FFPE Spatial Transcriptomics: Challenges, Solutions, and Platform Selection Strategies

Inquiry

Figure 1: The FFPE spatial transcriptomics challenge — formalin crosslinks fragment RNA, but probe-based and random-primer chemistries detect degraded transcripts when fixation and storage conditions are favorable.

More than 90% of clinical tissue specimens are preserved as FFPE blocks — yet FFPE was long considered incompatible with high-resolution spatial transcriptomics because of RNA crosslinking and fragmentation. That has changed. Probe-based capture chemistries, random-primer amplification, and imaging-based in situ methods now enable spatially resolved transcriptomic data from blocks stored for years. But not all FFPE samples perform equally, and not all platforms suit every archived-tissue project. This article addresses the core challenges of FFPE spatial transcriptomics, explains how different capture chemistries navigate RNA degradation, compares the major FFPE-capable platforms, and provides a framework for matching platform choice to project goals.

Why FFPE Demands a Different Approach

FFPE tissue is the foundation of clinical pathology archives. Billions of FFPE blocks exist worldwide, each linked to clinical outcomes, treatment histories, and longitudinal follow-up. For pharmaceutical research teams, translational scientists, and biobank managers, these archives represent the most valuable and most underutilized resource in spatial biology.

FFPE tissue cannot be treated as "fresh frozen but worse." Formalin crosslinks amine groups on RNA and proteins, creating a molecular mesh that preserves architecture but fragments nucleic acids. Over months to years — particularly at room temperature — RNA progressively degrades and crosslinks become harder to reverse. FFPE samples require different capture chemistries, different QC metrics, different sectioning protocols, and different expectations about transcript recovery than fresh frozen tissue.

The technology gap has closed. Between 2023 and 2025, every major spatial transcriptomics platform released or upgraded FFPE-compatible workflows. The remaining challenge is not "can we do spatial transcriptomics on FFPE" but "which FFPE blocks, on which platform, for which biological question." The fresh frozen vs. FFPE decision guide covers the high-level trade-offs between preservation paths; this article focuses on what happens after you have committed to FFPE.

How Crosslinking Blocks RNA Detection

Formalin fixation introduces methylene bridges between amino groups — primarily on lysine residues of proteins, but also on adenine bases of RNA. These crosslinks create two problems. First, they physically block probe access: a crosslinked protein sitting on an mRNA molecule prevents hybridization. Second, they introduce chemical modifications that interfere with enzymatic steps — reverse transcriptase cannot read through a formalin adduct, and ligases cannot seal nicks adjacent to crosslinked bases.

The secondary problem is fragmentation. Properly fixed blocks can retain RNA molecules several thousand bases long, but the combination of fixation, room-temperature storage, and mildly acidic conditions inside paraffin blocks gradually hydrolyzes RNA. Most FFPE RNA exists as fragments 100–500 nucleotides in length — too short for conventional poly(A)-dependent capture, but long enough for probe-based methods that target internal sequences.

This molecular reality explains why RIN, the standard RNA quality metric for fresh frozen tissue, is largely meaningless for FFPE. RIN measures the ratio of 28S to 18S ribosomal RNA — peaks that are absent or severely diminished in FFPE electropherograms. The relevant metric for FFPE is DV200: the percentage of RNA fragments longer than 200 nucleotides. A DV200 of 30% or above is the standard threshold for probe-based FFPE workflows. Below 30%, some platforms — particularly Stereo-seq with its random-primer approach — can still extract meaningful data, though with reduced sensitivity.

Figure 2: Four capture chemistry strategies for FFPE spatial transcriptomics — each navigates RNA degradation differently.

Capture Chemistries for Degraded RNA

The critical innovation that enabled FFPE spatial transcriptomics was the shift away from poly(A)-dependent mRNA capture. Four chemical strategies now dominate.

