The project will be done in the Predictive Oncology laboratory at IPC, which acquired during the last decades strong experience in translational research on BC, omics profiling (F. Bertucci et al. 2001), bioinformatics, pre-clinical models and ADCs (Cabaud et al. 2022; M-Rabet et al. 2017).

WP1 – Retrospective assessment of expression/co-expression of ADC targets in clinical and pre-clinical BC samples, and pharmacological proof-of-concept FoR ADC combination.

WP1.1. RNA expression/co-expression of ADC targets in clinical and pre-clinical BC samples.

We build a gene expression database, including >14.000 tumors (mainly early BC, but also advanced BC, 50 BC cell lines, 15 patient-derived xenografts (PDX)) profiled at the bulk level and 30 tumors profiled at the single-cell (sc) level. They were gathered from 353 patients treated in our institution and from 36 public databases ( Bertucci et al. 2018). For each sample, the gene expression profile (DNA microarrays or RNA-Seq) and the clinicopathological annotations are available. The mRNA profiling will be completed using RNA-sequencing (ThruSeq Illumina) for 35 additional PDX collected in our PERMED-01 precision medicine trial (Bertucci et al. 2021). All molecular subtypes are represented. We will interrogate these databases for a non-exhaustive list of ADC targets including HER2, HER3, LIV1, TROP2, Nectin-4, folate receptor alpha, tissue factor, ROR, ADAM9, MUC1, …. Analysis of bulk mRNA (“tumor resolution”) will define the percentage (%) of tumors expressing each target and % of tumors co-expressing targets. Analysis of sc-RNA (“single-cell resolution”) will provide the % of cancer cells “simple-positive”, “double-positive”, “triple-positive”… in each tumor, and will define the tumors co-expressing different targets on different cancer cells and on the same cancer cells. A score of heterogeneity for ADC targets expression will be established.

WP1.2. Protein expression/co-expression of ADC targets in clinical and pre-clinical BC samples.

We will use immunohistochemistry (IHC) and multiplex immunofluorescence (mIF). The CRCM recently acquired the MACSima Imaging Platform from Miltenyi allowing mIF. Of note that most antibodies-based assays for the mentioned ADC targets are already developed in our lab. We inventoried 500 FFPE BC samples from consenting patients treated in our institution (including those whose gene expression profile is available). We will perform 3 cores per FFPE sample and spot them onto tissue microarrays. All pre-clinical models will be profiled at the IHC and mIF levels. Analysis of IHC will correspond to the same data as bulk mRNA, and analysis of mIF data will provide the same data as sc-RNA, but at the protein level. Correlation between mRNA and protein levels will be established.

WP1.3. Proof-of-concept of dual targeting of ADC targets in pre-clinical BC samples.

For this proof-of-concept study, we will analyze 2 BC cell lines and 2 PDXs and treat them with ADCs alone (monotherapy) and in combination (bi-therapy). We will focus on ADCs available (Evidentic (https://evidentic.com), hospital pharmacy). Data from WP1.1./1.2. will allow the selection of the most relevant models (in term of ADC targets co-expression) on which the drug testing with the corresponding ADCs will be applied. We will compare the bi-therapy versus monotherapy results in terms of efficacy (synergy? antagonism?) in vitro and in vivo and of toxicity in vivo. The efficacy endpoint will be cell viability in vitro, and tumor growth and RFS time in vivo. Other parameters of treatment efficiency will be assessed in both in vitro and in vivo models, comparatively between mono- versus bi-therapy, such as degree of diffusion of ADCs within the tumor (Kopp et al. 2023) and degree of internalization within the cancer cells (M-Rabet et al. 2017).

WP2 – Retrospective assessment of expression of ADC targets with therapeutic vulnerability GES in clinical and pre-clinical BC samples, and pharmacological proof-of-concept.

WP2.1. RNA expression of ADC targets with therapeutic vulnerability GES in clinical and pre-clinical BC samples.

