Combining fluorescence-based image segmentation and automated microfluidics for ultrafast cell-by-cell assessment of biomarkers for HER2-type breast carcinoma

Direct three-dimensional segmentation of prostate glands with nnU-Net

Significance

In recent years, we and others have developed non-destructive methods to obtain three-dimensional (3D) pathology datasets of clinical biopsies and surgical specimens. For prostate cancer risk stratification (prognostication), standard-of-care Gleason grading is based on examining the morphology of prostate glands in thin 2D sections. This motivates us to perform 3D segmentation of prostate glands in our 3D pathology datasets for the purposes of computational analysis of 3D glandular features that could offer improved prognostic performance.

Aim

To facilitate prostate cancer risk assessment, we developed a computationally efficient and accurate deep learning model for 3D gland segmentation based on open-top light-sheet microscopy datasets of human prostate biopsies stained with a fluorescent analog of hematoxylin and eosin (H&E).

Approach

For 3D gland segmentation based on our H&E-analog 3D pathology datasets, we previously developed a hybrid deep learning and computer vision-based pipeline, called image translation-assisted segmentation in 3D (ITAS3D), which required a complex two-stage procedure and tedious manual optimization of parameters. To simplify this procedure, we use the 3D gland-segmentation masks previously generated by ITAS3D as training datasets for a direct end-to-end deep learning-based segmentation model, nnU-Net. The inputs to this model are 3D pathology datasets of prostate biopsies rapidly stained with an inexpensive fluorescent analog of H&E and the outputs are 3D semantic segmentation masks of the gland epithelium, gland lumen, and surrounding stromal compartments within the tissue.

Results

nnU-Net demonstrates remarkable accuracy in 3D gland segmentations even with limited training data. Moreover, compared with the previous ITAS3D pipeline, nnU-Net operation is simpler and faster, and it can maintain good accuracy even with lower-resolution inputs.

Conclusions

Our trained DL-based 3D segmentation model will facilitate future studies to demonstrate the value of computational 3D pathology for guiding critical treatment decisions for patients with prostate cancer.

Detection of Estrogen Receptor Alpha and Assessment of Fulvestrant Activity in MCF-7 Tumor Spheroids Using Microfluidics and SERS

Breast cancer is one of the leading causes of cancer death in women. Novel in vitro tools that integrate three-dimensional (3D) tumor models with highly sensitive chemical reporters can provide useful information to aid biological characterization of cancer phenotype and understanding of drug activity. The combination of surface-enhanced Raman scattering (SERS) techniques with microfluidic technologies offers new opportunities for highly selective, specific, and multiplexed nanoparticle-based assays. Here, we explored the use of functionalized nanoparticles for the detection of estrogen receptor alpha (ERα) expression in a 3D tumor model, using the ERα-positive human breast cancer cell line MCF-7. This approach was used to compare targeted versus nontargeted nanoparticle interactions with the tumor model to better understand whether targeted nanotags are required to efficiently target ERα. Mixtures of targeted anti-ERα antibody-functionalized nanotags (ERα-AuNPs) and nontargeted (against ERα) anti-human epidermal growth factor receptor 2 (HER2) antibody-functionalized nanotags (HER2-AuNPs), with different Raman reporters with a similar SERS signal intensity, were incubated with MCF-7 spheroids in microfluidic devices and spectroscopically analyzed using SERS. MCF-7 cells express high levels of ERα and no detectable levels of HER2. 2D and 3D SERS measurements confirmed the strong targeting effect of ERα-AuNP nanotags to the MCF-7 spheroids in contrast to HER2-AuNPs (63% signal reduction). Moreover, 3D SERS measurements confirmed the differentiation between the targeted and the nontargeted nanotags. Finally, we demonstrated how nanotag uptake by MCF-7 spheroids was affected by the drug fulvestrant, the first-in-class approved selective estrogen receptor degrader (SERD). These results illustrate the potential of using SERS and microfluidics as a powerful in vitro platform for the characterization of 3D tumor models and the investigation of SERD activity.

Microfluidics-assisted multiplexed biomarker detection for in situ mapping of immune cells in tumor sections

Because of the close interaction between tumors and the immune system, immunotherapies are nowadays considered as the most promising treatment against cancer. In order to define the diagnosis and the subsequent therapy, crucial information about the immune cells at the tumor site is needed. Indeed, different types or activation status of cells may be indicative for specific and personalized treatments. Here, we present a quantitative method to identify ten different immuno-markers in the same tumor cut section, thereby saving precious samples and enabling correlative analysis on several cell families and their activation status in a tumor microenvironment context. We designed and fabricated a microfluidic chip with optimal thermomechanical and optical properties for fast delivery of reagents on tissue slides and for fully automatic imaging by integration with an optical microscope. The multiplexing capability of the system is enabled by an optimized cyclic immunofluorescence protocol, with which we demonstrated quantitative sequential immunostaining of up to ten biomarkers on the same tissue section. Furthermore, we developed high-quality image-processing algorithms to map each cell in the entire tissue. As proof-of-concept analyses, we identified coexpression and colocalization patterns of biomarkers to classify the immune cells and their activation status. Thanks to the quantitativeness and the automation of both the experimental and analytical methods, we believe that this multiplexing approach will meet the increasing clinical need of personalized diagnostics and therapy in cancer pathology.

High-content, cell-by-cell assessment of HER2 overexpression and amplification: a tool for intratumoral heterogeneity detection in breast cancer

Immunohistochemistry and fluorescence in situ hybridization are the two standard methods for human epidermal growth factor receptor 2 (HER2) assessment. However, they have severe limitations to assess quantitatively intratumoral heterogeneity (ITH) when multiple subclones of tumor cells co-exist. We develop here a high-content, quantitative analysis of breast cancer tissues based on microfluidic experimentation and image processing, to characterize both HER2 protein overexpression and HER2 gene amplification at the cellular level. The technique consists of performing sequential steps on the same tissue slide: an immunofluorescence (IF) assay using a microfluidic protocol, an elution step for removing the IF staining agents, a standard FISH staining protocol, followed by automated quantitative cell-by-cell image processing. Moreover, ITH is accurately detected in both cluster and mosaic form using an analysis of spatial association and a mathematical model that allows discriminating true heterogeneity from artifacts due to the use of thin tissue sections. This study paves the way to evaluate ITH with high accuracy and content while requiring standard staining methods.

Special Section Guest Editorial: Translational Biophotonics

This guest editorial introduces the special section on Translational Biophotonics.