Cancer is a disease characterised by unregulated cell growth that ultimately leads to the development of tumours comprised of a solid mass of cancer cells. Primary tumours can become life-threatening once they affect the functioning of vessels or organs that they impact.¹  However, death usually results from the spread of a primary tumour to distant parts of the body, a process known as metastasis. Haematological tumours have a different presentation, with large numbers of white blood cells released into the blood and other body compartments rather than formation of a solid tumour mass.

In 2015, Cancer Research UK confirmed that there were more than 360 000 new cases of cancer in the UK, translating to 990 cases being diagnosed every day.²  More than one in two individuals will develop some form of cancer during their lifetime, and one in four will die from the disease. The 20 most commonly diagnosed cancers in the UK and their associated mortality rates are shown in Figure 1.1.² 

The history of development of cancer treatments began in the nineteenth century with surgery as the preferred option. The introduction of the nitrogen mustards in the 1940s, followed closely by the development of radiation therapy, significantly improved clinical outcome. Surgery is still the main treatment for cancer, provided it has not spread to neighbouring tissues or beyond, and if the tumour has well-defined margins and is reasonably accessible. In many cases, surgery is used in combination with other treatments such as chemotherapy or radiotherapy.³ 

The term chemotherapy was first coined by Paul Ehrlich in the early 1900s. He was at the time contributing to the development of treatments for syphilis and worked on a series of compounds which were used to treat this disease as well as cancer. The nitrogen mustards were the first agents to be used clinically, and this was followed closely by discovery of the folate analogues, first aminopterin and then amethopterin (methotrexate).⁴  Methotrexate, as a single agent, showed great promise as an antitumour treatment for various epithelial malignancies but also showed remarkable activity in the treatment of choriocarcinoma, which was the first solid tumour to be cured by drug therapy.⁴  In the early 1950s, the antitumour effect of the Vinca alkaloids was discovered and their ability to inhibit microtubule polymerisation and cell division revealed.⁵  This led to the introduction of combination therapies that began with a co-administration of 6-mercaptopurine (Purinethol®), vincristine (Oncovin®), methotrexate and prednisone (POMP) that could induce long-term remissions in children with acute lymphoblastic leukaemia (ALL).⁶  Hormone therapies, such as tamoxifen (Nolvadex®) and the aromatase inhibitors, were introduced in the late 1950s to treat or prevent the further reoccurrence of breast cancer. This discovery led to the development of drugs that block male hormones (e.g., finasteride, Proscar®) and are effective for the treatment of prostate cancer. Hormone-based agents, such as the aromatase inhibitors and the luteinizing hormone-releasing hormone (LHRH) analogues and inhibitors, have significantly changed the way that prostate and breast cancers are treated.⁷  Chemotherapy then rapidly evolved through the early 1970s and beyond, moving on to families of targeted agents, such as the kinase inhibitors, which are much more selective than older generations of cytotoxic drugs and are characterised by lower side-effect profiles. A good example is the development of the kinase inhibitor imatinib (Gleevec®) in the late 1990s that provides very high response rates with manageable side effects in patients with chronic-phase Philadelphia-chromosome-positive chronic myeloid leukaemia (CML).¹ 

The advent of hybridoma technology in 1975 enabled the production of monoclonal antibodies. Owing to their origins in mice, these monoclonal antibodies were typically immunogenic in humans and were poor at inducing human immune effector responses, thereby limiting their clinical applicability. Later advances in antibody engineering provided flexible platforms for the development of chimeric, humanized and, more recently, fully human monoclonal antibodies which satisfactorily addressed many of these problems.⁸ 

Following on from these developments, more than 20 monoclonal antibodies (mAbs) have been approved for use in many indications, including cancer.⁹  Currently, there are also four antibody-drug conjugates (ADCs) and two immunotoxins approved for clinical use which harness the specificity of a monoclonal antibody (mAb) to target the delivery of a cytotoxic agent to a tumour, and these are described in more detail below.¹⁰  Immunotherapy is one of the latest types of cancer treatment, designed to boost the body's natural defences to fight cancer. During recent decades, many new approaches to immunotherapy have been investigated. Although initially unsuccessful, in the past few years significant progress has been made in the immunological treatment of certain cancer types including melanoma, renal and lung tumours by the antibody-based programmed cell death 1 (PD-1), programmed death ligand 1 (PD-L1) and cytotoxic T-lymphocyte antigen 4 (CTLA-4) checkpoint inhibitors such as nivolumab (Opdivo®), ipilimumab (Yervoy®), pembrolizumab (Keytruda®) and atezolizumab (Tecentriq®).These recent successes in immunotherapy provide great promise for the future in the development of new types of targeted agents that act on specific proteins or receptors. However, the major challenge in the use of all types of therapeutic agents is that the tumour cells develop resistance. The advantage of immunotherapy agents is that the immune system has the capacity to remember antigens and is capable of attacking tumour cells even if they re-emerge many years later, which is a significant advantage.¹ 

