Novel Thienoduocarmycin–Trastuzumab ADC Demonstrates Strong Antitumor Efficacy with Favorable Safety Profile in Preclinical Studies

Antibodies–drug conjugates (ADC) rank among the most sophisticated pharmaceuticals ever developed and are constituted by three different components: a mAb, targeted to highly expressed tumor antigens, a linker and a cytotoxic drug (1). During the years, improvements in antibody engineering, reducing immunogenicity and in linker properties, increasing plasma stability and payload delivery in tumors, allowed to reach the success of the technology we are experiencing nowadays (2).

At present, 12 ADCs have been approved by FDA for the treatment of both hematological and solid tumors and more than one hundred additional molecules are in advanced clinical phases (3), being microtubule binding agents, such as auristatin and maytansine derivatives, the most popular payloads (4). More recently, topoisomerase 1 inhibitors, contained in Enhertu (5) and Trodelvy (6), emerged as an interesting new payload class (7). Despite the recent ADC success some issues remain unresolved, namely safety and resistance mechanisms (8, 9). Both questions can be addressed by optimizing existing payload–linkers or by introducing new cytotoxins with improved characteristics (10) and, indeed, innovation in this field is still much desired (11). Nowadays, most of the known cytotoxins have been tested for conjugation to generate ADCs, but just a small number of them have reached clinical trials (12, 13).

Duocarmycins (14) are one of the most attractive cytotoxin classes for ADC generation due to their high cytotoxic potency and activity in chemoresistant tumors; their success in the field, however, has been hampered by unfavorable properties such as poor solubility in aqueous solution, resulting in poor efficiency of conjugation, and tendency to induce antibody aggregation once conjugated (15).

Over the years some ADC products containing duocarmycin derivatives reached clinical testing (16) and seco-DUBA, developed by Synthon, now Byondis, is the most recent one, currently being included in four ADCs undergoing clinical trials, namely SYD985 (17), MGC018 (18, 19), BYON3521 (20), and SYD1875 (21). ADCs produced with seco-DUBA are all characterized by low DAR to avoid aggregation of the preparation and reduce possible immunogenicity (22). Moreover, some side effects are reported for SYD985, including ocular toxicity and some cases of interstitial lung disease, being eye and respiratory disorders the main causes of drug discontinuation (23).

With the aim to improve physicochemical properties and safety profile of duocarmycin-based ADCs, we started a class exploration, and we identified the highly potent toxin NMS-P528 with sub-nanomolar IC₅₀ values in proliferation assays across a large panel of tumor cell lines. NMS-P945, a drug linker containing NMS-P528 and a peptidase cleavable linker, coupled to antibodies showed favorable DAR in the absence of significant antibody aggregation and with full maintenance of antigen-binding capability. NMS-P945–containing ADCs targeting ALK (24) or HER3 (25) antigens have been previously described.

To fully characterize the properties of this novel payload–linker, we conjugated NMS-P945 to trastuzumab, a clinically approved mAb-targeting HER2 (26), which is a widely explored target for ADCs.

We evaluated the resulting trastuzumab–NMS-P945 ADC, reaching DAR>3.5 without aggregation, for in vitro cytotoxic activity across multiple cell lines in comparison with trastuzumab ADCs bearing deruxtecan DXd the payload included in the approved ADC Enhertu, or MMAE (monomethyl auristatin E), included in four FDA approved ADCs. Furthermore, the antitumor efficacy of NMS-P945–ADC was assessed in HER2 positive mouse xenograft models and pharmacokinetic (PK) studies in monkey were performed with the aim of generating a PK/pharmacodynamic (PK/PD) model for the prediction of the active dose in patients.

Given the positive results obtained, NMS-P945 demonstrates great potential as an improved payload–linker molecule suitable for the generation of well-tolerated and highly efficacious ADCs, possibly overcoming present challenges in clinically approved ADCs.

Pharmaceutical preparations of trastuzumab and rituximab antibodies were acquired from Roche and were reconstituted in water to a final concentration of 15 and 10 mg/mL, respectively.

MMAE and DXd ADCs were prepared following previously described conjugation protocols (27, 28).

Synthesis of NMS-P945 was performed by NMS as previously described (29).

For NMS-P945–ADC conjugation, trastuzumab and rituximab–NMS-P945 were prepared by partially reducing mAb interchain cysteines with 1.9 and 2.1 molar equivalents of TCEP (Aldrich C4706–2G), respectively, and then adding NMS-P945 drug linker (20% excess of free SH groups after reduction). 1 mmol/L solution of cysteine was added at the end of the conjugation process. Buffer exchange by desalting chromatography (Hiprep 26/10 column, GE 17–1088–01) performed in PBS used as storage buffer. Optimization experiments allowed to reach an average drug–antibody ratio (DAR) of >3.5 on the final ADC.

All final ADCs were sterile filtered through a 0.22 μmol/L filter (MILLEX-GV, Millipore SLGV033RS) and characterized by Size Exclusion Chromatography (SEC) using a Superdex 200 10/300 GL column (GE 17–5175–01) equilibrated in PBS buffer (DIAPATH T0023) and Hydrophobic Interaction Chromatography (HIC) on a Sepax HIC column (Proteomix Buthyl-NP5, 5 μm, 4.6×35 mm) using 1 mol/L ammonium sulphate (Sigma-Aldrich A4418) in the mobile phase A and 40% Acetonitrile (Aldrich Q34967) in the mobile phase B and 10 column volumes of gradient. Average DAR has been calculated by LC/MS and compared with HIC data.

For LC/MS analysis, 50 μg of each ADC were analyzed after reduction with 10 mmol/L DTT (Roche, 10708984001) performed for 10 minutes at 37°C on an Agilent 1200 HPLC instrument (Agilent Technologies) using a PLRP-S column (Agilent; 2.1×150 mm; 8 μm 1,000 Å) with a gradient from 20% acetonitrile to 50% in 22 minutes at 0.25 mL/min in the presence of 0.05% v/v trifluoroacetic acid, with a constant column temperature of 80°C. The eluate was sent to an ESI-ToF instrument 6230 TOF LC/MS (Agilent). The molecular weights and peaks area were analyzed with the Qualitative Analysis software B.07.00 (Agilent).

Before performing the cleavage assay, 7 μL of 1.5 mmol/L NMS-P945 solution in dimethyl sulfoxide were incubated with 2 μl of 100 mmol/L cysteine solution in the presence of 50 mmol/L Tris-HCl pH 7.4, to block the maleimide group. The resulting product was then incubated with 0.1 U cathepsin B (Sigma-Aldrich) or PBS as negative control, in the presence of 200 mmol/L sodium acetate pH 5.5 (PerkinElmer) buffer and 1 mmol/L EDTA (Sigma). The digestion was performed at 40°C for 1 hour before the analysis. At the end the digestion was stopped by adding 2 μL of 10% trifluoroacetic acid (Thermo Fisher Scientific) and the samples were subjected to HPLC-MS analysis with ESI-ToF-MS (Agilent) using Poroshell C3 5 μm column (Agilent). The gradient was set from 5% acetonitrile (PanReac) to 100% in 15 minutes at 0.2 mL/min in presence of 0.05% v/v trifluoroacetic acid, with a constant column temperature of 25°C.

