RNA sequencing from formalin-fixed, paraffin-embedded tissue has gone from a long shot to a practical reality, though the chemistry of preservation still imposes real constraints on what you can extract and how you should process it. FFPE blocks are the backbone of pathology archives worldwide, with billions of samples sitting in hospital storage, many linked to detailed clinical follow-up data. The ability to perform transcriptomic analysis on these specimens opens a door that fresh-frozen tissue alone cannot: retrospective studies on rare diseases, long-term treatment outcomes, and archival samples from historical outbreaks. Getting reliable gene-expression data out of fixed tissue, however, requires understanding what formalin does to RNA and choosing the right combination of extraction, library preparation, and computational correction to work around it.
What Formalin Does to RNA
Formaldehyde, the active ingredient in formalin, preserves tissue architecture by crosslinking proteins and nucleic acids. For RNA, the damage is twofold. First, formaldehyde adds small chemical groups called methylol adducts to the bases of the RNA strand, particularly adenine. Mass spectrometry studies of formalin-treated RNA showed that adenine residues accumulated these adducts at a rate of about 39% after 16 hours and over 62% after seven days, while uracil residues were far less reactive at roughly 4%.1Nucleic Acids Research. Analysis of chemical modification of RNA from formalin-fixed samples and optimization of molecular biology applications for such samples Second, the RNA strands themselves break. FFPE tissues generally yield low amounts of extractable RNA that show both covalent base modifications and strand cleavage.2PubMed Central. The effect of formaldehyde fixation on RNA: optimization of formaldehyde adduct removal Those two problems, chemical modification and fragmentation, are the central challenges every downstream step has to address.
The crosslinks are partly reversible. Heat treatment during extraction (often called decrosslinking) can remove many of the methylol adducts, freeing up enough RNA for sequencing. But strand breaks are permanent, and the fragments you extract will always be shorter than what you would get from fresh or frozen tissue. How much shorter depends heavily on how the block was handled before it reached you.
Storage Conditions Make or Break RNA Quality
If you have a choice of FFPE blocks to work with, the storage history matters more than almost any other variable. RNA isolated from FFPE specimens within a few days of embedding can be surprisingly intact, with RNA Integrity Numbers around 7 or higher. Even after a year at 4°C, ribosomal RNA bands remain visible and RIN values stay around 5 to 6. Blocks stored at room temperature tell a different story: mean fragment lengths drop well below 2,000 nucleotides, and at 37°C the fragments shrink below 100 nucleotides.3PLoS ONE. Determinants of RNA Quality from FFPE Samples Degradation is gradual and continuous, not a sudden event.
A longer-term study tracking both human and animal FFPE tissues confirmed that room-temperature storage causes RNA integrity to bottom out at a RIN of about 2 to 3 within the first six months to two years. Cold storage at 4°C slowed the decline but did not prevent it; those blocks also reached RIN 2 to 3, just over a longer timeline of roughly three years. Freezing blocks at −20°C or −80°C, however, essentially halted RNA degradation in both standard FFPE and paraformaldehyde-fixed tissue.4PLoS ONE. Impact of storage conditions on the quality of nucleic acids in paraffin embedded tissues A difference of 0.5 to 2 RIN values between 4°C and room temperature persisted even after nine years. The practical takeaway: if you are banking FFPE blocks for future molecular work, freezer storage is far superior to a shelf.
Pre-fixation handling matters, too. Optimized FFPE preparation involves keeping tissue cold during ischemia time (the interval between excision and fixation), ideally at 4°C for no more than 48 hours, followed by fixation for about 48 hours at 25°C. Sampling RNA from scrolls rather than thin sections also tends to yield better results.5PubMed Central. Optimization of FFPE preparation and identification of gene attributes associated with RNA degradation
Measuring RNA Quality Before Sequencing
The classic RIN score, which runs from 1 (completely degraded) to 10 (pristine), was designed for intact RNA and loses much of its discriminating power in the narrow, low-quality range typical of FFPE samples. A RIN of 2.1 and a RIN of 2.5 might reflect meaningfully different fragment-size distributions, but the number itself does not capture that well. The DV200, which measures the percentage of RNA fragments longer than 200 nucleotides, has emerged as a more informative metric for degraded material. In head-to-head testing, DV200 correlated more strongly with sequencing library yield than RIN and provided better sensitivity and specificity for predicting successful library construction. Samples with a DV200 above about 66% produced the most reliable libraries.6PubMed Central. DV200 Index for Assessing RNA Integrity in Next-Generation Sequencing
Other groups have pushed this further by looking at the fraction of fragments above 100 nucleotides rather than 200. One study found that a DV100 above 80% was the best predictor of gene diversity and read counts in sequencing data from both low- and high-quality FFPE samples, outperforming both RIN and the related paraffin-embedded RNA metric.7Toxicological Sciences. Enhanced Quality Metrics for Assessing RNA Derived From Archival Formalin-Fixed Paraffin-Embedded Tissue Samples The general principle: for FFPE, fragment-length-based metrics tell you more about whether your sequencing run will succeed than an overall integrity score does.
