1521-009X/44/7/897–910$25.00 http://dx.doi.org/10.1124/dmd.115.069021DRUG METABOLISM AND DISPOSITION Drug Metab Dispos 44:897–910, July 2016Copyright ª 2016 by The American Society for Pharmacology and Experimental Therapeutics

New Insights in Tissue Distribution, Metabolism, and Excretionof [3H]-Labeled Antibody Maytansinoid Conjugates in Female

Tumor-Bearing Nude Rats

Markus Walles, Bettina Rudolph, Thierry Wolf, Julien Bourgailh, Martina Suetterlin,Thomas Moenius, Gisela Peraus, Olivier Heudi, Walid Elbast, Christian Lanshoeft, and Sanela Bilic

Novartis Institutes for Biomedical Research, Drug Metabolism and Pharmacokinetics, Basel, Switzerland (M.W, B.R., T.W., J.B.,M.S., T.M., G.P., O.H., W.E., C.L.); and Translational Clinical Oncology, East Hannover, New Jersey (S.B.)

Received December 17, 2015; accepted April 26, 2016

ABSTRACT

For antibody drug conjugates (ADCs), the fate of the cytotoxic payload invivo needs to be well understood to mitigate toxicity risks and properlydesign the first in-patient studies. Therefore, a distribution, metabolism,and excretion (DME) study with a radiolabeled rat cross-reactive ADC([3H]DM1-LNL897) targeting the P-cadherin receptor was conducted infemale tumor-bearing nude rats. Although multiple components [totalradioactivity, conjugated ADC, total ADC, emtansine (DM1) payload,and catabolites] needed to be monitored with different technologies(liquid scintillation counting, liquid chromatography/mass spectrometry,enzyme-linked immunosorbent assay, and size exclusion chromatogra-phy), the pharmacokinetic data were nearly superimposable with thevarious techniques. [3H]DM1-LNL897 was cleared with half-lives of 51–62 hours and LNL897-related radioactivity showed a minor extent oftissue distribution. The highest tissue concentrations of [3H]DM1-LNL897–related radioactivity were measured in tumor. Complimentary

liquid extraction surface analysis coupled to micro-liquidchromatography–tandem mass spectrometry data proved that thelysine (LYS)–4(maleimidylmethyl) cyclohexane-1-carboxylate–DM1 (LYS-MCC-DM1) catabolite was the only detectable component distributedevenly in the tumor and liver tissue. The mass balance was completewith up to 13.8%6 0.482%of the administered radioactivity remaining incarcass 168 hours postdose. LNL897-derived radioactivity was mainlyexcreted via feces (84.5% 6 3.12%) and through urine only to a minorextent (4.15%6 0.462%). In serum, the major part of radioactivity couldbeattributed toADC,while smallmoleculedispositionproductswere thepredominant species in excreta. We show that there is a difference inmetabolite profiles depending on which derivatization methods for DM1were applied. Besides previously published results on LYS-MCC-DM1and MCC-DM1, maysine and a cysteine conjugate of DM1 could beidentified in serum and excreta.

Introduction

Antibody-drug conjugates (ADCs) constitute a therapeutic modalityin which a cytotoxic agent (payload) is chemically linked via a cleavableor noncleavable linker to amonoclonal antibody (mAb) that recognizes atumor-associated antigen. To date, three ADCs have been approved fortherapeutic use: 1) Gemtuzumab ozogamicin for acute myelogenousleukemia, which has been withdrawn from the market; 2) Brentuximabvedotin for Hodgkin’s lymphoma; and 3) the most recently approvedtrastuzumab ado-emtansine for second-line treatment of HER2-positivemetastatic breast cancer). While the goal of using ADCs as therapeuticsis to minimize or localize toxicity with the antibody intended to targetspecific tissue(s), toxic changes with these agents can been seen inmultiple tissues (van der Heiden et al., 2006; de Claro et al., 2012; Poonet al., 2013; Yan et al., 2016). Toxicities observed with ADCs may beassociated with either the components of the ADC (antibody, linker, and

payload), or the metabolites. Therefore, the fate of the ADCs in vivoneeds to be well understood tomitigate toxicity risks and properly designthe first in-patient studies. Up to now, little has been published about thedistribution and uptake of ADCs (Saad et al., 2015). Therefore, tissuedistribution as well as metabolism and excretion data can be very usefulto support pharmacokinetics and pharmacodynamics as well as safetyassessment in that they provide insights of drug levels at the site ofaction, e.g., in tumors (Wada et al., 2014). In addition, they may be usedfor mechanistic investigations, e.g., to explore sites of unexpected drugclearance or for characterization of target-mediated drug disposition.Therefore, these data can be considered as a powerful tool in relation tolearning about disposition, pharmacokinetics and pharmacodynamics,and toxicity of ADCs.The bioanalytical and distribution, metabolism, and excretion (DME)

properties of antibody maytansinoid–based ADCs have been recentlyreviewed (Erickson and Lambert, 2012; Kaur et al., 2013; Deslandes,2014; Han and Zhao, 2014; Thudium et al., 2013; Marcoux et al.,2015; Beck et al., 2016; Kraynov et al., 2016) The fate of antibodydx.doi.org/10.1124/dmd.115.069021.

ABBREVIATIONS: ADC, antibody-drug conjugate; AUC, area under the curve; AUClast, area under the curve from 0 hours to the last measuredtime point; DAR, drug-to-antibody ratio; DME, distribution, metabolism, and excretion; DM1, emtansine; ELISA, enzyme-linked immunosorbentassay; Fc, fragment crystallizable; LC, liquid chromatography; LESA-mLC-MS/MS, liquid extraction surface analysis coupled to micro-liquidchromatography–tandem mass spectrometry; LSC, liquid scintillation counting; LYS, lysine; mAb, monoclonal antibody; MCC, 4(maleimidylmethyl)cyclohexane-1-carboxylate; MS, mass spectrometry; m/z, mass-to-charge ratio; NEM, N-ethyl-maleimide; QWBA, quantitative whole-bodyautoradiography; RA, radioactivity; SEC, size exclusion chromatography; TCEP, tris(2-carboxyethyl) phosphine; UPLC, ultra-performance liquidchromatography.

897

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

maytansinoid conjugates was investigated in naive or tumor-bearingmice and naive rats with iodine labels on the antibody or a tritium labelon emtansine (DM1) (Xie and Blättler, 2006; Shen et al., 2012a; Saadet al., 2015). The measured tissue exposures were low compared withblood and the main components in serum were attributed to ADC. Themain catabolites in serum and excreta differed depending on whether theADC used a cleavable linker or a noncleavable linker. With a cleavablelinker, the main catabolite in mouse liver was DM1 (Sun et al., 2011),while with the noncleavable 4(maleimidylmethyl) cyclohexane-1-carboxylate (MCC) linker the main catabolites detected in rat wereMCC-DM1 and lysine (LYS)-MCC-DM1 (Sun et al., 2011; Saad et al.,2015). Since released DM1 has been reported to bind to proteins (Daviset al., 2012), derivatization methods using N-ethyl-maleimide (NEM)and tris(2-carboxyethyl) phosphine (TCEP) are usually applied (Shenet al., 2012a).This paper describes a DME study for an ADC with a tritium-

labeled DM1 and MCC linker conducted in female tumor-bearingnude rats against P-cadherin, which is overexpressed in mammarycarcinomas (Albergaria et al., 2011). A P-cadherin–targeted ADCapproach is currently explored in the clinic as a novel promisingtherapy for patients suffering from breast cancer (Menezes et al.,2015). The key questions we wanted to address in this study were ifthe tritium-labeled ADC with higher specific activity would besufficiently stable to perform quantitative whole-body autoradiogra-phy (QWBA) to study the tissue distribution in depth. We wereparticular interested in investigating the extent of distribution betweentumors and other tissues, since this can differ, and seeing if QWBAcan give additional insights compared with previously publishedtissue harvesting methods (Shen et al., 2012a). To investigate if theradioactivity (RA) in tumors can be attributed mainly to intact ADC orsmall molecule catabolites, tumors were analyzed at different timepoints to establish kinetics. In addition, complimentary investigationsusing the newly developed liquid extraction surface analysis micro-liquid chromatography–tandem mass spectrometry (LESA-mLC-MS/MS) technique (Lanshoeft et al., 2016) were performed on tumortissue for the first time. For the bioanalytical methods applied, wewanted to investigate if size exclusion chromatography (SEC)coupled to RA detection provides similar results for pharmacokineticassessments of ADC as can be obtained by the enzyme-linkedimmunosorbent assay (ELISA), and/or if SEC-RA can provideadditional benefits. Finally, since the catabolism of antibody may-tansinoid conjugates has been previously described (Shen et al.,2012a; Widdison et al., 2015), these current data compare themetabolic profiles with and without applying derivatization methodsto gain new insights into catabolism.

Material and Methods

Test Compound and Reference Standards

For the purpose of this study, an ADC with a MCC linker was labeled withtritium in the chemically stable aromatic methoxygroup of the DM1 moiety. Thereference standards LYS-MCC-DM1, MCC-DM1, as well as DM1 (structuresdescribed in Sun et al., 2011 and Fig. 8) were kindly provided by ImmunoGen,Inc. (Waltham, MA).

In Vivo Experiments for Radiolabeled ADC

All procedures on animals were conducted in Basel, Switzerland, inaccordance with the Swiss Animal Welfare Legislation and authorized by thecantonal veterinary office of Basel-Stadt (license number: BS-1975 and BS-2639;http://www.veterinaeramt.bs.ch/tierschutz/tierversuche.html). Nude rats wereprovided by TaconicArtemis (Cologne, Germany). Tumor transplantation wasperformed in accordance with animal license. For the tumor transplantation, the

human breast carcinoma cell line HCC70 was cultured under standard conditions.The cells were harvested, resuspended, and injected subcutaneously into the hindleft flank of the female nude rats. The tumors were allowed to grow until theyreached a mean volume of about;250–400 mm3 before dosing with [3H]DM1-LNL897 was initiated. For dosing, [3H]DM1-LNL897 was dissolved in watercontaining 10 mM citric acid and 135 mM sodium chloride (pH 5.5). The ratsreceived an intravenous bolus injection in the vena saphena of nominal 10 mgradiolabeled ADC/kg under light inhalation anesthesia using an oxygen/isoflurane mixture (97/3, v/v; isoflurane: Forene, Abbott AG, Zug, Switzerland).The total amount of RA administered was nominal 50 MBq/kg (where MBqdenotes megabecquerel).

