By Rashad Mammadov
The therapeutic window of an ADC highly depends on its (monoclonal) antibody component that determines the destiny of cytotoxic payload conjugated to it. Therefore, the antibody should have high target specificity and low cross-reactivity in order to limit undesired toxicity. As such clones are determined, it is critical to establish their homogeneity. This requires robust analytical methods (liquid chromatography/mass spectrometry) for characterization of the antibody. Otherwise, the heterogenous antibody profile can render the ADC prone to failure.
Another challenge in ADC development is the immunogenicity against the antibody. First-generation ADCs utilized antibodies of mouse origin which caused their failure because they were rejected by the human immune system. Later generations have chimeric, humanized, or even human antibodies instead to solve this problem. Apart from avoiding immune cells, the antibody should have enough stability to have long circulating life.
Chemical properties of the antibody is also a barrier in ADC development as they decide how many and which type of linkers can be conjugated to the antibody and to which part. Combinations of these parameters also significantly influence the stability of ADC.
In order to deliver cytotoxic payloads into tumor cells, antibody-antigen interaction should lead to efficient internalization of ADC. In the same context, having good retention after binding to target cells would serve to that end.
In some cases, the antibody binding to its target receptor may induce signaling and produce cytotoxic effect itself through mechanisms such as antibody-dependent cell-mediated cytotoxicity (ADCC), as it happens in the case of trastuzumab emtansine (Kadcyla). Such independent functioning of an antibody is not always considered beneficial for ADC efficacy, since it may reduce internalization of ADC into tumor cells. Such two-pronged killing (antibody and payload) is also a cause for concern since it may be too toxic for healthy cells.
Lastly, antibody isotype selection needs careful planning. Subclasses of IgG - IgG1, IgG2, or IgG4 (mostly IgG1) - is used in current ADC development. IgG3 is not employed as it has fast plasma clearance. The isotype selection defines the difficulty of conjugating linkers to antibody backbone. Antibody isotype also determines the potential immune effector functions such as ADCC.
In an ideal case, a target antigen should be highly expressed in the tumor (with homogeneous expression across the tumor cells), while having minimal or no expression in normal tissues. As an example, HER2 receptor, which is targeted by trastuzumab emtansine (T-DM1), have 100-fold higher expression in the tumor cells than the healthy cells. Accordingly, T-DM1 provides the greatest benefit to the patients with the highest expression of HER2.
Then, it should be internalized efficiently by receptor-mediated endocytosis and not recycled back to the surface. Non-internalized ADCs exert toxicity on neighboring cells. Also, its expression should not get downregulated. In this context, epitope on the target protein also can be important. For example, it was reported that different epitopes of the HER-2 receptor show differences in terms of internalization and degradation of antibody-antigen complex.
Antigens expressed on the surface of the tumor cells are preferred, as they are easier targets for circulating ADC. However, this may not be always possible or in some cases internalization may be hindered due to different reasons such as the high interstitial tumor pressure. For those cases, ADC could be targeted against antigens in the tumor microenvironment.
Shedding of the antigen could be a significant problem as free antigens wandering within the circulation will bind the ADC antibody and compromise its efficacy. Therefore, the target antigen should have minimal shedding.
In ADC design, one of the questions is that how many cytotoxic drugs will be conjugated to each antibody. In other words, what will be the drug‐to‐antibody ratio (DAR)? If DAR is lower than optimal, that will limit the efficacy of the ADC. However, when DAR is too high, ADC becomes unstable. Altered pharmacokinetic properties of ADC in that case reduces half-life and increases systemic toxicity. The optimal DAR depends on other ADC components as it ranges between 2 and 4 in clinically approved ADC systems.
Payload should be highly potent in terms of cytotoxicity because research shows that at best 0.01% of injected monoclonal antibody (or ADC) binds to target tumor cells. Moreover, optimal DAR also limits the amount of payload that is delivered into tumor cells. As a result, ADC payloads should have IC50 in nanomolar and even picomolar range. Consistently, classical chemotherapy drugs failed to show clinical benefit under ADC framework.
