In the last decade, owing to next-generation sequencing (NGS) techniques, our knowledge about the microbiota implications in health and disease has rapidly increased. As widely recognized, humans live in symbiosis with trillions of microorganisms (bacteria, fungi, viruses, and others) that colonize their different body sites, forming the so-called human microbiota, an adjunct ‘virtual organ’ that plays essential functions for healthy human physiology. The deep interconnection between host and microbiome has reconceived the human being as a superorganism, whose features can be understood in the context of the composite phenotype imparted by both host and microbial genomes [1]. The symbiotic relationship between host and microbial communities helps maintain homeostasis and control immune response, while many diseases, such as cancer, immune-dysregulated disorders, and cardiovascular diseases, can result from microbial dysbiosis [2]. As a result, to better understand human pathology, the study of the microbiome is currently considered to be an important component of the molecular pathologic epidemiology (MPE) research frameworks [3, 4]. Of note, considerable research has focused on the role of the host microbiome in human cancer development, a field that has been called “oncobiome” [5]. The tumor microenvironment represents a suitable niche for specific microbes and recent studies have launched the concept of an endogenous tumor microbiota, here referred to as oncobiota [6]. For instance, the vascularized surface of the tumors is suitable for aerobic species while the low oxygen levels inside solid cancers allow colonization of anaerobic species. Overall, despite the nourishment belonging from necrotic eukaryotic cells, the restricted access to nutrients and oxygen determines low biomass [7], with few exceptions as pancreatic cancer, which displays a higher microbial abundance than the normal tissue [8]. Besides the well-known infective etiology of some cancer [9], commensal bacteria may favor cancer initiation and progression in several ways, including the conversion of consumed substances into harmful metabolites, the direct secretion of toxic substances, and modulation of the anti-cancer immune response. Indeed, the microbiota can produce enzymes, microbe-associated molecular patterns (MAMPs), short-chain fatty acids (SCFAs), and other metabolites, capable of conditioning host cytokines secretion, cell cycle and survival, metabolic pathways, and DNA reparation [10]. Among others, bacteria like Fusobacterium nucleatum (F. nucleatum), Bilophila wadsworthia, and Desulfovibrio desulfurican [11] cause DNA oxidative stress by producing reactive oxygen species, while Campylobacter jejuni [12] generate double-stranded DNA breaks, driving tumorigenesis. What is more, intratumoral microorganisms can influence local immunity and halt anti-tumor response. For instance, F. nucleatum has been proven to inhibit the T lymphocyte (T-cell) mediated immune responses through the expressions of fibroblast activation protein 2 (Fap2) which binds and activates the inhibitory T cell immunoreceptor with Ig and ITIM domains (TIGIT) on natural killer (NK) cells and T cells [13]. Bacteria colonizing tumors not only influence their growth but also the response to treatment, for example transforming chemotherapy drugs, reprogramming tumor immunity, and influencing immunotherapy response [14, 15].

According to the METAgenomics of the human intestinal tract (MetaHit) project results, about 99% of the microbiome genes are bacterial [16]. Hence, the fine characterization of the oncobiota composition, in terms of which bacteria are present and what they do inside the tumor microenvironment, is important for the development of more tailored therapy regimens for a variety of malignancies, boosting the anti-cancer effect while reducing patient damage. Overall, awareness of the oncobiota role in cancer development could have a significant impact on cancer management, in terms of diagnosis, prevention, and treatment. Besides, it will be possible only as long as this field will develop a thorough understanding of the microbial-related tumor microenvironment and the causative links between cancer and bacteria.

In many cases, the oncobiota characterization is hampered by various technical problems, such as the analysis of samples with low microbial biomass, the problem of samples contamination, and most importantly, the availability of tumor patient-related samples. Moreover, way before the microbiota sequencing analysis, the first crucial step of the workflow is represented by sample choice, such as the tissue area, and its preservation method. Formalin-fixed paraffin-embedded (FFPE) tissue represents a precious source of cancer patient-related samples, that, if correctly employed, could increase the sample sizes and sites available for the oncobiota studies. Briefly, specimen preservation in FFPE consists in submerging in 10% neutral buffered formalin overnight followed by serial dehydration in gradually more concentrated alcohol ending with inclusion in paraffin [17]. The FFPE storage preserves the general integrity of the sample at room temperature allowing easier and cheaper handling, transport, and storage, compared to other storing methods representing the ideal choice in clinical routine [17, 18].

