Pharmacokinetic Considerations to Maximize an ADC’s Therapeutic Index

Featured Image: Ado-trastuzumab emtansine (Kadcyla; Genentech/Roche), known as T-DM1 during. Courtesy: Genentech/Roche.
Featured Image: Ado-trastuzumab emtansine (Kadcyla; Genentech/Roche), known as T-DM1 during. Courtesy: Genentech/Roche.

The Antibody-drug Conjugate (ADC) field is poised for several significant advances in solid tumors as the industry has built upon knowledge gained over the past several years.

The biggest challenge facing clinical translation of these drugs is improving the therapeutic index (TI). With the use of proven targeting agents (monoclonal antibodies) and proven cancer cytotoxins (e.g. microtubule inhibitors and DNA alkylating agents), increasing delivery to cancer cells while avoiding healthy tissue is the key issue in improving the TI.

While simple in concept, the implementation is complicated in practice. ADCs combine the specificity of antibodies and the potency of small molecule drugs, but they also combine the tissue penetration issues of antibodies and systemic toxicity of potent small molecules. Fortunately, the pharmacokinetic driving forces behind these delivery issues are known, and we can leverage this knowledge in a predictive manner to design more effective agents.

MabPlex
ADC Bio
Lonza
 

The ADC delivery problem can be described succinctly: the majority of the ADC never reaches the tumor, and what does is typically heterogeneously distributed. Because many antibodies are well-tolerated (e.g. the maximum tolerated dose of trastuzumab was never reached), the poor tumor uptake and distribution of unconjugated antibodies can often be overcome by increasing the dose. However, the toxicity of the payload prevents this approach with ADCs, and this can be target or non-target mediated, so selecting ‘cleaner’ targets does not completely solve the issue. Therefore, a limited amount of ADC can be safely delivered to the patient, treating only a fraction of the tumor. With a limited number of arrows in the quiver, each ADC dose needs to be designed to hit as many cancer cells as possible.

Matching Cellular Delivery and Potency to Improve the Therapeutic Index
A key concept to improve ADC efficacy at equal or lower systemic toxicity is to match the cellular-level potency with the cellular delivery. These factors are complex functions of the payload physicochemical properties, mechanism of action, drug-antibody-ratio (DAR), (accessible) target expression, cellular trafficking, and payload release efficiency, among other considerations. The requisite tools and models are available to assess these factors; however, the outcome of this matching can be counter-intuitive.

Using ado-trastuzumab emtansine (T-DM1; Kadcyla®; Genentech/Roche) as a model system, we demonstrated how co-administering unconjugated antibody (trastuzumab) with T-DM1 can increase the tissue penetration while still delivering a lethal number of payloads per cell. Using a constant dose of T-DM1 and increasing ratios of trastuzumab in a mouse model that is resistant to the antibody alone, the tumor growth inhibition and survival all increased. This is counter-intuitive because the addition of trastuzumab to T-DM1 in cell culture lowered the efficacy and the antibody has no measurable effect by itself in the xenograft, yet the combination improved efficacy in vivo. [1]

Due to rapid antibody binding compared to slow diffusion in tumor xenografts, perivascular cells receive many times more payload than required for cell death. Essentially, the unconjugated antibody improves tumor penetration by lowering the effective potency to better match cellular delivery in a high expression system. The combined dose reaches more cancer cells and as a result provides more efficient tumor killing. While the experiment was designed to isolate the impact of payload heterogeneity, additional benefits, such as reduced target-mediated drug disposition, increased Fc-effector function, and/or improved receptor signaling blockade would also be expected in other systems.[2]

The above approach combined unconjugated antibody with the ADC due to its simplicity and ability to adjust the ratio as a function of patient antigen expression (e.g. personalized medicine). For example, a patient with IHC 3+ staining may benefit from this combination while a patient with lower expression (e.g. IHC 2+) might receive just the ADC. However, the broader concept can be applied by reducing the average DAR (rather than combining with ‘DAR 0’ to achieve this effect) or using lower potency payloads. Decreasing the DAR also has the potential to reduce DAR-dependent clearance and DAR-dependent deconjugation to improve total tumor uptake, while the use of lower potency payloads, such as topoisomerase inhibitors, can increase tolerability, allowing higher doses to reach more cells.

While it is difficult to directly test this effect in the clinic, there are examples that suggest this result holds true (although alternative hypotheses cannot be ruled out). For example, the CA-125 ADC DMUC5754A (sofituzumab vedotin), given at a 2.4 mg/kg dose with DAR 3.5 resulted in a response rate of 17%. The next generation lower DAR (DAR 2) Thiomab version of this ADC, DMUC4064A, could be delivered at a 5.2 mg/kg dose resulting in a 45% response rate.

Improving Dosing Regimens to Optimize Therapeutic Index
Aside from the ADC design, the dosing regimen can be optimized to increase the TI. Several groups have explored fractionated dosing experimentally, and in collaboration with Prof. Jennifer Linderman at the University of Michigan, we have tested outcomes using an agent-based systems pharmacology model. A predictive computational model can quickly and inexpensively be used to focus experimental work in a few promising directions. The simulation accounts for systemic pharmacokinetics (clearance, deconjugation), tissue-level heterogeneity, cellular ADC processing, and single-cell division, drug-induced death, and removal.