Resistance mechanisms that cancer cells develop against cytotoxic drugs should also be considered as a preclinical challenge. These mechanisms may include increased expression of efflux pumps that remove drugs from cells and altered microtubule composition among others. Exploration of the susceptibility of the payload to drug resistance is crucial. Appropriate drugs with minimal susceptibility should be chosen.
Hydrophobicity/Hydrophilicity of drugs is another issue in ADC design. More lipophilic drugs tend to pass through the cell membrane and kill neighboring cells. This phenomenon is named as bystander effect. It may confer advantage for the treatment of certain solid tumors as their cells have heterogeneous expression of ADC target protein. However, in other cases bystander effect may cause off-target toxicity.
The other challenges in finding an appropriate drug are its stability in blood, solubility in water, and possession of chemical functionalities to conjugate it to a linker molecule.
It is obvious that finding appropriate drugs that meet these criteria is challenging. Currently, two main classes of cytotoxic drugs are used in ADC development. Microtubule inhibitors block assembly of tubulins and cause cell cycle arrest at mitotic phase. Auristatins and Maytansinoids are microtubule inhibitors that are used in three FDA approved ADCs (Adcetris, Kadcyla, and Polivy). The other class includes DNA damaging drugs. Calicheamicin, a drug used in two FDA approved ADCs (Besponsa and Mylotarg), binds DNA’s minor groove and induces double-strand DNA breaks and rapid cell death. Since they could function independently from cell cycle progression, DNA damaging drugs could be used against cancer stem cells which have lower proliferation rate.
Linker connects cytotoxic payload to monoclonal antibody, hence its properties play a significant role in pharmacokinetics and therapeutic window of ADC. The major preclinical challenge to the ADC efficacy and therapeutic index is high deconjugation rate. Ideally, it should have enough stability that allows ADC to circulate in blood without releasing the payload and causing systemic toxicity. On the other hand, they should release payload once they are inside the target cell. The two types of linkers used in ADC structure are non-cleavable and cleavable.
Non-cleavable linkers provide more stability to ADC and the release of payload happens by lysosomal degradation of the antibody. Even after degradation, payload is released as attached to the linker and the terminal amino acid residue of the antibody. Thus non-cleavable linkers are optimal for drugs that preserve their potency when bearing the moieties left from the degradation. For example, Trastuzumab emtansine (Kadcyla) has non-cleavable thioether linker its maytansinoid-based DM1 drug and antibody. Conversely, MMAE used in the Brentuximab vedotin (Adcetris) is a protein-based cytotoxic drug and has optimal potency in its unconjugated form hence the linker used with this drug is cleavable.
Cleavable linkers release their payload when they meet certain physiological stimulus present in their target site. For example, acid-sensitive linkers (e.g. hydrazone-based linker of Gemtuzumab ozogamicin/Mylotarg) are cleaved in the low pH environment of lysosomes/endosomes. However, these linkers have shown a certain level of plasma instability. As another example, valine-citrulline linker in Brentuximab vedotin is a protease-sensitive linker that is cleaved by cathepsins in lysosomes. Lastly, disulfide linkers used in the design of some ADCs are cleaved by high glutathione concentration found in tumor cells.
Linker’s hydrophilicity/hydrophobicity also affects ADC therapeutic index from many aspects. One of them is bystander effect, that, as it is mentioned above, may be beneficial against solid tumors. Depending on whether you want to enhance or reduce this effect, linker design may change. Non-cleavable linkers that stay with the payload together with a charged amino acid may prevent payload’s passage through membranes thus lowering bystander effect. This effect may be enhanced by using non-polar, more hydrophobic linkers. On the other hand, increased hydrophobicity is associated with high plasma clearance. Moreover, as the linker-drug molecule gets more hydrophobic, it becomes more prone to the work of MDR1 efflux pumps. Therefore, the question of which non-cleavable linker to select depends on which type of cancer cell you are targeting (i.e. does it have a drug-resistance mechanism), whether it is a solid tumor, and which drug you are using as a payload. Cleavable linkers allow drugs to leave alone thus here drug’s chemical properties determine its final destiny.
Beck, A., Goetsch, L., Dumontet, C. et al. Strategies and challenges for the next generation of antibody–drug conjugates. Nat Rev Drug Discov 16, 315–337 (2017).
