The FFPE mass spectrometry imaging question comes down to reversing chemistry rather than choosing a preference — formalin fixation crosslinks proteins to preserve morphology, and mass spectrometry detects exactly the molecular consequences of that crosslinking. Understanding the retrieval step that reverses it, in some detail, is more useful than a general comparison of the two preservation routes.
Key Takeaways
|
What Does Fixation Actually Change?
Formaldehyde crosslinks proteins to each other and to nearby molecules, which is precisely why it preserves tissue architecture well enough for a pathologist to read decades later. The same crosslinking is what a mass spectrometry workflow then has to contend with, because crosslinked material behaves differently at every subsequent step: extraction, digestion, and ionisation all interact with a molecule differently depending on what it is bound to.
The reversal step is called antigen retrieval or heat-induced epitope retrieval, and it was developed for an entirely different purpose. As the FFPE proteomics literature notes, it was originally introduced to improve the sensitivity of immunohistochemical methods and to unmask epitopes in fixed samples, where crosslinking otherwise hinders antibody binding. Its adoption into mass spectrometry workflows came later, once researchers recognised that the same chemistry limiting antibody access was also limiting enzymatic and physical access for MS-based methods.
Working in analytical science?
Register for a FREE Separation Science account to subscribe to the Separation Science Newsletter.
That shared mechanism is worth understanding, because it explains why one retrieval step serves both purposes. Heating fixed tissue in a buffered solution, commonly citric acid or EDTA-based, disrupts the crosslinks sufficiently to expose both antibody epitopes and enzyme cleavage sites. The chemistry does not distinguish between an antibody and a trypsin molecule looking for a site to bind.
How Much Does Retrieval Actually Recover?
Substantially more than a general improvement, and the scale is worth stating precisely because vague comparisons undersell how consequential this step is.
Close to an Order of Magnitude Foundational work on FFPE proteomic imaging, reviewed in the MALDI-MSI literature, reports a specific comparison. When FFPE breast cancer specimens were left untreated, heated in 0.1 M EDTA, or treated with trypsin alone, the numbers of identified proteins were 8, 70, and 22, respectively. Heat retrieval alone drove the largest single improvement, transforming an experiment that would identify almost nothing into one identifying dozens of proteins. The practical implication is unambiguous: skipping antigen retrieval on FFPE tissue destined for protein imaging is not a minor shortcut; it is close to disabling the experiment. This is not a step to omit for convenience or to treat as optional based on time pressure. |
A typical protocol sequence, established across multiple published workflows, runs: deparaffinisation in xylene, rehydration through graded alcohols, heat retrieval in a citric acid or EDTA buffer at close to boiling temperature for around 10 to 30 minutes, then on-tissue tryptic digestion before matrix application. A Nature Protocols method for FFPE tissue proteomes combines heat-induced retrieval with ultrasonic tissue disruption and detergent removal specifically to enable quantitative analysis from limited FFPE material, and demonstrates the approach on hepatocellular carcinoma combined with laser capture microdissection, quantifying intra-tumour heterogeneity from a single 10 micron section.
One caution worth building into any protocol: retrieval by boiling can detach tissue from the slide, a failure mode documented across the histology literature and easily mistaken for a staining or adhesion problem rather than a retrieval artefact. Using an adhesive slide surface appropriate for heat treatment, and checking a pilot section survives the full protocol before committing valuable material, avoids losing a sample to this specific and preventable failure.
The Finding Most Coverage Misses: Retrieval Unlocks Lipids Too
Antigen retrieval is almost universally discussed as a protein-recovery step — and the framing is incomplete. Formalin fixation creates lipid-protein crosslinks as well as protein-protein ones, and retrieval addresses both.
Work specifically on antigen retrieval and its effect on MALDI-MSI of lipids in FFPE tissue found an increased number of lipid signals detected after antigen retrieval, and reasoned that these additional species may derive from lipids implicated in the lipid-protein crosslinking that formalin fixation itself creates. Some proportion of lipids become confined within the covalent protein network formed during fixation and are not efficiently extracted without retrieval. The same study observed that the additional detected species better reflected the heterogeneous morphology of the tissue, suggesting the recovered lipids carry genuine spatial information rather than being extraction noise.
That reframes a familiar step. Antigen retrieval is not a protein-specific fix bolted onto an otherwise fixed-tissue-hostile workflow; it addresses a chemistry that constrains multiple analyte classes simultaneously, because formalin crosslinks whatever sits near a protein at the moment of fixation, lipid or otherwise. Method development for FFPE lipidomics should therefore test a retrieval step rather than assuming it is a protein-only apparatus, and evaluate coverage before and after to establish whether the gain justifies the additional processing time on a given tissue.
Are Metabolites Actually Conserved in FFPE Tissue?