Ethylenediamine spacer containing NMS-P528 has been synthesized and incubated at 37°C at a concentration of 50 μmol/L in 3 different buffers: Na bicarbonate 100 mmol/L, MeOH 15% pH 7.4; NaAcetate 300 mmol/L, MeOH 15% pH 5.5; and NaAcetate 300 mmol/L MeOH 15% pH 4.6. Aliquots of the solutions withdrawn at different time points 0, 2, 5, 10, and 24 hours submitted to LC/MS analysis.

The stability of NMS-P528 (30 μmol/L) and NMS-P945 (as cysteinyl derivative) was evaluated in the plasma of different species (mouse, rat, monkey, human) at 37°C up to 24 hours using a UPLC UV/MS analysis. At the end of the incubation time, plasma proteins were precipitated by adding 180 μL of acetonitrile/methanol 90/10 to 20 μL of plasma in an Eppendorf tube. Tubes were mixed for 3 minutes and centrifuged for 4 minutes at 14,000 rpm. Supernatants were then analyzed using an Acquity system (Waters) equipped with a PDA and TQD detectors. Separation performed on a BEH C18 (50×2.1 mm 1.7 μm) column (Waters) using a gradient of mobile phase A: 95:5–5 mmol/L ammonium formate pH 3.5: acetonitrile and mobile phase B: 5:95–5 mmol/L ammonium formate pH 3.5: acetonitrile.

The experiment was repeated in mouse plasma in the presence of a concentration of 1 mmol/L of two carboxylesterase 1 (CES1) inhibitors. Benzil (Diphenylethane-1,2-dione CAS number 134–81–6, Sigma-Aldrich) and BNPP [bis(4-nitrophenyl) phosphate CAS number 645–15–8, Sigma-Aldrich].

All cell lines used in this study were purchased from commercial sources; providers and growth media compositions are reported in Supplementary Table S1. Cell culture media, media additives, and FCS were purchased from Gibco (Thermo Fisher Scientific). Cell lines were authenticated using the AmpFLSTR Identifiler Plus PCR Amplification kit (Applied Biosystems) or the PowerPlex 16 HS System (Promega) following the manufacturer's instructions. Cell lines were tested for absence of Mycoplasma contamination during expansion for master banking using the MycoAlert PLUS Mycoplasma Detection Kit (Lonza, cat. #LT07–710). To perform experiments, aliquots of authenticated cells stored in liquid nitrogen were thawed each time; after thawing, cells were cultured for no more than four passages before use.

HCC1954 cells untreated or treated with NMS-P528, trastuzumab–NMS-P945 (TRZ/NMS-P945) ADC or free trastuzumab at 1 nmol/L for 24 to 72 hours were pulsed adding BrdUrd (bromodeoxyuridine; CAS n. 59–14–3, Sigma-Aldrich, cat. # B-5002) at 30 μmol/L final concentration in the medium for 30 minutes. Cells were harvested, washed, counted, and fixed in ethanol 70% at a density of 10⁶ cells/mL overnight. After washing with PBS (GIBCO, cat. # 10010) + 1% BSA (CAS n. 9048–46–8, Sigma-Aldrich, cat. # A7030), samples were incubated with 0.5 mL HCl 2 N for 20 minutes at room temperature, then acidic DNA denaturation was stopped by adding 1 mL Na₂B₄O₇ 0.1 mol/L (CAS n. 1330–43–4, Sigma-Aldrich, cat. # 221732) at pH 8.5. After centrifugation, cell pellets were resuspended with unconjugated mouse mAb anti-BrdUrd (Becton Dickinson, cat. # 347580) and incubated for 60 minutes at room temperature. After washing with PBS + 1% BSA, samples were incubated with FITC-labeled goat anti-mouse IgG (Jackson Immunoresearch, cat. # 115–096–008) and incubated for 60 minutes at room temperature in the dark. After a last wash in PBS + 1% BSA, samples were incubated with 5 μg/mL Propidium Iodide (CAS n. 25535–16–4, Sigma-Aldrich, cat. # P-4170) + 6 μg/mL RNAseA (CAS n. 9001–99–4, Sigma-Aldrich, cat. # R4875) in PBS for 1 hour at room temperature. At least 10⁴ events were acquired with a BD LSR Fortessa X-20 (BD Biosciences) excluding doublets and debris. Biparametric analysis was performed with FlowJo 10.7.1 software (Becton Dickinson).

Cells untreated or treated with NMS-P528, trastuzumab–NMS-P945 (TRZ/NMS-P945) ADC or free trastuzumab at 1 nmol/L for 24 to 72 hours were harvested, washed, counted, and fixed in 1% formaldehyde (Polysciences, cat. # 04018) for 15 minutes on ice and then permeabilized in methanol 90% (CAS n. 67–56–1, Carlo Erba, cat. # 412532) for 30 minutes on ice. After washing with PBS + 1% BSA, samples were incubated with unconjugated rabbit mAb anti–p-H2AX Ser139 (Cell Signaling Technology, cat. # 9718) and incubated overnight at 4°C. After washing with PBS + 1% BSA, samples were incubated with FITC-labeled goat anti-rabbit (Jackson Immunoresearch, cat. # 111–096–003) for 30 minutes at 4°C in the dark. After washing, DNA was counterstained with Propidium Iodide + 6 μg/mL RNAseA in PBS for 1 hour at room temperature. At least 10⁴ events were acquired with a BD LSR Fortessa X-20 (BD Biosciences) excluding doublets and debris. Biparametric analysis was performed with FlowJo 10.7.1 software.

NCI-N87, HCC1954, and MDA-MB-468 cells were detached with trypsin/EDTA (GIBCO, cat. # 25300), counted, and seeded at a density of a total of 2×10⁵ cell/sample/tube or 10⁵ cell/sample in 96-well plate and recovered in complete culture medium for at least 30 minutes. After washing with PBS + 1% BSA, cells were incubated on ice with trastuzumab–NMS-P945 ADC or trastuzumab at 0.1, 1.0, and 10 μg/mL for 1 hour. Cells were washed with PBS + 1% BSA and incubated with PE-labeled goat anti-human IgG antibody (Jackson Immunoresearch, cat. # 109–116–098) on ice for 1 hour. After washing with PBS + 1% BSA, samples were immediately analyzed with a BD LSR Fortessa X-20 (BD Biosciences) acquiring 10⁴ events gating out debris and aggregates. Analysis was performed with FlowJo 10.7.1 software. Median Fluorescence Intensity (MFI) of each sample was calculated and compared with that of untreated cells.