Chau, C. H. et al. Antibody–drug conjugates for cancer. The Lancet, 394, 10200, 793 - 804 (2019).
Diamantis, N., Banerji, U. Antibody-drug conjugates—an emerging class of cancer treatment. Br J Cancer 114, 362–367 (2016).
Hughes, B. Antibody–drug conjugates for cancer: poised to deliver?. Nat Rev Drug Discov 9, 665–667 (2010).
As we mentioned among preclinical challenges, it is mostly essential for an ADC to be delivered into lysosomes to release the cytotoxic payload. This is vital for the only approved ADC to treat a solid tumor, T-DM1 (Kadcyla), since it carries a non-cleavable linker. However, HER2 receptor, the target antigen for T-DM1, despite its selective expression on breast tumor cells is found to recycle back with high rate after internalization. Overall, this leads to only a small proportion of T-DM1 to reach lysosomes. This phenomenon may stand behind the limited efficacy of T-DM1 that involve its failure to show a benefit in a gastric cancer trial, whereas trastuzumab (unarmed antibody) passed approval. Also, in Phase II and III clinical trials for breast cancer it failed to show higher clinical benefit than standard therapies.
Its benefit seems to be restricted to breast cancer patients with high HER2 expression. Bispecific antibody technology delivers promising results in the way to solve this problem. These antibodies have two paratopes or they can recognize two different antigens, while conventional antibodies are “monospecific” (although bivalent). In the context of enhancing lysosomal trafficking, two particular bispecific ADC - based strategies stands out.
The first bispecific antibody has each paratope recognizing different epitopes on the same antigen. Such bispecific ADCs have been shown to induce receptor clustering on the target cells and increased internalization, lysosomal delivery and on-target cytotoxicity. ZW49 developed by Zymeworks is a bispecific ADC that recognizes two different epitopes on HER2. It is currently in Phase I for the treatment of biliary tract cancer.
Another approach, named as “drag and degrade”, is based on getting help from a receptor that regularly and efficiently traffics to lysosomes. Here, one of the paratopes of the antibody recognizes a “fast-internalizing” receptor, the other recognizes a tumor-specific antigen. HER2/CD63 and HER2/Prolactin receptor bispecific ADCs have been shown to enhance lysosomal delivery and on-target cytotoxicity when compared to HER2 monospecific ADC.
Both targets of bispecific ADC may not have similar contributions. Various parameters such as the affinity of the individual arms, the density of the target, the overall avidity and the valency of the bispecific format determines tumor selectivity. A study testing EGFR/cMet ADC showed that having low affinity EGFR paratope allowed to show toxicity against tumor cells while sparing normal keratinocytes which have moderate level EGFR expression.
Antibodies have relatively larger sizes that hinder the efficient delivery of ADCs to sequestered targets such as the interior of the solid tumors. The importance of this phenomenon is best reminded by the fact that most of the marketed ADCs are approved for the treatment of hematological cancers. Therefore, in recent years, attention is directed towards the use of smaller formats than antibody as a targeting unit of the conjugate. These include antibody fragments, peptides, scaffolds (e.g. centyrins), and even small molecules. Some of these are discussed in the next chapter with clinical applications.
Certain features of smaller format conjugates can potentially improve the therapeutic window. First, they have higher tumor penetration than ADCs due to their smaller physical radius and hence faster diffusion and extravasation coefficients. Second, they have a high plasma clearance rate that may reduce off-target toxicities. However, this may be a disadvantage as well since their concentration in tumors also decreases rapidly. That could be improved via repeated injections or further engineering to enhance their circulation time.
Smaller formats could especially make a difference by increasing the tolerated concentration via minimizing adverse effects. The lack of the Fc domain enables one to avoid cross-reactivity with Fc-receptors in healthy cells and this could improve the tolerability of higher doses. Fc domain of ADCs causes thrombocytopenia and neutropenia via binding to Fc receptors on non-target cells.
