Background:
Antibody–drug conjugates (ADCs) combine the precision targeting of monoclonal antibodies with the potency of cytotoxic payloads. While this targeted approach holds great promise, excessive payload potency can lead to dose-limiting toxicities, undermining the therapeutic potential.
The Challenge:
A client approached us with an ADC candidate that demonstrated strong anti-tumor efficacy but suffered from an unacceptably toxicity. The root cause was a payload whose potency, while effective against tumor cells, also produced measurable off-target toxicity. The objective was clear yet complex: reduce toxicity without compromising the therapeutic benefit.
Our Approach:
We used a data-driven, AI-assisted workflow to redesign the ADC payload:
Data Analysis – Mined preclinical datasets and applied predictive models to link payload potency with toxicity thresholds.
AI SAR Modelling – Predicted how structural tweaks would affect potency, stability, and selectivity, and filtered candidates via in silico simulations.
Payload Redesign – Made targeted chemical changes and optimized the linker to reduce potency, slow kill kinetics, and ensure tumor-specific release.
Outcome:
The optimized ADC exhibited:
Impact:
By integrating AI-driven predictive modelling with medicinal chemistry expertise, we transformed a high-risk ADC candidate into a safer therapeutic with a broader therapeutic index. This approach demonstrates how computational tools can accelerate and de-risk ADC development, enabling smarter, faster design cycles.
Key Insight:
In ADC development, raw potency is not always the goal. AI-enabled optimization can identify the balance point between efficacy and safety, turning a promising yet problematic candidate into a viable clinical contender.