After dosing, blood was collected from the sublingual vein of three tumor-bearing nude rats at 1, 24, 96, and 168 hours postdose. From each time point theblood was processed to serum by allowing it to clot for at least 30 minutes atroom temperature, followed by centrifugation at 2500g for 15 minutes at 4�C.A small aliquot was taken for total RA detection and the remainder of thematerial was snap frozen and stored at 280�C until further analysis. From asecond group of tumor-bearing nude rats, tumors were collected at the sametime points as for the rats where blood had been collected (N = 1 rat/time point).In addition, selected animals (N = 3 tumor-bearing nude rats) were housed inmetabolic cages for urine and feces collection, allowing determination of theroute of excretion and the mass balance. QWBA was performed in a last groupof animals at 1, 24, 72, 96, 168, and 264 hours postdose using one animal pertime point.

Analytical Methods

Radiometric Analysis. Radioactive signals in the different matrices werequantified by liquid scintillation counting (LSC), using Liquid ScintillationSystems 3500 TR from the Packard Instrument Co. (Meriden, CT). For quenchcorrection, an external standard ratio method was used. Quench correction curveswere established by means of sealed standards (Packard Instrument Co.).Background values were prepared for each batch of samples according to therespective matrix. The limit of detection was defined as 1.8 times the backgroundvalue. All determinations of RA were conducted in weighed samples. Todetermine possible formation of tritiated water, samples were analyzed twice:direct measurements of RA and after drying of the sample (at 37�C for at least 12hours). Dried or nondried serum and urine were mixed with scintillation cocktaildirectly, whereas whole blood and feces samples were solubilized before RAanalysis. Measured RA for blood, plasma, and tissues was converted intoconcentrations (mol Eq per volume or gram) considering the specific RA andassuming a matrix density of 1 g/ml.

Quantitative Whole-Body Autoradiography. Rats were sacrificed by deepisoflurane inhalation. After deep anesthesia, blood for RA determination by LSCwas collected from the sublingual vein. The animals were then submerged inn-hexane/dry ice at270�C for approximately 30minutes. The embedded carcasseswere stored at 218�C and all subsequent procedures were performed attemperatures below 220�C to minimize diffusion of radiolabeled materials intothawed tissues. Sections were prepared as follows: in brief, 40-mm-thicklengthwise dehydrated whole-body sections were prepared and exposed to FujiBAS III imaging plates (Fuji Photo Film Co., Ltd., Tokyo) in a lead-shielded boxat room temperature for 2 weeks. Afterward, the sections were scanned in a FujiBAS 5000 phosphor imager (Fuji Photo Film Co., Ltd.) at a 50-mm scanningstep. The concentrations of total radiolabeled components in the tissues weredetermined by comparative densitometry and digital analysis of the autoradio-gram as described in Schweitzer et al. (1987); blood samples of knownconcentrations of total radiolabeled components processed under the sameconditions were used as calibrators.

LESA-mLC-MS/MS. For mass spectrometric detection, whole-body sectionswere lyophilized at 240�C overnight and mounted on metal glass-slides plates,after which they were placed on a Perfection V370 flatbed scanner (Epson,Kloten, Switzerland) to acquire an optical image. A PAL Autosampler from CTCAnalytics AG (Zwingen, Switzerland) was used for extraction of the analytes fromthe tissue surface at predefined sampling spots using a standard 100 ml PALsyringe from Hamilton Company (Reno, NV). Droplets on the sections weredeposited and extracted using a modified CTC Autosampler (CTC Analytics AG,Schlieren, Switzerland). The extracted analytes were injected into an EkspertmicroLC 200 system from Eksigent (Dublin, CA) coupled to an API4000 triple

898 Walles et al.

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

quadrupole mass analyzer system (AB SCIEX, Toronto, Canada). The ex-traction with 2 ml of extraction solvent resulted in a spatial resolution of2 mm, observing specific transitions for MCC-DM1 and LYS-MCC-DM1in the mass-to-charge ratio (m/z) range from 100 to 1300. A detailed overviewof the principle experimental setup has been described recently (Lanshoeftet al., 2016).

Quantification of Whole and Conjugated ADC by ELISA. In the serumsamples, total antibody concentrations (reflecting conjugated and unconjugatedforms of the ADC) were quantified using an ELISA. With this ELISA, all ADCspecies available in the systemic circulation independent of the status of DM1conjugation [drug-to-antibody ratio (DAR) 0–8] were measured. In addition,conjugated antibody concentrations (ADC) of all ADCs bearing at least oneconjugated DM1 were also measured using an ELISA. The ELISA for themeasurement of conjugated ADC used an anti-maytansine antibody as thecapturing reagent (provided by ImmunoGen, Inc.), which was immobilized onmicrotiter plates. Digoxigenin-labeled anti-human–specific antibody followedby anti-digoxigenin/fragment antigen-binding/peroxidase was used for de-tection by colorimetric readout. The same concept was used for the totalantibody assay. However, in this case anti-human–specific antibodies wereused for capturing and detection.

For measurement of the ADC, biotinylated anti-maytansinoid mAbimmobilized on a streptavidin-coated microtiter plate was used as thecapturing antibody. A second antibody, digoxygenin-labeled anti-humanfragment crystallizable (Fc) mAb (R10) was used, which binds to the Fc partof the ADC and was detected by anti-digoxigen in the fragment antigen-binding region labeled with peroxidase and 3,39-5,59-tetramethylbenzidine asthe substrate. The optical density was measured at 450 nm. The same conceptwas used for the total antibody assay. However, in this case monoclonal R10antibody (anti-human Fc-specific mAb) was used to capture the ADC and apolyclonal goat anti-human Fc antibody was used for detection. The lowerlimit of quantification for both assays (ADC and total antibody) was 0.0991mg/ml DM1-LNL897 in a 100% matrix; the upper limit of quantification forboth assays (ADC and total antibody) was 2.13 mg/ml DM1-LNL897 in a100% matrix.

Determination of Payload DAR and Drug-Load Distribution in DosingSolution. The DAR as well as the distribution of the different ADC species(where D represents the number of DM1 molecules covalently attached to theantibody: D0, D1, D2, D3, D4, D5, D6, D7, and D8) in the dosing solution wasdetermined using high-performance liquid chromatography (LC) with massspectrometry (MS) detection.

To calculate the ADC distribution by MS, the dosing solution (20 ml) wasdiluted into 80 ml of mobile phase A (0.1% formic acid in water). The diluteddosing solution was injected (50 ml) and separated on a MassPrep on-linedesalting cartridge 10 � 2.1 mm (Waters, Milford, MA) with a flow rate of0.4 ml/min. Themobile phasewas composed of 0.1% formic acid in water (EluentA) (Fisher Scientific AG, Reinach, Switzerland) and acetonitrile containing 0.1%formic acid (Eluent B) (Sigma Aldrich GmbH, Buchs, Switzerland). The gradientcondition was maintained at 10% B for 1 minute, ramped to 80% B for 5 minutes,kept at 80% B for the next 4 minutes, and then returned back to 10% B for1 minute. After the chromatography, the effluent was split into a ratio of 1:3 withthe smaller portion directed into the MS interface and the bigger portion used forradiodetection. Mass measurements were made online using a Synapt G2-SiHDMS mass spectrometer from Waters operating under MassLynx (version4.1; Waters, Milford, MA) for instrument control, data acquisition, and dataprocessing. From 0 to 12 minutes, the mass spectrometer was operating inpositive electrospray ionization mode and scanning from m/z 400 to 4000 with aresolution of 9000.

The total ion chromatogram was selected and the corresponding time-of-flightMS mass spectrum was extracted at a representative time window (containingsignals from ADCs of interest) to determine the drug load distribution. Thecharacteristic mass spectrum was deconvoluted using the MaxEnt1 algorithm inMassLynx. The resulting processed mass spectrum (see Fig. 1) showed differentgroups of peaks corresponding to the different numbers of conjugated payloads.Various glycan forms were also observed for each group of a specific payload,since no deglycosylation procedure was applied. The area under each peak wasmeasured using the Radiostar software, version 5.12.0.3 (Berthold Technol-ogies, Bad Wildbad, Germany) and the sum of the peak areas for each groupwas determined. The relative percentage ratio of individual ADC species to

total peak area corresponds to the relative abundance of each ADC species(distribution):

DAR ¼ +n

i¼1

�DixRAAi

100

�ð1Þ

where RAAi denotes the percentage of relative abundance of each ADC species(see the aforementioned distribution calculation; for the assessment it wasassumed that all ADC species have the same ionization efficiency).

Determination of ADC in Serum and Excreta by SEC and Radiometry.A high-performance LC system (model 1100; Agilent Technologies, Santa Clara,CA) equippedwith a binary capillary pump, a column oven, an autosampler, and adiode array detector was used. Then, 15ml of each serum samplewas injected on aYarra SEC 3000; 300 � 7.8 mm, 3 mm SEC column (Yarra SEC 3000; 300 �7.8 mm, 3 mm, from Phenomenex Inc., Torrance, CA) equipped with a GFC3000m 4 � 3 mm i.d. guard column (GFC 3000m 4 � 3 mm i.d. fromPhenomenex Inc.). The column was maintained at 60�C. The ADC was elutedisocratically with 200mM ammonium acetate (pH 7) containing 15% isopropanoland an isocratic at a flow rate of 0.40 ml/min, which was maintained for 60minutes. The column effluent was collected in 6-second fractions on yttriumsilicate scintillator-coated 96-well plates (LumaPlates; Packard BioScience,

Fig. 1. Measurement of [3H]DM1-ADC species distribution in dosing solution. Thedosing solution was analyzed by LC-MS using a desalting cartridge. Representativedeconvoluted mass spectra at day 1 (day of dosing) and day 5 of the dosing solutionbeing stored at 4�C were compared and are shown here. The different groups ofpeaks were annotated by Dn, where n corresponds to the number of conjugated DM1molecules per antibody. Each group is composed of four main peaks and severalminor components that correspond to the different glycosylation forms. Since nodeglycosylation procedure was used for this analysis, the [3H]DM1-LNL897distribution in percentage refers to the relative area ratios of each group. Profileswere acquired by SEC coupled to RA detection.