NCI-N87 cells were detached with trypsin/EDTA (Gibco, cat. # 25300) and counted. A total of 10⁵ cells/sample in triplicate were distributed in two series of tubes for each time point in their complete culture medium and then incubated on ice with trastuzumab–NMS-P945 ADC or trastuzumab at 1 μg/mL for 30 minutes. Subsequently, one series of samples was placed in incubator at 37°C and another series left on ice. At several time points (1, 3, 5, 7, 24 hours), samples from both series were washed 3 times in PBS + 0.02% BSA, stained with PE-labeled goat anti-human antibody (Jackson Immunoresearch, cat. # 109–116–098) and then analyzed with a BD LSR Fortessa X-20 (BD Biosciences) acquiring 10⁴ events gating out debris and aggregates. Analysis was performed with FlowJo 10.7.1 software. MFI of each sample was recorded and comparison between the two temperatures at each time point was reported.

HCC1954 cells were seeded in 4-chambers microscopy glass slides (5×10⁴ cells/chamber) and were treated with trastuzumab–NMS-P945 ADC 1 μg/mL for increasing time points (from 1 to 120 hours), then washed with PBS and fixed for 30 minutes with paraformaldehyde 3.7% solution (Polysciences, cat. #04018). Cells were permeabilized for 30 minutes with PBS containing 1% BSA and 0.05% Triton X-100 (CAS n. 9036–19–5, Sigma-Aldrich, cat. #T9284), then incubated for 1 hour with a murine antibody against NMS-P945 custom-made by NMS. Cells were washed with PBS and incubated for 1 hour with Cy2-labeled anti-mouse antibody (Jackson Immunoresearch, cat. #115–225–166) and Cy5-labeled anti-human antibody (Jackson Immunoresearch, cat. #709–175–149), both diluted 1:500 in PBS containing 1% BSA, 0.5% Tween-20 (Bio-Rad, cat. #1706531) and 1 μg/mL Hoechst 33342 (CAS # 875756–97–1, Invitrogen, cat. #H3570). Cells were washed with PBS, and then the slides were mounted with coverslips with MOWIOL 4–88 Reagent (Millipore, cat. #475904). Fluorescence images were acquired in the Hoechst, Cy2 and Cy5 channels using a LSM 710 confocal microscope system (Zeiss).

In vitro bystander effect induced by trastuzumab–NMS-P945 ADC was determined as reported by Singh and colleagues (30). Briefly, MDA-MB-468 cells (HER2 negative) were transduced with Incucyte Nuclight Red Lentivirus Reagent (Sartorius, cat. # 4625) to express nuclear red fluorescent signal following the manufacturer's protocol (we named these cells MDA-MB-468 NucRed), and subsequently mixed at 1:1 ratio with unstained MDA-MB-468 or, alternately, with HCC1954 or NCI-N87 (Her2-positive cell lines). The three cell mixtures, as well as the monocultures of each cell line, were seeded in 96-well plates (Greiner Bio-One, cat. # 655087) at a density of 2,000 cells/well and left untreated or treated with increasing trastuzumab–NMS-P945 concentrations in the range 0.1–100 nmol/L. Growth was monitored up to 120 hours using an Incucyte system (Sartorius) to measure every two hours the area occupied by the cells (confluency) and the number of MDA-MB-468 NucRed cells per field (red nuclei count). We calculated the Bystander Effect Coefficient (φ_BE) for each co-culture according to Singh and colleagues (30) and reported the average φ_BE from 4 replicate experiments.

Ecto-CRT (ecto-calreticulin) was analyzed in NCI-N87, a Her2-positive/CD20-negative gastric cancer cell line. The cells were seeded in a 96-well black plate with clear bottom at a density of 2,000 cells/well and incubated overnight, then they were treated with 1 nmol/L trastuzumab–NMS-P945 ADC, 1 nmol/L rituximab–NMS-P945–ADC, as negative control, or left untreated. After 144 hours of treatment, cells were washed with PBS and exposed to FVS700 1:2,000 in PBS, a far-red impermeable DNA dye (Fixable Viability Stain 700, BD Horizon, cat. # 564997) for 15 minutes at 4°C. Cells were washed with PBS and fixed with paraformaldehyde 3.7% solution in the presence of 1 μg/mL Hoechst 33342 (Sigma-Aldrich) for 1 hour at room temperature. Cells were washed with PBS, blocked with BSA 3% solution in PBS and incubated for 1 hour with anti-calreticulin antibody 1:200 (Abcam, cat. # ab2907) in PBS + BSA 3%. After washing with PBS, the cells were incubated with Cy2-labeled goat anti-rabbit antibody 1:500 (Jackson Immunoresearch, cat. # 111–225–144) for 1 hour at room temperature. Cells were washed in PBS, then the plates were sealed and analyzed with an ArrayScan VTI high-content screening reader (Thermo Fisher Scientific) in the blue (Hoechst 33342), green (ecto-CRT), and far-red (FVS700) channels. At least 10 fields were acquired per well and biparametric single-cell analysis was performed to generate ecto-CRT versus FVS700 scatterplots. For each sample, the percentage of ecto-CRT–positive/FVS700-negative cells, that is, cells exposing CRT outside an intact membrane, was calculated. The experiment was repeated three times with similar results.

Exponentially growing cells were seeded in 384-well (80 μL/well; Greiner Bio-One, cat. #781093) or 96-well plates (200 μL/well; Greiner Bio-One, cat. #655087) at a density ranging from 10,000 to 30,000 cells per mL in appropriate culture media. After 24 hours, compound or ADC solutions were directly administered to cells using a D300e digital dispenser (Tecan) to obtain scalar concentrations (duplicate wells for each point). Final DMSO (Sigma-Aldrich, cat. # D2650), concentration in test wells was adjusted to 0.1% (v/v); control wells were mock-treated and contained the same final concentration of DMSO as test wells. Cell viability was determined after 72 or 144 hours by ATP determination in each well with a luciferase-based detection system (CellTiter-Glo, Promega, cat. # G7530) following the manufacturer's instructions. The intensity of emitted light was measured using an EnVision reader (PerkinElmer) and expressed as relative light units. Proliferation inhibition IC₅₀ values were calculated by four parameter sigmoidal regression analysis using GraphPad Prism software.

In vivo efficacy of trastuzumab–NMS-P945 ADC was tested using HCC1954 HER2-driven breast tumor model implanted in 4-week-old female CD1 nude mice (Charles River Laboratories). HCC1954 cells were maintained in vitro as a monolayer culture in RPMI-1640 medium supplemented with 10% FBS, 100 U/mL penicillin (Gibco, cat. # 15140122) and 100 μg/mL streptomycin (Gibco, cat. # 15140122) at 37°C in an atmosphere of 5% CO₂ in air. Cells growing in an exponential growth phase were harvested and counted for tumor inoculation.

