Fig. 1E). This subcellular localization was validated by immunofluorescence on BxPC3 and 1334 cells (Fig. 1F). The number of MET receptors at the plasma membrane was precisely evaluated by quantitative flow cytometry. We observed an average of 101,521 MET receptors per cell at the surface of BxPC3, while a significantly lower amount was found in FG cells with an average of 23,295 receptors per cell. Overall, these results confirm that MET is highly expressed at the plasma membrane of a significant fraction of human PDAC cells and, therefore, can be considered as a potential target for an ADC strategy.

To evaluate the efficacy of TR1801-ADC in vitro, we selected three human pancreatic cancer cell lines based on their level of MET expression and treated them for a period of 5 days at various concentrations (Fig. 2A). While the warhead, SG3199, alone efficiently induced cytotoxicity, even at low doses (IC₅₀ value, 3–27 pmol/L), a nontargeted negative control (secukinumab-ADC) led to cytotoxicity only at high doses (IC₅₀ value, 7–16 nmol/L). This result indicates that despite the high potency of the warhead, it does not induce a cytotoxic effect when combined with a nontargeting antibody. While the MET-negative cell line, MP2, was not affected by TR1801-ADC, BxPC3 and 1334 were sensitive with an IC₅₀ value of 60.7 and 8.8 pmol/L, respectively. Similar effects on cell proliferation were observed using the XCELLigence technology, as BxPC3 and 1334 cell proliferation was inhibited in a dose-dependent manner, while MP2 cells remained unaffected (Supplementary Fig. S2). These results demonstrate that the anti-MET TR1801-ADC shows specific cytotoxic effects on PDAC cells expressing MET at the plasma membrane.

To evaluate the efficacy of TR1801-ADC in vivo, PDX samples were assessed for MET expression to select those tumors expressing MET at the plasma membrane. A considerable variability in the overall protein expression level was observed across the PDX tumors tested by Western blotting (Fig. 2B). This reflects the variability observed in cell lines and human primary tumors, indicating that our PDX cohort represented the diversity observed in PDAC generally. We further confirmed results by IHC on 22 paraffin-embedded PDX tumors. While stromal cells were always negative, we found that the vast majority of epithelial tumor cells expressed MET, but with variable levels of immunoreactivity (Fig. 2C). Tumors generally were homogenous with respect to the presence of expression, but more variable with respect to the intensity, with some areas being strongly positive, while others being less immunoreactive. We also observed that two PDX tumors were found to be mostly negative with clusters of positive cells (score of 1). As in cell lines, plasma membrane staining was observed. On the basis of the signal intensity, a score from 0 to 4 was assigned to each tumor. Medium (3) to high (4) intensity was observed for half of the cohort of PDXs tested (Fig. 2D). Plasma membrane MET was also evaluated by quantitative flow cytometry on some PDX tumors. Interestingly, we observed that the average number of plasma membrane MET receptors per cell was highly variable and did not correlate with the IHC score. One third of tumors tested did not have any MET receptors at the plasma membrane, even though some of these tumors showed clear reactivity to MET immunostaining. Of MET-expressing tumors, the average number of receptors at the plasma membrane ranged from 800 to 36,000 per cell (Fig. 2D). On the basis of the content of MET at the plasma membrane, two PDX samples, PDX1334 and PDX1342, were selected for in vivo evaluation of TR1801-ADC efficacy.

PDX1334 was selected because it scored high for MET expression on both IHC and quantitative flow cytometry. Tumor pieces were grafted into the pancreata of NSG mice. Mice were randomized into treatment groups after tumor volumes reached 65 mm³ and received either one dose of TR1801-ADC at 0.5 (n = 5) or 1 mg/kg (n = 5), or one dose of the non-MET–targeted secukinumab-ADC at 1 mg/kg (negative control, n = 5). The efficacy of TR1801-ADC was compared with gemcitabine (n = 4), administered at 50 mg/kg on a weekly basis, and a group was treated with saline solution (gemcitabine vehicle, n = 4). Gemcitabine induced rapid and durable tumor shrinkage, achieving a significant overall reduction in tumor size compared with saline solution–treated tumors (P = 0.0005; Fig. 3A and Supplementary Fig. S3). In addition, complete response was achieved in all treated animals. TR1801-ADC also induced rapid and durable tumor shrinkage, reaching complete response in 70% of treated animals. Overall tumor size reduction was significant for the two doses when compared with secukinumab-ADC control (P = 0.0002 each). In the control groups, tumors grew very fast and some mice had to be euthanized after only 7 weeks (Supplementary Fig. S3). We calculated a TGI index by comparing the relative tumor volumes between treatment cohorts at different timepoints. After only 3 weeks, gemcitabine treatment resulted in very strong tumor inhibition with a TGI index above 80%, reaching a maximum of 98.2% after 7 weeks of treatment. The magnitude of tumor inhibition induced by TR1801-ADC at 0.5 and 1 mg/kg was very similar, with a TGI index slightly below gemcitabine at early timepoints, but reaching a maximum of 98.2% and 97.6%, respectively, after 9 weeks, comparable with the maximum observed with gemcitabine (Fig. 3B).