DME Properties of Antibody Maytansinoid Conjugates 899

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

Groningen, The Netherlands) from 0 to 40 minutes using a fraction collectorGX271 (Gilson, Villiers-le-Bel, France). The plates were dried and counted for 5minutes in triplicate on a Packard TopCount instrument (PerkinElmer, Waltham,MA). To determine the presence of intact monomeric ADC [3H]-DM1-LNL897in excreta, aliquots of urine and feces pools (group 2, T0–168h) were centrifuged at20,000g for 15 minutes at 12�C, and 15 ml of the respective supernatants weredirectly injected onto the SEC column as previously described.

Determination of DM1 and Catabolites in Serum, Urine, and Feces. Urineand feces samples from three animals collected between 0 and 168 hours werepooled by combining equal percentages (by weight) of individual fractions. Forserum, equal volumes taken from three rats at the same time postdose were mixed.Since DM1 has a thiol function, it can dimerize in solution under mild oxidativeconditions and/or form disulfide bonds with thiol-containing molecules such as

cysteine in serum (Shen et al., 2012a). Therefore, before extraction, serum poolsas well as urine pools and feces-homogenized pools were split into two equalparts, after which one was incubated with an equal amount (v/v) of 0.5 M TCEP(Sigma Aldrich GmbH, Steinheim, Germany) for 2 hours at room temperature inthe dark. The TCEP is used as a reducing agent to reduce any disulfides (includingS-bound DM1) in the sample. The resulting free thiol groups were then alkylatedby addition of a 4-fold amount (v/v) of 0.2MNEM (Sigma Aldrich GmbH) to theincubation solutions in order to prevent further reaction. The samples were storedfor two additional hours after addition of NEM (at room temperature in the dark)before further extraction.

The treated and untreated serum and feces samples were then extracted byprotein precipitation as follows. Aliquots of sample pools with and without TCEPtreatment were stirred for 1 hour with four volumes of acetonitrile followed bycentrifugation at 10,000g for 30 minutes. After decanting the supernatants, theRA in the pellets was determined and the recoveries of extraction werecalculated. The supernatants were evaporated to dryness under a stream ofnitrogen at room temperature and the obtained residues were reconstituted intwo volumes (equivalent to original feces/serum volume) of water/acetonitrile(95/5 v/v).

All extracts as well as TCEP treated/untreated urine samples were directlyinjected (injection volume 20–50 ml) into an ultra-performance liquid chromatog-raphy (UPLC) system (Waters) with off-line RA and MS detection. The UPLCsystem was equipped with two pumps, a column oven, a diode array detector, andan autosampler (CTC PAL 2777) from CTC Analytics AG. The chromatographicsystemwas in line with a fraction collector GX271 (Gilson) and a quadrupole time-of-flight tandem mass spectrometer, Synapt HDMS (Waters) operating underMasslynx, version 4.1, for instrument control, data acquisition, and dataprocessing. Samples were injected on a reverse-phase C18 column Acquity HSST3; 150 � 2.1 mm; 1.8 mm particles (Waters, Baden-Dättwil, Switzerland)equippedwith anAcquity guard column (5� 2.1mm; 1.5mmparticles;Waters). Thecolumn was maintained at 40�C, and the flow rate on the column was 0.45 ml/min.Eluent A was made up of 20 mM ammonium formate including 0.1% of tri-fluoroacetic acid (pH 3.5), and Eluent B was made up of acetonitrile. The gradientwas maintained at 5% B for 2 minutes, ramped to 35% B for 3 minutes, increasedslowly to 55%B for 15minutes and then to 90%B for 5minutes, kept at 90%B for5 minutes, returned back to 5% B in 5 minutes, and finally equilibrated for 5minutes before next injection (for a total run time of 40minutes). After elution fromthe column, the effluent was split into a ratio of 1:6with the smaller portion directedto the MS source, while the remaining flow was collected in 3-second fractions on96 LumaPlates (Packard BioScience) from 0 to 30 minutes. The plates were dried

Fig. 2. Off-line SEC radiochromatograms of serum samples. Upper graph contains timeconcentration serum profiles (logarithmic scale) acquired by LSC, ELISA, and SEC-RA.

TABLE 1

Measured concentrations and derived pharmacokinetic data of [3H]DM1-LNL897

Concentrations of total radiolabeled components measured by LSC, total and conjugated antibody measured by ELISA, conjugated antibody measured by SEC-RA (expressed in micromolars), andcorresponding pharmacokinetic parameters in serum of female tumor-bearing nude rats after a single intravenous administration of [3H]DM1-LNL897 (nominal dose: 10 mg/kg).

ParameterTotal RA (LSC)

ELISA

Conjugated Antibody (SEC)aTotal Antibody Conjugated Antibody

Mean S.D. CV Mean S.D. CV Mean S.D. CV

% % %

1 hour 1.39 0.0764 5.48 1.25 0.0557 4.46 1.42 0.0279 1.97 1.3124 hours 0.585 0.0286 4.89 0.577 0.0128 2.22 0.664 0.00873 1.31 0.48396 hours 0.197 0.0159 8.05 0.214 0.0034 1.61 0.268 0.012 4.47 0.147168 hours 0.0835 0.00683 8.18 0.107 0.00979 9.17 0.133 0.0133 9.98 0.0457Actual dose (mg/kg) 10 0.0808 0.807 10 0.0808 0.807 10 0.0808 0.807 10Tmax (hour) 1.0b (1.0–1.0)]c 1.0b (1.0–1.0)c 1.0b (1.0–1.0)c

Cmax (mM) 1.39 0.0764 5.48 1.25 0.0557 4.46 1.42 0.0279 1.97Cmax/dose (mM)/(mg/kg) 0.139 0.00717 5.16 0.125 0.0055 4.43 0.142 0.0021 1.49Tlast (hour) 168b (168–168)c 168b (168–168)c 168b (168–168)c

AUClast (h·mM) 62.4 2.08 3.34 62.3 0.749 1.2 73.4 1.33 1.81AUCinf (h·mM) 68.6 1.72 2.5 71.5 0.721 1.01 85.4 2.77 3.25T1/2 (hour) 51.3 2.45 4.77 59.2 4.02 6.79 62.2 4.09 6.58T1/2 range (hour) (24–168) (24–168) (24–168)Vss (l/kg) NC NC 0.617 0.00295 4.77

NC, not calculated.aAggregate peak and ADC peak values were summed to calculate concentrations.bMedian.cRange.

900 Walles et al.

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

and counted (up to 30minutes perwell, depending on the level of RA) on a PackardTopCount instrument (PerkinElmer).

From the relative peak areas, the concentrations of individual radiolabeledcomponents in serum and the amounts of individual radiolabeled components inthe excreta were calculated as follows:

Ci;serum ¼ CRA;serum ×RSP

100%×RPAi

100%ð2Þ

and

Ai;excreta ¼ ARA;excreta ×RSP

100%×RPAi

100%ð3Þ

whereCi,serum denotes the concentration of radiolabeled component i in the serum (ona molar basis); CRA,serum denotes the concentration of total radiolabeled components(RA) in the serum (on a molar basis); RSP denotes the recovery of RA after samplepreparation (%); RPAi denotes the relative peak area of radiolabeled component i inthe radiochromatogram (percentage of the total area under the radiochromatogram);Ai,excreta denotes the amount of radiolabeled component i in the excreta (urine, feces;percentage of the dose); and ARA,excreta denotes the amount of total radiolabeledcomponents (RA) in the excreta (urine, feces; percentage of the dose).

Determination of ADC and Its Catabolites in Tumor Tissues. Formetabolic profiling, collected tumor tissues were homogenized separately using a

Covaris CryoPrep (Woburn, MA). After extraction, samples were analyzed by LC-MS. Each tumor was weighed and placed in the center of the flexible pouch of aTT05 Tissue Tube (Covaris) using tweezers. The sample was frozen by immersingthe flexible pouch into liquid nitrogen, and the tissue was pulverized using aCovaris CryoPrep. The tissue particleswere then transferred into a 2-ml preweighedProtein LoBind tube from Eppendorf (Hamburg, Germany), weighed, and ho-mogenized with 10 parts of water (w/w) using a PT1200 Polytron (Kinematica,Eschbach, Germany). Afterward, an aliquot of each tumor homogenate was treatedwith TCEP (according to the derivatization method described previously) and asecond one was processed without treatment. Both samples were extracted asfollows: a 4-fold amount (v/w) of acetonitrile was added to the treated and untreatedtumor homogenates and the mixture was stirred (750 rpm) for up to 3 hours,followed by centrifugation at 20,000g for 30 minutes at 12�C. The remainingsupernatant was evaporated to dryness under a stream of nitrogen at roomtemperature. The residues of untreated tissues were then reconstituted in onevolume (equivalent to the original tumor homogenate volume) of water/acetonitrile(95/5, v/v), while residues of derivatized tissue samples were reconstituted in twovolumes of water/acetonitrile (95/5, v/v). Tumor extracts (80ml) were injected ontoan UPLC systemwith off-line RA andMS detection for determination of free drugsand catabolites (see the instrumentation and conditions for analysis of free DM1and catabolites in serum, urine, and feces). To determine the presence of intactmonomeric ADC [3H]DM1-LNL897 in tumors, homogenized tissues were centri-fuged at 18,000g for 20 minutes at 12�C and the recovered supernatants weredirectly injected (25 ml) on the SEC system described previously.

The RA amount in tumor homogenates was measured by LSC as describedpreviously. All determinations of RA were conducted in weighed samples. The

TABLE 2

Pharmacokinetic parameters of total radiolabeled components in selected tissues of female tumor-bearing nude ratsfollowing a single nominal 10 mg/kg i.v. dose of [3H]DM1-LNL897

The T/B ratio is the tissue to blood ratio of Cmax and AUClast, normalized to the respective blood value estimated by QWBA.