Each mouse was inoculated subcutaneously at the right front flank with HCC1954 tumor cells (1 × 10⁷ cells with 30% of matrix gel in 300 μL PBS). When the average tumor volume reached approximately to 100–200 mm³ at day 7 after inoculation, animals were randomized and each group, containing 8 mice, treated with test articles. The day of inoculation was considered as day 0. In vivo efficacy was confirmed in a second HER2 driven model, the gastric tumor cell line NCI-N87. Male CD1 nude mice (Charles River Laboratories) 5-week-old were subcutaneously inoculated with a total of 5×10⁶ cells in 200 μL buffer. When the average tumor volume reached approximately to 100–250 mm³ at day 12 after inoculation, animals were randomized into 8 mice groups and treated with test articles. To confirm target-dependent tumor growth inhibition (TGI) a HER2 negative model was selected. A total of 5×10⁶ Burkitt lymphoma Raji cells in 200 μL buffer were inoculated subcutaneously in female SCID mice (Harlan) 4-week-old. When the average tumor volume reached approximately to 100–250 mm³ at day 13 after inoculation, animals were randomized into 7 mice groups for treatment. For all studies, mice were weighed, and tumor growth was measured twice a week using calipers. All procedures related to animal handling, care and treatment in the studies were carried out under protocols approved by NMS Internal Committee for Animal Care and according to EU Directive 2010/63/EU and Italian law (Legislative Decree 26/2014). Tumor volume was calculated with the formula [½ (length × width²)].

TGI was calculated for each group using the formula: TGI (%) = [1−(T_i−T₀)/ (V_i−V₀)] ×100. T_i is the average tumor volume of a treatment group on a given day. T₀ is the average tumor volume of the treatment group on the first day of treatment. V_i is the average tumor volume of the vehicle control group on the same day of T_i and V₀ is the average tumor volume of the vehicle group on the first day of treatment. Tumor volumes at designated days were analyzed using one-way ANOVA with the Dunnett's Multiple Comparison test comparing each group with the control group with GraphPad Prism 6.0 and a P value of <0.05 was considered statistically significant.

For mouse PK studies, 4-week-old SCID female mice bearing Raji tumors were used. Animals were divided into 2 groups of 3 animals each and trastuzumab or trastuzumab–NMS-P945 ADC were injected at a single dose. 100-μL plasma were collected at the following time-points 3, 24, 72, 120, 192, 360, and 528 hours.

For monkey PK studies, 2 cynomolgus monkeys (Camarney), 1 male and 1 female about 3-years-old, were administered with a single dose of 5 mg/kg of trastuzumab–NMS-P945.

Blood samples for systemic exposure evaluation of trastuzumab–NMS-P945 in cynomolgus monkey were collected at 0.083, 0.25, 0.5, 1, 3, 6, 24, 72, 168, 360, and 552 hours after dosing.

All the procedures related to animal handling, care and the treatment in the study were performed according to EU Directive 2010/63/EU and the Italian law (D.Lgs. 26/2014).

PK data analysis was performed using R v4.2.3 (31, 32) with PKNCA v0.10.1 package.

Frozen samples were thawed at 4°C and analyzed by ELISA to check the concentration of the total and conjugated (NMS-P945 bearing) antibody using a total anti-human IgG (Bethyl laboratories A80 319P) and an anti–NMS-P945 antibody (custom-made by NMS), respectively. Briefly, for conjugated NMS-P945, concentration in serum, 96-well plates were coated with 1 μg/mL anti–NMS-P945 antibody, upon washing and blocking with blocking solution, plasma samples serially diluted in 2% BSA were added to the plates. After incubation, goat anti-human IgG HRP was added to 100 ng/mL. The ELISA was developed by addition of Substrate TMB (Thermo Fisher Scientific 34021) and reading at 450 nm. Low limit of quantitation for mouse plasma assay was 7.9 and 16.5 ng/mL whereas for monkey plasma was determined to be 6.7 and 7.7 ng/mL for anti-drug antibody and IgG, respectively.

A PK/PD model was developed using trastuzumab–NMS-P945 PK profiles and antitumor activity in a mouse xenograft model (HCC1954), based on literature (33). A two-compartment population PK model was then developed using NONMEM version VI (ICON Development Solutions) on monkeys’ plasma concentration using anti-NMS-P945 Ab data. Prediction of average human PK was performed through allometric scaling of the PK model developed on monkeys. A clinically relevant dose was predicted by simulations integrating the extrapolated human PK with the mice PD model.

The data generated in this study are available within the article and its Supplementary Data.

NMS-P528 is a novel thienoduocarmycin molecule (Fig. 1A) developed in Nerviano Medical Sciences’ (NMS) Laboratories.

Cytotoxic activity of NMS-P528 was tested in a panel of 30 diverse tumor cell lines (Supplementary Table S2) in parallel with MMAE, DXd, and doxorubicin, a reference potent chemotherapeutic agent. NMS-P528 showed high activity in all tested cell lines with an average antiproliferative IC₅₀ value at 72 hours of cell treatment of 0.202 nmol/L. This average IC₅₀ value outperformed DXd (average IC₅₀: 148 nmol/L) and doxorubicin (average IC₅₀: 200 nmol/L) activity and it was in the same range of MMAE (average IC₅₀: 0.393 nmol/L).

NMS-P528 ability to circumvent multidrug resistance (MDR) system was specifically tested using A2780/ADR, a chemoresistant cell line expressing high level of MDR system components. NMS-P528 demonstrated an antiproliferative activity similar to that observed in the parental A2780 cell line with a calculated resistance index (RI), corresponding to the IC₅₀ ratio of chemoresistant/ parental cell line, of 1.2/1.8 at 72/144 hours of treatment, respectively. In the same experimental setting, both MMAE and DXd showed clear dependency on MDR with an RI of 241/258 and 12/20 at 72 and 144 hours, respectively (Supplementary Table S3). The calculated RI for doxorubicin was 83/99, confirming both the validity of the cellular model and the MDR dependency for both MMAE and DXd.

NMS-P528 was then further characterized evaluating its plasma stability in different animal species. After incubation in plasma, NMS-P528 is spontaneously subjected to a spirocyclization reaction to generate a reactive cyclopropyl derivative. This reactive electrophilic form is then rapidly inactivated by reacting with nucleophilic species present in plasma. NMS-P528 half-life measured in plasma resulted to be 0.46, 0.22, 0.34, and 0.22 hours in mice, rats, cynomolgus monkeys, and humans, respectively (Supplementary Table S4). NMS-P528 poor stability in plasma is considered, in the ADC context, as a positive feature limiting the possible toxic effects of the free drug if released in the blood stream after unattended hydrolysis from the antibody.