Histologic analyses revealed the presence of scar tissue in all tumors treated with gemcitabine or TR1801-ADC (Fig. 3C). This was never observed in tumors treated with secukinumab-ADC or saline solution. IHC against MET was conducted on treated tumor tissue. IHC revealed that MET expression levels after gemcitabine, saline solution, or secukinumab-ADC treatment were comparable with pretreatment tumor. While, in a few instances, in which a cluster of tumor cells remained after TR1801-ADC treatment, MET staining was absent. These results indicate that all MET-expressing cells had been effectively targeted by the agent, and thus, a complete response in MET-expressing cells had been achieved.

Treated mice were followed up for a total of 247 days to evaluate the survival benefit of TR1801-ADC. Gemcitabine (P = 0.017) and TR1801-ADC at 0.5 (P = 0.002) or 1 mg/kg (P = 0.002) treatments resulted in a significant improvement in overall survival when compared with their corresponding control groups (Fig. 3D and E).

To evaluate TR1801-ADC efficacy in a different context, we selected PDX1342 for in vivo evaluation. This sample was derived from a patient who had received 12 cycles of gemcitabine/nab-paclitaxel and whose disease progressed on this regimen prior to tissue acquisition. PDX1342 expresses a relatively high level of MET with more than 14,000 receptors per cell detected at the plasma membrane. All mice grafted with PDX1342 and treated with gemcitabine (n = 5) progressed during treatment, with a lack of response comparable with the saline control group (n = 4; Fig. 4A and Supplementary Fig. S4). This lack of response was concordant with the patient's progression on treatment observed in the clinic. Interestingly, one dose of TR1801-ADC at either 0.5 or 1 mg/kg induced tumor shrinkage in all mice within 1–3 weeks, reaching complete response in one third of mice. Over a period of time ranging from week 1 to 11, TR1801-ADC induced a significant reduction in tumor volume compared with secukinumab-ADC (n = 5) for both 0.5 (P = 0.011; n = 5) and 1 mg/kg (P = 0.026; n = 4; Fig. 4A). TGI index calculated at various timepoints revealed that tumor inhibition increased rapidly during the first weeks and reached a maximum of 89% and 85.9% for the 0.5 and 1 mg/kg doses, respectively (Fig. 4B). After a variable period of tumor volume stabilization, most tumors progressed again around week 4–7 (Supplementary Fig. S4). We hypothesized that this tumor progression was the result of either acquired resistance and/or because of a limited quantity of ADC being active at the tumor site after a single dose. Therefore, we submitted the same mice to a second dose of TR1801-ADC, 12 weeks after the first dose was received. Most tumors achieved a period of disease control ranging from 2 to 4 weeks as manifested by either tumor stabilization or tumor shrinkage. (Fig. 4C and D). Following a period ranging from 2 to 4 weeks after the second dose was received, tumors escaped treatment control and grew again. These results suggested that after the first dose, some MET-positive cells remained alive and that some of them were likely targeted by the second dose of drug.

Histologic analysis confirmed that for TR1801-ADC–treated tumors, tumor cells remaining after treatment expressed MET (Fig. 5A). This result indicates that even two doses of TR1801-ADC were not sufficient to kill all MET-expressing tumor cells in PDX1342.

Mouse survival was evaluated for a total period of 232 days postdosing. While gemcitabine failed to improve survival, TR1801-ADC significantly extended survival of mice as compared with secukinumab-ADC at both 0.5 (P = 0.034) and 1 mg/kg (P = 0.004; Fig. 5B and C). Overall, these data demonstrate that TR1801-ADC was able to induce clinical response and improve survival in a gemcitabine-resistant PDX.