ADCs are mAbs attached to biologically active agents through chemical linkers. Currently, four ADCs have been approved by the US Food and Drug Administration (FDA): gemtuzumab ozogamicin (Mylotarg®), brentuximab vedotin (Adcetris®), ado-trastuzumab emtansine (Kadcyla®) and inotuzumab ozogamicin (Besponsa®); along with two immunotoxins [moxetumomab pasudotox (Lumoxiti®) and tagraxofusp (Elzonris®)] (Table 1.1).

The field of ADC discovery and development is expanding rapidly. At the time of writing there are over 70 ADCs at the clinical stage, the majority based on auristatin-, maytansin- and DNA-interactive payloads (Figure 1.2).

Interest in ADCs has changed dramatically over the past few decades. Historically, anticancer drug discovery focused solely on small-molecule chemotherapeutic agents, for example folate analogues, such as methotrexate, or DNA-damaging agents, such as the nitrogen mustards. These agents target rapidly dividing cancer cells, but at the same time other healthy dividing cells in the body are affected, causing severe side effects and limiting the administrable dose, thus narrowing the therapeutic window. In order to address this issue, one pathway explored by researchers was the development of ADCs, involving the use of antibodies to deliver highly cytotoxic agents directly to tumour cells without affecting other dividing cells in the body. Theoretically, this concept should provide a significant increase in the therapeutic window compared with chemotherapy and radiation (Figure 1.3).

The concept of ADCs was first described by Paul Ehrlich in the early 1900s who described them as ‘magic bullets’. Their development was not straightforward, and no significant progress was made until the early 1950s. ADCs studied in the 1980s and early 1990s also faced a number of challenges (Figure 1.4).

Poor target antigen selection was the most likely reason for the failure of early ADCs, as exemplified by the KS1/4 antibody–methotrexate conjugate for non-small-cell lung cancer and the BR96 antibody–doxorubicin conjugate for metastatic breast cancer. Both ADCs were evaluated in clinical settings but, despite localizing to tumours, provided little or no therapeutic benefit.¹²  Other factors that may have limited the success of these early ADCs was their use of either chimeric or murine antibodies, which elicited immunogenic responses, and the use of payloads with low cytotoxic potency.

The ADC research area has grown rapidly over the past few years, illustrated by an analysis of the number of research papers published on this topic since 1970 (Figure 1.5).

It is evident that the number of research articles has doubled between 2000 and 2010, partially due to both Kadcyla® and Adcetris® being in clinical development during this period. Between 2011 and 2018, over 60 000 research articles were published.

ADCs are mAbs attached to biologically active agents (payloads) by chemical linkers (Figure 1.6). An ADC binds to an antigen on the surface of a cancer cell and then internalises, after which the highly cytotoxic payload molecule is released, typically by lysosomal cleavage.

In order to design a successful ADC, it is crucial to understand the underlying mechanism of action. An effective ADC needs to retain the selectivity of the original monoclonal antibody while being able to release the attached cytotoxic payload in concentrations high enough to kill the targeted tumour cells. Each of these steps involves multiple unique challenges that complicate the design of ADCs¹⁴  (Figure 1.7).

The clinical development of ADCs requires careful development of several biological and pharmaceutical parameters. In particular, there are three key components to an ADC: the antibody, the linker and the payload, each of which requires careful optimisation, and this is discussed in detail in Chapter 2.¹⁰  Table 1.2 summarises some of the challenges associated with theoptimisation process.