To obtain the drug linker NMS-P945, NMS-P528 was then coupled to a self-immolative spacer and a valine citrulline linker bearing a maleimido moiety suitable to conjugation to interchain antibody cysteines (Fig. 1B). The peptidic linker is a substrate of lysosomal enzymes, and to confirm this hypothesis we evaluated the capability of NMS-P945 to release the free drug upon incubation with cathepsin B for 2 hours at pH 5.5. Complete cleavage of the peptidic linker has been observed confirming the sensitivity to cathepsin B activity; however, the release of free NMS-P528 resulted just partial (Supplementary Fig. S1) with a significant amount of the molecule maintaining the ethylenediamine self-immolative spacer bound to NMS-P528. To better understand the release kinetics, we synthesized NMS-P528 + the self-immolative ethylenediamine spacer and incubated the molecule at 3 different pH 7.4, 5.5, and 4.6 at 37°C and collected samples at different time points. The release of the spacer resulted particularly slow at acidic pH, whereas the half-life resulted to be 2.7 hours at pH 7.4 (Supplementary Fig. S2). The peculiar release kinetics allow to maintain a certain amount of inactive, spacer bound, drug inside the tumor to be slowly released continuing to feed the tumor cells with active drug over time.

NMS-P945 half-life was tested in the plasma of different animal species and found to be >24 hours in rats, cynomolgus monkeys and humans, whereas resulted much shorter (1.31 hours) in mouse plasma (Supplementary Table S5A).

HPLC/MS analysis indicated that free drug NMS-P528 was released in the plasma of mice but not in human plasma. We reasoned that this mouse-specific mechanism of release could be attributed to the activity of carboxylesterase 1 (CES1), a highly abundant enzyme in mouse but not in human plasma. Consistently, incubation with 1 mmol/L Benzil or BNPP, known CES inhibitors, increased the mouse plasma stability of NMS-P945 up to approximately 36% or 63%, respectively, at 24 hours (Supplementary Table S5B).

Trastuzumab was used as biological proof of concept and conjugated to NMS-P945, MMAE, and DXd through interchain cysteines upon reduction of interchain disulfide bonds. Trastuzumab conjugation to NMS-P945 and MMAE was obtained by partial interchain disulfide reduction whereas trastuzumab DXd was produced by total reduction of interchain cysteines as previously reported. SEC analysis showed that all final products were mainly monomeric. HIC analysis indicated a similar profile for MMAE and NMS-P945–ADCs with DAR2 and DAR4 mainly represented. The average DAR of trastuzumab–NMS-P945 preparation was 3.9, as judged by LC/MS and HIC analysis, with no need for chromatographic purification or other enrichment steps (Fig. 1C–E). This remarkable result confirmed the peculiar suitability of NMS-P945 as duocarmycin derivative for ADC production. This molecule is indeed, in our knowledge, the only duocarmycin derivative able to produce ADCs with DAR >3.5 without any need of further purification after the conjugation step. Comparable results were obtained for trastuzumab MMAE, whereas nearly complete interchain disulfide derivatization and DAR≈8 was obtained for trastuzumab DXd as reported in literature (Supplementary Fig. S3).

NMS-P945–ADC antigen-binding properties were assessed in parallel with MMAE and DXd ADCs. Cytofluorimetric assay performed on HER2-positive cell lines (NCI-N87 and HCC1954) confirmed similar binding properties for the three different ADCs (Fig. 2A). As expected, no binding was detected for any of the three ADCs and naked trastuzumab in the HER2-negative cell line MDA-MB-468 confirming their specificity.

Cytotoxic activity of trastuzumab conjugated with NMS-P945, MMAE, and DXd was tested side-by-side (Tables 1 and 2) in various human cancer HER2 positive or negative cell lines (34, 35).

In all the cell lines, trastuzumab ADCs showed target related, potent cell growth inhibition after 144 hours cell treatment. NMS-P945–ADC showed picomolar IC₅₀ values (range, 0.055–0.069 nmol/L) across HER2 positive cell lines, comparable with MMAE ADC, with strong selectivity versus cells that did not express the target, whereas DXd ADC resulted less potent (IC₅₀ range, 0.186–1.66 nmol/L).

DXd ADC, peculiarly, showed relatively lower cytotoxicity toward HCC1569 HER2 positive cell line with an average IC₅₀ value of 1.66 nmol/L compared with 0.065 and 0.058 nmol/L for NMS-P945 and MMAE ADCs, respectively.

The HCC1569 cell line bears a nonsense mutation of SLX4 gene producing a protein truncated form at Q934, as reported in cBioportal database (36, 37), predicted resulting in a loss of function. SLX4 mutations are reported in 20% of patients resistant to trastuzumab DXd treatment in the Daisy II study and SLX4 depletion in two breast cancer cell lines led to 5–20-fold increased IC₈₀ values for DXd (38). Our results indicate that NMS-P945 trastuzumab ADC, due to peculiar properties of NMS-P528 payload, could overcome resistance to trastuzumab DXd induced by SLX4 mutations and the use of ADCs containing NMS-P945 payload–linker could be possibly considered in patients relapsed upon treatment with DXd based ADCs.

Internalization kinetics of NMS-P945–ADC analyzed by flow cytometry in HCC1954 cells recapitulated those of naked trastuzumab with about 40%–50% internalization after 7 hours incubation at the permissive temperature of 37°C (Fig. 2B). Furthermore, trastuzumab–NMS-P945 ADC internalization and release of the cytotoxic moiety were analyzed in HER2-positive HCC1954 cells by immunofluorescence and confocal microscopy. Cells were treated with ADC for 1, 7, 72, 96 and 120 hours, fixed and immunostained with a custom-made antidrug antibody to detect NMS-P945 as well as an anti-human IgG1 to detect total trastuzumab (Fig. 2C). ADC internalization was evident at 7 hours after treatment, when both immunofluorescence signals, associated to the payload and to the total antibody, colocalize in the membrane and in the endosomal vesicles. Interestingly, upon internalization the IgG1 signal detecting trastuzumab was clearly present in the endosomes even at later time points, whereas the signal of bound NMS-P945 disappeared during time as expected after drug release from the antibody due to lysosomal proteases cleavage.

Trastuzumab–NMS-P945 ADC mechanism of action was explored in parallel to trastuzumab in HCC1954 cells by flow cytometry (Fig. 2D). DNA damage, cell-cycle block, and apoptosis induction were clearly observed with trastuzumab–NMS-P945 ADC whereas trastuzumab results were superimposable with the ones of untreated cells. Clear p-H2AX induction, indicative of DNA damage, was observed, with 39% positive cells with the ADC whereas just 3% positivity was present with the naked trastuzumab or in untreated cells. Moreover, cell-cycle block, with the presence of inactive S phase and BrdUrd incorporation reduced from 21% to 7%, as well as apoptosis induction, with increased sub-G₁ cell-cycle phase from 7% in untreated cells to 21% in cells treated with the ADC, were observed.

