Hydrophobic interaction chromatography (HIC) can be used for the determination of drug-to-antibody ratio (DAR) and to monitor drug distribution for cysteine-linked ADCs; molecules with more drug conjugated exhibit increased hydrophobicity [59, 146]. DAR calculations can also be performed following the reduction of the ADC and the analysis of the composite LC and HC by RPLC with UV detection [147, 148]. MS, however, is currently emerging as a new method of choice for ADC characterization and evaluation of DAR due to the additional structural information offered [21, 56, 84, 149, 150].
In the case of lysine- or engineered cysteine-conjugated ADCs (types 1 and 2, respectively), the interchain disulfide bonds between the HC and LC of the antibody remain intact. This means that DAR can be determined using LC-MS methods employing organic mobile phases [63, 151]. ADCs conjugated at the interchain cysteine residues produce a mixture of noncovalent mAb tetramers (two LC and two HC with a variable number of drug molecules attached) after reduction of interchain disulfide bonds; hence the application of more classic LC-MS analytical strategies would result in the dissociation of this noncovalent ADC. The organic solvents used disrupt the noncovalent nature of the ADC structure producing conjugated LC and HC fragments, making it hard to define average DAR . Careful sample preparation needs to be employed for the analysis of interchain cysteine-linked ADCs, especially if using denaturing conditions, to avoid ADC denaturation, which in turn could affect average DAR calculations.
There are a few different methods reported in the literature that employ native MS for the intact mass and DAR analyses of these cysteine-linked ADCs [56, 153, 154]. In 2012, Valliere-Douglass et al.  presented the first method for the rapid determination of intact mass and DAR for interchain cysteine- conjugated ADCs. Following ESI-MS analysis of deglycosylated ADCs, the authors observed some dissociation of the noncovalent ADCs into conjugated LC and HC but reported levels sufficiently low enough to not impact the relative levels of the intact drug-loaded species. The results acquired by MS were similar to those obtained from the orthogonal HIC method.
In 2013, Chen et al.  reported an MS method engaging limited enzymatic digestion, nESI, and native MS for the direct determination of intact mass and calculation of the average DAR of cysteine-linked ADCs. Utilizing an nESI source provided improved ionization, sensitivity, and increased confidence in DAR determination. The cytotoxic drug monomethyl auristatin E (MMAE) is reported to have high hydrophobicity. This meant that high-drug-load species were underestimated due to reduced affinity for protons causing inefficient ionization. To minimize this ion suppression and equalize ionization efficiency among species, proteolytic drug removal was induced. The hydrophobic moiety was selectively cleaved, using cysteine protease, from the ADC while leaving the linker attached indicative of the original drug load. The DAR values obtained following enzymatic digestion were more comparable with those determined by HIC methods.
Hengel et al.  have described a customized sample preparation method for cysteine-linked ADCs from an in vivo source using native SEC LC-MS for accurate drug load distribution determination. The results confirmed that higher-drug-loaded species were undetectable shortly after dosing; however, after 30 min the predominant species was the ADC with four conjugated drug molecules. The observed shift was likely due to clearance of the higher-loaded species and/or deconjugation. The MS dimension of the analysis was also capable of quantifying the in vivo increase in ADC mass heterogeneity. Overall this method provides useful PK and stability insights into the in vivo changes in drug load distribution and ADC structure.
Shortly afterward, Debaene et al.  developed a semiquantitative method for the determination of average DAR and DAR distributions based on native IM-MS data. The authors utilized TWIMS to conformationally characterize and separate each drug-loaded species and observed constant drift time/CCS shifts between consecutive DAR variants. This indicated that no conformational changes were induced upon drug binding and that all shifts were attributable to changes in mass. By plotting the intensities for each drift peak for the different drug binding stoichiometries against charge state, the areas under each curve were found to be representative of the relative abundances of the different ADC species present. This enabled the calculation of an average DAR value with excellent agreement toward HIC and native MS obtained data as well as accurate DAR distribution profiling.
An Orbitrap Exactive® mass spectrometer with extended mass range (EMR) [Thermo/Finnigan] enabled the Heck group to isolate native ADCs as part of tandem MS/MS experiments for localizing bound drug moieties . The cysteine-linked ADC brentuximab vedotin was studied and the authors present a direct approach for the assessment of drug distribution and localization. In the same paper, insights into hexamerization of IgG1 mutant assemblies (the mechanism by which IgG mAbs facilitate complementary activation) were reported. Their tandem MS/MS workflow enabled insight into the stoichiometry of antigen binding, even for protein assemblies with molecular weights over 1 million Da, and concluded that IgG hexamerization does not significantly affect bivalent antigen binding. These results were expected to an extent as IgG hexamer interactions have previously been shown to be confined to the Fc region .
Charge reducing agents have been used to reduce spectral complexity. Recently, Pacholarz et al.  have used the charge reducing agent triethyl- ammonium acetate (TEAA) to help preserve intact mAb structure during
ADC analysis and DAR calculation (paper submitted). Marcoux et al.  also used native MS in combination with the charge state reducing reagent imidazole to reduce the charge states observed for the lysine-linked ADC tras- tuzumab emtansine (T-DM1) to below 20+. With the addition of imidazole and use of a high-resolution Orbitrap® mass spectrometer, the authors were able to avoid the overlapping of peaks corresponding to the D0 and D8 species of sequential charge states (zero and eight drug moieties linked, respectively), even without deglycosylation. It was emphasized also that DAR calculations should be performed from raw data to avoid the introduction of bias upon the application of deconvolution parameters. A solution to this type of bias was recently published by Firth et al.  whereby use of enzymatic digestions meant that default deconvolution parameters could be applied to the raw data. MaxEnt1 (UNIFI, Waters) parameters, applied to data for intact or fragmented proteins greater than 40 kDa, can require significant manipulation, which often leads to skewed results that would invariably affect DAR calculations. However, the addition of middle-up enzymatic cleavage steps using IdeS and DTT reduction meant that three different fragments (all ~25 kDa) for thiol-linked ADCs were generated and then analyzed using their in-house LC-MS method with minimal chromatographic separation. Analysis of LC, Fc, and Fd fragments meant that default parameters could be used, which offered mass accuracy within less than 20 ppm (<1 Da).
In terms of MS developments for ADC analysis, the most recently published method was proposed by Gautier et al.  who demonstrated the use of a tandem mass tag (TMT) to identify “hotspot” lysine residues on IgG1 mAbs. TMT labeling relies upon the same chemistry as drug conjugation in ADCs, and therefore this approach enables straightforward identification of mAbs suitable for lysine-linked ADC development. Attachment of the TMT label was confirmed using LC-MS/MS, and the comparison of glycosylated and deglycosylated IgG1s concluded that glycan presence can sterically protect lysine residues from TMT labeling and hence block cytotoxic drug conjugation. This technique could provide fundamental information with regard to controlling drug load and specificity within lysine-linked ADCs.