Comer F., Gao C., Coats S. (2018) Bispecific and Biparatopic Antibody Drug Conjugates. In: Damelin M. (eds) Innovations for Next-Generation Antibody-Drug Conjugates. Cancer Drug Discovery and Development. Humana Press, Cham. Maruani A. (2018) Bispecifics and antibody–drug conjugates: A positive synergy. Drug Discovery Today: Technologies, 30, 55-61. https://www.zymeworks.com/pipeline
Deonarain, M.P.; Yahioglu, G.; Stamati, I.; Pomowski, A.; Clarke, J.; Edwards, B.M.; Diez-Posada, S.; Stewart, A.C. Small-Format Drug Conjugates: A Viable Alternative to ADCs for Solid Tumours? Antibodies 2018, 7, 16. He, R.; Finan, B.; Mayer, J.P.; DiMarchi, R.D. Peptide Conjugates with Small Molecules Designed to Enhance Efficacy and Safety. Molecules 2019, 24, 1855.
Antibody fragments have caught much interest as they constitute only a portion of an antibody, but still retains the paratopes necessary to bind to the target antigen. Employing this technology could minimize immunogenicity and heterogeneity problems, while their smaller size seems to provide better solid tissue penetration. Various engineered formats of antibody fragments fused to cytotoxic payloads have been proposed up to date, some of which have made strides through clinical studies.
Moxetumomab pasudotox (Lumoxity) is an antibody fragment - drug conjugate has been approved recently by the FDA “for the treatment of adult patients with relapsed or refractory hairy cell leukemia (HCL) who have received at least two prior systemic therapies, including treatment with a purine nucleoside analog.” Its targeting component is a disulfide stabilized variable fragment (dsFv) of a monoclonal antibody that recognizes CD22 on malignant B cells. dsFv is fused to a 38-kDa fragment of Pseudomonas exotoxin A (PE38) that interferes with protein synthesis in cells and inducing cell death eventually.
Oportuzumab monatox (Vicinium) employs a slightly different antibody fragment, named as single-chain variable fragment (scFv), that is conjugated to a Pseudomonas exotoxin A fragment. scFv fragments are generated by conjugating N and C terminals of variable regions of the heavy (VH) and light chains (VL) with a short linker peptide. Vicinium targets Epithelial Cell Adhesion Molecule (EpCAM) with the help of its scFv. EpCAM is overexpressed in >98% of high-grade non-muscle invasive bladder cancer (NMIBC), while minimally expressed in healthy bladder tissue. Vicinium was successful in Phase 1 and 2 studies for Bacillus Calmette-Guérin (BCG)-unresponsive, high grade non-muscle invasive bladder cancer (NMIBC). It showed a good safety profile and the complete response (CR) rate at 3 months in 29- 40% of subjects. Now, it is tested in a Phase 3 study to treat BCG-unresponsive NMIBC.
The action mechanisms of Lumoxity and Vicinium rely on receptor-mediated endocytosis and subsequent cleavage of the peptide linker with endosomal proteases that release the cytotoxic payload. However, the anti-cancer mechanism of the Daromun differs from them in that it is expected to stimulate the anti-tumoral effects of immune cells.
Daromun is a combination of two different immunocytokine drugs, Darleukin and Fibromun. Darleukin is a conjugate between a diabody (i.e. a noncovalent dimer of 2 scFv) and two IL-2 molecules. On the other hand, Fibromun is a conjugate of an scFv and a single TNFalpha molecule. Both of their targeting moiety recognizes the extra-domain B of fibronectin has been shown to facilitate tumor accumulation. On the other hand, cytokines are expected to induce local immune cells, especially T cells. Currently, it is tested as an intralesional therapy in patients with fully resectable stage IIIB/C melanoma in a Phase III trial.