A much-desired feature of ADCs is the ability to induce immunogenic cell death (ICD), which is a type of cell death capable of triggering the activation of the immune system. The ICD involves modifications of the cell membrane and release of the so-called DAMPs (damage-associated molecular patterns), including exposure of calreticulin (ecto-CRT). We analyzed ecto-CRT by immunofluorescence and high-content analysis in NCI-N87 cells. In membrane-intact cells (confirmed by lack of staining with FVS700 impermeable fluorescent dye) significant induction of CRT was detected at the outer plasma membrane in cells treated with trastuzumab–NMS-P945 ADC, but not with rituximab-NMS-P945–ADC, used as negative control, confirming the specificity of the signal and the target-dependent ICD induction capability of NMS-P945–ADCs (Fig. 3A).

The Bystander effect was determined in vitro by analysis of the cytotoxic effect of trastuzumab–NMS-P945 ADC toward antigen-negative cells in co-culture with antigen-positive cells. We calculated the Bystander Effect Coefficient (φ_BE) according to Singh and colleagues (30): A φ_BE greater than zero denotes bystander effect. The experiment was performed co-culturing MDA-MB-468 cells (HER2 negative) transfected with Nuclight Red for nuclear labeling of live cells (MDA-MB-468 NucRed) in 1:1 ratio with unstained MDA-MB-468 or, alternatively, with HCC1954 or NCI-N87 (HER2 positive) cell lines. From the analysis of NucRed fluorescent signal upon treatment with trastuzumab–NMS-P945, detecting MDA-MB-468 NucRed-surviving cells, a φ_BE of about 20 was calculated when these cells were co-cultured with HER2 positive cell lines (HCC1954 and NCI-N87), whereas a negative φ_BE was detected in the co-culture with unstained MDA-MB-468 (HER2 negative), demonstrating that NMS-P945–ADCs exert bystander cytotoxic effect in vitro (Fig. 3B).

Next, the in vivo antitumor activity of trastuzumab–NMS-P945 ADC was evaluated in HER2 expressing HCC1954 breast cancer mouse xenograft model. Mice were treated with intravenous injection of a single or double dose of trastuzumab–NMS-P945 ADC or rituximab-NMS-P945, as negative control ADC, in parallel with naked trastuzumab.

A dose of 3, 7.5 or 15 mg/kg of trastuzumab–NMS-P945 was applied in single administration. With 3 mg/kg, initial tumor stabilization was observed whereas with 7.5 and 15 mg/kg tumor regression can be appreciated (Fig. 4A). With one dose of 15 mg/Kg trastuzumab–NMS-P945, 4 out of 8 animals showed complete tumor regression and remained tumor free at the end of the experiment (Fig. 4B) and with two administrations of trastuzumab–NMS-P945 at 15 mg/kg, 6 out of 8 animals resulted tumor-free at the end of the experiment (day 118); none of the animals in any of the experimental groups in the in vivo efficacy study showed significant change in body weight (Fig. 4C), confirming strong efficacy at well-tolerated doses. Considering the instability of the NMS-P945 in mouse plasma, the free NMS-P528 generated from 15 mg/Kg NMS-P945 did not cause any apparent toxicity, consistent with the design of reducing systemic toxicity by the short half-life of the free drug.

Additional confirmation of in vivo activity of trastuzumab–NMS-P945 ADC was obtained in the NCI-N87 HER2-positive gastric cancer mouse xenograft model. Animals were treated with a single dose of 6 or 12 mg/kg trastuzumab–NMS-P945 ADC, 12 mg/kg rituximab–NMS-P945–ADC or 12 mg/kg naked trastuzumab. Both trastuzumab–NMS-P945 ADCs doses induced transient tumor regression whereas no or mild TGI was observed with rituximab–NMS-P945 or naked trastuzumab (max TGI 27%), respectively (Fig. 4D).

Target-driven activity of the ADC was confirmed by performing an experiment in the CD20-positive, HER2-negative Raji Lymphoma model, subcutaneously implanted in female SCID mice. In this case, single treatment with anti-CD20 rituximab–NMS-P945–ADC at 15 mg/kg was compared with trastuzumab–NMS-P945–ADC as negative control and NMS-P528–free payload at 0.5 mg/kg. No TGI was observed with NMS-P528–free payload or untargeted trastuzumab–NMS-P945–ADC, whereas complete tumor regression was obtained upon administration of 15 mg/kg rituximab-NMS-P945–ADC with 3/7 cured mice at the end of experiment confirming that the effect of trastuzumab–NMS-P945 in HCC1954 model is target related and, in the absence of the target, no TGI is obtained either by injecting the free payload or the untargeted ADC (Fig. 4E and F). Moreover, the activity of rituximab–NMS-P945–ADC in Raji CD20-positive Lymphoma model reinforced the concept that NMS-P945 conjugated to different antibodies can induce tumor regression with cured mice at the end of the experiment.

As discussed earlier, NMS-P945 is characterized by a reduced stability in mouse plasma due to the well-known CES1 activity in this species. PK studies were performed in mice and non-human primates (NHP; Fig. 5A). As expected, upon single-dose administration of 15 mg/kg trastuzumab–NMS-P945 to mice, whereas the total antibody AUC was comparable with naked trastuzumab, the conjugated ADC showed a very steep decrease in the concentration corresponding to a low AUC. The mean clearance of total and conjugated Ab was 0.325 versus 29.0 mL/h/kg, respectively, confirming faster elimination of the conjugated antibody from circulation, half-life of conjugated antibody was 26.4 hours whereas the one of total antibody resulted to be 245 hours. This behavior is compatible with hydrolysis of the drug linker in plasma, as expected due to CES1 activity in mouse. Consistently, in NHP, where the CES1 enzyme is not present, a half-life of the conjugated trastuzumab–NMS-P945 more similar to the one of total antibody was observed being 55.2 and 113 hours, respectively. Accordingly, in NHP a low clearance of 0.701 and 1.37 mL/h/kg was observed for both total and conjugated Ab, respectively. The volumes of distribution (Vss/Vz) of conjugated and total Ab were limited, in the range of approximately 3.02/3.2 and 4.02/4.2-fold, respectively, versus, the total monkey plasma volume (35 mL/kg). The mean C_max ratio between anti-drug Ab and total Ab was 0.991 whereas in terms of AUC∞, the mean ratio was 0.520 (Supplementary Table S7). No signs of toxicity were observed both in mice and in NHP at the tested doses.

Mice PK and efficacy data were used for building a PK/PD model suitable for prediction of the active dose for tumor eradication in humans.

As previously described, PK results indicated that in mice trastuzumab–NMS-P945 accounted for less than 1% of total trastuzumab (including conjugated and unconjugated antibody), whereas the total antibody AUC was close to the naked trastuzumab one.

Noteworthy, the monkey trastuzumab–NMS-P945 AUC accounted for approximately 50% of the total antibody one and a similar C_max was observed for total and conjugated antibody.

Predicted human tumor growth following different trastuzumab–NMS-P945 administration schedules was simulated (Fig. 5B). On the basis of this model, two administrations of trastuzumab–NMS-P945 0.5 mg/kg, one week apart, are expected to considerably reduce the tumor growth.