Tissue Tmax CmaxCmax

T/B RatioAUClast

AUClast

T/B RatioTlast T1/2 T1/2 Range

hour nmol/g h·nmol/g hour hour hour

Blood (LSC) 1.0 0.902 0.784 44.4 0.800 264 76.6 96–264Blood (QWBA) 1.0 1.15 1.0 55.5 1.0 264 75.5 96–264Adrenal gland (cortex) 1.0 0.333 0.290 23.3 0.420 264 91.4 72–264Adrenal gland (medulla) 1.0 0.573 0.498 52.7 0.950 264 NC NCBone marrow 1.0 0.390 0.339 20.6 0.371 264 85.7 72–264Choroid plexus 24.0 0.190 0.165 15.9 0.287 264 81.3 72–264Esophagus 72.0 0.142 0.123 12.8 0.231 264 80.9 96–264Eye (choroid) 24.0 0.174 0.151 17.0 0.306 264 94.2 72–264Eye (ciliary body) 72.0 0.0719 0.0625 12.7 0.229 264 67.5 96–264Fat (brown) 1.0 0.106 0.0922 12.8 0.231 264 NC NCHair (follicle) 24.0 0.129 0.112 28.2 0.508 264 NC NCHair (tactile) 168 0.139 0.121 24.7 0.445 264 NC NCHeart 1.0 0.391 0.340 19.7 0.355 264 81.3 72–264Intestinal wall (colon) 24.0 0.109 0.0948 11.7 0.211 264 NC NCIntestinal wall (small intestine) 24.0 0.281 0.244 24.8 0.447 264 65.3 96–264Kidney (CM junction) 1.0 0.377 0.328 29.4 0.530 264 NC NCKidney (cortex) 1.0 0.358 0.311 33.8 0.609 264 NC NCKidney (medulla) 1.0 0.494 0.430 38.7 0.697 264 73.1 96–264Liver 1.0 0.595 0.517 48.5 0.874 264 85.8 72–264Lung 1.0 0.679 0.590 39.3 0.708 264 85.4 96–264Lymph node (submandibular) 1.0 0.443 0.385 42.2 0.761 264 88.4 96–264Ovarian tissue 1.0 0.556 0.483 46.1 0.831 264 68.7 96–264Pancreas 1.0 0.125 0.109 10.2 0.184 264 74.2 72–264Pineal body 1.0 0.265 0.230 17.1 0.308 264 85.9 72–264Pituitary gland 24.0 0.279 0.243 28.4 0.512 264 80.6 72–264Preputial gland 24.0 0.211 0.183 21.3 0.384 264 NC NCSkin 72.0 0.0676 0.0588 10.7 0.193 264 76.3 96–264Spleen 1.0 0.475 0.413 43.4 0.782 264 78.3 96–264Spleen (red pulp) 1.0 0.506 0.440 49.2 0.887 264 84.7 96–264Spleen (white pulp) 1.0 0.203 0.177 19.9 0.359 264 86.7 24–264Stomach (glandular) 24.0 0.121 0.105 10.7 0.193 264 62.4 96–264Thyroid gland 1.0 0.144 0.125 15.9 0.287 264 NC NCTongue 24.0 0.145 0.126 12.4 0.223 264 81.8 96–264Tooth (pulp) 1.0 0.478 0.416 41.6 0.750 264 NC NCTumor 24.0 0.575 0.500 85.6 1.54 264 NC NCUterus 72.0 0.177 0.154 27.5 0.496 264 83.7 96–264Uterus mucosa 72.0 0.265 0.230 43.7 0.788 264 88.0 96–264

NC, not calculated (due to the limited data set available or the adjusted r2 , 0.75); T/B, tissue to blood.

DME Properties of Antibody Maytansinoid Conjugates 901

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

RA was transformed into concentrations of total radiolabeled componentsconsidering the specific RA of [3H]-DM1-LNL897, which was derived by thedetermination of the RA in the dose solution

CRA;tumor ¼ RAtumor � 1000Mmass� RAspec�Wtumor�60

ð4Þ

whereCRA,tumor denotes the concentration of radiolabeled component in the tumor(pmol·g21); RAtumor denotes the total RA measured in the tumor (dpm); Mmass

denotes the molecular mass of the ADC (g·mol21); RAspec denotes the specificRA in the dosing solution (MBq·mg21);Wtumor denotes the tumor weight (g); and60 denotes the conversion factor (Bq to dpm).

Peaks in the radiochromatogramswere integratedmanually using the Radiostarsoftware (Berthold Technologies). From the relative peak areas, the concentra-tions of individual radiolabeled components in tumor were estimated as follows:

Ci;tumor ¼ CRA;tumor � DAR� RPAi

100%� RSP

100%ð5Þ:

where Ci,tumor denotes the concentration of radiolabeled component i in the tumor(pmol·g21);CRA,tumor denotes the concentration of radiolabeled component in thetumor (pmol·g21); RSP denotes the recovery of RA after sample preparation (%);and RPAi denotes the relative peak area of radiolabeled component i in theradiochromatogram (percentage of total area under the radiochromatogram).

Pharmacokinetic Analysis. Pharmacokinetic parameters in serum and tissueswere calculated using Phoenix WinNonlin software (Pharsight Corp., MountainView, CA). The pharmacokinetic estimations were based on a noncompartmentalanalysis model. The tissue-to-blood Cmax and the area under the curve (AUC)from the 0 hour to the last measured time point (AUClast) ratios were calculatedwhen possible. The AUClast was calculated using the linear trapezoidal rule. Thehalf-life was calculated for those tissues and matrices where at least threemeasurable data points were available in the terminal phase of the tissueconcentration time course.

Results

Purity, Composition, and Stability of Dosing Solutions

The radiopurity of [3H]DM1-LNL897ADC in the dosing solution wasdetermined by SEC to be 94.5% pure (ADC monomer) with 2.94%

accounted for by soluble aggregates and 2.56% by small moleculecomponents (data not shown). In addition to radiochemical puritymeasurements of [3H]DM1-LNL897 ADC in the dosing solution, therelative distribution of different ADC species (0–8 DM1 moleculesattached to antibody) aswell as theDARwere determined byMS (Fig. 1).The deconvoluted mass spectrum revealed relative area ratios of 3.3%,11.7%, 19.7%, 22.7%, 18.2%, 12.5%, 6.3%, 3.9%, and 1.8% for ADCsD0, D1, D2, D3, D4, D5, D6, D7, and D8, respectively (where Dncorresponds to the number of covalently bound payloadDM1). After fivedays at 4�C, the sample was remeasured and a comparable distributionwas observed (1.2%, 10.4%, 20.5%, 25.2%, 19.9%, 13.6%, 5.8%, 2.6%,

Fig. 3. Selected whole-body autoradiograms after a single nominal10 mg/kg i.v. dose of [3H]DM1-LNL897 in female tumor-bearingnude rats. Selected lengthwise whole-body sections (A–H) throughtumor-bearing nude rats are displayed: (A and B) 1 hour postdose;(C and D) 24 hours postdose; (E and F) 72 hours postdose; and (Gand H) 168 hours postdose. Selected major organs are labeled in thefollowing order: a, tumor; b, blood; c, kidney; d, stomach; e, liver; f,heart; g, thymus; h, brain; j, intestinal tract; k, tongue/mouth; l,spinal cord; and m, salivary gland.

Fig. 4. (A and B) Radiochromatograms of serum samples 1 hour postdose acquiredby high-performance liquid chromatography with off-line RA detection.

902 Walles et al.

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

and 0.8% for ADCs D0–D8, respectively). The calculated DAR at day 1and after five days was found to be 3.3. These results demonstrate that[3H]DM1-LNL897 ADC was stable (over a period of 5 days at 4�C) inthe dosing solution without significant loss of DM1 from the ADC.

Pharmacokinetics of Total Radiolabeled Components and Totaland Conjugated Antibodies after Intravenous Dosing of Nominal10 mg/kg [3H]DM1-LNL897 ADC

The concentrations of total radiolabeled components, as well as thetotal and conjugated antibodies in serum were measurable throughoutthe observation period up to 168 hours postdose (Fig. 2; Table 1). TheADC was eliminated with mean half-lives ranging from 51.3 hours(based on RA measurements by LSC) to 62.2 hours (based onconjugated antibody measured by ELISA). The AUClast of total andconjugated antibodies amounted to 62.3 and 73.4 hours/mM, respec-tively (Table 1), which represented ;99.8% and 117% of the AUClast

value of total radiolabeled components, respectively. The concentrationdata acquired by SEC-RA measurements were in good agreement withthe ELISA-derived data (Fig. 2; Table 1).

Tissue Distribution of Total Radiolabeled Components asDetermined by QWBA after Intravenous Dosing of Nominal 10mg/kg of [3H]DM1-LNL897 to Female Tumor-Bearing Nude Rats

The tissue distribution data are summarized in Table 2. RepresentativeQWBA images are shown in Fig. 3. Total radiolabeled components wereextravascularly distributed throughout the body to a minor extent. Inselected tissues given in Table 2 and shown in Fig. 3, total radiolabeledcomponents were quantifiable up to 264 hours postdose. In 20 out of 36tissues, Tmax was rapidly reached at 1 hour postdose, the first time pointanalyzed. In 10 out of 35 tissues, Tmax was reached at 24 hours postdose, thesecond time point analyzed. In the remaining tissues,Tmaxwas reached at 72hours postdose. Only hair (tactile) showed a Tmax value 168 hours postdose.The highest Cmax values were determined for lung (0.679 nmol/g), tumor(0.575 nmol/g), liver (0.595 nmol/g), adrenal gland (medulla) (0.573 nmol/g),ovarian tissue (0.556 nmol/g), spleen (red pulp) (0.506 nmol/g), andkidney (medulla)(0.494 nmol/g). In all tissues analyzed, exposure basedon Cmax was lower in comparison with blood (QWBA) (1.15 nmol/g).The highest AUClast valueswere determined for tumor (85.6 h·nmol/g),

adrenal gland (medulla) (52.7 h·nmol/g), spleen (red pulp) (49.2

TABLE 3

Elemental compositions and mass spectral data of [3H]-DM1-LNL897 metabolites and derivatized DM1

Name LC Rt Full-MSa Type of Ion Elemental Composition Fragments Observed in MS/MS

min m/z m/z

M7 (cysteine conjugate) 7.0 (857) [M+H]+ C38H54N4O12S2Cl 795, 547, 485, 467, 435, 299, 140

M2b (LYS-MCC-DM1)9.0 1125 [M+Na]+ C53H76N6O15SCl 1085, 1041, 1023, 1009, 557, 547, 539, 529, 521, 511, 485, 467, 453, 4359.5 (1103) [M+H]+

M6 (maysine)13.0 569 [M+Na]+ C28H36N2O7Cl 529, 453, 425

(547) [M+H]+

M1b (MCC-DM1)

997 [M+Na]+ C47H64N4O14SCl 957, 913, 881, 547, 529, 485, 467, 453, 435, 411, 383, 326, 298, 280, 14014.0 (975) [M+H]+

14.5 957 [M+H]+-H2O913 [M+H]+-H2O-CO2

M5 (derivatized DM1)

15.0 885 [M+Na]+ C41H56N4O12SCl 845, 801, 547, 485, 453, 435, 299, 271, 214880 [M+NH4]

+

(863) [M+H]+

845 [M+H]+-H2O801 [M+H]+-H2O-CO2

M4 (DM1)

15.8 760 [M+Na]+ C35H49N3O10SCl 720, 676, 658, 644, 626, 547, 529, 485, 467, 453, 435, 174, 146(738) [M+H]+

720 [M+H]+-H2O676 [M+H]+-H2O-CO2

Rt, retention time.aMass to charge used for MS/MS fragmentation is shown in brackets.bMetabolites M1 and M2 consist of two diastereoisomers that result in the formation of double peaks.