Better than the field’s general caution suggests, at least for one well-characterised comparison, and this is a case where a specific number is more useful than a general assurance.
The same study that examined lipid recovery also addressed metabolites directly, using high-mass-resolution MALDI-FT-ICR imaging on FFPE material following deparaffinisation and matrix coating with 9-aminoacridine. Across a mass range of roughly 50 to 1,000 daltons, the method detected approximately 1,500 mass species in FFPE samples, with 72 percent overlap compared with fresh frozen samples from the same tissue. The authors concluded that metabolites are largely conserved in FFPE material.
That 72 percent figure deserves a caveat rather than blind acceptance: it describes overlap in detected species at a defined mass resolution and mass range, on one tissue type, with one matrix and one instrument configuration. It is not a universal conservation rate, and the missing 28 percent likely includes both genuinely lost metabolites and species that shifted just enough in relative intensity to fall outside the overlap criterion. Treat it as evidence that FFPE metabolomics is viable and worth piloting on your tissue, not as a guarantee transferable across every specimen type.
The rapid-degradation risk that dominates metabolite work in fresh material is largely absent from archival FFPE tissue, because fixation happened long ago and whatever survived it is chemically stable now. That is a genuine advantage over fresh frozen material, where a study of harvest-induced hypoxia in mouse liver found the broad metabolome reaching half its total measured change within about 3.6 minutes of dissection. An archival FFPE block carries no equivalent handling risk, since whatever degradation was going to happen already happened during and shortly after fixation.
Does Retrieval Only Matter for FFPE?
No, which is a genuinely underappreciated point and worth correcting explicitly, since most discussion of antigen retrieval implicitly frames it as a fixed-tissue problem.
Work on thermal denaturation of fresh frozen tissue investigated the same family of heat treatment applied to unfixed, fresh frozen human colon, ovary, and pancreas tissue, and found a tissue-dependent effect on peptide signal: a 22.5 percent improvement in colon, 73.3 percent in ovary, and 96.6 percent in pancreas. Biochemical analysis showed the treatment specifically facilitates identification of hydrophobic peptides, which is a mechanistically distinct benefit from reversing formalin crosslinks, since there is no crosslinking to reverse in unfixed tissue.
The mechanism there is different: heat denatures native protein structure and improves enzyme access to buried cleavage sites, independent of any fixation chemistry. The practical consequence is that thermal treatment is worth piloting on fresh frozen tissue too, particularly where hydrophobic membrane-associated proteins are of interest, rather than reserving it exclusively for FFPE protocols.
Choosing by Target Analyte
The decision is genuinely asymmetric across molecular classes, and the practical guidance differs accordingly.
Analyte Focus | If Only FFPE Is Available | If You Have a Genuine Choice |
Proteins and peptides | Viable, but heat-induced antigen retrieval is mandatory rather than optional, given close to a tenfold identification gain | Fresh frozen avoids the retrieval step entirely, at the cost of the cold chain and archival access FFPE provides |
Lipids | Viable, and retrieval improves coverage measurably by releasing lipids trapped in fixation-driven crosslinks | Fresh frozen likely gives somewhat better native coverage, but the gap has not been extensively quantified |
Metabolites | Viable for many species, with roughly 72 percent overlap against fresh frozen reported in one high-resolution study | Fresh frozen avoids any question of fixation-driven loss, but introduces the handling-speed risk FFPE does not carry |
Morphology and region selection | Excellent, which is the whole reason FFPE dominates archival pathology collections | Fresh frozen morphology is poorer, which can complicate region selection for microdissection |
Table 1. Practical guidance by analyte focus. Figures are drawn from the specific studies cited above and should be validated on your own tissue rather than assumed.
Where the choice is not yours, because the specimen is archival, the practical response is to plan the retrieval and extraction protocol around your target analyte from the outset, using the figures above to set expectations rather than discovering the constraint mid-experiment. Where the choice is genuinely open, weigh the cold chain and prospective collection burden of fresh frozen material against the retrieval step FFPE requires, in light of which analyte matters most.
This decision is also covered from a laboratory operations and sample handling perspective, rather than a method development one, in our sister publication\u2019s FFPE vs. Fresh Frozen for Spatial Biology: Sample Handling Decisions, which addresses fixation chemistry and platform quality thresholds for a spatial biology programme rather than a single imaging experiment. Sectioning and mounting decisions that follow from whichever route is chosen are covered in Tissue Sectioning and Mounting for Spatial MS, and the wider preparation decision sequence is set out in Sample Preparation for Spatial Analysis: From Tissue to Data. For where preparation sits alongside ionisation and instrument choice, see MALDI Imaging Mass Spectrometry: How It Works and, for how these constraints interact with microdissection specifically, Laser Capture Microdissection + LC-MS Proteomics.
This article was produced under Separation Science's AI Editorial Guidelines.