With more and more diverse ADCs entering clinical trials, there is increasing awareness of the impact on clinical outcomes of the general properties of the complete ADC, as well as of its single components (39).

As a matter of facts, big improvements have been made over the years in antibody engineering, with humanized and fully human antibodies replacing mouse or chimeric antibodies, in conjugation technology, decreasing ADC heterogeneity in terms of drug distribution, and in linker technology, increasing plasma stability and targeted delivery in tumors of the free payloads (40).

Moreover, in outlining the future landscape of the ADC field, one must consider that present and future therapeutic strategies in oncology will increasingly be based on combinations of drugs with different mechanisms of action, possibly synergizing to achieve greater tumor inhibition with a larger therapeutic window. For this reason, next-generation ADCs need to be tailored to be synergistically combined with small-molecule–targeted therapies, classical chemotherapeutic regimens and immunotherapy in specific tumor settings (41). In view of this, a portfolio of diversified payloads in terms of mechanism of action, PK properties, and mechanism of resistance will allow to choose the optimal match with respect to the tumor type, previous and undergoing therapy regimens.

The recent success of topoisomerase I inhibitors included in the approved Trodelvy and Enhertu ADCs has strengthened the idea that new payloads, with different mechanism of action, may work in tumor settings where microtubule-binding agent bound to the same targeting mAb previously failed. Kadcyla, progenitor HER2 driven ADC bound to DM1, a microtubule-binding agent with uncleavable linker, successfully reached approval in HER2 high metastatic breast cancer, whereas it failed to meet the criteria in HER2 low patients. Enhertu, instead, with a payload based on topoisomerase I inhibitor and a cleavable linker bearing strong bystander effect, resulted successful in HER2 low metastatic breast cancer (42).

At this stage, ADC-targeting antibodies previously tested into clinical trials and resulting unsuccessful can be reconsidered in view of mechanism of action or favorable properties of new payloads and linkers, opening to new tumor indications and overcoming liabilities of first-generation ADCs against the same targets.

Among the new payloads, duocarmycins are considered of high interest due to their DNA-damaging mechanism of action, possibly synergizing with currently approved oncological therapies, including immuno-stimulating agents. Moreover, thanks to their activity in chemoresistant tumors overexpressing MDR system, they can be indicated in patients heavily pretreated with topoisomerase inhibitors such as irinotecan or antimitotic therapies such as paclitaxel.

The use of duocarmycins as ADC payloads has been impaired, or at least limited, so far, by the physicochemical parameters of the class causing poor antibody conjugation level and antibody aggregation.

Some duocarmycin-based ADCs have reached clinical trials. The first example is BMS-936561 (MDX-1203) (43), an anti-CD70 ADC containing a duocarmycin prodrug via lysine conjugation. The amount of drug loaded on the antibody was rather low and a large fraction of unconjugated antibody was present in the final preparation and the molecule was abandoned in phase I due to concerns about efficacy and safety.

More recently, Byondis, formerly Synthon, developed vc-seco-DUBA, a second duocarmycin with a cleavable linker suitable for cysteine conjugation through reduced interchain disulfides and entered clinical trials with SYD985, a trastuzumab duocarmycin conjugate. In July, 2022, FDA granted the approval of the Company's BLA to treat patients with HER2-positive unresectable locally advanced or metastatic breast cancer. This application was supported by the findings from the pivotal Phase III TULIP clinical trial (23). The vc-seco-DUBA payload–linker is included in 3 additional clinical stage ADCs: MGC018-targeting B7-H3 under development by Macrogenics and currently in phase I/II clinical trials (19), BYON3521, targeting cMet starting phase 1 clinical trials in 2022 (20), and SYD1875 (21), targeting 5T4 that currently completed recruitment of phase 1 clinical trial. These ADCs showed preclinical and, in some cases, clinical activity in different tumor settings; however, seco-DUBA payload still presents some issues and a new generation of duocarmycin payload–linkers is much desired to fully leverage the value of this class.

ADC generation with vc-seco-DUBA is challenging as a low DAR around 2 must be maintained either by site-specific conjugation, as was done for BYON3521 (20) reaching a DAR 1.8, or through a preparative chromatographic step to eliminate the remaining large amount of unconjugated antibody and the highly conjugated species to avoid aggregation (22). The preparative column purification and the low DAR obtained limit vc-seco-DUBA broad applicability, impairing the use of this payload–linker with more hydrophobic antibodies usually more prone to aggregation. Moreover, some side effects limited the use of vc-seco-DUBA ADCs, most notably ocular toxicity and respiratory problems are the main reported one for the most advanced ADC SYD985 (44).

Duocarmycin (14) mechanism of action relies on activation by intramolecular cyclization followed by alkylation of nitrogen in position 3 of adenine present in DNA minor groove. Kinetics of intramolecular cyclization and subsequent alkylation reaction are fundamental to retain antitumoral activity.

NMS-P528 design includes a pyrrolidine-solubilizing moiety to improve physicochemical profile of the molecule and an indolic minor groove binder moiety coupled to thienoindole scaffold (45) to reach the best balance between reactivity and stability as previously reported by Boger group studies (46).

The resulting thienoduocarmycin molecule presents shorter plasma stability in comparison with reported data for seco-DUBA (17) and different kinetics of conversion between the seco and the closed form. Seco-DUBA in vitro and in vivo is immediately converted to the closed active form and seco-DUBA per se cannot be isolated whereas the seco form of NMS-P528, still inactive, is isolated in plasma but, upon cyclization, the closed form is rapidly converted to inactive forms. Reported half-life for DUBA are 5.3, 0.6, 1.6, and 1 hours whereas NMS-P528 half-life are 0.46, 0.22, 0.34, and 0.22 hours in mouse, rat, monkey, and human plasma, respectively, indicating much lower stability. It can be expected that this different kinetic mechanism could result in a lower systemic nonspecific exposure to the toxic molecule, increasing the safety of the ADC if some free drug is released in plasma.

In addition, upon lysosomal protease cleavage of the payload–linker, the ethylenediamine self-immolative spacer has to be released to free up NMS-P528 in tumor cells. We reasoned that a slower release could allow a prolonged delivery of the payload, maintaining the effect for longer time inside the tumor cells and stabilizing NMS-P528 as a prodrug. We compared NMS-P528 spacer release half-life and it resulted to be 2.7 hours at pH 7.4, 37°C whereas a much shorter half-life has been reported for seco-DUBA being of 27 minutes at 25°C and even shorter at 37°C (17) confirming once more the superiority in terms of kinetics of release and inactivation of NMS-P528.

NMS-P945 payload–linker containing NMS-P528, an ethylenediamine self-immolative spacer, a valine citrulline linker, and a maleimido moiety to bind reduced antibody interchain cysteines, was accurately designed with the aim of decreasing hydrophobicity and allowing its easy conjugation to a large number of antibodies.