Probe-pair ligation (10x Visium FFPE and Visium HD). Two adjacent oligonucleotide probes hybridize to each target mRNA. When both probes bind correctly, a ligase seals the junction — requiring perfect complementarity and ensuring high specificity. One probe carries a poly(A) sequence captured by spatially barcoded oligo-dT on the slide. Because both probes must bind for signal, the approach tolerates fragmented RNA: as long as an intact 50–100 nucleotide stretch remains in the target region, the probe pair can bind and ligate. Visium HD extends this to 2-micron resolution with near-whole-transcriptome coverage.

Padlock probes and rolling circle amplification (10x Xenium). Padlock probes hybridize to target mRNA and are circularized by ligation — again requiring perfect base-pairing. The circular probe is amplified by RCA, producing a concatemer that fluorescently labeled secondary probes bind to across multiple imaging cycles. RCA generates thousands of signal copies per transcript, enabling strong detection from single molecules. The 5,000-gene Xenium panel provides the highest per-gene sensitivity among imaging-based platforms, with negative control signal rates below 0.1%.

Branched DNA amplification (NanoString CosMx SMI). Five primary probe pairs per gene bind to the target mRNA, each carrying a readout domain. Secondary fluorescent probes with multi-fluorophore branched structures bind to these readout domains across 16 imaging cycles. The branched architecture inherently amplifies signal — important for degraded FFPE RNA — but increases non-specific background relative to padlock-probe approaches. The CosMx 6K panel covers 6,175 genes.

Random-primer capture (Stereo-seq V2 FFPE). Random hexamer primers capture all RNA species — mRNA, lncRNA, miRNA, and microbial RNA — without requiring poly(A) tails. This is the most degradation-tolerant strategy: fragments as short as 50 nucleotides can be captured. The trade-off is that non-targeted capture generates more background, and spatial diffusion can blur transcript localization in lipid-rich tissues. Stereo-seq provides whole-transcriptome coverage at 500-nanometer resolution on capture areas up to 13.2 x 13.2 cm.

For projects starting with FFPE specimens, CD Genomics FFPE spatial transcriptomics provides sample QC, probe panel selection guidance, library preparation, and sequencing across multiple platforms. The spatial transcriptomics sample preparation guide covers the tissue handling and sectioning steps that precede library preparation.

Platforms Optimized for FFPE Spatial Profiling

The table below summarizes the major FFPE-capable spatial transcriptomics platforms as of mid-2026.

Platform maturity. Visium HD and Xenium have the longest FFPE track records. A 2026 benchmarking study across six cancer types found Visium v2 with CytAssist detected approximately 2,000 unique genes per 55-micron spot — roughly 10-fold more than Visium v1 — and Xenium consistently showed lower background and stronger spatial signal than CosMx. A second benchmarking study of imaging platforms on 33 FFPE tissue types confirmed Xenium produced the highest transcript counts per gene without sacrificing specificity.

For researchers considering targeted approaches, 10x Xenium spatial RNA analysis provides subcellular-resolution transcript detection. For whole-transcriptome discovery, 10x Visium HD combines probe-based FFPE chemistry with near-single-cell resolution.

Matching Platforms to Research Questions

Platform selection is a project-design decision — the platform that wins a benchmarking study is not necessarily the one that answers your biological question. Below is a decision framework organized by research goal.

Whole-transcriptome discovery. To survey the full transcriptomic landscape without prior assumptions about which genes matter, choose Visium v2 for 55-micron spot resolution or Visium HD for near-single-cell resolution. Both use probe-pair ligation optimized for FFPE and detect approximately 18,000 genes. Visium v2 is more cost-effective for cohort-scale discovery; Visium HD is preferable when 2-micron spatial resolution is required.

Targeted validation and high-sensitivity profiling. When the research question centers on a defined gene set — immune checkpoint molecules, drug targets, cell-type markers — choose Xenium (5K panel) or CosMx (6K panel). Xenium offers the highest per-gene sensitivity and lowest background; CosMx offers the largest targeted panel and more precise cell boundary segmentation via combined membrane and nuclear staining. Both deliver subcellular resolution.

