Abstract 661: Proteomics and selecting the right combination of target and toxin for antibody-drug-conjugate (ADC) development

With the successes of Kadcyla and Adcetris, and the emerging data from early phase trials of many other ADCs our understanding of the critical attributes for target selection and ADC development has improved significantly. Clinical experience indicates that the dose-limiting toxicities observed are predominately off-target and thus are similar across ADCs using the same linker and drug combinations and independent of the target antigen. Furthermore, the maximum tolerated doses (MTDs) in humans, monkeys, and mice for many ADCs seem to be very similar regardless of the protein being targeted. Thus, the therapeutic index of new ADCs using the current drug/linker technologies (tubulin and DNA targeting toxins) will be driven predominately by efficacy, specifically the ability to achieve antitumor activity at exposure levels below the MTD. Therefore, the focus of future target selection should be on efficacy within the therapeutic window rather than on antigen-dependent (i.e. on-target) toxicity arising from normal tissue expression. We have selected targets and toxins for ADC development based on the likelihood of robust potency against cancer cells. The properties favored are : 1) high expression and high prevalence in selected malignancies, 2) normal tissue expression profiles known to be tolerable for ADCs ( e.g. her2neu, gpnmb, nectin4), 3) targets enabling rapid intracellular delivery of ADC and release of active toxin, 4) clinical indications sensitive to the toxin of choice. This process and its success are demonstrated here through the development of two ADCs. Firstly, Ox-1476 for triple negative breast cancer (TNBC). The target was selected based on consistent high expression in TNBC, the MAb selected from a large screen of greater than 1000 candidates based on activity as an actual ADC, and the toxin selected based on the positive experience with maytansine in breast cancer (Kadcyla) and the preclinical data presented here. The Ox-1476 ADC shows greater and broader activity than Kadcyla in vitro and an initial toxicity profile consistent with pursuing clinical development. Secondly, Ox-1425 is being developed for small cell lung cancer (SCLC). The target and MAb were selected by the same processes but for Ox-1425 the toxin screen resulted in the selection of a DNA alkylating agent. Our proteomics database has provided a unique data set of candidate cell surface targets for the development of antibodies compatible with the current state of the art ADC payloads. The selected targets show high and prevalent expression in cancers, are in protein families shown to internalize efficiently, and have normal tissue expression profiles similar to other proteins already being targeted by clinical stage ADCs. The preclinical safety and activity data shown here is supportive of our strategy for selecting ADC targets. Citation Format: Jonathan A. Terrett, Rachel Dusek, Dee Aud, Rahel Awdew, Sudha Swaminathan, San Lin Lou, Michael Trang, Arnima Bisht, Mary Do, Jim Ackroyd, Robert Boyd, Lindsey Hudson, Phuoc Pham, Nickolas Attanasio, Ami Antani, Carmel Lynch, Christian Rohlff. Proteomics and selecting the right combination of target and toxin for antibody-drug-conjugate (ADC) development. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 661. doi:10.1158/1538-7445.AM2014-661