Gut Mycobiome in Inflammatory Bowel Disease: A Systematic Review of Case-Control Studies

Abstract

Background

The bacterial composition of the microbiome in inflammatory bowel disease (IBD) has been the focus of substantial interest. In contrast, the fungal part of the microbiome, the mycobiome, has only rarely been investigated—although anti-Saccharomyces cerevisiae antibodies, an antibody against fungal mannan, have been known for years as a biomarker for Crohn’s disease (CD), but not for ulcerative colitis (UC).

Methods

A systematic review of case-control studies on the human gut mycobiome in IBD was conducted using searches in EMBASE and MEDLINE.

Results

Twenty-seven studies, with a total of 1406 IBD patients and 1060 controls were identified. The differences in alpha diversity varied across studies and were related to geography, whereas differences in beta diversity between cases and controls were found in a large majority of the studies. Overall, the results were inconsistent at different taxonomic levels, and the studied populations were heterogeneous, as were the methodological approaches. The most consistent finding was an increase of Candida for both CD and UC and of Malassezia in CD, where it was often linked to a decrease of Saccharomyces.

Conclusions

The mycobiome is altered in IBD, as differences in beta diversity were found between cases and controls consistently. Future studies should carefully standardize every step from sample collection through analysis and data processing, allowing external validation of findings. Inclusion of treatment naïve patients and symptomatic controls could further advance this field.

Graphical abstract

Graphical Abstract.

Key messages.

What is already known?

Multiple studies have suggested that changes in the mycobiome are associated with inflammatory bowel diseases.

What is new here?

This is the first systematic review of case-control studies on the mycobiome in inflammatory bowel disease, and it finds that differences in beta diversity are found in a majority of the studies; however, specific genus- or species-level alterations are rarely identical.

How can this study help patient care?

This review points to important considerations in future studies that investigate the role of fungi in disease pathogenesis as well as their potential as new treatment targets.

Introduction

Inflammatory bowel disease (IBD), primarily ulcerative colitis (UC) and Crohn’s disease (CD), is characterized by chronic gastrointestinal inflammation. The currently accepted hypothesis regarding IBD is that environmental risk factors trigger an immune response in genetically susceptible individuals. Alterations of the gut microbiota have been linked to the disease onset, and bacteria, which represent the most abundant microorganisms, have been studied extensively.^1,2 Interventional studies with fecal microbiota transplantation (FMT) have shown that restoring microbial balance can induce remission in patients with UC, which highlights the significance of addressing the microbiome.³ Other microorganisms present in the gut with lower abundance have been studied to a lesser extent, but increasing amounts of data are emerging on the presence and impact of fungi, known as the gut mycobiome.⁴ Intriguingly, the efficacy of FMT was recently linked to the composition of the pre-FMT mycobiome, thus highlighting clinically relevant inter-kingdom interactions.^5,6

Almost 3 decades ago, Sendid et al. reported that anti-Saccharomyces cerevisiae antibodies (ASCA) were present in CD. Examining ASCA can be helpful in discriminating between CD and UC and has been shown to identify patients with CD up to 5 years before diagnosis.^7–9

