1. When Discovery Is Mistaken for Development: An Early Red Flag in Antibody Pipelines
One of the most common and costly sources of confusion in antibody programs is the blurred line between discovery and development. While discovery is about generating and optimizing options, development is about committing to one defined molecular entity and turning it into a reproducible, safe, and manufacturable product. Toxicology marks the point at which that commitment becomes explicit and testable.
For bispecific and multispecific antibodies, this transition is more challenging than it appears. Their novel architecture introduces additional layers of molecular complexity that can amplify risks to immunogenicity, stability, and manufacturability if not fully understood at the point of candidate lock. Molecular identity is therefore not defined by amino-acid sequence alone, but by a constellation of attributes, including correct heavy- and light-chain pairing, preservation of higher-order structure, surface charge distribution, hydrophobicity, and the consistency of post-translational modifications such as glycosylation, oxidation, or deamidation [1].
These properties, along with the molecule’s structural format and stability under relevant formulation and dosing conditions for toxicology and early clinical studies, ultimately determine whether the “same” molecule behaves consistently in vivo [1]. Locking a candidate before these dimensions are sufficiently characterized and controlled only displaces risk downstream. Liabilities frequently surface during toxicology evaluation and ultimately come under regulatory scrutiny at IND review.
2. How Oversimplified Rankings Create Late Stage Antibody Failures
Drug quality does not exist on a single axis. Potency, exposure, stability, safety, manufacturability, and analytical clarity all matter. For complex antibody formats, these attributes are tightly coupled and often in tension with one another. Unlike conventional monoclonals, these formats introduce additional structural, biophysical, and functional dimensions that directly affect developability. Correct chain pairing, higher-order structure, domain orientation, linker design, and local charge or hydrophobicity can all influence not only binding activity, but also expression yield, aggregation propensity, stability, pharmacokinetics, and immunogenic risk [2]. Optimizing one dimension, such as improving binding affinity or dual target engagement, can easily degrade another, such as manufacturability or safety margin. Many of the most consequential liabilities in bispecific and multispecific molecules are emergent properties that cannot be predicted from any single assay or metric. For example, modest increases in hydrophobic surface area may enhance target binding but simultaneously increase aggregation, alter clearance, or complicate purification. Similarly, format specific features that improve functional potency may narrow the therapeutic index by amplifying on target biology in non diseased tissues [2]. These trade-offs cannot be meaningfully captured by single number rankings without obscuring where real risk resides. Experienced developers, therefore, adopt staged, multidimensional decision frameworks, deliberately re-prioritizing parameters as programs mature. Early discovery may emphasize biological relevance and functional proof, while later stages shift weight toward exposure, safety margin, stability, and process robustness [1].
3. The Potency Trap: How “Strong” Antibodies Fail in Toxicology
Understanding why potency fails as a proxy for safety requires clarity on what the toxicology stage is fundamentally designed to answer. The primary goals of nonclinical toxicology for biotherapeutics extend beyond identification of target organs of toxicity to establish the relationship between exposure, target engagement, and safety margins that inform first-in-human dose selection and clinical monitoring strategies [3]. Achieving these objectives relies heavily on a thorough understanding of target biology, mechanisms of action, exposure-response relationships, pharmacokinetics and pharmacodynamics, and the selection of a relevant animal model for pharmacological studies. For biologics, the availability of pharmacologically relevant species is often a limiting factor, which can impact both study design and interpretation of toxicology findings. Critically, the absence of toxicity findings in a pharmacologically irrelevant species cannot be interpreted as evidence of safety. For antibody-based therapeutics, toxicology findings are meaningful only when observed in an appropriate species, most commonly the cynomolgus monkey, where target engagement and biological responses are conserved [3]. What toxicology ultimately probes is therapeutic index: the margin between pharmacological effect and unacceptable toxicity. Potent molecules with narrow safety margins frequently fail not because the science is flawed, but because early programs prioritized activity while under‑examining the safety signals that ultimately determine clinical viability [4,5]. This distinction helps explain why strong on‑target activity, while compelling, is often misleading in early antibody discovery. For immune cell engagers, this risk is magnified as mechanisms designed to activate immune responses can mask emerging safety liabilities until systemic exposure reveals them, such as on-target off-tumor toxicities, cytokine release syndrome, and immune effector cell-associated neurotoxicity syndrome (ICANS) [6,7]. By the time overt toxicity appears in vivo, teams may already be deeply invested in a candidate that cannot be safely dosed. decisions.
4. The Toxicity Distinction That Determines Whether an Antibody Is Developable
Off-target toxicity arises when an antibody drug binds unintended molecules outside its intended target, but these liabilities are not uniform. This type of non-specific polyreactivity, driven by unfavorable biophysical features, such as exposed hydrophobic or charged surfaces, is often detectable early and can sometimes be mitigated through engineering [8].
Specific off-target binding, in contrast, is far less predictable and may emerge only in vivo, where it can manifest as unexpected toxicity, altered pharmacokinetics, or reduced therapeutic index, often late enough to constrain program flexibility [8].
On-target toxicity presents a more fundamental challenge. Many antibody targets are not uniquely expressed on diseased cells but are also present, albeit to varying degrees, on healthy tissues. When target expression is insufficiently differentiated, engagement of the intended target itself drives toxicity (“on-target, off-tissue” effects), reflecting a biological incompatibility with systemic dosing [9].
Unlike many off-target liabilities, on-target toxicity is often driven by underlying target biology and is rarely resolvable through molecular engineering alone. Attempts to rescue such programs typically require targeted delivery or conditional activation strategies, effectively redefining the modality and resetting the development, CMC, and regulatory path.
For bispecifics and multispecific antibodies, simultaneous engagement of multiple targets further narrows safety margins, making early differentiation between off target risk and on target limitation one of the most consequential decisions a program will face.
5. CMC Red Flags That Surface as Safety Issues in Tox and IND Review
In next generation antibody programs, CMC choices are not neutral with respect to safety. Poor solubility, aggregation, heterogeneity, or instability can directly influence toxicology outcomes and interpretation. Red flags often appear early:
- Inherent molecular complexity presents a key challenge: controlling heterogeneity and ensuring consistent product quality
- High aggregation propensity under stress conditions
- Charge or glycosylation heterogeneity that complicates analytical clarity
Toxicology studies do not test abstract biology; rather, they test specific materials produced through defined processes, formulations, and controls. When toxicology studies are conducted using material that differs meaningfully from what is later manufactured under GMP, regulators may question both the relevance and reliability of safety conclusions, particularly for complex formats. For next generation modalities, toxicology success is often determined months earlier by decisions around cell line strategy, analytical depth, formulation readiness, and process robustness.