TABLE 4

Concentrations (micromolar) and AUC values (micromolar/hour) of [3H]DM1-LNL897 in serum

Concentrations and pharmacokinetic parameters of intact ADCs, ADC complexes, and/or aggregates as well as free small molecule components in serum of tumor-bearing nude rats following a single intravenous administration of 10 mg/kg [3H]DM1-LNL897. Data were derived from SEC-LC with off-line radiodetection obtainedafter direct injection of serum.

Metabolite/ComponentSample Collection Time Exposure

1 Hour 24 Hours 96 Hours 168 Hours AUC0–168h Percentage of AUC0–168h AUCinf

mM mM mM mM mM·h % mM·h

Aggregates/complexesa 0.119 0.102 0.0558 0.0200 11.0 17.3 12.9LNL897 monomer 1.31 0.483 0.147 0.0457 51.6 81.0 54.4Total ADCb 1.42 0.664 0.268 0.133 73.4 NC 85.4Free small molecule componentsc 0.0294 0.0106 0.00472 0.00199 1.27 1.99 1.44Sum of additional metabolites ND ND ND NDTotal components detected 1.46 0.593 0.207 0.0679 63.7 100 68.2Total components in original sample 1.46 0.593 0.207 0.0679 63.7 100 68.2

NC, not calculated; ND, not detected.aTotal RA corresponding to ADC aggregates and/or complexes with endogenous proteins in the serum.bValues in italics denote conjugated antibody (LNL897 monomer + potential aggregates/complexes) concentrations measured by the ELISA method.cRA corresponding to small molecule–containing components of [3H]DM1-LNL897.

DME Properties of Antibody Maytansinoid Conjugates 903

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

h·nmol/g), liver (48.5 h·nmol/g), ovarian tissue (46.1 h·nmol/g), anduterus mucosa (43.7 h·nmol/g). In almost all tissues analyzed, exposurebased on AUClast was lower in comparison with blood (QWBA) (55.5h·nmol/g). The only exception was tumor tissue, where the exposurebased on AUClast was 1.54-fold higher than in blood; indicating cleardistribution into the tumor as the target tissue of the ADC.At 264 hours postdose, total radiolabeled components were still

detected in the analyzed tissues, which is in line with the high fraction ofRA determined in the carcass at the end of the excretion experiment(13.8%6 0.482% of the dose) (Table 8). Inmany tissues a half-life (T1/2)was calculated. The tissues showed varying half-lives ranging from 62.4hours for stomach (glandular) to 94.2 hours for eye (choroid).Metabolic Stability of [3H]-Label for DM1 and ADC. Since we

have synthesized radiolabeled ADC with high specific activity, theformation of tritiated water was retested and calculated according to Tseand Jaffe (1991). Total radiolabeled components measured in urinecondensates of individual rats after lyophilization indicated that a verylow fraction of the administered RA was transformed into tritiated water(0.826%–1.01% of the dose), indicating that the [3H] label showedacceptable metabolic stability (data not shown).

Characterization of [3H]DM1-LNL897 and Catabolites

In Fig. 1, the deconvolutedmass spectrum of [3H]DM1-LNL897 in thedosing solution is depicted. The ADC peaks were distributed in a m/z

range of 1800–3800 atomic mass units, corresponding to a charge-statedistribution of 40–80 charges on the intact [3H]DM1-LNL897 (data notshown). The resolving power of the quadrupole time of flight (approx-imately 9000) allowed baseline separation of the various glycosylationforms of the antibody with molecular weight differences of 160 Da(nominal mass) representing galactose units (Beck et al., 2013). Byapplying a collision energy of 45 eV in the trapping cell (MS/MSexperiments with collisional energy switching), two characteristicfragment ions at m/z 547 and 485 were observed, corresponding to intactmaysine andmaysine after loss of water and carbon dioxide, respectively,as described in Liu et al. (2005). The presence of these two signature ionsconfirmed that the detected proteins correspond to the [3H]DM1-LNL897since they are characteristic for the fragmentation of the payload DM1.In general, the metabolite identification was challenging due to the

low abundance of catabolites. The structural characterization of metab-olites in different matrices (serum, urine, feces, and tumor) was carriedout by LC-MS/MS and peaks were assigned as shown in Figs. 4–7.However, the poor ionization by electrospray of this class of DM1-containing species, combined with their very low concentration inbiologic samples, created significant technical challenges for theiridentification using LC-MS/MS. As a result, no mass spectral data couldbe acquired for components P1 and P1.8 (front peaks, Fig. 4), and as wellas for peaks summarized as unidentified residual RA in Fig. 6.Nevertheless, structural proposals and identification of metabolites suchas MCC-DM1 and LYS-MCC-DM1, which have been characterized

TABLE 5

Concentration and AUC data of catabolites in serum

Concentrations (nanomolar) and pharmacokinetic parameters of metabolites in serum of tumor-bearing nude rats following a single intravenous administration ofnominal 10 mg/kg [3H]DM1-LNL897. Data were derived from SEC-LC with off-line radiodetection obtained after injection of extracted serum.

Metabolite/ComponentSample Collection Time Exposure

1 Hour 24 Hours 96 Hours 168 Hours AUC0–168h Percentage of AUC0–168h AUCinf

nM nM nM nM nM·h % nM·h

P1 (front peak) 5.81 2.47 0.754 0.215 249 0.395 262P1.8 (front peak) 0.700 0.408 0.158 0.0321 40.3 0.0640 42.1M7 (cysteine conjugate) 1.58 0.515 0.166 0.0463 57.1 0.0907 59.9M2a (LYS-MCC-DM1) ND 0.408 0.232 0.0676 38.5 0.0611 42.5M6 (Maysine) 1.06 0.419 0.116 0.0248 41.9 0.0664 43.1M1a (MCC-DM1) 1.72 0.327 ND ND 36.2 0.0575M5 (derivatized DM1)b 16.3 2.93 0.821 0.189 400 0.635 410Sum of additional metabolitesc 2.23 0.199 0.230 0.0895 56.0 0.0888 60.6Total components detected 13.1 4.74 1.66 0.475 519 0.824 549Lost during sample processingd 1443 588 205 67.4 62,469 99.2 NCTotal components in original sample 1456 593 207 67.9 62,988 100 67,500

NC, not calculated; ND, not detected.aPeaks were identified by retention time only.bDM1-NEM measured after TCEP treatment is shown in italic (corresponds to possible disulfide-bonded DM1 with endogenous components in serum).cCorresponds to the sum of the residual RA measured in each radiochromatogram.dIncluding ADCs and/or aggregates/complexes.

TABLE 6

Concentrations of total radiolabeled components and metabolite M2 (LYS-MCC-DM1) in rat tumors

Amount (percentage of the dose) and concentrations (nanomolar/gram) of total radiolabeled components as well as concentrations(nanomolar/gram) of LNL897 and the principal metabolite M2 (LYS-MCC-DM1) in tumors at different time points after a singleintravenous administration of nominal 10 mg/kg [3H]DM1-LNL897. Data were derived from metabolic patterns obtained by SEC analysiswith off-line RA detection and UPLC/MS/off-line radio-detection for small molecules.

Rat Number Time Total Radiolabeled Component LNL897 Concentration M2 (LYS-MCC-DM1)

hour % nmol/ga nmol/ga nmol/gb nmol/gb

1 1 0.262 0.0858 0.0377 0.125 0.001592 24 0.590 0.282 0.0516 0.170 0.2633 72 0.351 0.244 0.0275 0.0907 0.3074 168 0.242 0.142 0.00935 0.0309 0.188

aConcentration expressed as protein.bConcentration expressed as DM1 conjugated (DAR of 3.3).