Martini et al. connected the B cell-mediated ASCA response to a fungal epitope with a CD4⁺ T cell reactivity against both food-derived and gut fungi, thus linking the mycobiome to pathological changes of the adaptive immune system.^10,11 Studies of the mycobiome in patients with IBD have, however, been inconsistent in findings, diverging in methodology and heterogeneous with respect to patient characteristics. Several methodological challenges may explain the conflicting results in the previous literature. First, the sturdy fungal cell wall decreases DNA yield, which necessitates a mechanical or enzymatic step in DNA extraction.^12,13 Second, bacterial DNA prevails in stool, as does human and bacterial DNA in biopsies, which means that getting acceptable yields of fungal reads in untargeted approaches, such as shotgun metagenomic sequencing, requires deep sequencing.¹⁴ Alternatively, targeted approaches, which focus on specific regions of ribosomal RNA (rRNA), such as the internal transcribed spacer (ITS) regions or 18S rRNA,¹³ can be applied to provide sufficient coverage. However, due to several factors such as low variability of sequences among fungal species in targeted regions, amplification biases, or poorly curated databases, the amplicon-based approaches often suffer from limited species-level resolution and identification uncertainty.^14,15 As a result, metagenomic knowledge of the mycobiome is often limited to genus-level information, whereas much of what is known about gut fungi at species level or even functional level is from in vitro studies of specific species. The most studied fungi of the gut are Candida albicans, S. cerevisiae, and Malassezia restricta. C. albicans is an abundant commensal fungus, well known for its pathogenic potential in immunocompromised hosts, as seen in glucocorticoid steroid (GCS)-treated patients with IBD and individuals undergoing antibiotic (AB) treatments.¹⁶ It has also been linked to exacerbation of inflammation in murine colitis models,¹⁷ as has M. restricta, a well-known skin commensal present in the gut of the vast majority of healthy donors.¹⁸ It has been suggested that its release of lipases, which produce unsaturated free fatty lipids, may contribute to pro-inflammatory effects. In contrast, S. cerevisiae has been shown to ameliorate chemically induced colitis in murine models, but its abundance in the diet may also reflect that it is not always a true commensal.¹⁸

Here, we present the findings of our systematic review of case-control studies evaluating the gut mycobiome in patients with CD, UC, or IBD, compared to symptomatic controls or healthy controls.

Methods

Search strategy

A systematic search of MEDLINE and EMBASE (ovidSP) was performed from inception to June 7, 2024 on studies investigating the gut mycobiome in patients with IBD (CD and/or UC) in comparison with healthy or symptomatic controls. No study protocol was submitted to PROSPERO before this review was conducted. The search strategy included the keywords: mycobiome, gut fungi, mycobiota, commensal yeast, or fungal microbiome in combination with IBD, CD, or UC. The detailed search strategy can be found in the Supplemental Data.

Study selection

We included studies reporting on the mycobiome of patients with IBD and healthy controls or symptomatic controls in a case-control design. We included both pediatric (the definition varied among studies from 16 to 20 years of age) and adult populations and accepted fecal samples, biopsies (small and large intestine), and colonic lavage as suitable material. Mycobiome analyses that warranted genus- or species-level relative abundances were required. These were defined as metagenomics, ITS sequencing, 18S sequencing, targeted qPCR (quantitative Polymerase Chain Reaction), and culture-based approaches. Studies that presented significant genus- or species-level differences between cases and controls as linear discriminant analysis effect size were included but not depicted in the quantitative presentation in the figures. A flow diagram of the study inclusion is shown in Figure 1.

Figure 1.

PRISMA flow diagram of study inclusion.

Outcomes

The primary outcome was significant differences in the relative abundance of fungi in the cases and the controls. The kingdom fungi has 8 taxonomic levels; we focused on the species (level 7), genus (level 6), or phylum (level 2) levels. When information about relative abundance was available, we only included results in which at least one group had an abundance of >1% to reduce the risk of false positive results due to contamination and noise. If no specific data for UC or CD were available, we reported the outcomes of the IBD group vs controls. Secondary outcomes were alpha and beta diversity. Alpha diversity is a measure for the mycobiome diversity within a sample and can thus increase or decrease, whereas beta diversity captures the dissimilarity or similarity between 2 communities. We only considered differences in alpha- or beta diversity between populations within each study, as comparisons across studies with distinct methodologies are too dissimilar to be considered comparable. Only statistically significant (P < .05) results were included, and we noted if these results were corrected for multiple testing.

Eligibility assessment and data extraction

Two of the authors (J.D.F., O.B.) screened all the retrieved studies by title, abstract, or full-text (Figure 1). The data extraction from the included studies was done by one author, collected in a spreadsheet and independently validated by a second author. Any discrepancies between the authors were handled by re-checking the data and obtaining a consensus.