Indeed, conjugation to NMS-P945 with DAR >3.5 without any final purification was previously reported for EV20, a HER3-targeting antibody, and ALK-targeted antibodies.

HER3-targeting antibody, EV20–NMS-P945–ADC-bearing DAR 3.5 was reported to have good target-driven antiproliferative activity in cells and tumors, acceptable PK, and early safety profile in monkey where no safety concerns were identified up to 8 mg/kg. The antitumor activity of EV20-NMS-P945 was indeed in line with AMT-562 (47), an exatecan-based HER3-targeted ADC currently undergoing clinical trials, confirming once more the validity of NMS-P945 as payload–linker. As previously mentioned, the duocarmycin payload could allow to overcome chemoresistance induced by previous treatments with topoisomerase I inhibitors.

ADCs generated with NMS-P945 mainly contain DAR2 and DAR4 species but also a small amount of free antibody that is never present in vc–seco-DUBA–containing ADCs, whereas similar drug distribution is observed in MMAE-approved ADCs. This feature has been explored in the paper of Singh and colleagues (48) where the impact of free antibody coadministration with ADC preparations has been evaluated concluding that antibody coadministration may be helpful for ADCs in certain circumstances helping in diffusing the ADC inside the tumor mass.

In this study, we conjugated NMS-P945 to trastuzumab to further validate this payload–linker and compare payload properties with approved MMAE and DXd payloads.

Strong antiproliferative activity for trastuzumab–NMS-P945 was observed in all the HER2 positive cell lines tested, including the breast cancer cell line HCC1569. This model bears a mutation in SLX4, a protein involved in DNA damage response, which was found mutated in 20% of patients with breast cancer resistant to trastuzumab DXd in the Daisy II study (38). In vitro results clearly indicated the possibility for trastuzumab–NMS-P945 ADC to overcome resistance either induced by MDR system genes overexpression, as NMS-P528 is not a substrate of MDR, or by nonsense mutation in SLX4 as depletion of this protein is not affecting the activity of trastuzumab–NMS-P945 ADC.

In preclinical xenograft mouse models, trastuzumab–NMS-P945, despite low plasma concentration linked to species-specific expression of CES1 enzyme, leading to artifactual ADC cleavage and consequent toxin release and inactivation, showed strong target-driven efficacy with cured mice in HER2-positive tumors at well-tolerated doses. Target-driven activity was further confirmed by the fact that free payload and untargeted ADC administered at similar doses did not result in significant TGI. Moreover, despite release of the free toxin in plasma, no body weight loss or other signs of toxicity were appreciated in treated animals, confirming that the drug is likely fast inactivated in plasma, as expected.

PK parameters confirmed greater stability of trastuzumab–NMS-P945 in monkey compared with mouse; however, conjugated antibody clearance, AUC and half-life values resulted lower than the ones measured for the total antibody, in line with the observation previously made with EV20 NMS-P945. Reduction of PK parameters for conjugated ADC in comparison with total antibody values in NHP has been reported in a number of approved ADCs. Kadcyla's conjugated antibody clearance is indeed 10 mL/d/kg whereas the total antibody has a clearance of 4.6 mL/d/kg; similar behavior is observed with Adcetris with a conjugated antibody clearance of 18.5 mL/d/kg and a total antibody value of 14.6 mL/d/kg (49) as for Trodelvy an half-life of 7–21 hours has been reported for the conjugated antibody with a corresponding half-life for the total antibody between 73 and 152 hours.

In our case, in view of the fast clearance of the free toxin NMS-P528 from circulation by inactivation, this slight instability should not be considered negatively as a reduced half-life of the conjugate ADC also reduces the impact of very long-term exposure to normal tissues with the possibility of unattended capture of the ADC and release of the free payload in normal cells.

The developed PK/PD model for trastuzumab–NMS-P945 resulted in prediction of a very low dose of 0.5 mg/kg trastuzumab–NMS-P945 ADC given twice as expected therapeutic dose in patients. This dose is much lower than the currently administered doses for approved ADCs: Kadcyla (50) 3.6 mg/kg or Enhertu (5) 5.4 mg/kg, both administered every three weeks, and it is also lower than SYD985 (51) dose of 1.2 mg/kg every three weeks applied in phase III clinical studies. This PK/PD prediction for an intermitted schedule administering only two doses of trastuzumab–NMS-P945 represents a significant safety feature associated with the use of NMS-P945 for the generation of novel ADCs, possibly helping in reducing side effects.

In summary, we present here an in-depth characterization of NMS-P945, obtained through its conjugation with the model mAb trastuzumab and its characterization in vitro and in vivo, in comparison with reference ADC derivatives.

Overall, the data indicate that NMS-P945 is a highly versatile payload–linker, easy to conjugate to different antibodies with reproducible, significant DAR without ADC aggregation. The toxin insensitivity to multidrug resistance system and fast inactivation in circulation, representing highly desirable safety features, together with the ability to induce immunogenic cell death, bystander effect and strong antitumor activity in target-expressing tumor cells warrant further exploration of the payload–linker NMS-P945 in different contexts for the generation of novel ADCs with enhanced therapeutic index in the clinics.

P. Orsini reports a patent for WO2013/149948 and WO2017/012924 issued. M. Caruso reports a patent for WO2013/149948 and WO2017/012924 issued. M. Salsa reports a patent for WO2013/14994 issued. No disclosures were reported by the other authors.

B. Valsasina: Conceptualization, resources, data curation, supervision, validation, investigation, visualization, writing–original draft, writing–review and editing. P. Orsini: Data curation, formal analysis, supervision, investigation, methodology, writing–review and editing. M. Caruso: Conceptualization. C. Albanese: Data curation, formal analysis, investigation, visualization, methodology. A. Ciavolella: Investigation, methodology. U. Cucchi: Data curation, validation, investigation, visualization, methodology. I. Fraietta: Investigation, visualization, methodology. N. Melillo: Data curation, software, formal analysis, investigation, visualization, methodology. F. Fiorentini: Data curation, software, validation, investigation, visualization, methodology. S. Rizzi: Data curation, validation, investigation, visualization, methodology, writing–review and editing. M. Salsa: Validation, investigation, visualization, methodology. A. Isacchi: Resources, formal analysis, supervision, writing–review and editing. F. Gasparri: Conceptualization, data curation, formal analysis, supervision, investigation, visualization, methodology, writing–review and editing.

The authors thank Daniele Donati for his strong support in the overall evolution of the project, Rita Perego and Sonia Troiani for assistance in NMS-P945–ADC conjugation and characterization, Aurelio Marsiglio for assistance with cell proliferation assays, Carlo Visco for assistance with ELISA assay, Federico Riccardi Sirtori, and Nicoletta Colombo for assistance in payload characterization, and Marina Ciomei for assistance with in vivo experiments.

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.