From the data mentioned in WP1.1., we will also establish the vulnerability GES profile of the >14.000 tumors and pre-clinical models. The non-exhaustive list of assessed GES includes signatures predictive for response to drugs marketed or in development in BC (hormone therapy, chemotherapy, HER2 inhibitors, CDK4/6 inhibitors, ICI, PARP inhibitors, AKT inhibitors, PIK3CA inhibitors…). Analysis of bulk RNA data will provide the same data as WP1.1., but in term of co-expression of ADC target with vulnerability GES.

WP2.2. Spatial expression of ADC targets and associated therapeutic vulnerability GES in clinical and pre-clinical BC samples.

Here, this analysis will concern only the tumor samples profiled at both mRNA and protein levels (IPC samples) and all our pre-clinical models based on the results established with WP1.2 and WP2.1. In addition, we will analyze heterogeneity of therapeutic vulnerability GES between different regions of 20 selected tumors heterogeneous with respect to expression of ADC target (for example, Trop2-low and Trop2-high tumor regions as presented in Figures 3C2 and 3C3; unpublished preliminary data). This will be done using the RNA/protein multiplex spatial detection system or GeoMx Digital Spatial Profiler Platform from NanoString (whole transcriptome level). This technology now allows to integrate high-plex and high-throughput data about intra-tumor spatial heterogeneity to characterize tumors in depth . AI-assisted image analysis for cell dimensionality reduction, clustering, and subtyping will be done. Multivariate regression models linking staining patterns identified by machine learning and vulnerability GES will be undertaken.

WP2.3. Proof-of-concept of dual targeting of ADC targets and non-ADC drugs in pre-clinical BC samples.

Data from WP2.1./2.2. will allow selection of the most relevant (ie. with heterogeneous expression for a given ADC target) models on which the corresponding drugs (combinations versus monotherapy) will be tested in vitro and in vivo (as in WP1.3). The idea is to identify a non-ADC drug, based on GES, potentially most efficient on cells that do not express the ADC target and to test its combination with this ADC, that will target the other cells. By targeting more tumor cells, we expect to increase the tumor’s response and limit the recurrence.

1. Significance and Statement of Relevance

ADCs are rapidly changing the therapeutic landscape of oncology where they are among the fastest-growing drug classes. Beyond the unbridled expansion of ADCs’ market, it is incumbent upon the academic community to build on pre-clinical and biological data to design new drug combinations and clinical trials. In BC, ADCs appear to have a spectrum of activity across different molecular subtypes, and we already face the situation where two ADCs might have survival benefits in similar setting without available predictive biomarkers or information regarding the best sequence of use (Modi et al. 2022; Rugo et al. 2022). Resistances to ADCs are inevitable, and efforts to limit their emergence are warranted. Combinations of ADCs with other systemic therapies such as immune checkpoint inhibitors or DNA-damaging agents are being assessed in clinical trials (Drago, Modi, and Chandarlapaty 2021). To our knowledge, the combination of ADCs with different targets and/or payloads has not been assessed in the literature. However, we hypothesize that such ADC combination might increase the efficacy of treatment, compared to monotherapy. Better assessment of intra-tumor BC heterogeneity may help to design more adapted drug combinations, and now may be evidenced using single-cell level analyses. Given the lack of data concerning i) expression/co-expression of ADC targets in BC, ii) co-expression of ADC targets with therapeutic vulnerability signatures, and iii) the testing of ADC combinations, we designed this project. Our results on clinical samples will provide for the first-time information about the proportions and the type of tumors showing co-expression and consequently being potentially candidates for therapeutic targeting by ADCs-based combinations. The fully characterized pre-clinical models will allow us to assess the efficacy and toxicity of ADC combinations. If the results are positive, it is planned (advanced discussions) to test our hypothesis in clinical trials with the French UniCancer Breast cancer Group (UCBG). These models will represent precious tools for further analyses of mechanisms of resistance to ADCs alone and/or in combination.

In conclusion, this original project has the potential to improve the use of ADCs and the development of ADC-based combination strategies in BC.