For each study, we extracted data on country of origin, publication year, number of patients with UC, CD, or IBD, number of controls, age group (pediatric, adult), percentage of patients with active disease (definition varied across studies), and percentage of patients treated with GCS and AB. Furthermore, we extracted information on the material studied (feces, biopsies of large and small intestines, and colonic lavage), the method for analysis of the mycobiome (ITS, 18S, targeted qPCR, and culture-based), and the statistical method for multiple comparison correction in addition to the primary and secondary outcomes.

Quantitative assessment

When available, we recorded the relative abundance of any genus- or species-level fungi with significant differences between cases and controls, demonstrated in at least 2 study populations. To analyze the effect of our annotations on significant differences in alpha or beta diversity, we applied Cramér’s V, with predefined levels of association (0 to 0.1 extremely weak association, 0.1 to 0.3 weak association, 0.3 to 0.5 moderate association, and above 0.5 strong association).

Quality assessment

To assess the quality of the case-control studies included in this review, we used the Newcastle-Ottawa scale (Table S1). Scoring was handled as data extraction. In the Newcastle-Ottawa scale, the 3 domains of participant selection, participant comparability, and participant exposure are scored with up to 9 points in total.¹⁹ The case definition and the representativeness of cases as well as the selection and definition of controls were evaluated in the “participant selection” domain. The baseline characteristics (sex and age) were assessed in the “comparability” domain, and the ascertainment of exposure, the method for ascertainment, and the non-response rate were considered in the “exposure” domain.

Results

Study selection

We retrieved 219 records and excluded 173 after reviewing the title and abstract and removing duplicates. The remaining 46 articles were reviewed in full text, after which 19 were excluded (Figure 1). The remaining 27 studies, covering 590 patients with UC, 816 with CD, and 1060 controls, were incorporated in this review.^5,6,20–44 Of the 27 studies, 16 reported results for patients with CD (Table 1), 13 for patients with UC (Table 2), and 4 for patients with IBD only (Table 3), with regard to the primary outcome.

Table 1.

Study overview of the 16 studies reporting on patients with Crohn’s disease.

						CD patients				Controls
Year	First author	Country	Ref	Type of specimen	Age group	Active disease	GCS treated	AB treated	No. of participants	No. of participants	Assessment method of mycobiome	Multiple comparison correction
2006	Bougnoux	Belgium	20	Stool	NR	NR	NR	NR	79	98	Culture, Candida PCR	NR
2016	Hoarau	France and Belgium	27	Stool	Adult	27%	NR	NR	20	28	ITS1-2	NR
2017	El Mouzan	Saudi Arabia	25	Stool and biopsies	Pediatric	NR	NR	7%	15	20	ITS1-2	Benjamini–Hochberg
2017	Sokol	France	42	Stool	Adult	45%	19%	0%	149	38	ITS2	NR
2019	Limon	USA	31	Colonic lavage	Adult	0%	NR	NR	55	108	ITS1	FDR
2019	Imai	Japan	28	Stool	Adult	0%	15%	0%	20	20	ITS1-2	NR
2020	Qiu	China	34	Stool	Adult	76%	12%	0%	25	20	ITS1-2	NR
2021	Frau	Britain	26	Stool and biopsies	Adult	48%	13%	13%	23	20	18S	NR
2021	Frau	The Netherlands	26	Stool	Adult	NR	NR	0%	26	15	18S	NR
2021	Wang	China	37	Stool	Pediatric	100%	0%	0%	29	20	ITS2	FDR
2021	Nelson	Britain	32	Stool	Adult	0%	3%	47%	34	47	ITS1-2	FDR
2022	Zeng	China	40	Stool	Adult	76%	7%	5%	45	17	ITS2	NR
2022	Olaisen	Norway	33	Ileal biopsies	Adult	73%	25%	0%	44	40	ITS	NR
2022	Cimická	Czech Rep.	41	FFPE biopsy	Adult	62%	NR	NR	8	9	ITS	NR
2023	Krawczyk	Poland	29	Stool	Pediatric	63%	0%	0%	105	40	ITS1	NR
2023	Catalán-S.	Norway	21	Stool	Adult	66%	42%	0%	38	21	ITS1	Benjamini–Hochberg

Abbreviations: AB, antibiotics; CD, Crohn’s disease; FFPE, formalin-fixated paraffin-embedded; GCS, glucocorticoids; NR, not reported; Ref, reference.

