For many antibody programs, the target is compelling and the biology is convincing. The difficult part is the margin between an effective dose and activity in healthy tissue. Engineering can widen that margin, but the choices that improve selectivity can also leave some tumor cells beyond the therapy’s reach. 

Lower affinity may spare cells with low antigen expression while weakening activity against less densely expressing tumor populations. Requiring two antigens may sharpen discrimination while excluding cells that carry only one. Conditional activation may restrict activity to the tumor, provided the activating conditions are present where the drug needs to work. 

The challenge is deciding which tradeoff the disease biology can support. A useful selectivity strategy preserves enough tumor coverage to deliver meaningful activity while reducing the exposure or activation that limits treatment. That requires a closer look at how the whole molecule behaves, beyond the properties of its individual binding arms. 

Affinity needs to leave room for discrimination 

When tumor and healthy cells express the same antigen at different densities, stronger binding can work against the design. An interaction that is already very stable on a low-expressing cell leaves less opportunity for additional tumor-binding arms to create meaningful discrimination. 

Reducing individual-arm affinity can make stable engagement more dependent on multiple interactions. On a densely expressing tumor cell, avidity can sustain binding; on a lower-expressing cell, those interactions may be less likely to form. The engineering opportunity lies in the difference between those responses, rather than the strongest possible binding to the tumor. 

The complication is that tumor expression rarely falls into neat high and low categories. A design that performs well against a highly expressing model may lose activity against a substantial part of the intended tumor population. Conversely, a molecule that appears selective at one concentration may begin to affect lower-expressing cells as exposure rises. 

Affinity, valency, and geometry therefore need to support the same biological objective. Domain placement, flexibility, and epitope accessibility influence whether additional binding arms improve discrimination or simply increase engagement across both populations. 

Two targets can sharpen selectivity and narrow coverage 

Dual targeting is attractive when neither antigen offers enough selectivity alone. A combination may distinguish malignant cells more effectively, especially when each antigen is also present in a different healthy tissue. But the value of an AND-gated design depends on how consistently both targets appear together on the cells that matter. 

That makes coexpression a central development question. A tumor can test positive for both antigens overall while containing separate populations that express only one. Even within dual-positive cells, an imbalance in expression or an unfavorable spatial arrangement may limit productive engagement. Single-cell and spatial information can reveal constraints that bulk measurements leave hidden. 

OR-gated recognition takes a different approach to heterogeneity by allowing activity through either target. It can broaden coverage and reduce opportunities for antigen escape, but each additional recognition route brings another healthy-tissue expression pattern into consideration. 

The choice between these approaches turns on the program’s main limitation. Where unwanted activity prevents adequate dosing, stricter recognition may be worth some loss of coverage. Where incomplete coverage is the larger problem, adding another target may be useful, provided the broader recognition remains tolerable. 

Conditional activation depends on the conditions actually present 

Masking and pH-sensitive binding offer another route when antigen expression alone provides too little separation. They make activity depend partly on the local environment, creating an opportunity to pursue targets that would otherwise be difficult to use safely. 

The important question is how reliably that environment distinguishes the tumor. Protease activity and extracellular acidity can vary within and between tumors. A construct that activates strongly under a favorable laboratory condition may encounter regions where activation is incomplete. Residual activity outside the intended setting matters too, particularly for a potent immune engager. 

This is why the size of the change between activating and nonactivating conditions tells only part of the story. A large difference can coexist with biologically meaningful activity in the nominally inactive state. It can also conceal weak activity under the less favorable conditions found within the tumor. 

pH-dependent engineering deserves particular care because it serves several purposes. Binding more strongly in an acidic tumor environment and releasing a target in an acidic endosome address different problems. Improved intracellular release may change recycling or pharmacokinetics without establishing better discrimination between tumor and healthy tissue. 

The immune response adds another layer of selectivity 

For immune-cell engagers, selectivity also depends on which cells respond and how they are stimulated. Tumor recognition establishes where an interaction can occur, but the effector population and signaling pathways influence what happens next. 

Approaches that favor selected T-cell populations, natural killer cells, or other effectors aim to shape that response more precisely. Costimulatory domains add another opportunity to strengthen activity in settings where the initial activation signal is insufficient. Their value depends on whether the added stimulation remains tied to the intended target context. 

That connection can be sensitive to architecture. Changing the position or affinity of an immune-binding domain may alter tumor killing, proliferation, and cytokine release differently. A gain in one response does not establish a favorable balance across the others. 

The most informative comparison is therefore broader than peak cytotoxicity. It considers whether the molecule sustains useful activity while limiting activation without the intended tumor context. This keeps the focus on the quality and location of the response, rather than treating stronger stimulation as an improvement in itself. 

Layered selectivity needs a clear purpose 

Combining recognition, conditional activation, and localized delivery can address limitations that one mechanism cannot resolve. The temptation is to assume that more layers will produce a better therapeutic window. Each layer, however, also adds conditions that must be satisfied for the drug to work. 

A construct might recognize the right antigen pair but activate poorly in part of the tumor. Another may show strong cellular selectivity yet distribute unevenly through the tissue. More elaborate formats can also introduce expression, assembly, or stability problems that limit the practical benefit of their targeting logic. 

Added complexity is most useful when it addresses a specific weakness in the simpler design. That gives the team a concrete reason to retain it and a fair basis for judging the tradeoff. It also keeps development focused on the biological problem rather than the number of functions the molecule carries. 

Look for selectivity that survives biological variation 

The most persuasive evidence shows how a candidate behaves between idealized tumor-positive and target-negative conditions. Intermediate antigen expression, uneven coexpression, relevant healthy cells, and mixed-cell environments help reveal where the intended discrimination holds and where it begins to weaken. 

These comparisons give meaning to a selectivity claim. They connect the exposure needed for useful tumor coverage with the conditions that produce unwanted activity. They also help explain whether the next engineering step should adjust affinity, change the architecture, or reconsider the underlying targeting strategy. 

Antibody engineering offers considerable freedom to shape therapeutic activity. The strongest designs use that freedom to preserve efficacy across the variation the drug is likely to encounter while creating enough separation from healthy-tissue effects to make the target therapeutically useful. 

Continue the conversation at PEGS Summit Europe and PEGS AI

Explore antibody selectivity, multispecific engineering, and immune-cell engagement at PEGS Summit Europe. Join scientists and industry leaders discussing how molecular design can translate into a stronger therapeutic window. 

Take the insights and excitement from PEGS Europe right into PEGS AI in January. Register today.