Choosing a Library Preparation Strategy
Standard poly(A) capture, the default for fresh-tissue RNA-seq, selects messenger RNA by grabbing the polyadenylated tails. Since formalin fragmentation often shears off those tails, poly(A) enrichment performs poorly on degraded FFPE samples. That leaves three main categories of library preparation, each with distinct trade-offs.
Ribosomal RNA depletion removes the abundant ribosomal RNA and sequences everything that remains. In paired comparisons between fresh-frozen and FFPE RNA, ribosomal depletion methods like Ribo-Zero removed rRNA with efficiency comparable to poly(A) capture and provided equivalent or less biased coverage across gene bodies.8PubMed Central. Comparison of RNA-Seq by poly (A) capture, ribosomal RNA depletion, and DNA microarray for expression profiling Ribo-depletion is a solid general-purpose choice for moderately degraded FFPE RNA, though it can struggle with very low-quality samples where the ratio of informative to uninformative reads drops.
Exome capture-based methods (sometimes called hybridization capture or RNA Access-type protocols) use probe panels to pull down transcripts of interest. For low-quality FFPE material, a capture-based approach outperformed ribosomal depletion in a direct comparison on breast cancer subtyping. The capture method produced higher concordance between FFPE and fresh-frozen expression profiles and less score dispersion across molecular subtypes.9PLoS ONE. RNA-seq RNAaccess identified as the preferred method for gene expression analysis of low quality FFPE samples A broader comparison of three commercial exome capture platforms found high Spearman correlations (0.72 to 0.9) between FFPE capture-based expression profiles and matched fresh-frozen TruSeq data, and all platforms reliably detected clinically actionable gene expression outliers and fusions.10npj Genomic Medicine. Functional comparison of exome capture-based methods for transcriptomic profiling of formalin-fixed paraffin-embedded tumors Capture methods also showed a clear advantage for fusion gene detection, achieving roughly two- to six-fold enrichment of fusion fragments compared to ribosomal depletion, with the benefit being greatest in highly degraded samples.11PubMed Central. A comprehensive assessment of exome capture methods for RNA sequencing of formalin-fixed and paraffin-embedded samples
A third option, 3′ mRNA-Seq, only sequences the 3′ ends of transcripts. Because fragmented RNA still retains 3′ ends more reliably than full-length structure, this approach works even with severely degraded FFPE material. Combined with unique molecular identifiers to correct for PCR amplification bias, 3′ mRNA-Seq generated high-quality gene quantification data from both fresh-frozen and FFPE inputs.12PubMed Central. Application of the 3′ mRNA-Seq using unique molecular identifiers in highly degraded RNA derived from formalin-fixed, paraffin-embedded tissue The trade-off is that you lose information about full transcript structure, alternative splicing, and fusion breakpoints. For straightforward gene-level quantification, though, it is an efficient and cost-effective route.
How Close Does FFPE Get to Fresh-Frozen Results
The honest answer is surprisingly close for protein-coding genes, and noticeably worse for certain other RNA classes. Whole-transcriptome comparisons of fresh, frozen, and FFPE cardiac tissue showed Spearman correlations above 0.94 for protein-coding transcripts across all three conditions. Mitochondrial transcripts remained highly correlated between fresh and frozen tissue but showed some false overexpression in FFPE. Long non-coding RNA correlations were lower, dropping to around 0.53.13PubMed Central. Comparison of whole transcriptome sequencing of fresh, frozen, and formalin-fixed, paraffin-embedded cardiac tissue
In glioblastoma samples, paired FFPE and fresh-frozen tissues showed gene expression correlations around 0.9, with one heavily degraded outlier sample dropping to about 0.35.14PLoS ONE. A Comparison of RNA-Seq Results from Paired Formalin-Fixed Paraffin-Embedded and Fresh-Frozen Glioblastoma Tissue Samples Principal component analysis consistently showed that paired FFPE and fresh-frozen samples from the same patient cluster together, meaning the biological signal dominates over the preservation-induced noise. The pattern across studies is consistent: if the FFPE RNA is not catastrophically degraded, you can recover the major expression patterns faithfully. The risk lies with badly stored or very old blocks where fragmentation has crossed a threshold past which the library preparation cannot compensate.