Table 2.

Study overview of the 13 studies reporting on patients with ulcerative colitis.

							UC patients					Controls
Year	First author	Country	Ref	Type of specimen	Age group		Active disease	GCS treated	AB treated	No. of participants		No. of participants		Assessment method of mycobiome	Multiple comparison correction
2017	Sokol	France	42	Stool	Adult		45%	19%	0%	86		38		ITS2	NR
2019	Imai	Japan	28	Stool	Adult		0%	6%	0%	18		20		ITS1-2	NR
2020	Leonardi	Australia	6	Stool	Adult		100%	NR	NR	39		14		ITS1	NR
2020	Rühlemann	Germany	43	Stool	Adult		NR	NR	NR	38		66		ITS2	NR
2021	Sharifinejad	Iran	36	Stool	Adult		NR	0%	0%	79		58		Culture, PCR for 10 spp.	NR
2021	Frau	Britain	26	Stool and biopsies	Adult		45%	5%	0%	20		20		18S	NR
2022	van Thiel	The Netherlands	5	stool	Adult		100%	3%	NR	31		7		ITS	NR
2022	Li	USA	30	Colonic lavage	Adult		NR	NR	NR	40		38		ITS1-2	Benjamini– Hochberg
2022	Cimická	Czech Rep.	41	FFPE biopsy	Adult		100%	NR	NR	10		9		ITS	NR
2023	Del Chierico	Italy	24	Stool	Pediatric		74%	41%	0%	27		26		ITS2	NR
2023	Chen	China	23	Stool	Adult		100%	NR	0%	22		9		Metagenomics	Benjamini– ­Hochberg
2023	Catalán-S.	Norway	21	Stool	Adult		46%	30%	0%	46		21		ITS1	Benjamini– ­Hochberg
2024	Scanu	Italy	44	Stool	Adult		100%	77%	NR	25		17		ITS2	Benjamini– Hochberg

Abbreviations: AB, antibiotics; FFPE, formalin-fixated paraffin-embedded; GCS, glucocorticoids; NR, not reported; Ref, reference; spp., species; UC, ulcerative colitis.

Table 3.

Study overview of the 4 studies reporting on patients with IBD.

						IBD patients				Controls
Year	First author	Country	Ref	Type of specimen	Age group	Active disease	GCS treated	AB treated	No. of participants	No. of participants	Assessment method of mycobiome	Multiple comparison correction
2015	Chehoud	USA	22	Stool	Pediatric	19%	16%	0%	32	90	ITS1-2	Bonferroni
2021	Schierova	Czech Rep.	35	Stool	Adult	35–100%	NR	NR	27	37	ITS1	Benjamini–Hochberg
2023	Yu	China	39	Stool	Adult	63–81%	31%	22%	51	40	Metagenomics	Benjamini–Hochberg
2023	Yoon	Korea	38	Stool	Adult	49%	54%	NR	67	97	ITS2	FDR

Abbreviations: AB, antibiotics; GCS, glucocorticoids; IBD, inflammatory bowel disease; NR, not reported; Ref, reference.

Study material

Colonic lavage was used in 2 studies,^30,31 the combination of stool samples and biopsies in 2 studies^25,26 ileal biopsies in one study,³³ formalin-fixed paraffin-embedded biopsies in one study,⁴¹ and the remaining 21 studies reported on stool samples alone. Seven studies included only patients with active disease,^{5,6,23,35,37,41,44} 3 studies included patients in remission,^28,31,32 and the remaining 17 studies included patients with active as well as inactive disease (range 19%-81%). A GCS-free population was reported in only 3 studies (range of GCS use 3%-77% in the remaining populations),^29,36,37 an AB-free population in 12 out of 23 studies that reported on AB (range 5%-47% in the remaining populations),^{21–24,26,28,29,33,34,36,37,42} and a pediatric population in 5 studies.^{22,24,25,29,37}

