At Thursday morning’s AMP conference session, attendees filled the room to hear Rong Fan, PhD, professor of pathology at Yale University, present a sweeping vision for how single-cell and spatial omics technologies may redefine the future of molecular pathology.
Fan framed his talk, Decoding RNA Biology in Space: Toward the Future of Molecular Pathology, around a central challenge: the limitations of traditional histopathology and the enormous opportunity in layering molecular information directly onto tissue images. His frequent collaborator, Yale pathologist Mina Xu, MD, provided a running example. “She diagnoses patients by looking at the tissue histology image, and she can tell me so much about just a single tissue slide,” Fan explained. Yet in difficult cases “that requires a lot more molecular information…she has to order a lot more IHC, but potentially also some genetic testing as well.” The “holy grail,” he said, would be to “add molecular information directly on top of the same histology image slide.”
Spatial omics, a rapidly advancing area of technology, attempts to deliver exactly that. “Over the past 10 years, we have seen the explosion of the technology,” Fan said. Spatial transcriptomics remains a “mainstay tool,” but Fan’s group has long sought to push beyond gene expression alone, toward what he calls spatial multi-omics.
His laboratory’s approach turns fixed tissue itself into a reaction chamber—“much more versatile,” he noted—capable of tagging not only RNA and proteins but “very important epigenetic features, such as open chromatin regions, or histone modification marks, or even DNA methylation.” They have published multiple papers demonstrating spatial epigenomics and other modalities, but on Thursday, Fan narrowed his focus to RNA biology.
RNA, he reminded the audience, is far more than messenger molecules. “From your freshman biology class, you already learn RNA is not just the messenger. Every messenger RNA molecule has very dynamic life cycle,” he said. Understanding those RNAs directly in tissue, he argued, may be key to unlocking deeper biology and providing pathologists with new diagnostic precision.
His team adapted an in-tissue polyadenylation strategy originally developed by Stanford’s Stephen Quake, enabling them to add poly(A) tails to “whatever RNA molecules” remain in the tissue section. After barcoding, “you now can see all kinds of RNA species,” Fan said, including long non-coding RNAs, small non-coding RNAs, and even microRNAs. Most strikingly, the method enables spatial mapping of tRNAs—“the bridge between the RNA and the protein synthesis.” Fan emphasized the significance: “Turns out tRNA was the first noncoding RNA discover and recognized by the Nobel Prize.”
The clinical potential of this molecular richness came alive through a lymphoma case study. A patient with prolonged stomach pain underwent biopsy that produced two distinct regions of disease. Xu could distinguish low grade B cell lymphoma (mucosa-associated lymphoid tissue, or MALT) from diffuse large B cell lymphoma (DLBCL), but that alone could not guide therapy. Transformations from low-grade to high-grade disease worsen outcomes, and effective targeted treatments exist—but come with high toxicity.
Fan’s team applied their technology to the samples, generating spatial clusters and cell-type maps. Crucially, AI machine learning tools, including the iStar pipeline from the University of Pennsylvania, allowed them to integrate FFPE histology with spatial transcriptomics to achieve what Fan termed “super resolve, almost single cell” data across the tissue.
That resolution allowed them to ask previously inaccessible questions. For instance, comparing macrophages across low- and high-grade regions, they found that in high-grade lymphoma, macrophages were “much more polarized to the M2 macrophage alternative activation pathway,” with additional pathway differences that were fascinating. These molecular clues, he explained, “can potentially give rise to better treatment ideas.”
The same datasets also held genomic information. Because their approach captures RNA across the length of transcripts, “we can actually lead to a lot of genomic operation information,” including single nucleotide variants and copy number alterations. This allowed them to reconstruct the evolutionary phylogenetic tree of the different tumor clones and, critically, place those clones back into their precise spatial context within the tissue.
Fan also highlighted their ability to interrogate microRNAs, noting that the team detected roughly 1,800 human microRNAs—“approaching the whole pool”—of 2,000 human microRNAs in a cell. Integrated analyses suggested a mechanistic chain in this patient’s tumor involving chronic inflammation, NF-κB activation, which ultimately led to the activation of the PI3K–AKT pathway. “Some of the patients might respond to PI3K–AKT,” he noted, pointing to a potential biomarker-driven therapeutic route.
As he concluded, Fan reflected on the broader significance of the work. FFPE samples—the everyday materials of clinical pathology—can now yield unprecedented molecular depth. “For the first time, the human clinical tissue specimen…you can see so much molecular biology information,” he said. “Nowadays, with the emerging tools like what I showed you today and many others, we’re at the beginning of the new era in molecular pathology.”