Image of a mouse xenograft treated with 3:1 co-administered trastuzumab:T-DM1 (white) surrounding tumor blood vessels in red (left). The simulations (middle) capture individual cell division and payload death to predict response. The results show that fractionated dosing reduces efficacy due to lower tissue penetration but can improve efficacy if higher doses are tolerated (right).
Image of a mouse xenograft treated with 3:1 co-administered trastuzumab:T-DM1 (white) surrounding tumor blood vessels in red (left). The simulations (middle) capture individual cell division and payload death to predict response. The results show that fractionated dosing reduces efficacy due to lower tissue penetration but can improve efficacy if higher doses are tolerated (right).

The modeling results have two main findings backed by experiments from the literature. First, fractionated dosing by itself will lower efficacy due to the lower Cmax from reducing the individual doses. Cmax is correlated with the tissue penetration of the ADC, which is why a higher bolus dose can be more effective in animal models. However, the tolerability of ADCs often improves with dose fractionation. The higher total dose delivered with fractionation can more than compensate for the lower bolus dose, thereby resulting in an improved TI with dose fractionation. Overall, if a higher total dose can be given, dose fractionation is predicted to be beneficial, while if the same total dose is administered (e.g. if healthy tissue recovery is not fast enough to allow higher dosing), dose fractionation is detrimental.

Tools for Matching Efficacy
What is truly exciting is to see how quickly the tools and application of these concepts, based on lessons learned in the clinic, have influenced the benchtop and cycled back to the clinic. Payloads with a range in potencies are now available, not just across classes (e.g. microtubule inhibitors, topoisomerase inhibitors, etc.) but also within classes (e.g. DNA alkylating agents). Linker technologies and conjugation strategies have improved to enable precise and controllable DAR. These technologies are now employed in multiple clinical trials including the use of lower potency topoisomerase inhibitors and reducing the DAR from 8 to 4 (and 4 to 2) for several ADCs to better match delivery with potency and improve the TI.

The complexity of ADC design remains challenging, and no ‘universal rules’ apply to all targets and agents. Early ADCs used traditional chemotherapeutics (e.g. methotrexate) that lacked sufficient potency, and current low expression targets require high potency and/or higher DAR to deliver a toxic dose. However, the tools are now available to optimize rather than simply maximize properties. By leveraging predictive computational approaches, the design of ADCs can be approached comprehensively rather than a linear in vitro potency to in vivo efficacy to toxicity design. The clinical endgame can therefore be considered at the initial stages of discovery so that decisions made early in the program aren’t focused on in vitro potency but rather focused on maximizing the TI and clinical efficacy.

This article is based on a workshop “Modelling & Simulation of Antibody-Drug Conjugate Pharmacokinetics & Pharmacodynamics (PK/PD) in Drug Development” during the World ADC, held October 8 – 11, 2019 in San Diego, CA,

References
[1] Cilliers C, Menezes B, Nessler I, Linderman J, Thurber GM. Improved Tumor Penetration and Single-Cell Targeting of Antibody-Drug Conjugates Increases Anticancer Efficacy and Host Survival. Cancer Res. 2018;78(3):758-68.
[2] Nessler I, Khera E, Thurber GM. Quantitative pharmacology in antibody-drug conjugate development: armed antibodies or targeted small molecules? Oncoscience. 2018;5(5-6):161-3.


DOI: https://doi.org/10.14229/jadc.2019.11.12.001


How to cite:<
By: Thurber GM. Pharmacokinetic Considerations to Maximize an ADC’s Therapeutic Index – J. ADC. November 12, 2019. DOI: 10.14229/jadc.2019.11.12.001


Last Editorial Review: November 11, 2019

Creative Commons License

Article History:

  • Original Manuscript Received: November 4, 2019
  • Review results received: November 7, 2019
  • Manuscript accepted for publication: November 11, 2019
  • First Published Online: November 12, 2019.

 

Previous articleFrom First-in-Human to Post-Commercial: ADC Reference Materials and Specifications
Next articleAntibody-drug Conjugates with Novel Kinesin Spindle Protein Inhibitor Payloads and a Tailor-made Linker Chemistry
Greg Thurber
Greg M. Thurber is Associate Professor of Chemical Engineering and Biomedical Engineering at the University of Michigan. His work focuses on applying fundamental biotransport principles to design novel therapeutics and molecular imaging agents. These widely applicable techniques have found use in the Thurber laboratory in diverse projects ranging from self-administration of near-infrared molecular imaging agents for early disease detection to the modular design of stabilized alpha helices using bioorthogonal chemistry. The concepts for monoclonal antibodies and small molecule payloads have found application in designing improved antibody-drug conjugates for cancer therapy. Prof. Thurber received his training in protein therapeutics at MIT under the guidance of Dr. Dane Wittrup. He completed his in vivo training in molecular imaging in the laboratory of Dr. Ralph Weissleder at Mass General Hospital and Harvard Medical School. During his career, he has delivered over 50 invited talks at major pharmaceutical companies, national and international conferences, and university departmental seminars. He also has consulting/research contract affiliations with more than 15 different companies. Prof. Thurber has authored 40+ peer-reviewed journal publications, 3 book chapters, numerous conference proceedings, and his work has been featured in popular news outlets including NPR’s “All Things Considered” and Smithsonian Magazine. He has received several awards including an NIH K01 award and the National Science Foundation CAREER award