Engineering Tumor Distribution of Antibody-Drug Conjugates

Abstract

Antibody-drug conjugates (ADC) are sophisticated pro-drugs conceptualized nearly four decades ago as Ehrlich’s elusive ‘magic bullet’ against cancer. ADCs consist of an antibody backbone that specifically targets tumor cells, conjugated to a potent cytotoxic small molecule payload efficient at killing cells. Despite the relative conceptual simplicity, its lagging clinical and commercial success has proved that ADC development is anything but simple. However, a recent surge in clinical success, evidenced by FDA-approval of five new ADCs in the last 12 months compared to four in the previous 19 years shows promise for the future of this field. This recent clinical success can be attributed to a paradigm shift towards more mechanistically driven development compared to existing empirical dogma. However, three decades of ADC failures has made one thing clear – no ‘one-size-fits-all’ strategy exists for ADC design, particularly against solid tumors. In this thesis, I incorporate theoretical dimensional analysis (Thiele Modulus, Saturation Potential, Damköhler number), in silico models (mechanistic Krogh Cylinder model), in vitro tools (3D tumor spheroids and pharmacodynamic biomarkers) and in vivo systems (primary human tumor xenograft models) to underscore how mechanistically tailored design and evaluation of ADCs is the key to sustained clinical success. The work in this thesis is categorized by approaches to design ADCs with the lens of either ‘armed antibodies’ or ‘targeted payloads’. From the first perspective, I highlight how current pharmaceutical design and screening strategies for ADCs are in direct contradiction with each other due to oversights on ADC behavior in vivo. Poor ADC penetration in a well-known issue, and a common strategy to overcome this issue is the use of smaller scaffolds as the ADC backbone to afford faster intratumoral diffusion. On the other hand, ADCs are also designed to exhibit maximum potency in 2D in vitro assays (driven by rapid internalization), and lower potency ADCs are often rejected from further development. However, as I show through my comprehensive mechanistic analysis, rapid internalization directly negates any benefits of faster diffusion and results in the same poor efficacy in vivo as a conventional IgG. I show that ADC penetration and receptor occupancy is a function of several interdependent factors such as receptor density, binding affinity, internalization rate, etc. that must be balanced for uniform ADC penetration. While these findings may appear straightforward with the benefit of hindsight, it is critical to remember that previous three decades of ADC development had unquestioningly relied on such empirical strategies. ADCs can also mediate efficacy via bystander effects, where the payload escapes ADC-targeted cells and diffuses into nearby untargeted tumor cells to mediate cytotoxicity. This work provides the first direct quantification of bystander effects with cellular resolution. Furthermore, I show for the first time, via both computational and experimental evidence, that ADC payloads possess a quantifiable bystander potential that can readily be predicted by the Damkohler number (Da). Payloads that exhibit Da ranging from 1-3 possess the ideal bystander properties and exhibit maximum bystander efficiency. This can enable mechanistically driven designs of bystander payloads to help compensate for ADC distribution heterogeneity and target antigen expression heterogeneity that is common in clinical tumors. Overall, this thesis highlights the use of several high resolution and rigorous approaches to mechanistically quantify nuanced pharmacokinetics and pharmacodynamics of ADCs, which can aid in rational design and development of ADCs for improved clinical success.