Abstract PO1-14-04: A multivariate biomarker to guide antibody-drug conjugate selection and provide insight on response differences across breast cancer subtypes

Abstract

Abstract Background: Antibody-drug conjugates (ADCs), including trastuzumab deruxtecan (T-DXd; targeting HER2) and sacituzumab govitecan (SG; targeting TROP2), have transformed the management of metastatic breast cancer, with additional ADCs approved in other solid tumors or in late-stage development. To date, expression of ADC target alone has been poorly predictive of objective response rates (ORRs) in both breast cancer and other tumors. Hence, there is an urgent need to develop better predictive biomarkers to guide ADC vs. other treatments, T-DXd vs. SG treatment, and optimized predictive medicine opportunities involving other ADCs. Additionally, whether ADCs are similarly effective in less common breast cancer subtypes—such as invasive lobular carcinoma (ILC) vs. invasive ductal carcinoma (IDC), is unclear. Given the limited availability of trial tissue samples and cohorts with long-term follow-up, herein we sought to determine whether a tissue based, pan-solid tumor, multivariate biomarker algorithm could predict observed ORRs across ADCs and tumor types, as has been used previously to associate tumor mutation burden with immunotherapy ORRs. Methods: From 15,032 FFPE tumor tissue samples (from 21 tumor types) tested by clinical comprehensive genomic profiling plus RNA based quantitative transcriptional profiling (qTP) as part of the observational Strata Trial (NCT03061305), the ADC treatment response score (TRS) was discovered and validated to predict published ORRs across tumor types and approved/late-stage ADCs (n=16 observed ORRs [from 7 tumor types and 8 ADCs; SG not included]). The best performing 3-factor algorithm (by Pearson correlation coefficient of observed ORRs vs. tumor type and ADC specific predicted biomarker positivity rates) included only qTP components and combined ADC target expression, cell proliferation, and extracellular matrix adhesion (the latter being negatively associated with ORRs). Importantly, predictive biomarker positivity rates of TRS was more correlated vs. observed ORRs (n=16, r=0.81, p=0.0001) than target expression alone (n=16, r=0.54, p=0.03). TRS was then validated using held out SG ORRs from nine tumor types in the IMMU-132-01 basket trial and two ADC ORRs from ASCO 2023 abstracts (enfortumab vedotin [EV; targeting NECTIN-4] in head and neck cancer and patritumab vedotin [targeting HER3] in breast cancer), with TRS predictive biomarker positivity rates again being more significantly correlated vs. observed ORRs than target expression alone (n=11, TRS r=0.91, p=0.0001; target expression alone r=44, p=0.17). Lastly, the locked TRS model was then applied to the DESTINY-PanTumor02 T-DXd dataset presented at ASCO 2023 (n=21 tumor type/HER2 expression groups matched to the trial groups), with TRS predictive biomarker positivity rates again being highly correlated vs. observed ORRs (n=21, TRS r=0.80, p< 0.0001). Across the 21 tumor types in the Strata Trial dataset, breast cancer had the highest percentage (76%) of patients predicted positive for at least one ADC, with 23% and 55% positive for the approved ADCs mirvetuximab soravtansine (targeting FOLR1) and EV, respectively. Lastly, patients with ILC vs. IDC had significantly greater TRS positivity rates for T-DXd (61% vs. 47%) and SG (57% vs. 48%) due to significantly decreased extracellular matrix adhesion in ILC vs. IDC. Clinical outcomes data for patients treated with ADCs and available TRS enrolled in the Strata Trial are maturing and will be presented at the meeting. Conclusion: We have developed and validated TRS, a multivariate RNA based tumor tissue algorithm that predicts observed ORRs across tumor types and approved/late-stage ADCs. More than 75% of all patients with metastatic breast cancer are predicted to be responsive to one or more ADCs, including those approved in other tumor types. Citation Format: Laura Lamb, Jonathan Mowers, Azadeh Nasrazadani, Mark Burkard, Nickolay Khazanov, Daniel Hovelson, Kat Kwiatkowski, Scott Tomlins, D. Bryan Johnson, Daniel Rhodes. A multivariate biomarker to guide antibody-drug conjugate selection and provide insight on response differences across breast cancer subtypes [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO1-14-04.