Antibody-drug conjugates

An antibody-drug conjugate is an attempt to solve chemotherapy’s central problem — cytotoxics kill dividing cells without asking which tissue they belong to — by making delivery, rather than the drug itself, carry the selectivity. The construct has three parts, and the antibody is the least interesting of them.

The path a payload must survive

The antibody binds a surface antigen, the antigen-antibody complex is internalised, and the vesicle matures into a lysosome, where either the linker is cleaved or the whole antibody is proteolysed and the payload is released inside the cell. Every step is a filter. Targets must be internalising: an antigen that sits on the surface without endocytosing makes a poor ADC target however abundant it is. HER2, TROP2, nectin-4 and CD30 are used because they internalise, not merely because they are overexpressed.

The efficiency of that path is brutal. Biodistribution work on solid tumours consistently finds that only a low single-digit percentage of an injected antibody dose reaches the tumour, and only a fraction of that is internalised. That is why ADC payloads are not ordinary chemotherapy drugs but ultra-potent agents — auristatins, maytansinoids, calicheamicin, camptothecin analogues — with sub-nanomolar cytotoxicity, far too toxic to give systemically on their own.

The linker is the design

Two failure modes bound the field from opposite sides. A linker that is too stable releases too little drug where it matters; one that is too labile sheds payload in circulation and reproduces untargeted chemotherapy. Cleavable designs exploit a difference between compartments: the valine-citrulline dipeptide in brentuximab vedotin is cut by lysosomal cathepsin B, and hydrazone or disulfide linkers respond to endosomal acidity or intracellular glutathione. Non-cleavable designs, such as the thioether in ado-trastuzumab emtansine, require full antibody catabolism and release a charged lysine- or cysteine-adduct that cannot cross membranes.

That difference decides whether an ADC has a bystander effect. A membrane-permeable released payload diffuses into neighbouring antigen-negative cells and kills them too, which is why cleavable, permeable-payload conjugates work against heterogeneous tumours and even in tumours classified as low-expressing. It is also why they are less selective.

Drug-to-antibody ratio is the same trade in another variable. Higher loading gives more payload per binding event but increases hydrophobicity, aggregation and hepatic clearance; site-specific conjugation exists because heterogeneous cysteine or lysine chemistry produces a mixture with species that behave very differently in vivo.

Where the ceiling actually sits

Dose-limiting toxicities of approved ADCs are largely explained by payload released where it was not intended, not by antigen expression on healthy tissue. Auristatin conjugates give peripheral neuropathy and neutropenia; the ocular surface toxicity of certain maytansine and MMAF conjugates reflects uptake by rapidly renewing corneal epithelium; interstitial lung disease is a recognised, labelled and occasionally fatal risk of deruxtecan-based conjugates and remains mechanistically incompletely explained. Resistance follows the same logic: antigen downregulation, impaired internalisation or trafficking, lysosomal dysfunction, and efflux-pump expression against the payload each break a different step, and a change of payload class is often more useful than a change of target.

See also the technology article at /technology/antibody-drug-conjugates-adcs/.