The formaldehyde, which is the main component of the formalin solution, has fixative proprieties, due to its chemical features: such a small aldehyde has a strong electrophilic behaviour, because of the low dispersal of the partially positive charge of the carbon engaged in double bond with oxygen. This charge leads first to formalin transformation in methylol and then to a covalent bond of dehydrated methylol with another molecule [19, 20]. Being nucleophiles both the initiators and the final targets of this reaction, the formalin acts as a cross-linker between nucleophilic molecules such as amino acids and DNA bases resulting in sample fixation and long-term preservation. Moreover, the inclusion of paraffin wax block allows an easy cut of slices of desired size and the choice of specific areas for histological, proteomic, genomic, or microbial examinations. The possibility to choose a well-defined tissue area is crucial in oncobiota characterization because the oncobiota inside the tumor mass may be dissimilar from the outer surface [10].

Given its usefulness, many kits regarding nucleic acid extraction from FFPE are available and specifically designed for purifying DNA or RNA from cross-linkers caused by formaldehyde. Studying microbial composition and functionality starting from FFPE tumor tissues would enable us to efficiently deepen our understanding of microbial implications in tumor events thanks also to the possibility of comparing specimens collected even at a distance of years. However, there are some drawbacks when using FFPE specimens that can’t be fully avoided leading to technical challenges which sum up to the ones already related to microbiota analysis.

The FFPE storage of tumor tissues can bring many biases in microbiota characterization due to 1) the low biomass environment [21]; 2) the degradation and mutation to the DNA undergoes during the fixing and embedding process; 3) the contamination inherent to this biobanking process.

The microbiota identification in low biomass specimens runs into complications due to the strong contamination from host DNA and other microbes located in the reagents or the working place. For example, in this context, among the possible sequencing strategies for microbiota characterization, the whole genome sequencing (WGS) method could be less effective because 97% of generated reads will map on the human genome [7, 22] and solely the 3% will reflect the searched microbiota. Conversely, the 16S amplicon targeted sequencing can outperform WGS in bacterial microbiota studies, even if lacking species level resolution, detection of rare species, and other kingdom organisms (viruses and fungi) [23]. That said, also amplicon sequencing should follow certain precautions, since the use of primer pairs targeting the variable (V)3–V4 regions has been associated with an off-target amplification of human DNA, that can be avoided using primer pairs targeting the V1–V2 regions [24]. Nevertheless, the formalin damages to microbial DNA lead to other adv that, as explained below, directly counteract the last one.

In presence of formaldehyde, the DNA bases establish cross-linkers with other bases, histones, and proteins. In addition, formaldehyde reactions generate products which decrease the pH and raise both the DNA depurination rate and the C > U transition after C deamination leads to misplaced thymine (T) during polymerase extension [19, 25]. As a result, the DNA stability is considerably corrupted and errors are committed during polymerase chain reaction (PCR) and sequencing. Moreover, it causes the progressive fragmentation of the sample’s DNA sugar-phosphate backbone, which is a crucial issue in the amplicon sequencing method [26, 27]. Of note, the DNA damages increase over time, significantly after four years of storage [27], potentially affecting the microbial community inferred from the sequenced reads. Amplicon sequencing targeting the 16S V4 region, being shorter than other setups (about 254 bp), could mitigate the probability of stochastic mutations and fragmentation in the DNA region [26], but, as said above, can lead to human DNA amplification in low microbiota environment.

In addition, the microbiota characterization in FFPE samples is also complicated by contamination of microbes that normally grow in paraffin [26]. Indeed, together with kit reagents’ contaminants, they may hide the true microbial composition of the specimen with low biomass [23, 26]. While the contaminants from kit reagents and fresh frozen storage are usually water and soil associated and then easily recognizable [28], microbial taxa deriving from FFPE storage, such as Proteobacteria and Actinobacteria, can overlap with those truly populating the tumor sample [29]. Pinto-Ribeiro and colleagues [29] found that 12.9% of operational taxonomic units (OTUs) in their analysis were shared in paraffin controls and FFPE tissues and their removal significantly altered the inferred tissue taxonomic profile.

Conflicting results are reported in the literature about the comparison of paired microbiota between FFPE and fresh frozen preserved samples. In the same work, Pinto-Ribeiro et al. [29] did not find significant differences in alpha- or beta-diversity between the two methods of sample preservation, even if the microbial community present in FFPE tissues 12 months post-collecting was slightly different from that of frozen and FFPE tissues preserved for 6 months and 18 months. Also, DebesaTur and colleagues [30] reported comparable data between fresh frozen and FFPE tissues of colorectal cancer patients. Conversely, in a recent study Hockney et al. [26] showed that the fetal membranes-associated microbiota of FFPE and frozen stored samples were not comparable, and additionally that the microbiota of FFPE membranes was indistinguishable from the surrounding paraffin wax, highlighting the impact and biases of the processing methods on microbiota characterization. Likewise, Borgognone et al. [28] reported significant differences between the inferred microbiota compositions of paired FFPE and frozen colon rectal cancer tissues, which, following the authors’ suggestion, greatly depends on contamination issues. Despite every potential measure taken to ensure results displaying only the true oncobiota profile, the analysis of microbiota from FFPE tumor tissues is often associated with inevitable background noises.

