Early, precise, and safe clinical evaluation of the pharmacodynamic effects of novel agents in the intact human tumor microenvironment

Abstract

Introduction: Drug development is systemically inefficient. Research and development costs for novel therapeutics average hundreds of millions to billions of dollars, with the overall likelihood of approval estimated to be as low as 6.7% for oncology drugs. Over half of these failures are due to a lack of drug efficacy. This pervasive and repeated low rate of success exemplifies how preclinical models fail to adequately replicate the complexity and heterogeneity of human cancer. Therefore, new methods of evaluation, early in the development trajectory, are essential both to rule-in and rule-out novel agents with more rigor and speed, but also to spare clinical trial patients from the potentially toxic sequelae (high risk) of testing investigational agents that have a low likelihood of producing a response (low benefit). Methods: The clinical in vivo oncology (CIVO^®) platform was designed to change this drug development paradigm. CIVO precisely delivers microdose quantities of up to 8 drugs or combinations directly into patient tumors 4-96 h prior to planned surgical resection. Resected tissue is then analyzed for responses at each site of intratumoral drug exposure. Results: To date, CIVO has been used safely in 6 clinical trials, including 68 subjects, with 5 investigational and 17 approved agents. Resected tissues were analyzed initially using immunohistochemistry and in situ hybridization assays (115 biomarkers). As technology advanced, the platform was paired with spatial biology analysis platforms, to successfully track anti-neoplastic and immune-modulating activity of the injected agents in the intact tumor microenvironment. Discussion: Herein we provide a report of the use of CIVO technology in patients, a depiction of the robust analysis methods enabled by this platform, and a description of the operational and regulatory mechanisms used to deploy this approach in synergistic partnership with pharmaceutical partners. We further detail how use of the CIVO platform is a clinically safe and scientifically precise alternative or complement to preclinical efficacy modeling, with outputs that inform, streamline, and de-risk drug development.

Keywords: drug development; intratumoral microdosing; multidrug analyses; pharmacodynamics; phase 0; spatial profiling; tumor microenvironment.

FIGURE 1

CIVO Phase 0 Clinical Workflow (A). Drugs are mixed with CIVO GLO (Yellow or Red) under aseptic conditions in the site pharmacy, per study-specific instructions in vials and loaded into the MID transfer vessel using luer-lock syringes. (B). Once loaded, the injector is transferred from the pharmacist to the investigator performing the injection. Immediately prior to an injection, ultrasonography is used to record the dimensions of the mass, evaluate the tumor’s internal architecture, identify the optimal injection placement, and customize the MID to the patient’s tumor (adjusting the depth of needle insertion and length of deposited column). During an injection, needles are carefully inserted into the tumor, and then, upon lever actuation, are retracted slowly within the tumor tissue to simultaneously deposit distinct and trackable drug columns containing minute volumes (up to 8.3 µL) of each drug, drug combination, or control. This is performed as an outpatient procedure. (C). Following injection, 4–96 h later, the patient returns to the site for the scheduled surgical resection of the injected tumor, per the patient’s standard of care plan. (D). The excised tumor is then transferred to pathology, where the injected portion of the tumor sample is identified, using custom blue light and yellow filter lens, then cut out, sectioned transverse to the injection columns in ∼4 mm sections and placed into 10% buffered formalin containing 0.92 mg/mL sodium orthovanadate, 1.5 mg/mL sodium glycerophosphate, 1 mg/mL sodium fluoride, and 2.2 mg/mL sodium pyrophosphate and shipped at room temperature.

FIGURE 2

CIVO Phase 0 Analysis Workflow (A). Formalin-fixed tissue sections are received by Presage, processed, embedded, and cut at 4 μm onto glass slides. (B). For IHC/ISH staining, slides are stained using fluorescent detection of antibody/probe and scanned on a whole-slide scanner. Cell segmentation and biomarker analysis is performed using HALO^® (Indica Labs) to attain cell-level data. This is then measured over radial distance from the injection site to visualize the changes in drug effect across the gradient of drug exposure from the injection site. (C). For GeoMx^® DSP, slides are stained for morphology markers and a GeoMx probe mix of barcoded RNA probes, then probes are photocleaved and collected from areas of interest (AOIs) around each injection site. Probe counts are quantified using next-generation sequencing and matched to their corresponding genes, which are compared across treatments to determine differentially expressed genes between injection sites. Common signatures of drug exposure are generated across patients in addition to patient specific responses. These signatures, as well as publicly available signatures, can then be used to score the injection sites by patient and drug. (D). For CosMx^® SMI, slides are prepped with CosMx Universal Cell Characterization Panel RNA probe set, then probes are counted at a single cell level within predefined fields of view (FOVs) at each injection site. Cells are defined and clustered based on differential RNA expression. This data can then be used to spatially reconstruct the tumor areas, allowing comparisons of parallel IHC images to cell type, drug response and drug exposure. Cell type or expression data can also be used to compare FOV composition or signature scores to drug exposure quantitatively.

FIGURE 3

CIVO Compatibility with Classes of Drugs. Demonstration of delivery and efficacy of drug classes using the CIVO platform. Each column is an example of a drug class delivered clinically or preclinically, as well as sample quantitation using HALO^® of the pictured injection site compared to controls. Quantitation is % positivity unless otherwise stated. Small Molecule: Trabectedin injection site in a clinical soft tissue sarcoma patient, stained for cleaved caspase 3 (IHC, CC3, red), phospho-ERK (IHC, pERK, yellow), phospho-STAT3 (IHC, pSTAT3, cyan) and nuclei (DAPI, blue). Antibody: Nivolumab injection site in a clinical soft tissue sarcoma patient, stained for nivolumab (IHC, Nivo, yellow), CXCL9 (ISH, red), IDO1 (ISH, cyan) and nuclei (DAPI, blue). Top axis is Nivolumab % positive. Combination: Vehicle (V), subasumstat (A), cetuximab (B), avelumab (C), and combination (A + B, A + C) injection sites in a clinical head and neck squamous cell carcinoma patient, stained for IFNB1 (ISH, red). Antibody-Drug Conjugate (ADC): Injection sites containing a tumor-targeted, myeloid-stimulating linker-payload drug in preclinical humanized xenograft models, one negative (left) and one positive (right) for the antibody target. Stained for ADC distribution (IHC, ADC, white) and IL6 (ISH, red). Top axis is ADC % positive. Anti-sense Oligo (ASO): Stat6 ASO injection site in preclinical murine melanoma model stained for Nos2 (ISH, red), Tnfa (ISH, yellow), and nuclei (DAPI, blue). Top axis is % Nos2 positive. Control is a background region. Exosome: Injection site for an exosome containing STING agonist in preclinical murine lymphoma model stained for an exosome surface protein (IHC, Exosome, yellow), Cxcl10 (ISH, red), Ifnb1 (ISH, cyan), and nuclei (DAPI, blue). Top axis is % Exosome positive. Control is empty exosome. CAR-T: Human CAR-T injection site, in a preclinical xenograft model positive for CAR-T target, stained for CD3 (IHC, yellow), cleaved caspase 3 (IHC, CC3, red), and nuclei (DAPI, blue). CC3 is represented as percent positive as a fold change over the vehicle injection site. For the control site, the same CAR-T cells were injected in a preclinical xenograft model negative for CAR-T target.