ZEISS @ Society for Immunotherapy (SITC) Annual Meeting 2026

Presented by: Peter F. Favreau, Ph.D., Business Development Manager, Biotech & Pharma, ZEISS Microscopy

Abstact:

Three-dimensional (3D) tumor models such as organoids recapitulate tissue architecture, cell-cell interactions, and drug penetration more faithfully than two-dimensional (2D) monolayers, making them increasingly central to oncology drug discovery and to immuno-oncology co-culture assays. Extracting quantitative 3D data at screening scale remains a bottleneck: conventional confocal imaging reconstructs each organoid from a serial stack of optical sections, which is slow across multiwell plates. A volumetric method that captures an entire organoid in a single acquisition would remove this constraint while preserving full 3D morphology.

Green fluorescent protein (GFP)-labeled HCT-116 colorectal carcinoma cells were co-cultured with red fluorescent protein (RFP)-labeled human dermal fibroblasts in a 384-well organoid plate at two tumor-to-stroma seeding ratios. Organoids were grown for 6 days, treated on day 4 with the topoisomerase I inhibitor SN-38 across four concentrations (0.5-500 nM) alongside vehicle and media controls, then cleared and counterstained. Whole organoids (200-325 µm) were imaged volumetrically on a ZEISS LSM 990 with Lightfield 4D, which captures one complete volume per well in a single snapshot without z-sectioning; selected wells were re-imaged at higher resolution on the same instrument. 3D analysis in ZEISS arivis Pro used deep-learning segmentation to extract per-well, per-channel cell counts.

Of 384 wells, 322 were analyzed; tissue loss during clearing and handling, primarily from 500 nM-treated spheroids, accounted for the remainder. SN-38 produced a clear dose-dependent reduction in GFP-positive tumor cells, reaching approximately 57% at 500 nM relative to media controls. RFP-positive fibroblast counts remained comparatively stable, declining only modestly at the highest dose, consistent with a tumor-selective effect and demonstrating compartment-resolved readouts from dual-labeled co-cultures. Analyzing the same volumetric data as a 2D maximum-intensity projection (MIP) systematically undercounted cells: full 3D counting recovered roughly 2.3-fold more tumor cells and 2.5-fold more nuclei than the MIP, because overlapping cells collapse along the projection axis.

Single-snapshot volumetric imaging enables quantitative 3D drug-response screening of tumor co-culture organoids in 384-well format, removing the throughput penalty of serial z-sectioning while delivering compartment-resolved, full-volume cell counts. Because the complete volume is captured in one acquisition, the more accurate 3D readout incurs no additional throughput cost. The platform is directly extensible to tumor-immune co-culture screening, offering immuno-oncology researchers a scalable route to 3D-resolved phenotyping and biomarker quantification.

Presented by: Peter F. Favreau, Ph.D., Business Development Manager, Biotech & Pharma, ZEISS Microscopy

Abstract:

Multiplex immunofluorescence (mIF) provides a powerful approach for visualizing complex biomarker signatures while preserving their spatial context within tissue. However, many high-plex mIF methods rely on repeated cycles of staining and imaging, increasing workflow complexity and requiring either labor-intensive manual handling or specialized instrumentation to automate multistep procedures. Single-round mIF panels using carefully selected fluorophores can simplify this process by enabling the simultaneous detection of multiple protein biomarkers with standard fluorescence filter configurations.



Here, we evaluated a streamlined high-plex mIF workflow combining multiplexed reagent panels with the ZEISS Axioscan 7 spatial biology platform, which is designed for automated, high-throughput slide scanning across an expanded range of fluorescence channels. Fluorophores were selected to minimize spectral overlap and maximize detection specificity, enabling clear separation of individual signals without spectral unmixing. Signal quality and spatial phenotypes were assessed using an AI-enabled image analysis workflow to quantify marker-positive cell populations, signal intensity, cell density, and spatial distribution.



The expanded fluorescence panel demonstrated strong signal-to-noise performance and minimal crosstalk between channels. The workflow supported the simultaneous detection of multiple biomarkers in tumor tissue while reducing staining and imaging complexity. The additional channel capacity also increased flexibility in panel design and facilitated the inclusion of markers with different expression levels within a single experiment.



Together, multiplexed mIF reagent panels, automated high-throughput imaging, and AI-enabled image analysis provide an efficient and scalable workflow for spatial phenotyping. This approach reduces manual steps, supports reproducible acquisition across larger studies, and enables deeper investigation of tumor biology and the spatial organization of complex tissue microenvironments.

Presented by: Peter F. Favreau, Ph.D., Business Development Manager, Biotech & Pharma, ZEISS Microscopy

Abstract:

Multiplex immunofluorescence (mIF) has become a cornerstone technique in spatial biology for investigating the tumor microenvironment. Despite its widespread use, existing mIF workflows are often hindered by variability in image acquisition and dependence on labor-intensive manual steps. To overcome these limitations, we developed a fully integrated, high-throughput spatial biology workflow combining the highly standardized and automated ZEISS spatial biology platform. Together, these deliver high-throughput, fully automated, and reproducible biomarker detection across a broad dynamic range, enhancing both efficiency and scalability.

To validate these claims, we conducted a multi-site verification study. A 4-plex panel including CD3, Ki67, Granzyme-B and CK/Sox10 was utilized to stain multiple tissue indications on formalin-fixed paraffin-embedded (FFPE) sections. On all sites, the slides underwent the same predefined and automated routine for staining using Leica Bond RX instruments and imaging using ZEISS Axioscan 7 spatial biology slide scanners. Acquired images were seamlessly uploaded to Mindpeak and analyzed with integrated algorithms optimized for mIF assays, providing quantitative analysis for densities and fluorescence intensities of positive cells. Analysis of the results demonstrated excellent concordance between sites, confirming the precision and accuracy of the combined workflow across multiple operators and instruments. Qualitatively, staining was highly comparable across all sites, and the limited variations in fluorescence signal and background did not affect the accuracy of the AI algorithms. Quantitatively, the presented workflow exhibits high reproducibility and repeatability of results across locations, providing a reliable solution for translational research with amplified speed and quality of data generation. As a result, it offers significant potential to advance spatial biology from translational research to clinical applications.