ADCs are designed to directly target and kill cancer cells, and so the antibody has to be able to recognise and bind to its corresponding antigen localized on the tumour cell. Once bound to the antigen, the entire antigen–ADC complex is then internalized through receptor-mediated endocytosis. The internalization process proceeds with the formation of a clathrin-coated early endosome containing the ADC–antigen complex. Once inside the lysosome, the ADC is degraded and free cytotoxic payload released into the cell, leading to cell death (Figure 1.7). The mechanism of cell death will depend on the type of cytotoxic payload. For example, tubulin inhibitors, such as the auristatins or maytansines, cause disruption of cytokinesis by interfering with tubulin, whereas DNA-interacting agents, such as the PBD dimers or duocarmycins, cause DNA damage leading to apoptosis. New classes of payloads presently under development interfere with other cellular processes, such as RNA processing. One important aspect of the mechanism of action of ADCs is the bystander effect by which free drug is exported from the tumour cell through the cell membrane into the tumour environment. This has the potential therapeutic benefit of killing neighbouring tumour cells, including those that may not have the relevant antigen on their surface.¹² 

Another crucial requirement is to ensure that a sufficient concentration of payload reaches the interior of the cancer cells to guarantee their death. In reality, this is a complex process which is difficult to ensure. It has been estimated that, even if the overall mechanism of action of an ADC works at an efficiency of 50%, only 1–2% of the administered payload will reach the tumour cells.¹⁴  Therefore, it is important that the chosen payload is sufficiently cytotoxic to exert an effect at very low concentrations. It is now recognised that the choice of antigen target, and the antibody, linker and payload components, and how effectively they work together, are crucial for the success of an ADC. Although the choice of target antigen is beyond the scope of this chapter, the antibody, linker and payload components are discussed in more detail below.

One of the most critical factors when designing an ADC is the choice of antibody. Even a “clean” tumour antigen (i.e. an antigen with a high copy number on cancer cells and a low copy number on normal cells) cannot be targeted unless the antibody selected has several attributes,¹²  the most essential being a high specificity for the antigen. An antibody that lacks high specificity and cross-reacts with other antigens can act unpredictably, for example causing off-target toxicities by interacting with healthy tissues or causing premature elimination from the body before reaching the tumour site. It is also important that the antibody binds to the target antigen with high affinity while being minimally immunogenic. Another important characteristic is favourable pharmacokinetic (PK) properties.¹²  Lastly, it is helpful if the mAb has inherent antitumour activity through direct modulation of the biological activity of the target antigen, as is the case with the anti-human epidermal growth factor receptor 2 (HER2) mAb trastuzumab (Herceptin®) which is the antibody component of trastuzumab emtansine (Kadcyla®).¹⁵ 

Antibodies are grouped into five classes based on the sequence of their heavy chain constant regions: immunoglobulin M (IgM), IgD, IgG, IgE and IgA. Of the five classes, IgG is most frequently used for cancer immunotherapy. A typical IgG1 antibody consists of two heavy (H) and two light (L) chains which comprise constant (C) regions, constituting the Fc domain, and variable (V) regions, constituting the Fab domain, which provide the antigen specificity (Figure 1.8).

Subclasses of IgG, more specifically IgG1 and IgG3, are potent activators of the classical complement pathway. The binding of two or more IgG molecules to a cell surface leads to high-affinity binding of the complement component 1q (C1q) to the Fc domain, followed by activation of C1r enzymatic activity, and subsequent activation of downstream complement proteins. The result of this cascade is the formation of pores on the tumour cell surface by the membrane attack complex (MAC) with subsequent tumour cell lysis.¹⁷ 

In order to guarantee efficient internalisation, the antigen binding sites should have the highest possible target affinity. Figure 1.9 shows how antigens coated with IgG antibodies can bind to Fc receptors and initiate signalling through immunoreceptor tyrosine-based activation motifs (ITAMs) or immunoreceptor tyrosine-based inhibitory motifs (ITIMs). IgG can also bind to neonatal Fc receptors (FcRn) on endothelial cells to maintain serum IgG levels, and to tumour cells where they can recruit complement component 1q (C1q) to initiate the complement cascade, resulting in tumour cell lysis by the MAC.¹⁷ 

Bearing in mind the high target-specificity, target-affinity and prolonged exposure required at the tumour site, antibody selection should ideally ensure minimal cross-reactivity with healthy tissues, sub-nanomolar affinity to the target antigen and a long pharmacokinetic half-life combined with minimal immunogenicity.¹⁴  The degree of immunogenicity of an ADC is a crucial factor and an important determinant of circulatory half-life. In particular, an antibody that is deficient in tumour specificity may be rapidly eliminated from the circulation due to immunogenicity, leading to sub-optimal target exposure and decreased therapeutic effect.¹⁸  In the early years of ADC development, research was based on murine mAbs, which resulted in the formation of human anti-mouse antibodies within a few weeks of a single dose. Therefore, murine antibodies were quickly replaced by chimeric IgG antibodies followed later by humanised IgGs. In recent years, ADCs have been mostly based on either humanised or fully human antibodies.¹⁴ 