Centyrins are proteins based on a fibronectin type III (FN3) domain sequence from human tenascin. They have simple structures (~100 amino acids) that lack disulfide bonds and glycosylation so that they can be produced in homogenous batches. They have the potential to overcome limitations of antibodies. While antibodies can bind one or two antigens, centyrins can be engineered via genetic fusion to bind multiple targets. Furthermore, they can be conjugated to cytotoxic drugs, oligonucleotides, and nanoparticles for their targeted delivery. They have good thermal and low pH stability. Also, they are highly soluble enabling to prepare solutions of high concentration. Centyrins have much smaller size (10 kDa) than antibodies (~1/15th of an antibody) that can provide better penetration and concentration in solid tumors. For the same reason they have short in vivo half-life as they are cleared through renal filtration. This may reduce liver toxicity that is associated with ADC clearance. Half-life of centyrin can be extended by adding moieties specific for serum proteins such as albumin binding domain. Centyrins have no cysteine residues naturally, and this allows to incorporate cysteines at specific sites for controlled conjugation of cytotoxic/drug payloads. A study, which used high-throughput methods to test the tolerance of each residue of centyrin (total ~100 residues), found that cysteine mutations in 26 sites did not have significant negative effects on biophysical properties or biological activity.
Peptide Drug Conjugates (PDC) employ peptides, much smaller biological molecules (~1-3 kDa) than antibodies and centyrins, as a targeting unit. Hence, they can penetrate the tumors faster than larger formats while plasma clearance rates are higher. PDCs are composed of a peptide molecule covalently conjugated to a drug molecule by using various linker chemistries. Peptide conjugation helps small molecule drugs to surmount challenges such as poor aqueous solubility, drug-drug interaction, fast metabolism, and cellular impermeability. The amino acid sequence of the peptide part can be custom-tailored to increase physicochemical properties of the conjugate (e.g. including charged/hydrophilic amino acids to increase solubility) or make the peptide mimic receptor binding domain of the proteins with an aim of targeting it to specific cells.
Since chemical synthesis of peptides allow higher molecular diversity and accuracy than what antibody production allows, better structural precision and optimization is possible. Having low molecular weight, PDCs allow purification with HPLC and obtaining homogenous products. All these facilitate commercial synthesis and compliance with regulatory requirements.
177Lu-Dotatate (Lutathera) is a PDC approved by FDA for the treatment of gastroenteropancreatic neuroendocrine tumors (GEP-NETs). The peptide part of 177Lu-Dotatate is a somatostatin analogue and shows high affinity to SSTR2, a somatostatin receptor that is overexpressed in many tumors. Somatostatin naturally inhibits growth hormone secretion and its activity suppress growth of cancer cells. Hence, besides improving targeting of the drug to tumor cells, peptide of 177Lu-Dotatate itself is thought to induce anti-tumor signaling as well. Peptide is conjugated to a radioactive chemotherapeutic drug via an amide linker. This system allows targeting radiation to tumor and induce selective killing of cancer cells. Clinical trials showed that 177Lu-Dotatate improves progression-free survival and complete/partial shrinkage of tumor in somatostatin-receptor positive patients.
Another example of PDCs that use SSTR2-targeting peptide for solid tumor penetration is PEN-221. Its octreotide peptide is conjugated to a cytotoxic payload DM1 maytansine with a cleavable linker. It is currently investigated in a Phase 2a trial in patients with small-cell lung cancer to evaluate its efficacy, safety, and pharmacokinetics.
Phage display technology can be utilized to select peptides that have a high affinity to tumor-specific antigens. The bicyclic peptide of the PDC BT1718 (also named as Bicycle Drug Conjugate) has been discovered via phage display selection. When three cysteines at fixed positions of the peptide made to covalently bond with a reagent such as TBMB, a bicycle-like structure emerges. This peptide has a strong affinity to MT1-MMP (human matrix metalloprotease 14) that is highly expressed in multiple cancers including triple-negative breast, non-small cell lung, and soft tissue sarcoma. The cytotoxic payload of BT1718 is DM1. After successful anti-tumoral effect at in vitro and in vivo lung tumor xenograft mouse models, it is now tested in Phase I trial in patients with advanced solid tumors and Phase II trial in patients with non-small cell lung cancer.
Oligonucleotides (ON) are highly anionic (i.e. negatively charged) molecules with potential therapeutic applications. They can be designed as single-stranded antisense ON or double-stranded siRNAs for silencing of genes characterizing progression of a particular disease. However, the anionic nature of ONs hinders their internalization into cells [their site of action]. Moreover, there is a huge need to improve their stability and targeting. Lastly, achieving the cytosolic delivery of the ON (rather than being trafficked to the lysosome) is a critical barrier for the use of ON as therapeutics. In order to circumvent these barriers, conjugation of ONs to polymers, peptides, proteins and antibodies has been suggested. For example, conjugation of VEGFR2 siRNA to cyclic RGD, a tripeptide, improved its in vitro and in vivo performance. The conjugate knocked down the target gene, decreased angiogenesis and tumor growth in a mouse model.