Mycobiome assessment methods

A majority of the studies used ITS sequencing, including ITS1, ITS1-2, and ITS2. Two studies reported on 18S sequencing,²⁶ 2 on culture with successive qPCR,^20,36 and 2 on metagenomics data.^23,40

Changes in relative abundance: Primary outcome

Increase of Malassezia and Candida in CD

In CD, the phylum Ascomycota was increased in one study³² and decreased in another.³¹ The Basidiomycota-to-Ascomycota ratio was reported by 12 studies. Five of the 12 studies reported no change (n = 233 participants),^{21,25,28,34,35} in 3 studies, the ratio was increased (n = 434 participants),^31,33,42 whereas it was decreased in 4 studies (n = 305 participants).^29,32,40,41 The only other significant results at phylum level were a decrease of Rozellomycota and Mortierellomycota in patients with CD, as reported in a single study.⁴⁰

At the genus level, 5 studies reported an increase of Malassezia of between 1.1 and 20.4-fold relative abundance.^26,29,31,33 Four studies reported an increase in Candida of 1.8 to 20-fold, ^26,28,29,34 whereas 2 studies reported Candida to be decreased (0.5-fold).^26,40  Saccharomyces were decreased (0.9-0.7-fold) in 3 studies^26,29,33 and increased (3.2-fold) in one.⁴⁰ The genera Exophiala, Mucor, and Trichosporon were both increased and decreased in one study each.^27,37,40

At the species level, 2 studies reported an increase of Candida tropicalis,^27,29  M. restricta,^31,33 and C. albicans.^20,33 One study found a decrease of C. albicans (Figure 2A, and Table S4).⁴⁰

Figure 2.

Fungal taxa with altered relative abundance in Crohn’s disease and ulcerative colitis, and associations of study variables with diversity metrics. (A) Relative abundance (y-axis, 0%-100%) of 6 fungal taxa that were reported to differ significantly between patients with CD and healthy controls in at least 2 independent studies. (B) Relative abundance (y-axis, 0%-100%) of 7 fungal taxa with significant differences between patients with UC and healthy controls, also reported in at least 2 studies. In both panels, the x-axis lists the fungal taxa (at genus or species level), and each circle represents a study group (either patients or controls). The size of each circle is proportional to the number of participants in the respective group, providing a visual indication of sample size. The direction of change in relative abundance from controls to patients is indicated by gray arrows (upward for increase, downward for decrease). The name of the first author of each study is displayed adjacent to the corresponding data point. Only taxa with a relative abundance greater than 1% in at least one group are shown. Studies that did not report relative abundance data were excluded from this figure. Detailed study characteristics are provided in Tables 1 and 2. (C) Associations between 8 study-level variables and reported findings on alpha and beta diversity, quantified using Cramér’s V. Variables include country of study, method of mycobiome assessment, age group of participants, disease type, type of specimen analyzed, disease activity status, and treatment with glucocorticosteroids or antibiotics. Cramér’s V values are interpreted as follows: 0-0.1 = very weak association, 0.1-0.3 = weak association, 0.3-0.5 = moderate association, and >0.5 = strong association. CD, Crohn’s disease; UC, ulcerative colitis.

Increase of Candida and Alternaria in UC

One study found a decrease in the phylum Basidiomycota in UC.²⁴ The Basidiomycota-to-Ascomycota ratio was reported by 9 studies. Five studies found no differences between cases and controls (n = 328 participants), whereas 2 reported an increased ratio in cases with UC (n = 162 participants), and 2 observed a decreased ratio (n = 86 participants).

Three studies reported a 1.2-3.6-fold increase in Candida,  ^21,24,30 whereas 2 studies reported a species-level increase of C. albicans.^21,36 Similarly, the genus Alternaria was increased in 2 studies.^30,44

Contradicting results were found for the genera Malassezia, Saccharomyces, Clavispora, and for the species Candida glabrata that were all found to have increased in one study^21,36,44 and decreased in another (Figure 2B, and Table S2).^23,24,30,44

Heterogenous results for patients with IBD

In the 4 studies reporting results for UC and CD patients combined as IBD patients, no significant findings could be reproduced (Table S3).