One of the most important benefits of using mAbs rather than small-molecule chemotherapeutic agents for cancer therapy is that mAb-based agents can have favourable pharmacokinetics with respect to duration, metabolism and elimination, provided certain characteristics have been optimised as described above. Once mAbs are administered into the bloodstream, they can distribute into tumour tissue either through extravasation via pores in the endothelium or through pinocytosis.¹⁴  However, the distribution of ADCs into tumour tissue is limited by the size of the antibody, which typically represents approximately 95% of the mass of an ADC. However, unlike normal blood vessels that have a monolayer of endothelial cells forming tight junctions with each other, tumour endothelium is usually characterized by excessive branching and sprouting, resulting in a “leaky” monolayer. Therefore, the relatively large size of ADCs means that they may distribute into tumour tissue through the leaky vasculature while their distribution into metabolizing and eliminating organs such as the liver, intestines, muscle and skin is restricted, thus potentially extending their half-life and limiting systemic toxicity.¹⁴  More recently, antibody fragments are being investigated. These can have improved tumour penetration characteristics due to their smaller size, although they typically have a shorter half-life compared to full-size antibodies.

The design, structure and chemistry of the linker that connects the cytotoxic payload to the antibody are important in contributing to the characteristics of the ADC important for specificity, potency and safety. Usually, linkers are designed to be stable in the bloodstream (i.e. to support an increased circulation time), but sufficiently labile to efficiently release the cytotoxic payload at the tumour site.¹⁹  It is also crucial for the conjugate to remain stable in buffered aqueous-based solutions to allow formulations to be developed for intravenous administration.²⁰  Linkers are classified into different types based upon their mechanism of payload release, the main two being “cleavable” or “non-cleavable”. These are described below in more detail.

Cleavable linkers utilise the differences in conditions between the bloodstream and the cytoplasm within tumour cells.²¹  The change in environment once the ADC–antigen complex has internalized triggers cleavage of the linker and release of the active payload.²¹  Cleavable linkers are divided into three main sub-categories: (1) Acid-labile (e.g. hydrazones), (2) Reducible (e.g. disulphides) and (3) Enzyme-cleavable (e.g., peptides). An example of a hydrazone linker, as utilised in Gemtuzumab ozogamicin (Mylotarg®), is shown in Figure 1.10a.²²  However, there is now a significant body of research based on clinical studies indicating that acid-cleavable linkers are associated with non-specific release of the payload, which can lead to systemic toxicities. This was one of the reasons behind Mylotarg's initial withdrawal from the market in 2010.²³ 

An example of a reducible linker is shown in Figure 1.10b. The principle behind this type of linker is the ability to respond to a difference in reduction potential between the intracellular compartment of a tumour cell versus the blood based on glutathione concentration gradients.

An example of an enzyme cleavable linker is shown in Figure 1.10c. This type of linker takes advantage of hydrolytic enzymes capable of recognising and cleaving particular peptide sequences contained within linkers, thus ensuring that the ADC only undergoes cleavage in the lysosomal environment and not in the plasma.²² 

The other type of enzyme-cleavable linker is based on the β-glucuronide moiety. The enzyme β-glucuronidase, which can release payloads from β-glucuronide-containing linkers, is present in lysosomes and is over-expressed in some tumour cell types. An important characteristic of a β-glucuronide linker is its hydrophilic properties, which can potentially reduce aggregation during conjugation compared with constructs containing dipeptide-based or other linker types.^22,27  The β-glucuronide unit has been used as a linker moiety in a number of ADCs utilising payloads including the auristatin derivatives monomethyl auristatin E and F (MMAE and MMAF), and doxorubicin. An example of a β-glucuronide-based doxorubicin-containing ADC is shown in Figure 1.10d.^22,25 

Non-cleavable linkers are beginning to be recognised as potentially more attractive than cleavable ones due to the advantage of increased plasma stability. Also, there is evidence that non-cleavable linkers perform significantly better in in vivo studies, with release of the payload occurring mainly in the lysosome after internalisation of the ADC and degradation of both the antibody and linker. This leads to a lower risk of systemic toxicity due to premature release of the payload. Therefore, non-cleavable linkers can potentially provide a better therapeutic window, with improved stability and tolerability.²² 