Despite being few, there are preclinical studies regarding antibody-ON conjugates (AOC) reported in the literature. Some of them compared non-covalently and covalently conjugated versions of siRNA and antibody in terms of cellular uptake, nuclear translocation, and gene silencing. Although conjugation method did not matter when it comes to cellular uptake, non-covalent conjugations appeared superior than the covalent ones for the other two outputs. Inferiority of covalent conjugation stayed same for both cleavable and non-cleavable bonds. It can be inferred that covalent bonding leads to poorer intracellular trafficking at least for some receptors. However, another study reported contrasting results. They targeted CD19 receptor (ALL biomarker) with an antibody conjugated to an antisense ON with (covalent) disulfide bond. This AOC knocked down ALL fusion protein effectively both in vitro and in vivo, also doubled the survival time of human ALL. This story suggests that not all receptors are created equal at least when targeting them with AOC and endosomal escape of AOC is receptor-specific.
Sheridan, C. Ablynx's nanobody fragments go places antibodies cannot. Nat Biotechnol 35, 1115–1117 (2017).
Deonarain, M.P.; Yahioglu, G.; Stamati, I.; Pomowski, A.; Clarke, J.; Edwards, B.M.; Diez-Posada, S.; Stewart, A.C. Small-Format Drug Conjugates: A Viable Alternative to ADCs for Solid Tumours? Antibodies 2018, 7, 16.
List T, Neri D. Immunocytokines: a review of molecules in clinical development for cancer therapy. Clin Pharmacol. 2013;5(Supplement 1):29-45
Kim, Ji-Sun et al. Critical Issues in the Development of Immunotoxins for Anticancer Therapy. Journal of Pharmaceutical Sciences, Volume 109, Issue 1, 104 - 115
https://sesenbio.com/wp-content/uploads/2019/01/BLADDR_Congress_2018_Posterl.pdf
https://www.fda.gov/news-events/press-announcements/fda-approves-new-kind-treatment-hairy-cell-leukemia
http://www.philogen.com/en/products/pipeline/oncology/daromun_26.html
https://www.arobiotx.com/research-development
S. D. Goldberg et al. Engineering a targeted delivery platform using Centyrins, Protein Engineering, Design and Selection, Volume 29, Issue 12, 28 December 2016, Pages 563–572
C. Shi et al. Bioanalytical workflow for novel scaffold protein–drug conjugates: quantitation of total Centyrin protein, conjugated Centyrin and free payload for Centyrin–drug conjugate in plasma and tissue samples using liquid chromatography–tandem mass spectrometry, Bioanalysis 2018 10:20, 1651-1665
He, R.; Finan, B.; Mayer, J.P.; DiMarchi, R.D. Peptide Conjugates with Small Molecules Designed to Enhance Efficacy and Safety. Molecules 2019, 24, 1855.
Wang Y., Cheetham A.G., Angacian G., Su H., Xie L., Cui H., Peptide–drug conjugates as effective prodrug strategies for targeted delivery. Advanced Drug Delivery Reviews 2017, 110–111,112-126.
https://www.tarvedatx.com/pipeline
https://www.bicycletherapeutics.com/wp-content/uploads/22_AACR-BT1718_-01-04-2017-Abstract-No-1167.pdf
https://clinicaltrials.gov/ct2/show/NCT03486730
M. Gooding et al. (2016) Oligonucleotide conjugates – Candidates for gene silencing therapeutics, European Journal of Pharmaceutics and Biopharmaceutics, 107, 321-340.
Winkler J. (2013) Oligonucleotide conjugates for therapeutic applications. Therapeutic Delivery,4(7):791-809. I. Dovgan et al. (2013) Antibody−Oligonucleotide Conjugates as Therapeutic, Imaging, and Detection Agents, Bioconjugate Chemistry 2019 30 (10), 2483-2501.