Archived cohorts with variable RNA quality. When FFPE blocks span a wide range of DV200 values, block ages, and fixation conditions — typical for retrospective clinical cohorts — Stereo-seq V2 is the most degradation-tolerant option. Its random-primer approach captures RNA fragments regardless of poly(A) status and generates interpretable data from blocks with DV200 as low as 18%. The trade-off is higher spatial diffusion in lipid-rich tissues and a shorter validation track record.

Region-of-interest analysis with pathologist input. When the question involves specific histological structures — tumor-stroma interfaces, lymphoid follicles, necrotic zones — GeoMx DSP allows a pathologist to select regions of interest based on fluorescent markers and H&E morphology, then profile each independently. GeoMx is less suited to unbiased discovery because ROI selection introduces sampling bias, but it excels when the hypothesis is spatially precise.

Large-area and organ-scale mapping. Stereo-seq's capture area of up to 13.2 x 13.2 cm enables profiling of whole mouse organ sections or large human tissue slices in a single run. For organ-scale atlases, cross-species comparisons, or host and microbial RNA detection from the same section, Stereo-seq has no equivalent alternative.

The spatial transcriptomics platform comparison guide provides additional detail on sequencing-based vs. imaging-based platform trade-offs. For researchers still defining their project scope, the spatial transcriptomics project design guide covers biological question definition, replicate strategy, and sample planning.

Pre-Analytical Variables and Quality Gates

Platform chemistry can compensate for FFPE RNA degradation — up to a point. Pre-analytical variables set the ceiling on what any platform can achieve.

Fixation time: the dominant variable. A 2025 study of 16 archived FFPE tumor samples across breast cancer, lung cancer, and lymphoma found that blocks fixed for less than 24 hours in 10% NBF consistently outperformed those with longer fixation, regardless of block age. Fixation beyond 48 hours progressively reduces probe accessibility — crosslinks accumulate faster than decrosslinking protocols can resolve them. Blocks fixed for over six years showed near-complete loss of spatial transcriptomic signal in one study of archived nervous tissue.

Block age and storage conditions. Block age and DV200 are correlated but not interchangeable. Blocks stored at 4 C in a dry environment degrade more slowly than those at room temperature. A 2025 study comparing Visium v2, GeoMx DSP, and single-nucleus RNA-seq on blocks with a median age of 57 months (range 22–103 months) found that all three methods produced reproducible data — a well-fixed, well-stored 10-year-old block may outperform a poorly fixed 2-year-old block.

DV200 as a guide, not a cutoff. A DV200 of 30% or above is optimal, but the Stereo-seq V2 workflow has produced interpretable data from blocks with DV200 as low as 18%. Borderline samples should be tested with a pilot section rather than discarded based on a single QC number.

Section thickness and adherence. FFPE sections are cut at 5 um — thinner than fresh frozen (10 um) because probe penetration is more efficient through thinner sections. Dense, collagen-rich, or calcified tissues benefit from hydrogel-coated slides to prevent detachment during high-temperature decrosslinking.

Batch design for cohort studies. The strongest source of unwanted variation in multi-sample FFPE studies is block-level: fixation differences, storage conditions, and handling history. Randomize block allocation across capture areas or sequencing runs, and include technical replicate sections from a subset of blocks.

Figure 3: FFPE spatial transcriptomics decision framework — block assessment, platform selection by research goal, and pre-submission QC.

FFPE samples can vary significantly depending on fixation history, storage conditions, and RNA quality. If you are unsure whether your archived tissue blocks are suitable for spatial transcriptomics, CD Genomics experts can help assess sample suitability and recommend the most appropriate FFPE spatial profiling strategy for your research goals.

FAQ

Q: Can I use FFPE blocks older than five years for spatial transcriptomics?