For example, the ADC huC242–SMCC–DM1 (Figure 1.11) utilises N2′-deacetyl-N2′-(3-mercapto-1-oxopropyl)-maytansine (DM1) as a payload linked to the monoclonal antibody huC242 via an N-succinimidyl-4-(N-maleimidomethyl)cyclohexane-1-carboxylate (SMCC) moiety joined through a non-cleavable thioether bridge.²⁸ 

In summary, choosing the most appropriate linker is crucial for the ADC development process, as it is not only responsible for efficient and selective delivery of the cytotoxic payload to the tumour cell followed by successful release of the payload, but it can also influence the degree of aggregation during the conjugation process, the pharmacokinetics of the ADC and its stability during both preparation and storage.

Early antibody–drug conjugates relied on small-molecule payloads that were already FDA-approved as stand-alone anticancer agents, such as doxorubicin and vinblastine, however these conjugates had relatively poor cytotoxicity. Therefore, researchers turned to more potent cytotoxic agents as payloads, most of which were too toxic to be used as stand-alone therapeutic agents. This led to the concept that cytotoxic agents for use as payloads need to be active in the low nanomolar or even picomolar region, as well as possessing favourable physicochemical properties, such as an acceptable hydrophilic/hydrophobic balance and good stability. A degree of selectivity in the ability to effectively kill tumour cells while leaving healthy cells intact is also now considered to be desirable property, along with the ability to retain cytotoxicity in target tumour cells that develop resistance. In this context, it is also considered desirable for payloads to have limited cellular permeability in order to minimise collateral action on peripheral healthy tissues should premature release from an ADC occur, although sometimes a bystander effect is desirable. From a commercial and clinical perspective, payload molecules should also be chemically tractable, with a clear path for scale-up production and purification. They should also have cost-effective synthetic routes with clear intellectual property (IP) and/or licensing pathways. All these factors have played roles in the development of the various classes of payloads currently used in ADCs.

The majority of ADCs presently in clinical use contain the auristatins (i.e. MMAE, MMAF) and maytansines (i.e., DM1, DM4), which are microtubule inhibitors. However, over the past few years there has been a surge of interest in using DNA-interactive agents such as the calicheamicins (used in two approved ADCs, Mylotarg® and Besponsa®), duocarmycin derivatives (e.g. SYD985, which is currently in Phase II clinical trials) and the PBD dimers (used in the form of tesirine and talirine, with the former currently in a number of clinical trials), all of which target the minor groove of DNA and work by cleaving mono-alkylating adenine bases or cross-linking guanine bases, respectively. There are also multiple new DNA-interactive agents in development such as the indolinobenzodiazepines (IGNs) (e.g. DGN462 currently undergoing clinical evaluation, Chapter 15) and the pyridinobenzodiazepines (PDDs, Chapter 16) (e.g. FGX-2-62) which work by mono-alkylating guanine bases in the DNA minor groove. There is also ongoing research into protein-based toxins such as SarcinDI, the De-Bouganin and diphtheria toxins, pseudomonas exotoxin A, gelonin and saporin.

There are many issues to consider when compiling comparative cytotoxicity data for different payloads, including the type of cell line used to evaluate them, the incubation time used, the starting number of cells per well and the type of assay used (e.g. a metabolic assay, such as MTT, or an ATP-based assay, such as MTS), which can all affect the concentrations giving 50% of maximum inhibition (IC₅₀) values observed for any given payload. Therefore, it is very challenging to compare the potency of different payloads, or groups of payloads, with any degree of meaningfulness. However, Figure 1.12 provides an approximate representation of the cytotoxicity ranges of various payload groups.

As shown in Figure 1.12, the most cytotoxic payload types include the cyclopropabenzindole (CBI) monomers and dimers, the IGNs, the PBD dimers and PDDs. A detailed discussion of each payload family can be found in the remaining chapters of this book.