Clinical Fusion Detection and Molecular Subtyping
One of the most impactful clinical uses of FFPE RNA-seq is detecting gene fusions, which drive treatment decisions in cancers like lung adenocarcinoma, sarcomas, and brain tumors. A neuropathology department that implemented routine RNA-seq on FFPE specimens over 26 months identified clinically relevant fusions in 65% of the 101 cases analyzed.15PubMed. Routine RNA sequencing of formalin-fixed paraffin-embedded specimens in neuropathology diagnostics identifies diagnostically and therapeutically relevant gene fusions At larger scale, a diagnostic laboratory that ran whole-transcriptome sequencing on over 800 clinical FFPE samples across diverse tumor types detected 121 fusions, with turnaround times comparable to targeted panel approaches.16British Journal of Cancer. Diagnostic whole transcriptome sequencing in a series of 1233 FFPE solid tumor samples Bioinformatic analyses of fusion detection pipelines on FFPE material broadly support the clinical reliability of these calls.17PubMed Central. Targeted RNA-sequencing analysis for fusion transcripts detection in tumor diagnostics
Beyond fusions, FFPE RNA-seq enables molecular subtyping of tumors and pathway-level analyses using archival tissue that would otherwise sit unused. Work on archival FFPE breast cancer samples demonstrated that RNA-seq pipelines could faithfully recover molecular pathway information, supporting efforts to distinguish indolent from aggressive disease and identify mechanisms of treatment resistance.18PubMed Central. RNA-seq from archival FFPE breast cancer samples: molecular pathway fidelity and novel discovery A proof-of-concept study in clear cell renal cell carcinoma confirmed that FFPE biopsy RNA-seq data were comparable to results from freshly preserved material, supporting biomarker development from archival tissue.19PLoS ONE. Transcriptome Sequencing (RNAseq) Enables Utilization of Formalin-Fixed, Paraffin-Embedded Biopsies with Clear Cell Renal Cell carcinoma for Exploration of Disease Biology and Biomarker Development
Computational Correction of FFPE Artifacts
Even with optimized wet-lab protocols, FFPE introduces systematic biases into sequencing data that benefit from computational cleanup. One prominent artifact is false-positive somatic mutation calls. Formalin induces deamination of cytosine bases, creating artificial C-to-T transitions that can mimic real mutations. A machine-learning tool called FFixR, trained on FFPE melanoma samples with matched DNA for ground truth, removed up to 98% of artifactual mutations while retaining about 92% of true variants, and it works without needing a matched normal sample.20Bioinformatics. FFixR: a machine learning framework for accurate somatic mutation calling from FFPE RNA-seq data in cancer
Another computational challenge arises when you want to compare FFPE RNA-seq data generated with one library preparation protocol against data generated with a different one, or against public datasets from fresh tissue. The batch effects can be substantial. A machine-learning algorithm called Procrustes was designed to align expression profiles across protocols, raising the fraction of genes with high concordance from 26% to 61% across the full transcriptome, and from 36% to 84% for cancer-specific and microenvironment-related genes.21Communications Biology. Procrustes is a machine-learning approach that removes cross-platform batch effects from clinical RNA sequencing data Tools like these are becoming essential as clinical laboratories integrate FFPE RNA-seq into diagnostic workflows that already have years of fresh-tissue data.