Given that single-agent therapy for pancreatic cancer has been poorly effective and that chemotherapy is a standard of care for all stages of PDAC, we sought to evaluate the potential benefits of combining TR1801-ADC and gemcitabine. A cytotoxic assay of all possible combinations over a wide range of concentrations for the two drugs was conducted in vitro on the 1334 cell line (Fig. 6A). The level of drug interaction was evaluated using Combenefit software under the HSA synergy model (17). Notably, we observed synergism between the drugs at several dose combinations suggesting that, when combined, both treatments might lead to an enhanced effect on tumor growth. We then sought to evaluate the combination of TR1801-ADC and gemcitabine in vivo. NSG mice orthotopically grafted with the gemcitabine-resistant PDX1342 received one dose of TR1801-ADC at either 0.5 or 1 mg/kg and weekly doses of gemcitabine (n = 5 for both groups). Strikingly, tumor shrinkage was observed for all mice treated with the combination (Fig. 6B; Supplementary Fig. S5). Combination with TR1801-ADC led to a strong reduction of tumor volume within the first 6 weeks, reaching complete clinical response in all animals. Tumors treated with TR1801-ADC at 0.5 mg/kg remained under treatment control for an additional 6 weeks before most of them started to grow again. Gemcitabine combined with TR1801-ADC at 1 mg/kg resulted in more durable responses as only a single tumor progressed on treatment by the end of the follow-up period. Overall, the combination with gemcitabine provided superior TGI to TR1801-ADC alone at 0.5 mg/kg, which was statistically significant, despite a relatively small sample size (P = 0.018; Fig. 6B).

TGI index calculated at different timepoints reached very high levels of growth inhibition for the two doses with a maximum of 98.6% for the combination at 0.5 mg/kg and 97.3% for the combination at 1 mg/kg. In addition, the combination therapy always provided superior disease control as compared with the corresponding single dose of TR1801-ADC, leading to an overall significant improvement of tumor inhibition (Fig. 6C and D).

The overall survival of cohorts treated with the combination was evaluated over a period of 238 days posttreatment. When compared with the corresponding dose of single treatment with TR1801-ADC, the combination with gemcitabine extended the overall survival of mice at the corresponding ADC dose, although likely due to the relatively small cohort sizes, it did not reach statistical significance (Fig. 5E). Altogether, these data suggest that TR1801-ADC plus gemcitabine is an active combination even in the setting of gemcitabine-refractory disease.

Here, we characterized MET expression in human pancreatic cancer cell lines, tumors, and PDXs to utilize expression as a biomarker for response to a MET-targeted ADC. We found that a substantial portion of PDACs express MET, and thus, we went on to evaluate the efficacy of TR1801-ADC, a newly developed anti-MET ADC, both in vitro and in vivo. Despite heterogeneity of expression, we show that MET is highly expressed in pancreatic cancer and localized at the plasma membrane, corroborating previous report on the expression of MET in PDAC (6, 7). We show that TR1801-ADC has anticancer efficacy both in vitro and in vivo, and that it had activity in the setting of gemcitabine resistance. Importantly, we also observed synergy between TR1801-ADC and gemcitabine, with the combination resulting in durable complete responses in nearly all treated PDXs. Of note, we also observed early activity of secukinumab-ADC with smaller relative tumor volume as compared with saline solution at weeks 2–5 (Fig. 3A). This off-target effect of control ADC is most likely due to nonspecific uptake of the ADC by tumor cells and has been reported previously by Gymnopoulos and colleagues using either secukinumab-ADC or rituximab-ADC in head and neck, gastric, and colorectal preclinical models (15). These cancer types also responded very well to TR1801-ADC with TGI ranging from 40% to 100% using similar drug doses to the ones we used in this study. Other anti-MET ADCs have been developed recently, such as SHR-A1403, which showed efficacy in xenograft models of gastric and liver cancers, and telisotuzumab vedotin, which induced variable levels of TGI, with the greatest being observed in xenograft models of gastric and lung cancers (30, 31). Telisotuzumab vedotin was further evaluated in a dose escalation study on patients with lung cancer and resulted in partial responses in one third of the patients while being well tolerated (32). SHR-A1403 has entered a phase I trial for advanced solid tumors, but no results have been disclosed so far.