A number of methodologies have been developed to attach linker–payload constructs to antibodies, with most work to date focused on the IgG1 and IgG4 families. One of the first conjugation methods developed involved reaction of an electrophilic group [e.g., a maleimide or N-hydroxysuccinimide (NHS) moiety] positioned at the terminus of a linker–payload assembly with the exposed nucleophilic –NH₂ groups of lysine residues within the antibody. This, so-called “stochastic” method of conjugation gave a heterogenous mixture of ADC species. An improvement to the homogeneity of the ADC species produced was achieved by initial partial or full reduction of the four interchain disulphide bridges of the antibody to produce an average number (i.e., 2, 4, 6, 8) of nucleophilic thiol groups for reaction with the electrophilic linker–payload construct. Although an improvement, this method still produced a mixture of ADC species [i.e., drug to antibody ratios (DARs) of 1–8], which can have a negative effect on parameters such as pharmacokinetics, tolerability and efficacy. Therefore, site-specific conjugation methodologies, such as the THIOMAB™ technology, have been developed, whereby linker–payload constructs could be attached to engineered sites within the antibody containing nucleophilic thiols or other coupling partners (e.g., cysteines, click chemistry partners) allowing well-defined ADC products comprised of just one uniform species to be produced. Examples of the various conjugation technologies are briefly described below.

Conjugation to the terminal amines of lysine residues was an early conjugation method developed, but is the least selective as approximately 80–100 lysine amines are scattered throughout the structure of an antibody (Figure 1.13a).²⁰ 

However, these lysine amines are very effective for payload conjugation as they are well-exposed and their amino groups are good nucleophiles.³¹  A popular lysine conjugation reaction involves formation of a stable amide bond using activated esters of the payload, typically an O-succinimide reagent such as NHS, as shown in Figure 1.13b(1). Other approaches to lysine conjugation utilise maleimide, or haloacetamides. Traut's³²  reagent has also been used to form stable amidine bonds. In addition, conjugation via isothiocyanate chemistry has been used to form stable thiourea linkages, and this has been widely used to fluorescently label antibodies with, for example, fluorescein isothiocyanate.^31,33 

Monoclonal IgG antibodies, such as IgG1, IgG2 or IgG4, contain multiple disulphide bonds, which were once considered a uniform and homogenous structural feature.³⁴  Classical disulphide bond arrangements are shown in Figure 1.14. Each IgG contains a total of 12 intra-chain disulphide bonds, each of which is associated with different regions of the antibody.³⁴ 

The inter-chain disulphide bonds are the most desired sites of attachment for cytotoxic payloads. For human IgG1, the four inter-chain disulphide bonds are typically reduced with agents such as tris(2-carboxyethyl)phosphine (TCEP), dithiothreitol (DTT) or 2-mercaptoethylamine (2-MEA) prior to conjugation. Once the free thiols have been generated, they are reacted with a linker–payload complex containing a suitable electrophilic moiety [e.g. maleimide, NHS, 3,4-dibromomaleimide etc. (see Figures 1.13b and 1.15)]³⁰  resulting in a population of conjugates, with mixed DARs and varying locations of payload.

In the case of an IgG1 antibody, up to four interchain disulphide bonds can be reduced, thereby exposing up to eight nucleophilic thiol groups for conjugation. Typical conditions developed for thiol-based conjugation lead to either complete or partial reduction of these disulphide bonds, and the ADCs produced using these methodologies contain up to eight payloads per antibody (i.e., a DAR of up to 8).³⁵ 

Due to problems associated with the disulphide bridge reduction technologies, a number of site-specific conjugation methods have been developed to improve homogeneity of the final ADC product. The best-known example of site-specific conjugation is the THIOMAB^TM technology developed by Genentech Inc.³⁶  THIOMABs are antibodies with engineered reactive cysteine residues. This allows payloads to be conjugated with defined stoichiometry without disruption of the interchain disulphide bonds.³⁶  THIOMAB–payload conjugates are superior to non-site-specific ADCs because they contain a uniform distribution of the attached linker–drug construct, normally with a DAR of 2. ADCs made with this technology are efficacious and have also shown superior safety characteristics compared with stochastically-conjugated ADCs in numerous in vivo studies, especially with regard to liver and bone marrow toxicities (Figure 1.16).^37,38 

Other site-specific engineered antibodies have been developed. For example, Selenomab™ ADCs were developed at the Scripps Research Institute (USA) and contain one or more strategically positioned selenocysteine residues (Figure 1.17). The unique reactivity of the selenol group of a selenocysteine residue permits highly efficient site-specific conjugation of payloads. Compared with other natural and unnatural amino acids and carbohydrate residues that have been used for the generation of site-specific ADCs, the selenocysteine residues are particularly reactive, permitting fast, single-step efficient reactions under near physiological conditions.⁴⁰ 

A variety of click chemistry methodologies have been used to attach payloads to antibodies. In this approach, an azide-containing reagent such as sodium (difluoroalkylazido) sulfonate (DAAS-Na), is used to attach an azide group to the terminus of a linker–payload construct.⁴¹  The azide functionality can then be reacted with a dibenzylcyclooctyne (DBCO) grouping incorporated into the antibody (Figure 1.18a). One advantage of click chemistry is that the reaction is very efficient and high-yielding, and occurs under mild conditions, typically using copper catalysts.