Spatial Transcriptomics on FFPE Tissue
The most exciting recent frontier is spatial gene expression profiling directly on FFPE tissue sections, preserving the relationship between transcripts and their physical location in the tissue. One early approach demonstrated unbiased genome-wide spatial expression profiling in FFPE for the first time, opening the door to studying disease progression across both time and space using archived biobank material.22PubMed Central. Genome-wide spatial expression profiling in formalin-fixed tissues
A more recent method called Patho-DBiT combines in situ polyadenylation with computational tools tailored for FFPE, enabling spatial co-profiling of gene expression and RNA processing. It can reveal region-specific splicing patterns and achieve high-sensitivity transcriptomic mapping of clinical tumor samples stored for five years.23PubMed Central. Spatially exploring RNA biology in archival formalin-fixed paraffin-embedded tissues Another platform, spRandom-seq, uses random primers to capture total RNA spatially, which lets it detect not only messenger RNA but also long non-coding RNAs, alternative splicing events, and even bacterial and viral RNA within FFPE tissue sections. In hepatitis B virus-positive liver cancer FFPE samples, this approach simultaneously mapped host and viral transcripts, revealing that complement and coagulation pathways were activated specifically in virus-infected tumor regions.24PubMed Central. Spatial Total RNA Sequencing of Formalin-Fixed Paraffin-Embedded Tissue by spRandom-seq
Single-Cell and Single-Nucleus Approaches
Bulk RNA-seq averages the expression of every cell in a tissue sample, which masks the cellular heterogeneity that drives tumor behavior and treatment resistance. Getting single-cell resolution from FFPE has been a significant technical hurdle because formalin crosslinks make it difficult to dissociate intact cells. The workaround is to isolate nuclei instead, and several single-nucleus RNA-seq protocols now work on FFPE material.
One protocol, snPATHO-seq, successfully isolated intact nuclei from FFPE tumors and produced cell-type annotations that matched the morphological features of the tissue. The cell-type proportions were generally comparable to those obtained from frozen tissue using conventional single-nucleus methods, with the best concordance seen between snPATHO-seq and the 10x Flex assay on frozen tissue. The number of genes and unique molecular identifiers detected per nucleus was lower from FFPE than from frozen tissue, reflecting the reduced RNA quality, but the biological signal came through clearly enough to identify expected cell populations including myoepithelial cells in breast tissue and hepatocytes in liver.25Communications Biology. snPATHO-seq, a versatile FFPE single-nucleus RNA sequencing method to unlock pathology archives
A method called snRandom-seq, which uses random primers rather than oligo(dT), detected a much broader spectrum of RNA types from FFPE nuclei, including long non-coding RNAs, small nucleolar RNAs, and microRNAs. It reported median gene detection of over 3,000 genes per nucleus from FFPE samples, substantially higher than other FFPE single-nucleus methods available at the time, and total RNA profiles correlated well between FFPE and fresh samples (Pearson R around 0.9).26Nature Communications. High-throughput single nucleus total RNA sequencing of formalin-fixed paraffin-embedded tissues by snRandom-seq A separate study performing single-cell RNA-seq on FFPE breast cancer tissue found that FFPE- and fresh-frozen-derived libraries had similar sequencing quality metrics, identified the same cell populations with no exclusive populations found by either approach, and that cell-type proportions agreed with standard immunohistochemistry on adjacent sections.27PubMed Central. Single-Cell RNA Sequencing on Formalin-Fixed and Paraffin-Embedded (FFPE) Tissue Identified Multi-Ciliary Cells in Breast Cancer
Applications Beyond Cancer
While oncology dominates FFPE RNA-seq research for obvious reasons (cancer biopsies are routinely fixed), the technology has non-oncology uses that are worth knowing about. One striking example comes from infectious disease: researchers sequenced the full genome of the 1918 influenza virus at 3,000-fold coverage from a single FFPE autopsy lung sample, along with bacterial sequences from secondary pneumonias and well-represented host transcripts.28PubMed Central. High-throughput RNA sequencing of a formalin-fixed, paraffin-embedded autopsy lung tissue sample from the 1918 influenza pandemic This kind of work demonstrates that FFPE archives are not just repositories of human tissue but also of the pathogens those patients carried, extending the utility of archived samples into epidemiology and virology. As spatial methods mature, the ability to simultaneously map pathogen and host transcripts within the same tissue section adds another dimension, potentially revealing how infections reshape local tissue environments in ways that were previously inaccessible from fixed material.
Laser-capture microdissection, which allows researchers to isolate tiny regions of interest from a tissue slide, can also be coupled with FFPE RNA-seq. Optimized extraction protocols for small microdissected FFPE sections have demonstrated RNA quality sufficient for next-generation sequencing.29PubMed Central. An optimised protocol for isolation of RNA from small sections of laser-capture microdissected FFPE tissue amenable for next-generation sequencing This is particularly useful for studying rare cell populations in heterogeneous tissue, or for isolating tumor from adjacent normal tissue in samples too small for flow sorting or nucleus isolation. Combined with 3′ mRNA-Seq or exome capture methods that tolerate low input, microdissected FFPE samples are increasingly tractable for transcriptomic analysis.