ADC biomarker development is outgrowing target expression alone
The rapid spread of antibody–drug conjugates across oncology has exposed a familiar problem: the biomarker strategy has lagged behind the drugs themselves. Semi-quantitative immunohistochemistry still dominates clinical selection, yet ADC activity depends on far more than whether a target is simply present on the tumour surface. Antigen abundance, accessibility, intracellular trafficking, linker cleavage, payload susceptibility and the surrounding tumour microenvironment all shape whether a patient responds or fails therapy.
One sentence captures the central problem. A positive stain does not guarantee drug delivery, drug release or tumour kill.
The review argues for a more modular biomarker framework because the biology of ADCs is modular by design. The antibody guides the conjugate to its target, the linker governs when and where the payload is liberated, and the payload determines the cytotoxic mechanism. Each element can be disrupted by tumour heterogeneity, variable antigen internalization, lysosomal processing differences or resistance mechanisms within the cancer cell itself. Under those conditions, reliance on a single staining threshold becomes a blunt instrument for a treatment class that is anything but blunt.
Why immunohistochemistry is no longer enough
Conventional immunohistochemistry remains useful, but it only captures a narrow slice of ADC biology. It gives a mostly static view of expression and usually compresses complex spatial patterns into coarse categories. In practice, two tumours with the same nominal target score may behave very differently if one expresses the antigen diffusely and the other displays patchy, inaccessible, or functionally irrelevant protein.
Spatial and functional assays offer a more complete read on whether an ADC can actually do its job. Quantitative proteomics, spatial profiling and other emerging platforms may better measure antigen abundance and distribution, as well as the cellular context that shapes uptake and processing. These approaches are attractive because they can distinguish between mere presence of a target and biological accessibility to the conjugate. They also better reflect the real sequence of events required for efficacy: binding, internalization, processing and payload release.
Payload biology is becoming central to sequencing decisions
The framework also places unusually strong emphasis on payload susceptibility and cross-resistance. This is not an abstract laboratory concern; it is becoming a bedside issue as more ADCs share similar payload classes, especially topoisomerase I inhibitor payloads. When two agents rely on related cytotoxic mechanisms, prior exposure to one may blunt the effect of the next even if the target is different.
Sequential use of ADCs therefore cannot be planned around target expression alone. The biology of the payload matters when a tumour has already adapted to one cytotoxic pressure and has begun to evade another with the same downstream mechanism. Resistance may emerge at the level of drug transport, DNA damage response, cell-cycle handling or survival signaling. A biomarker framework that ignores these mechanisms leaves clinicians guessing when they switch from one ADC to another.
Toxicity biomarkers belong in the same conversation
The review gives toxicity the same conceptual weight as efficacy, and that is an important correction. Toxicity has often been treated as a dosing nuisance rather than a biomarker problem, yet adverse effects can determine whether an ADC remains usable at all. If a patient cannot tolerate the agent, theoretical efficacy never reaches the clinic.
Biomarkers of toxicity could improve dose optimization, therapeutic index and patient-centred treatment decisions. They may also help identify patients at disproportionate risk before avoidable harm occurs. This is especially relevant for ADCs because the same design features that improve tumour delivery can still create off-target effects or payload-related injury in normal tissues. A biomarker program that measures only response and ignores toxicity is incomplete by design.
Modular biomarkers could reshape trial design and regulation
The review proposes that biomarker development should move in step with ADC trials rather than trailing them. Prospectively embedded biomarkers could help define who should enter a study, which cohorts should expand, how combinations should be selected and whether sequencing rules should change after early signals emerge. Regulatory strategy is part of the same equation, because a biomarker that is useful in one tumour type may not transfer cleanly to another if antigen biology, microenvironmental context or payload sensitivity differ.
This approach also creates a stronger rationale for harmonization. Trials conducted across different tumour types and geographic regions have to contend with variation in pathology workflows, assay platforms and interpretation thresholds. Without alignment, an ADC biomarker may look promising in one dataset and unreliable in another simply because the measurement strategy changed. Standardization is not a bureaucratic add-on; it is what turns a biomarker from an interesting correlative tool into something that can guide care.
Prospective validation remains the hard gate. Quantitative and functional assays may outperform semi-quantitative IHC in principle, but they still need to prove that they improve patient selection, sequencing or outcomes before they can replace entrenched practice. The field has enough proof-of-concept biology; what it lacks is consistent clinical validation across settings.
The path forward depends on matching assays to ADC biology
The core message is straightforward: ADC biomarkers must measure more than target expression if they are going to match the complexity of the drugs themselves. A modular framework that incorporates antigen biology, linker processing, payload susceptibility and toxicity risk offers a more realistic model of treatment response. It also aligns better with the way resistance actually develops in solid tumours, where one biomarker rarely explains both benefit and failure.
The practical consequence is immediate for drug development and drug sequencing. If biomarker programs continue to rely on a single expression threshold, they will continue to miss patients who could benefit and expose others to ineffective therapy. If they expand to include spatial, quantitative and functional readouts, they may finally capture the determinants that decide whether an ADC works, fails, or causes harm.
The remaining obstacle is not conceptual. It is the lack of harmonized, prospectively validated biomarker systems that can travel across ADC trials, tumour types and regulatory settings without losing clinical meaning.