904 Walles et al.

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

before for ADCs using the same payload/linker combination (Ericksonet al., 2010; Sun et al., 2011; Shen et al., 2012a), were supported bycomparison with synthetic reference standards whenever possible. Othermetabolites such as M6 (maysine), which has been previously publishedas a disposition product of maytansine (Suchocki and Sneden, 1987) andM7 (cysteine conjugate), were characterized in serum samples followingprotein precipitation by LC-MS/MS (Fig. 4).In excreta samples (Fig. 6, urine; Fig. 7, feces) metabolitesM1 (MCC-

DM1) and M2 (LYS-MCC-DM1) were identified. Mass spectral data ofall characterized components are summarized in Table 3. Additionalinformation about the chemical structures of metabolites was acquiredby reduction of the disulfide bonds using TCEP and by alkylation of theresulting free thiol groups with NEM. This derivatization process waspreviously tested using a spiked rat serum with a DM1 dimer (positivecontrol) by LC-MS. The presence or absence of NEM-conjugated DM1in derivatized biologic samples indicated that DM1 was conjugatedto endogenous thiol-containing molecules via disulfide bonds. Thestructures of all identified disposition products of LNL897 are shownin Fig. 8.Metabolite Profiles in Serum. Size exclusion off-line radioprofiles

of [3H]DM1-LNL897 in serum ([3H]-DM1-MCC-Ab, Fig. 2) wererecorded up to 168 hours after dosing. Intact ADC was the mostabundant radiolabeled compound in serum (81.0% of [3H]-AUC0–168h

of total radiolabeled components). Additional peaks in front of the maincompound, suggesting the presence of soluble ADC aggregates and/orother ADC/protein complexes, together accounted for 17.3% of the[3H]-AUC0–168h. The sum of [3H]-AUC0–168h for the circulating intactADC monomer and aggregates accounted for 62.6 mM·h (Table 4),which is slightly lower than the 73.4 mM·h measured by the ELISAmethod (Table 1). The total radiolabeled components corresponding tofree small molecule components together accounted for only 1.99% of[3H]-AUC0–168h.The concentration of total radiolabeled components measured at 1, 24,

96, and 168 hours in serum supernatant samples following proteinprecipitation were 13.1, 4.74, 1.66, and 0.475 nM, which resulted in[3H]-AUC0–168h of 519 nM·h or 0.824% of detected small moleculecomponents (Table 5). These results suggested that free small moleculecomponents in serum represent only a minor proportion compared withthe intact ADC or other protein-bound DM1-containing components inprecipitated pellets. The metabolite profiles of the serum supernatantfractions (Fig. 4) showed two early eluting components: P1 and P1.8.For both of these, no mass spectral data could be acquired. Since thefront peaks were still present after drying and solvent evaporation, priorto off-line RA measurement, the presence of tritiated water in thefront peaks could be excluded. The [3H]-AUC0–168h for P1 and P1.8accounted for 0.395% and 0.0640% of total radiolabeled components(Table 5).MCC-DM1 (M1), LYS-MCC-DM1 (M2), maysine (M6), anda DM1-cysteine conjugate (M7) were identified by MS and accountedfor 0.0575%, 0.0611%, 0.0664%, and 0.0907% of [3H]-AUC0–168h,respectively. Additional unidentified peaks were observed in serumextracts and together accounted for 0.0888% of [3H]-AUC0–168h

(Table 5).After derivatization, a new peak corresponding to derivatized DM1

(M5, DM1-NEM) appeared (Fig. 4) and accounted for 0.635% ofAUC0–168h (Table 5). The radiochromatographic backgroundwas higherafter derivatization between 5 and 15 minutes postinjection, whichpotentially indicates that the antibody degraded during the procedure,forming artifacts. The recoveries of RA after extraction of untreatedserum samples were found to be 0.9%, 0.8%, 0.8%, and 0.7%, for timepoints 1, 24, 96, and 168 hours, respectively. For the TCEP-derivatizedsamples, the recoveries were 4.1%, 3.9%, 4.8%, and 5.2% at theequivalent time points.

Metabolite Profiles in Tumor. In tumor, the concentration of totalradiolabeled components was between 110 and 137 pmol·g21 for thefour rats. The derived concentrations for [3H]DM1-LNL897 and LYS-MCC-DM1 are listed in Table 6. Analysis of rat tumors collected after 1,24, 72, and 168 hours also demonstrated that the intact [3H]DM1-LNL897 was the major component at 1 hour postdose, increased slightlyat 24 hours postdose, and then decreased with time. The fractionrepresenting the small molecules increased from 1 to 72 hours postdose.After 168 hours postdose, the fraction representing the intact ADC wassmall compared with the fraction representing the small moleculecatabolites (Fig. 5). The metabolite profiles following protein pre-cipitation (Fig. 5) showed LYS-MCC-DM1 (M2) to be the majorcatabolite, which had a concentration in the tissue between 159 and 307pmol·g21 (Table 6). An additional minor radiolabeled component,eluting between 7 and 8minutes, was detected but could not be identifiedby MS due to the low concentration and the poor ionization byelectrospray. Radiochromatograms of tumor homogenates treated withTCEP (reducing agent) resulted in a new small peak eluting at 15minutes, which corresponds to the retention time of DM1 derivatizedwith NEM (data not shown).

Fig. 5. (A and B) Size exclusion radiochromatograms and high-performance liquidchromatography radioprofiles (embedded) of tumors homogenates at T1h and atT168h postdose after a single intravenous administration of nominal 10 mg/kg[3H]DM1-LNL897.

TABLE 7

Peak area of LYS-MCC-DM1 (M2) in tumor sections obtained with LESA-mLC-MS/MS

Plate SampleTime Postdose

1 Hour 24 Hours 72 Hours 168 Hours

cps cps cps cps

1 1 93 2550 3410 15501 2 0 2490 2800 16101 3 29 1740 2750 24302 4 0 2540 3010 17802 5 0 3620 1400 22202 6 0 3030 2280 1310

Mean 20 2662 2608 1817S.D. 37 626 697 427CV 184.1 23.5 26.7 23.5

DME Properties of Antibody Maytansinoid Conjugates 905

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

Concentrations of Catabolites in Tumor and Liver TissuesDetermined by LESA-mLC-MS/MS. Complimentary tumor dataanalyzed by LESA-mLC-MS/MS from tumors of different animals wereacquired and compared with RA measurements (Fig. 9). Three differentsampling locations on two separate slices could be probed and the resultssuggested a homogenous distribution of LYS-MCC-DM1 catabolite(Table 7). Although these tumors were collected from different animals,and measured by a different technique, the measured concentration dataare in alignment with the radioprofiling data (Fig. 9). The sensitivity ofLESA-mLC-MS/MS allowed specific detection of the main cataboliteLYS-MCC-DM1 (M2) in tumor sections.Metabolite Profiles in Urine. Analysis of rat urine (0–168 hours) by

SEC (data not shown) showed no peak corresponding to the retentiontime of the intact [3H]DM1-LNL897 (based on the retention time ofthe standard). The remaining part of excreted RA represented smallmolecule components, which were analyzed by UPLC/MS with off-lineRA detection in the presence and absence of reducing agent TCEP. Theresults showed that only a small fraction of the dosed RA, equivalent to4.15% between 0 and 168 hours (Table 8), was recovered in urine. Themetabolite profile in the urine pool showed M2 (LYS-MCC-DM1) asthemajor identified radiolabeled component (1.69% of the dose) andM7(DM1-cysteine conjugate) representing 0.621% of the dose (Table 8).Other minor radiolabeled components such as M1 (MCC-DM1) andM6(maysine) accounted for 0.0946% and 0.0265% of the dose, respec-tively. In addition to the aforementioned components, the radioprofileshowed unresolved baseline RA and a front peak (P1), which combinedaccounted for 1.41% of the dose (Fig. 6; Table 8). Treatment of the urinepool with TCEP resulted in a new small peak in the radiochromatograms(0.127% of the dose). This eluted after M1 and corresponded to DM1derivatized with NEM (M5).Metabolite Profiles in Feces. In feces, 84.5% of the dose was

excreted between 0 and 168 hours (Table 8). The most abundantmetabolites in feces were M1 (MCC-DM1), which accounted for 42.5%of the dose, andM2 (LYS-MCC-DM1), which represented 12.7% of thedose (see Fig. 7; Table 8). MetaboliteM7 (DM1-cysteine conjugate) wasidentified based on retention time and accounted for 1.96% of the dose.

Additional unknown radiolabeled components, such as front peak P1(0.320% of the dose) and P24 representing 1.79% of the dose, weredetected in rat feces between 0 and 168 hours. The sum of residualunresolved RA accounted for 16.8% of the dose.

Fig. 7. Metabolite profiles in feces. Off-line radiochromatograms in the (A) absenceand (B) presence of reducing agent TCEP of feces pools (0–168 hours) following asingle intravenous administration of nominal 10 mg/kg [3H]DM1-LNL897 to femaletumor-bearing nude rats (n = 3). Proteins contained in the feces pools werepreviously precipitated using acetonitrile and only the soluble metabolites wereprofiled by LC/MS with off-line RA detection.

Fig. 6. Metabolite profiles in urine. Off-line radiochromatograms in the (A) absenceand (B) presence of reducing agent TCEP of urine pools (0–168 hours) following asingle intravenous administration of nominal 10 mg/kg [3H]DM1-LNL897 to femaletumor-bearing nude rats (n = 3). Proteins contained in the urine pools werepreviously precipitated using acetonitrile and only the soluble metabolites wereprofiled by LC/MS with off-line RA detection.

TABLE 8

Amounts (percentage of the dose) of LNL897 catabolites excreted in urine and fecesof rats after a single intravenous administration of nominal 10 mg/kg [3H]DM1-

LNL897

Data were derived from metabolic patterns obtained by LC analysis with off-line RA detection(pools of n = 3 rats).

Metabolite/ComponentExcretion (Percentage of the Dose)

Urine Feces Total Excretion

Excretion period (hours) 0–168 0–168 0–168P1 (front peak) 0.0943 0.320 0.414M7 (cysteine conjugate) 0.621 1.96 2.58M2 (LYS-MCC-DM1) 1.69 12.7 14.4M6 (maysine) 0.0265 ND 0.0265M1 (MCC-DM1) 0.0946 42.5 42.6P24 ND 1.79 1.79Sum of additional componentsa 1.32 16.8 18.2Total components detected 3.85 76.1 80.0Lost during sample processing 0.303 8.37 8.67Total components in original sample 4.15b 84.5c 88.7Carcass 13.8d

Cage wash 2.55e

ND, not detected.aCorresponds to the sum of the residual RA measured in each radiochromatogram.bS.D. was 0.462; CV was 11.1%.cS.D. was 3.12; CV was 3.69%.dMeasured after 168 hours; S.D. was 0.482; CV was 3.49%.eMeasured after 168 hours; S.D. was 0.818; CV was 32%.

