NEW YORK – The Mayo Clinic and Alpenglow Biosciences are developing 3D spatial transcriptomic methods to make biopsies more informative with respect to predicting things like therapy response and the need for immediate intervention versus observation.
After presenting data from a feasibility study using archival tissue samples at the recent American Association for Cancer Research annual meeting, the collaborators are gearing up for a prospective study using fresh tissue samples.
"3D spatial [biology] and transcriptomics really come into play because we can understand at a cellular level how these tumors are behaving," said Janani Reisenauer, a thoracic surgeon with the Mayo Clinic and the project's principal investigator. "It allows us to not only predict response to targeted therapies, but also biologic behaviors or patterns of aggressiveness, who needs intervention right away versus observation, and who has a higher likelihood of recurrence."
The feasibility study presented at AACR represents the first phase of an initiative to develop a diagnostic method that combines 3D microscopy with transcriptomics — and possibly other omics — in analyzing tissue biopsies.
Seattle-based Alpenglow has developed an imaging and analysis platform based on a patented open-top light-sheet microscopy technology and coupled to cloud-based artificial intelligence data analysis and novel chemistry. One of the technology's key features is that it can be used to nondestructively image tissues at subcellular resolution.
"We render the tissue transparent using a process called chemical clarification," said Alpenglow CEO Nicholas Reder.
Clearing consists of decoloring a tissue sample (often via methanol-mediated dehydration), removing lipids, and treating the tissue with a liquid (in this case ethyl cinnamate) that closely matches the refractive index of most tissue components in order to reduce optical scattering and minimize light absorption.
Following clarification, a beam or sheet of light is then shone across the sample and perpendicular to the direction of view, illuminating molecules within that section. Stacking sections then generates a 3D image.
Reder explained that imaging tissue samples without destroying them is key to combining microscopy and omics in the same sample.
The proof-of-concept study was performed using archived tissue samples from the Mayo Clinic.
"We're very fortunate that through benefactors, we have a tissue specimen registry where, when patients went to surgery, if they consented, we had the capability of storing a piece of their tissue and freezing it for further research avenues," Reisenauer said.
Reisenauer said that the next phase of the study, the details of which are still being planned out, will utilize fresh tissue samples. Key questions that the Mayo Clinic will focus on in that phase are whether the spatial and transcriptomic information they gathered from archived samples can be collected at the time of biopsy, how effective their methods are in treatment-naïve patients, and how to leverage artificial intelligence into their workflow, to better understand and predict treatment response.
Doug Farrell, VP of investor relations and corporate communications at NanoString Technologies, commented that computationally analyzing 3D spatial microscopy images presents a key challenge in any spatial biology workflow, partly stemming from the terabyte-sized datasets they generate.
NanoString, which has entered the spatial biology field with its CosMx Spatial Molecular Imager and GeoMx Digital Spatial Profiler platforms, offers customers a cloud-based informatics tool called AtoMx Spatial Informatics Platform to analyze their own data.
"No coding experience is required to use AtoMx SIP," Farrell said, adding that the company can create custom analysis modules and pipelines for researchers with computational experience.
NanoString Technologies recently began a collaboration with Weill Cornell Medicine to create spatial maps of healthy human tissue from multiple organs using its CosMx Spatial Molecular Imager and GeoMx Digital Spatial Profiler platforms.
Alpenglow's data analysis tools are also cloud-based, but the company, rather than its clients, performs the analysis using custom in-house artificial intelligence tools. The firm has developed tools for parallelized, cloud-based, GPU-accelerated data processing and for 3D spatial statistical analytics. It also uses an "enhanced" version of the publicly available CytoMap cell position and phenotype data tool, developed by Alpenglow's director of research and development, Caleb Stoltzfus.
10x Genomics also offers spatial biology solutions, such as its Xenium and Visium platforms, which both combine spatial microscopy with transcriptomics.
"Unlike open top light-sheet microscopy, Xenium is an integrated system that combines fluidics, optics, and chemistry to cyclically image and detect RNA localization within the tissue," said Nikhil Rao, director of product management at 10x.
Visium, which can measure spatial gene expression from at least 10 cells at a time, can analyze tens of thousands of genes simultaneously, while maintaining the spatial and morphological context of the tissue, without the use of a microscope.
Rao commented that 10x is currently developing Visium to achieve single-cell, whole-transcriptome analysis using next-generation sequencing workflows.
Alpenglow, 10x, and NanoString's spatial biology platforms are compatible with FFPE, fresh frozen, tissue microarray, and organoid samples. Farrell and Rao both commented that their platforms are not yet compatible with live tissue imaging.
Alpenglow's Reder, meanwhile, said that using its platform to track live cells is possible and that some of its customers are investigating that use.
"Our focus is on transforming clinical trials and clinical diagnostics," he said, "so live cell imaging is not a space that we explore internally at Alpenglow."
Reisenauer said that it was too early to say when the next batch of results might come out, but that her team at the Mayo Clinic would be doing more prospective and treatment-based work on patient samples over the course of the year.
For now, the project doesn't aim to include other broad scale analyses, such as proteomics, although Reisenauer wouldn't rule it out as a future possibility.
"Transcriptomics is a great place to start," she said, adding that the collaborative work would help shed light on several biological features that are obstacles to precision therapy and remain incompletely understood, such as the effects that tumor fibrosity and the background tissue stroma have on the penetrability of and response to targeted therapies.
"I think it is all the preliminary background and data that we have to have before we start thinking about local drug delivery," she said.