Click chemistry has also been used to attach azide-functionalised porphyrins to trastuzumab.³⁰  For example, the reaction of reduced trastuzumab with N-propargyl-3,4-dibromomaleimide to produce an alkyne-containing antibody which was then conjugated to porphyrin derivatives through CuAAC click chemistry has been described (Figure 1.18b).³⁰ 

Apart from the well-known conjugation methodologies described above based on thiols, amines or click partners, there are other conjugation technologies based on alcohols (e.g. those that can form carbonates, ethers and esters) and aldehydes [e.g. conjugation via formylglycine-generating enzyme (FGE)], engineered aminoacyl-tRNA synthetase (aaRS), oxidized sialic acids and transamination reagents.^30,42  Additionally, new methodologies are emerging, such as GlycoConnect™, developed by Synaffix, based on initial enzymic modification of two naturally-occurring glycan anchor points on an antibody. Metal-free click chemistry is then used to attach payload molecules in a site-specific manner.⁴³ 

A number of methodologies have been developed for determining DARs, including whole-antibody mass spectrometry,⁴⁴  spectrophotometric absorbance ratios⁴⁵  and various chromatographic methods.⁴⁶ 

High-performance liquid chromatography (HPLC) methodologies that resolve the antibody or its constituent heavy and light chains on the basis of increases in hydrophobicity resulting from payload conjugation have proven particularly useful for cysteine-conjugated ADCs. However, for lysine-conjugated antibodies, these chromatographic methods often provide inadequate resolution of the various payload-conjugated species, due to the positional isomerism arising from conjugation at multiple sites.⁴⁶  For cysteine-conjugated ADCs, the limited number of available cysteine residues and their localization in the hinge region appears to limit the peak-broadening effect.

The simplest methodology for ADC characterisation relies on ultraviolet/visible light (UV/VIS) spectroscopic analysis⁴⁶  given that the payload and antibody have different maximum absorbance (λ_max) values. Using the measured absorbance of the whole ADC, the individual extinction coefficients of the mAb and the payload and their distinctive λ_max values (the former usually at 280 nm), the individual concentrations of mAb and payload can be determined through the solution of two simultaneous equations, thereby providing the molar ratio (i.e. moles of payload per mole of antibody). This approach has been widely used, even for ADCs with a relatively small difference in λ_max between the mAb and payload. Examples of payloads for which this methodology has been used include maytansinoids (e.g. DM1), methotrexate, CC-1065 analogues, doxorubicin, calicheamicins and dipeptide-linked auristatins such as maleimidocaproyl-valine-citrulline-p-aminobenzyloxycarbonyl-MMAE (vc-MMAE).^46,47 

Stochastically conjugated ADCs are typically highly heterogeneous, consisting of a population of conjugated species, each with different numbers of payload molecules at various conjugation sites. Therefore, analytical techniques capable of characterizing and monitoring ADC payload distribution are of significant value, as these species may have very different pharmacokinetic, biological and toxicological properties, which can directly affect the efficacy, toxicity and safety profile of the parent mixture. ADC species with a significant number of payload molecules attached (i.e. more than four payloads per mAb) often exhibit a higher aggregation propensity, which may result in an increased toxicity profile in patients. The payload component of an ADC is typically a small hydrophobic molecule conjugated at an exposed region of the mAb. Therefore, ADC species containing larger numbers of conjugated payload molecules have more regions of hydrophobicity, making hydrophobic interaction chromatography (HIC) analysis uniquely suited to monitoring ADC payload distribution. HIC is gaining in popularity, and many reports of the successful use of HIC for ADC payload distribution studies have appeared in the literature.⁴⁸  HIC is able to effectively separate differentially substituted ADCs on the basis of the hydrophobic interactions between the stationary phase and the native, folded structure of the mAb.⁴⁷ 