906 Walles et al.

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

Elimination of [3H]DM1-LNL897

Analysis of rat excreta (urine and feces extracts) by SEC (data notshown), showed negligible peaks corresponding to the retention time ofthe intact [3H]DM1-LNL897 (based on the retention time of the standard).The remaining RA excreted represented metabolites. Hence, metabolism/catabolism of [3H]DM1-LNL897 is one major elimination pathway. Theprimarymetabolic pathways of [3H]DM1-LNL897were cleavage before theterminal LYS leading to metabolite M2 (LYS-MCC-DM1) and cleavageafter the terminal LYS leading to M1 (MCC-DM1). Additional minormetabolites were characterized in urine and feces such as the cysteineconjugate (M7) andM6 (detected in serum and urine only) corresponding tomaysine (Fig. 8). Several additional minor components could be detected(P1, P24, and unresolved baseline RA) (see Fig. 6; Table 6), but could not becharacterized with the applied analytical methods. No free DM1 or DM1dimer could be detected in untreated excreta samples.

Excretion of RA and Mass Balance

Within 168 hours after administration, 4.15%6 0.462% and 84.5%63.12% (Table 8) of the administered RAwas excreted in urine and feces,respectively. At the end of the experiment, i.e., at 168 hours postdose,13.8%6 0.482% of the administered RAwas determined in the carcass.Together with the RA recovered in the cage wash (2.55%6 0.818%) therecovery of RA was complete (105% 6 3.27%).

Discussion

This study presents the DME properties of the rat cross-reactive[3H] DM1-LNL897 ADC after intravenous dosing to female tumor-bearing nude rats.Pharmacokinetics and Serum Profiles. The pharmacokinetic

parameters of the radiolabeled ADC were determined by different

Fig. 8. Identified disposition products of[3H]DM1-LNL897.

DME Properties of Antibody Maytansinoid Conjugates 907

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

analytical methods (ELISA, SEC, and MS) as described by Shen et al.(2012a). By comparison of AUClast of total radiolabeled componentswith ELISA data (Table 1), it is obvious that the total radiolabeledcomponents mainly reflect the conjugated antibody, suggesting that theMCC linker of the ADCwas stable in circulation as previously reported,which is different from previous observed results with trastuzumab-MCC-DM1 maytansinoid conjugate (Shen et al., 2012a). In our opinionpossible changes in the manufacturing process could contribute to thesefindings; however, further investigations are needed. The high stabilityof LNL897 in circulation is further supported by catabolite profiling andrecovery measurements, which were below 1% for all time points. NoDM1 could be directly detected at 1 hour postdose and about 16.3 nM ofthe DM1 derivate (DM1-NEM, M5; data not shown) could be detectedat the same time point after derivatization, which indicates that DM1waslikely bound to endogenous substrates (glutathione and cysteine) orproteins via S-S bonds. The treatments of ADC dosing solution withincreasing concentrations of TCEP have shown increased release ofDM1. At the same time, the control analyses have shown that stability ofthe thioether bond is not affected by the presence of TCEP and that onlyspecies with disulfide bonds are affected. Based on this experiment, weconcluded that minor amounts of DM1 are directly bound to cysteines.This cysteine-bound DM1 species could be released as cysteine-DM1conjugates (such as M7) after lysosomal processing. However, gluta-thione conjugates of DM1 after intravenous dosing of DM1 to rats havebeen reported recently (Shen et al., 2015). Further degradation ofglutathione conjugates to cysteine could also lead to formation of M7.Compared with the ELISA data, the SEC-RA profiles provided more

insights about the circulating components. As shown in Fig. 2, SEC-RAallows the differentiation between ADC and higher molecular weightaggregates/complexes, while for the ELISA it is unclear how theseadditional aggregates/complexes impact the detection. When the mainADC peak and earlier eluting small peaks were summed, serum concen-trations were approximately the same as those measured by the ELISAmethod (Fig. 2; Tables 1 and 4).With SEC-RA profiles, it is only possible to distinguish between

fractions of large molecules and small molecules. For the late elutingpeaks, which are attributed to small molecules, it is possible that thesepeaks contain multiple coeluting components. Therefore, the superna-tants after protein precipitation were injected and reprofiled on ananalytical LC-MS system.The metabolite profiles in serum and urine showed qualitative and

quantitative differences depending on whether or not the previouslydescribed derivatization method was applied (Figs. 4 and 6; Table 5).A front peak as well as metabolites M7, M6, and M1 were detectedwith both sample preparation methods, whereas M5 was only detectedafter applying the derivatization method. It is worth mentioning thatonly limited control experiments were conducted for the distribution,metabolism, and excretion studies. Since possible S-S bond formationwas identified as the main liability, the sample methods without TCEPwere tested first with the DM1 dimer to make sure not to cause anyreduction. A front peak was also detected during the distribution,metabolism, and excretion study in rats for the trastuzumab-MCC-DM1 maytansinoid conjugate and was attributed to tritiated water(Shen et al., 2012a). Since our samples were dried during processing,the front peak cannot be fully attributed to tritiated water. Hence, thepresence of further polar catabolites eluting in the front peak cannot beexcluded.Under the conditions of derivatization, M7 could form DM1-NEM

(M5). M5 was only detected after derivatization, and the recoveries ofRA from serum were also slightly higher (up to 5%), which couldindicate that protein-bound DM1 was released by the derivatizationmethod. On the other hand, it cannot be excluded that the derivatization

method destroys antibodies to a minor extent, which could also lead to aslightly higher amount of RA in the solution.Distribution. In previous studies, either tritium labels on the payload

with lower specific activity (Shen at al., 2012a) or with 125I labels on theantibody were synthesized, of which only the iodinated material wasused to perform QWBA studies (Saad et al., 2015). In our study, theconfirmed stability of the ADC in the dosing solution allowed for theQWBA experiment to be performed with a tritium label on the DM1payload.In rats, P-cadherin is mainly expressed in tumors and tissues such as

the tongue, skin, esophagus, and bladder (internal data not shown). Afterintravenous dosing of the ADC, the tissue distribution as determined byQWBAwas comparably low in all tissues analyzedwith the exception oftumor tissue, where the highest AUClast value was determined to be1.54-fold higher than that in blood (Table 2). In the remaining tissues,the exposure based on AUClast was 0.184- to 0.950-fold higher than theAUClast value of blood. This demonstrates a clear distribution of ADC-related radiolabeled components into the tumor as the target tissue. Theexposure in other tissues also indicates that non-target–specific distri-bution occurs, which is also nicely illustrated in Fig. 3. Besides tumor,there was no accumulation or tissue retention of RA over time in any ofthe tissues analyzed (data not shown). Similar to small molecules,biotherapeutics are also cleared through kidney and liver (Hamuro andKishnani, 2012). Although these tissues are highlighted in the depictedQWBA pictures, it needs to be recognized that the organs are also highlyperfused by blood, which contains large amounts of ADC (Fig. 3).However, since QWBA data only show the distribution of total radio-labeled components and only the DM1 part of ADC was radiolabeled,it is unclear whether the observed RA in tissues could be attributed tointact ADC, released DM1, or another catabolite. Therefore, furtherinvestigations were conducted.Figure 5 shows the tumor profiles acquired by SEC-RA and it is

obvious that after 1 hour tumor tissuemainly containedADC, while after24 hours the catabolite LYS-MCC-DM1 (M2) was the predominantcomponent (Table 6). To see if the catabolite M2 was evenly distributedin tumor tissue, the newly developed LESA-mLC-MS/MS (Lanshoeftet al., 2016) was applied. The acquired quantitative LESA data showedno significant difference over tumor tissue (Table 7) and the measuredconcentration data were well in line with the results fromRA profiling intissues (Fig. 9), considering that analyzed tumor samples were collectedfrom different rats with different tumor sizes.In line with previous results, the LYS-MCC-DM1 catabolite was

confirmed as the main component in the liver (data not shown) byapplying the LESA-mLC-MS/MS method on QWBA slices, which is in

Fig. 9. Comparison of tumor concentration data acquired by LC with RA detectionand LESA-mLC-MS/MS.

908 Walles et al.

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

line with early observations (Erickson and Lambert, 2012). The mainadvantages of the LESA-mLC-MS/MS method are that no radiolabel isultimately needed and no sample processing is required, thus it can providecomplimentary data. For quantitative assessments without radiolabel, morethorough validation using internal standards to correct for different tissueresponses would need to be applied (Lanshoeft et al., 2016).Elimination. Similar to serum, differences in the catabolite profiles

in urine were observed with and without derivatization. In untreatedurine LYS-MCC-DM1 (M2) and the cysteine-conjugated DM1 (M7)were the main components, followed by maysine (M6) and MCC-DM1(M1). After applying the derivatization method, the peak correspondingto M7 was reduced and M5 appeared in the sample (Fig. 6). This couldindicate that either the derivatization was not complete or that furtheruncharacterized components coelute under the M7 peak. These findingsare in line with previous observations (Shen et al., 2012a), except thatmaysine (M6) as well as the cysteine conjugate of DM1 (M7) have notbeen reported before. In a recent rat in vivo study with DM1, maytansinolhas been reported as a catabolite (Shen et al., 2015), but it remains unclearif the detected maysine is a downstream product of maytansinol.The presence of MCC-DM1 in plasma and excreta suggests that this is

probably cleaved off in plasma or other tissues from the ADC. Differentmechanisms for hydrolysis of the maleimide ring, which could lead toloss of DM1, have been recently described for cysteine-linked ADCs(Shen et al., 2012b), but not for LYS-linked ADCs. It also cannot beexcluded during synthesis that reactive N-Hydroxysuccinimid esters canreact and couple with other amino acids such as serine and threonine(Hermanson, 2013), which could lead to formation and release of MCC-DM1. Further investigations are required to investigate how MCC-DM1is released.The elimination of LNL897 was moderately fast and the results

were aligned with previously reported results (Shen et al., 2012a).After 1 week, a complete mass balance was obtained (105% 6 3%);88.7% of the dose was detected in the collected excreta and 13.8% inthe carcass (data not shown). To obtain a complete excretion profile(to reduce the amount of the dose retained in the carcass), a longersampling time up to 2 weeks is recommended for future DME studiesof ADCs.