Interestingly, we failed to observe a correlation between the overall level of protein expression detected by IHC, by Western blotting, and the content of MET at the plasma membrane detected by quantitative flow cytometry. In addition, we observed a response to TR1801-ADC in PDX1342 despite the fact that this sample displayed only modest reactivity by MET IHC. This finding suggests that the overall level of expression measured by IHC cannot be considered a reliable biomarker for patient stratification and selection. Plasma membrane expression may be a more reliable biomarker, but this would require validation, and practically it remains difficult to perform quantitative flow cytometry on biopsy samples. Other gaps in our knowledge include the threshold of MET expression at the plasma membrane required for response to TR1801-ADC and how the native microenvironment of PDAC may influence the delivery and efficacy of TR1801-ADC. As an example, we observed more complete responses after treatment of PDX1334, which has a slightly lower quantity of plasma membrane MET protein. This lack of correlation between TR1801-ADC efficacy and plasma membrane MET has been described previously in PDXs derived from other solid tumors (15). We believe it is thus fair to speculate that tumors with moderate to low amount of plasma membrane MET expression might respond well to this ADC. Clearly additional experiments are required to better define the best biomarker selection criteria and to identify a low-end threshold necessary to induce TR1801-ADC activity.

PDX1334 was derived after surgical resection of a patient who did not receive any neoadjuvant therapy. This treatment-naïve sample was highly sensitive to gemcitabine in our experiment. As complete responses were achieved in all treated animals, it was not possible to assess any benefit of the ADC over standard of care in this context. While gemcitabine often leads to strong responses in certain murine models of pancreatic cancer, it produces only a minor benefit on overall survival as a single agent in patients (33). Acquired resistance to gemcitabine and to all currently available chemotherapies is essentially guaranteed to occur in the clinic and it is, therefore, critical to develop new treatment regimen to be used as second and later lines of therapy. In this context, the response to TR1801-ADC observed in the gemcitabine-resistant PDX1342 is very encouraging.

In our study, TR1801-ADC treatment resulted in complete responses in PDX1334 and partial responses in PDX1342. Residual tumor cells in mice grafted with PDX1342 expressed MET levels comparable with pretreatment tumor, indicating that these cells were not targeted by TR1801-ADC. We therefore hypothesize that it may be possible to achieve a complete response by adjusting the dose and schedule of TR1801-ADC. None of the animals treated in this study experienced toxicity, and thus, it might therefore be possible to provide additional and/or higher doses of TR1801-ADC to target all MET-expressing cells in the tumor.

Interestingly, the combination of TR1801-ADC with gemcitabine significantly improved tumor response. This is in agreement with the synergistic effects we observed between the two drugs in vitro (Fig. 5A). We also observed that the combination resulted in improved survival as notably, all mice survived to the end of the follow-up period. Nevertheless, the small cohort sizes combined with the high response to TR1801-ADC monotherapy precluded statistical significance. The multi-drug regimens of FOLFIRINOX or gemcitabine/nab-paclitaxel, have led to improvements in overall survival of patients as compared with gemcitabine alone (34, 35). It would thus be interesting to evaluate the efficacy of TR1801-ADC in combination with these more effective treatment regimens. Finally, it has also recently been demonstrated that ADC-bearing PBD payloads induce an immunogenic cell death and can synergize with immunotherapies (36). We can then speculate that, in addition to the activity on tumor cells that we describe here, therapies such as TR1801-ADC could induce responses in patients that could form part of a combinatorial strategy with immuno-oncology drugs.

A. Cazes reports grants from Tanabe Research Laboratories U.S.A., Inc during the conduct of the study. D. Jaquish reports grants from Tanabe during the conduct of the study. A.M. Lowy reports grants from Tanabe Research Laboratories and Syros during the conduct of the study; personal fees from Rafael; and personal fees and other from Merck outside the submitted work. No disclosures were reported by the other authors.

A. Cazes: Conceptualization, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. A.M. Lowy: Conceptualization, supervision, funding acquisition, project administration, writing–review and editing. O. Betancourt: Formal analysis, validation, investigation, visualization, methodology. E. Esparza: Resources, investigation. E.S. Mose: Resources, methodology. D. Jaquish: Resources, methodology. E. Wong: Resources, investigation. A.A. Wascher: Investigation. H. Tiriac: Writing–review and editing. M. Gymnopoulos: Conceptualization, supervision, funding acquisition, project administration, writing–review and editing.

This study was supported, in part, by a sponsored research agreement from Tanabe Research Laboratories U.S.A., Inc. (to A. Cazes and A.M. Lowy), as well as by NIH grant CA155620, a generous support from the Research for a Cure of Pancreatic Cancer Fund, and Ride the Point (to A.M. Lowy). We thank all members of the Lowy laboratory for insightful discussions. We thank the Tissue Technology Shared Resource at UCSD for tissue and slide processing and H&E staining of tumors.

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.