Early work on the use of HIC to study loading was carried out by researchers at Seattle Genetics.⁴⁷  A typical methodology involved the use of a column such as the Ether-5PW column (Tosoh Bioscience) with a linear gradient.⁴⁷  HIC methodology can be complex, and many parameters need to be optimised including the column (i.e. hydrophobic ligands and support matrix), mobile phase, salt type and concentration, temperature and pH.⁴⁹  The initial mobile phase conditions used for HIC usually start with a high ionic strength, often using ions such as ammonium sulfate which appear early in the Hofmeister series⁵⁰  and which thus have the greatest effect in strengthening hydrophobic interactions. Since the method is non-denaturing, ADCs are eluted from the column as intact antibodies, even if all the interchain disulphides have been reduced.⁵¹  Thus, the method can reveal not only the average number of payloads per antibody but can also provide information about their distribution, including the percentage of total antibody with any given level of payload loading.⁵¹ 

The payload loading is calculated as the weighted average using the integrated areas of the constituent peaks, with the payload loading of each peak as the weighting factor. An example of a HIC chromatogram of an ADC with an average DAR of 4 prepared by partial reduction is shown in Figure 1.19. As anticipated, the chromatogram shows antibody species containing exclusively even numbers of payloads, since each reduced disulphide bridge produces two thiol functionalities for conjugation.⁵¹ 

One of the crucial steps in a conjugation procedure is to determine the content of unconjugated payload. For traditional maleimide (MC) coupling, the unconjugated payload content depends on the purification step performed immediately after linking the maleimide drug–linker complex to the available cysteine thiols generated during the reduction step. The conjugation reaction is typically quenched with a small-molecule thiol-containing agent such as N-acetyl cysteine in order to neutralize any excess of the unconjugated payload–linker. Therefore, in the majority of cases, there will be a number of impurities in the final ADC preparation. The ADC itself is usually isolated by precipitation with an organic solvent to prevent the co-elution of proteinaceous species, and to reduce fouling of the HPLC column. Reverse phase (RP)-HPLC can be used to quantify unconjugated payload–linker content based on an external standard curve. Rather than reporting the molar concentration of unconjugated payload, the result is usually converted to a fractional value of the total payload present in the ADC preparation.⁵²  This is important, as the ADC will typically be studied initially in in vitro and/or in vivo assays in which it is important to be able to measure the ADC concentration in terms of micrograms per millilitre or milligrams per kilogram, respectively. It is important to note that the unconjugated payload as a percentage of the total ADC does not change upon dilution, whereas absolute molar concentrations will change.⁵¹ 

Hydrophobicity plays an important role in the development of ADCs. Despite extensive research, its molecular mechanism remains controversial, and there are still no reliably predictive models for its role in drug discovery.⁵³  Some of the most potent ADC payloads available (e.g., the PBD dimers and the duocarmycins) are also the most hydrophobic, and this can cause difficulties for both conjugation and bulk manufacture. Also, it is recognised that reducing the hydrophobicity of homogeneous ADCs can improve both their pharmacokinetics and therapeutic index (TI). Therefore, it is crucial that, during the discovery and development phases of new ADCs, the molecular weights of payload molecules and associated linkers are kept as low as possible. It is also helpful to include as many heteroatoms as possible in ring structures and chains, and as many hydrophilic functional groups as possible as substituents (e.g., –NH₂, –OH, –COOH) to optimise hydrophilicity. For this reason, glycol chains [e.g., polyethylene glycol (PEG)] are often included in linker–payload complexes,⁵⁴  and there are multiple examples of these in the literature. An example of a PBD dimer payload connected through its N10 position to an antibody via a PEG unit is shown in Figure 1.20.⁵⁵ 

ADCs have significant potential as anticancer therapies, but have many technical challenges associated with them. Despite many decades of research in this area, only four products: gemtuzumab ozogamicin (Mylotarg®), brentuximab vedotin (Adcetris®), ado-trastuzumab emtansine (Kadcyla®) and inotuzumab ozogamicin (Besponsa®); along with two immunotoxins [moxetumomab pasudotox (Lumoxiti®) and tagraxofusp (Elzonris®)] have been approved. However, ‘next generation’ ADCs are increasingly entering the pre-clinical and clinical pipelines as a result of ongoing research aimed at improving efficacy and reducing side effects. Multiple novel payload families, new linker technologies and engineered antibodies are currently being researched that provide hope for future generations of improved ADCs with the potential for safer and more effective cancer treatments.