Conclusions

We have demonstrated that the use of a tritium label on the payload ofan ADC is suitable even for longer-term DME and QWBA studies. Wehave also shown that a diverse set of analytical methods is required toinvestigate all DME properties of an ADC and that the combination ofQWBA and LESA-mLC-MS/MS is a powerful tool, which can give newinsights into the distribution of these constructs. Label-free distributionstudies may also be possible in the future if the quantitative abilities ofthe LESA method can be further improved. For metabolism investiga-tions of antibody maytansoid conjugates, it is important to pay attentionto the sample preparation method, especially to the use of reductiveagents, since this has an impact on the observed catabolite profile. In thisstudy, we identified a DM1-cysteine conjugate and maysine as addi-tional catabolites. MCC-DM1 was also identified as a major componentin excreta. This suggests that there must be additional mechanismspresent that cause its release, which need to be further investigated. Themethods presented provide useful perspectives for the further develop-ment of in vivo applications of ADCs as therapeutics.

Acknowledgments

The authors sincerely thank Wayne Widdison, XiuXia Sun, and Kate Lai ofImmunoGen, Inc., for discussions on the study design of radiolabeled synthesisand support for release analysis. The authors also acknowledge Prakash Mistry

from Oncology, Basel, for implementing the tumors in female rats; AlexanderDavid James for thorough review of the manuscript; and Wolfgang Marterer forgiving insight into the synthetic strategies.

Authorship ContributionsParticipated in research design: Rudolph, Walles, Wolf, Peraus, Heudi,

Bilic, Moenius.Conducted experiments: Wolf, Lanshoeft, Bourgailh, Suetterlin.Contributed with new reagents or analytic tools: Elbast, Lanshoeft, Moenius.Performed data analysis: Rudolph, Walles, Wolf.Wrote or contributed to the writing of the manuscript: Walles, Wolf,

Rudolph, Heudi, Lanshoeft, Elbast.

References

Albergaria A, Ribeiro AS, Vieira AF, Sousa B, Nobre AR, Seruca R, Schmitt F, and Paredes J(2011) P-cadherin role in normal breast development and cancer. Int J Dev Biol 55:811–822.

Beck A, Terral G, Debaene F, Wagner-Rousset E, Marcoux J, Janin-Bussat MC, Colas O, Dorsselaer AV,and Cianférani S (2016) Cutting-edge mass spectrometry methods for the multi-level structural char-acterization of antibody-drug conjugates. Expert Rev Proteomics 13:157–183.

Beck A, Wagner-Rousset E, Ayoub D, Van Dorsselaer A, and Sanglier-Cianférani S (2013)Characterization of therapeutic antibodies and related products. Anal Chem 85:715–736.

Davis JA, Rock DA, Wienkers LC, and Pearson JT (2012) In vitro characterization of the drug-druginteraction potential of catabolites of antibody-maytansinoid conjugates. Drug Metab Dispos 40:1927–1934.

de Claro RA, McGinn K, Kwitkowski V, Bullock J, Khandelwal A, Habtemariam B, Ouyang Y,Saber H, Lee K, and Koti K, et al. (2012) U.S. Food and Drug Administration approvalsummary: brentuximab vedotin for the treatment of relapsed Hodgkin lymphoma or relapsedsystemic anaplastic large-cell lymphoma. Clin Cancer Res 18:5845–5849.

Deslandes A (2014) Comparative clinical pharmacokinetics of antibody-drug conjugates in first-in-human phase 1 studies. MAbs 6:859–870.

Erickson HK and Lambert JM (2012) ADME of antibody-maytansinoid conjugates. AAPS J 14:799–805.

Erickson HK, Widdison WC, Mayo MF, Whiteman K, Audette C, Wilhelm SD, and Singh R(2010) Tumor delivery and in vivo processing of disulfide-linked and thioether-linked antibody-maytansinoid conjugates. Bioconjug Chem 21:84–92.

Hamuro LL and Kishnani NS (2012) Metabolism of biologics: biotherapeutic proteins. Bioanalysis4:189–195.

Han TH and Zhao B (2014) Absorption, distribution, metabolism, and excretion considerations forthe development of antibody-drug conjugates. Drug Metab Dispos 42:1914–1920.

Hermanson GT (2013) The reactions of bioconjugations, in Bioconjugate Techniques, (Audet J,Preap M, East K, and Halterman K, eds) pp 229–258, Academic Press/Elsevier, London.

Kaur S, Xu K, Saad OM, Dere RC, and Carrasco-Triguero M (2013) Bioanalytical assay strategiesfor the development of antibody-drug conjugate biotherapeutics. Bioanalysis 5:201–226.

Kraynov E, Kamath AV, Walles M, Tarcsa E, Deslandes A, Iyer RA, Datta-Mannan A, Sriraman P,Bairlein M, and Yang JJ, et al. (2016) Current approaches for absorption, distribution, metab-olism, and excretion characterization of antibody-drug conjugates: an industry white paper. DrugMetab Dispos 44:617–623.

Lanshoeft C, Stutz G, Elbast W, Wolf T, Walles M, Stoeckli M, Picard F, and Kretz O (2016)Analysis of small molecule antibody-drug conjugate catabolites in rat liver and tumor tissue byliquid extraction surface analysis micro-capillary liquid chromatography/tandem mass spec-trometry. Rapid Commun Mass Spectrom 30:823–832.

Liu Z, Floss HG, Cassady JM, and Chan KK (2005) Metabolism studies of the anti-tumor agentmaytansine and its analog ansamitocin P-3 using liquid chromatography/tandem mass spec-trometry. J Mass Spectrom 40:389–399.

Marcoux J, Champion T, Colas O, Wagner-Rousset E, Corvaïa N, Van Dorsselaer A, Beck A,and Cianférani S (2015) Native mass spectrometry and ion mobility characterization oftrastuzumab emtansine, a lysine-linked antibody drug conjugate. Protein Sci 24:1210–1223.

Menezes D, Abrams TJ, Karim C, Tang Y, Chi Y, Miller K, Fanton C, Ghoddusi M, Wang Z,and Patawaran M, et al. (2015) Abstract 1682: Development and activity of a novel antibody-drug conjugate for the treatment of P-cadherin expressing cancers. Cancer Res 75:1682.

Poon KA, Flagella K, Beyer J, Tibbitts J, Kaur S, Saad O, Yi JH, Girish S, Dybdal N, and ReynoldsT (2013) Preclinical safety profile of trastuzumab emtansine (T-DM1): mechanism of action of itscytotoxic component retained with improved tolerability. Toxicol Appl Pharmacol 273:298–313.

Saad OM, Shen BQ, Xu K, Khojasteh SC, Girish S, and Kaur S (2015) Bioanalytical approachesfor characterizing catabolism of antibody-drug conjugates. Bioanalysis 7:1583–1604.

Schweitzer A, Fahr A, and Niederberger W (1987) A simple method for the quantitation of 14C-whole-body autoradiograms. Int J Rad Appl Instrum [A] 38:329–333.

Shen BQ, Bumbaca D, Saad O, Yue Q, Pastuskovas CV, Khojasteh SC, Tibbitts J, Kaur S,Wang B, and Chu YW, et al. (2012a) Catabolic fate and pharmacokinetic characterization oftrastuzumab emtansine (T-DM1): an emphasis on preclinical and clinical catabolism. CurrDrug Metab 13:901–910.

Shen BQ, Bumbaca D, Yue Q, Saad O, Tibbitts J, Khojasteh SC, and Girish S (2015) Non-clinicaldisposition and metabolism of DM1, a component of trastuzumab emtansine (T-DM1), inSprague-Dawley rats. Drug Metab Lett 9:119–131.

Shen BQ, Xu K, Liu L, Raab H, Bhakta S, Kenrick M, Parsons-Reponte KL, Tien J, Yu SF,and Mai E, et al. (2012b) Conjugation site modulates the in vivo stability and therapeutic activityof antibody-drug conjugates. Nat Biotechnol 30:184–189.

Suchocki JA and Sneden AT (1987) Characterization of decomposition products of maytansine.J Pharm Sci 76:738–743.

Sun X, Widdison W, Mayo M, Wilhelm S, Leece B, Chari R, Singh R, and Erickson H (2011)Design of antibody-maytansinoid conjugates allows for efficient detoxification via liver me-tabolism. Bioconjug Chem 22:728–735.

Thudium K, Bilic S, Leipold D, Mallet W, Kaur S, Meibohm B, Erickson H, Tibbitts J, Zhao H,and Gupta M (2013) American Association of Pharmaceutical Scientists National Biotechnology

DME Properties of Antibody Maytansinoid Conjugates 909

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from

Conference Short Course: Translational Challenges in Developing Antibody-Drug Conjugates:May 24, 2012, San Diego, CA. MAbs 5:5–12.

Tse FLS and Jaffe JM (1991) Use of radioactivity in drug disposition studies, in Preclinical DrugDisposition, Marcel Dekker, New York.

van der Heiden PL, Jedema I, Willemze R, and Barge RM (2006) Efficacy and toxicity ofgemtuzumab ozogamicin in patients with acute myeloid leukemia. Eur J Haematol 76:409–413.

Wada R, Erickson HK, Lewis Phillips GD, Provenzano CA, Leipold DD, Mai E, Johnson H,and Tibbitts J (2014) Mechanistic pharmacokinetic/pharmacodynamic modeling of in vivo tumoruptake, catabolism, and tumor response of trastuzumab maytansinoid conjugates. Cancer Che-mother Pharmacol 74:969–980.

Widdison W, Wilhelm S, Veale K, Costoplus J, Jones G, Audette C, Leece B, Bartle L, Kovtun Y,and Chari R (2015) Metabolites of antibody-maytansinoid conjugates: characteristics and in vitropotencies. Mol Pharm 12:1762–1773.

Xie H and Blättler WA (2006) In vivo behaviour of antibody-drug conjugates for the targetedtreatment of cancer. Expert Opin Biol Ther 6:281–291.

Yan H, Endo Y, Shen Y, Rotstein D, Dokmanovic M, Mohan N, Mukhopadhyay P, Gao B, PacherP, and Wu WJ (2016) Ado-trastuzumab emtansine targets hepatocytes via human epidermalgrowth factor receptor 2 to induce hepatotoxicity. Mol Cancer Ther 15:480–490.

Address correspondence to: Dr. Markus Walles, Novartis Pharma AG, NovartisInstitutes for Biomedical Research, Drug Metabolism and Pharmacokinetics,Fabrikstrasse 14, 1.12, CH-4002 Basel, Switzerland. E-mail: markus.walles@novartis.com

910 Walles et al.

at ASPE

T Journals on February 24, 2021

dmd.aspetjournals.org

Dow

nloaded from