Despite the above presented technical limits and concerns, the use of FFPE samples represents an extremely beneficial resource in oncobiome studies. First of all, FFPE preservation allows for the maintenance of the tissues at room temperature for a long time, requiring far less space and costs compared to other methods, and an easy transfer between facilities. Routine collection of histopathological FFPE samples provides a large availability of specimens essential for affordable and impactful retrospective analyses and to obtain robust statistical results. As already stated, the tumor microbiota could vary along the tumor mass, and FFPE samples offer the possibility to easily dissect tumor areas to compare the oncobiota associated with more or less vascularized and necrotic regions. These kinds of studies could help define the tumor-associated microbiota, allowing the understanding of microbes’ implication in carcinogenesis and treatment response. Finally, applying in situ hybridization approaches, FFPE samples offer the possibility of studying the spatial localization of intratumoral microorganisms, unveiling their interaction with host cells in the tumor microenvironment and suggesting the mechanisms used to drive cancer promotion [31].

In order to mitigate the existing issues and to successfully infer the oncobiota composition, many precautions should be taken by researchers during their workflows and here they are presented as a unique hypothetical protocol. First of all, the sample size and the sequencing depth should be increased in research projects using FFPE samples compared to the number necessary when utilizing fresh tissue [30]. Debesa-Tur and colleagues [30] suggest comparing when it is possible, the metagenomic analysis of paired FFPE and frozen tissue samples to assess the validity of the results from FFPE specimens, since they may be not optimal as a first choice, but complementary when fresh samples are unavailable.

Then, it will be necessary to include in the analysis a variety of negative controls to assess the degree of contamination introduced during processing steps (i.e. DNA extraction blank control, no-template control, sampling blank control) [32]. In particular, the inclusion of paraffin wax block controls, surrounding the analyzed tissue, is fundamental to detecting and removing paraffin contaminants [26, 29]. Increasing the sample size and including negative controls, could statistically reduce the bias from human and environmental contamination and then automatically refine results arising from the true comparison of the interest.

In addition, analyses such as quantitative PCR (qPCR), digital PCR, and electrophoresis gel would be helpful to control the DNA integrity before amplicon sequencing analysis [19, 26, 33]. Specific kits to repair DNA damage can also be employed [23].

To specifically avoid contaminant DNA, a number of precautions should be taken: 1) personnel should wear protections to cover exposed human surfaces; 2) the first few scrolls of the FFPE blocks should be discarded, the microtome regularly cleaned, and the selected slides collected in sterile tubes; 3) the DNA extraction must be performed under the sterile condition through specific extraction kit for FFPE sample designed to solve the cross-linkers and mutations [28, 29, 33]. In particular, protocols based on uracil DNA glycosylase (UDG) in combination with heat denaturation and lesion repair enzymes should be applied to further solve formaldehyde-induced mutation [8, 23]. In addition, the frequency of formaldehyde inducted C > T transitions can be evaluated through global imbalance value (GIV) score which is computed comparing forward and reverse reads assuming that in vitro damages to DNA are more probably present only in one strand [34].

The DNA sequencing method should be carefully evaluated, since amplicon sequencing, compared to WGS, could avoid the problem of the proportionally more abundant human DNA in low biomass samples [23]. Nevertheless, the fragmentation caused by formaldehyde could make the choice of hypervariable target region difficult [24]. Alternatively, if using the WGS sequencing, host DNA depletion strategies are strongly recommended [35, 36]. The amplicon sequenced DNA reads should be clustered as OTUs and not as amplicon sequence variants (ASVs) that are distinguishable by a single nucleotide difference, since the punctiform mutations in DNA, occurring if formalin preservation, would increase the data noise [23, 37].

Finally, specific bioinformatics approaches will help detect and remove contaminants in silico after the sequencing as long as the number of filtered OTUs is acceptable and comparable among samples [30]. For example, host DNA can be aligned to the relative reference database and then removed [24, 29] while bacterial contaminants should be identified through direct OTU comparison with negative samples [37] or using tools such as SourceTracker and Decontam [28, 38, 39].

Microbiome research has enormous potential in health and diseases, and the use of FFPE specimens could be helpful, especially in oncobiome research. Anyway, methodological pitfalls and limits of this approach may negatively affect this research field. For this reason, it is imperative to adopt a precaution and a standardized protocol to acceptably reduce the related bias and exploit the potential behind this storage method.