A: Yes, in many cases. Block age alone is not the determining factor — fixation quality and storage conditions matter more. Multiple 2025 studies generated high-quality spatial transcriptomic data from FFPE blocks stored for 5 to 14 years. The critical variables are fixation duration (less than 24 hours in 10% NBF is optimal), storage temperature (4 C in a dry environment preserves RNA better than room temperature), and DV200. Run a pilot section before committing the entire cohort.

Q: Which platform gives the best sensitivity for FFPE samples with low DV200?

A: Stereo-seq V2, which uses random-hexamer primers rather than poly(A) capture, is the most tolerant of severely degraded RNA and has produced interpretable data from blocks with DV200 as low as 18%. Among probe-based platforms, Visium HD and Xenium both perform well on moderately degraded FFPE — Visium HD because probe-pair ligation requires only a short intact RNA stretch per gene, and Xenium because RCA generates strong signal from single transcripts. CosMx's branched DNA amplification also compensates for low RNA input, though with higher background than Xenium.

Q: Should I use Visium HD or Xenium for my FFPE project?

A: Choose Visium HD for whole-transcriptome discovery — it detects approximately 18,000 genes at 2-micron resolution and suits projects where the key biology may involve genes not in any targeted panel. Choose Xenium for targeted validation of 5,000 genes with subcellular resolution, the highest per-gene sensitivity, and the lowest background among imaging platforms. Both are fully validated for FFPE with published benchmarking data across multiple cancer types.

Q: How do I select the best FFPE blocks from my archive for spatial transcriptomics?

A: Prioritize blocks meeting four criteria: fixation in 10% NBF for less than 24 hours, storage at 4 C in a dry environment, DV200 of 30% or above, and intact morphology on an adjacent H&E section. Blocks that fail one criterion — particularly DV200 — should not be automatically discarded; run a pilot section to assess actual performance. Blocks that fail two or more criteria should be deprioritized unless they are irreplaceable.

Q: Is GeoMx DSP or Visium better for FFPE cohort studies?

A: A 2025 head-to-head comparison of Visium v2, GeoMx DSP, and single-nucleus RNA-seq on archival FFPE blocks found that Visium and single-nucleus RNA-seq were better suited for discovery-driven cohort studies because of their unbiased tissue coverage, while GeoMx DSP excelled at targeted spatial questions — such as comparing gene expression at tumor-stroma interfaces or within immune-infiltrated regions. For most cohort-scale FFPE studies, Visium v2 provides the best balance of gene coverage, resolution, cost, and reproducibility.

References

1. Cervilla S, Grases D, Perez E, et al. A technical comparison of spatial transcriptomics platforms across six cancer types. Genome Biology. 2026;27:22. doi:10.1186/s13059-026-03937-y
2. Wang H, Wei K, Goods BA, et al. Systematic benchmarking of imaging spatial transcriptomics platforms in FFPE tissues. Nature Communications. 2025;16:10215. doi:10.1038/s41467-025-64990-y
3. Ozirmak Lermi N, Molina Ayala M, Hernandez S, et al. Comparison of imaging based single-cell resolution spatial transcriptomics profiling platforms using formalin-fixed paraffin-embedded tumor samples. Nature Communications. 2025;16:8499. doi:10.1038/s41467-025-63414-1
4. Dong Y, Saglietti C, Bayard Q, et al. Transcriptome analysis of archived tumors by Visium, GeoMx DSP, and Chromium reveals patient heterogeneity. Nature Communications. 2025;16:4400. doi:10.1038/s41467-025-59005-9
5. Vo T, Prakrithi P, Jones K, et al. Assessing spatial sequencing and imaging approaches to capture the molecular and pathological heterogeneity of archived cancer tissues. Journal of Pathology. 2025;265(3):274-288. doi:10.1002/path.6383

The information provided in this article is for research use only and is not intended for use in diagnostic or therapeutic procedures. CD Genomics provides sequencing and bioinformatics services for research purposes. Researchers should consult the appropriate regulatory guidelines for their specific applications.