Alpha- and beta diversity: Secondary outcome

In total, 23 studies reported on alpha diversity, but the results were not consistent. To calculate alpha diversity, the Shannon index was applied by all but 2 studies. Bray-Curtis dissimilarity was applied for the analysis of beta diversity in all but 6 studies and revealed more consistent results.

The country of study was strongly associated with differences in both alpha- and beta diversity. Significant differences in beta diversity between patients with IBD and controls were more frequent in studies of stool samples than biopsies, as the type of specimen was strongly associated with beta diversity (Figure 2C).

In patients with UC, differences of alpha- and beta-diversity compared to controls are heterogenous throughout the studies

For UC, mixed results were reported for alpha- and beta diversity. One study reported an increase in alpha diversity in patients with UC compared to the controls, ²³ whereas 2 studies found a decrease,^42,44 and 5 studies found no significant differences in alpha diversity.^{6,24,28,30,45} Differences in beta diversity were significant between patients with UC and controls in 5 studies,^{23,24,28,30,44} whereas one study only observed differences between patients with active UC compared to controls.⁴² Three studies did not observe differences in beta diversity (Figure S1).^6,26,43

In patients with CD, significant differences in beta diversity compared to controls were observed

Alpha diversity was reported in 10 of 15 studies of CD and found to be decreased in patients with CD compared to controls in only 2 studies,^26,33 whereas 8 studies showed no differences.^{27,28,32,34,37,40–42} Beta diversity between patients with CD and controls differed significantly in 7 studies,^{26,28,29,32,33,40,42} while 4 studies did not observe such differences (Figure S1).^25,26,34,37

IBD patients have an altered fungal community compared to controls

In patients with IBD compared to controls, alpha diversity was increased in 2 of 5 studies,^38,39 whereas one study found a decrease,²² and 2 studies observed no difference.^21,35 Out of 5 studies, 4 found significant differences in beta diversity between patients with IBD and controls,^21,22,38,39 while one study observed no such difference (Figure S1).³⁵

Quality assessment

The quality assessment applying the Newcastle-Ottawa scale revealed that the case definition was well described in all studies, but that 11 of 29 studies did not report sufficiently on the representativeness of the cases included. Baseline characteristics for sex were either with a > 10% difference between groups or not reported in 17 of 29 studies. In 18 of 29 studies, there was a > 10% difference in age between groups, or this was not reported (Table S1).

Discussion

In this comprehensive review of the gut mycobiome, we report consistent differences in beta diversity between patients with IBD and controls underlying general mycobiome alterations in the inflamed gut compared to non-IBD controls. However, the mycobiome assessment method, the specimen type, as well as the study’s country of origin were strongly associated with beta diversity.

The most consistent findings at the genus level were an increase of Candida and Alternaria in UC and an increase of Candida and Malassezia in CD, whereas the genus Saccharomyces was decreased (Figure 3). Notably, a majority of the studies included patients with active disease as well as GCS-treated patients. In most of the studies, alpha diversity did not reveal differences in diversity between patients with IBD and controls.

Figure 3.

The mycobiome in IBD is altered as differences in beta diversity are consistently found across most studies when patients with ulcerative colitis or Crohn’s disease are compared to controls. The results for alpha diversity are more heterogeneous and do not allow any conclusion. The most consistent findings at the genus level were an increase of Candida and Alternaria in UC and an increase of Candida and Malassezia in CD, whereas the Saccharomyces genus was decreased. CD, Crohn’s disease; IBD, inflammatory bowel disease; UC, ulcerative colitis.

Differences in the applied methodologies for mycobiome analysis across the included studies may have influenced the results of this review. One study did not include a step of mechanical impact in the DNA extraction.³⁸ This resulted in very low reads, which underscores the importance of a methodologically stringent approach. Only 2 culture-based studies of 158 patients with IBD were identified. Both culture-based studies utilized qPCR on the cultured material to allow for comparison of relative abundance, but these studies could not contribute to diversity measures. Although metagenomics could yield more detailed species-level information, the 2 studies taking this approach seem disconcerted by the low total abundance of fungi in the gut compared to bacteria in their analysis of fecal samples. ITS sequencing was applied in all but 6 studies, and there was no observable pattern with regard to the primary outcome, depending on whether ITS1, ITS2, or ITS1-2 was targeted. It has been suggested that ITS1 sequencing yields marginally more Basidiomycota in contrast to ITS2 sequencing which yields more Ascomycota.^46,47 We did not find evidence of this in the results of the Basidiomycota-to-Ascomycota ratio in the included studies.

In UC, an increase of the genus Alternaria was observed in 2 studies.^30,44  Alternaria was the first fungal pathogen related to autoimmune disease and has a well-established role in atopic dermatitis and asthma pathogenesis.⁴⁸

The most notable finding was an increase in the genus Candida in patients with UC, reported in 3 studies.^21,24,30 These 3 studies included patients that had active disease (46%-74%) and were treated with GCS (30%-41%). The high rate of GCS usage in these 3 studies has to be seen in the context of the increased risk of opportunistic Candida infections previously reported in patients with IBD and treated with GCS.⁴⁹ Nevertheless, the increased risk of infection may not necessarily be mediated through higher relative abundance at the genus level, but rather through species-level virulence factors.⁵⁰

Interestingly, the results for the genus Candida were conflicting in patients with CD, with a significant increase in 4 studies^26,28,29,34 and a decrease in 2.^26,40 The patients were heterogeneous in the 4 studies with an increase (0%-76% active disease, 0%-15% GCS treated), and a decrease (48%-76% active disease, 7%-13% GCS treated), which does not underscore the increase in the genus Candida found in patients with UC, treated with GCS. In the 2 studies with a decrease in Candida, 5%-13% of the patients received ABs, compared to 0% in all 4 studies with an increase of Candida. This contradicts existing evidence that Candida abundance increases after AB treatment.⁵¹ The very small control groups (n = 15 and 17) of the 2 studies that found a decrease in Candida possibly increase the risk of coincidental findings and may explain the discrepancies between these findings and our previous knowledge on GCS and AB treatment effects on Candida.

In CD, an increase of the genus Malassezia was reported in 5 studies.^26,29,31,33 The fact that different biological materials, including stool samples, colonic lavage, and ileal biopsies, were analyzed in the 5 studies and that both ITS and 18S sequencing were applied strengthens the validity and generalizability of these findings. In addition, the pronounced differences in disease activity, with 0%-73% of patients having active CD across the studies, suggest that the increase in the genus Malassezia is not only caused by active inflammation.

Malassezia, one of the core fungal genera of the human gut,⁵² exacerbated a dextran sodium sulfate-induced colitis in a mouse model and is linked to CARD9, a well-described IBD risk gene.³¹ Strikingly, Olaisen et al. found that increased abundance of Malassezia was associated with an increased risk of need for treatment escalation in patients with CD over one year of follow-up.³³ Sokol et al. correlated an increase of the order Malasseziales in CD with disease activity. A decrease of Saccharomyces in CD was also linked to activity in disease (and an increase of the genus Malassezia, in 3 studies^26,29,33). This tendency was also found for patients with UC in 2 studies^30,42 and this was supported by a decrease of S. cerevisiae in a third study.³⁶ These findings are particularly interesting in light of a recent report that patients with CD had high levels of CD4+ T cells reacting to S. cerevisiae, an abundant fungus not only in the gut mycobiome, but also in food (bread, beer).¹⁰ The closely related Saccharomyces boulardii was shown to ameliorate inflammation through commensalism with butyrate-producing bacteria in an animal study,⁵³ and S. cerevisiae was correlated with an increase of the butyrate-producing Roseburia, Blautia, and Ruminococcus genera, thought to be protective of gut immune homeostasis.^42,52,54 In addition, the established relationship between ASCA antibodies and CD points toward a complex inter-kingdom microbiome and host interaction, involving different pathways of host immunity.⁷

Few studies reported significant differences in alpha diversity. Five studies pointed toward a reduced alpha diversity in patients with IBD when compared to controls, whereas only 3 studies suggested an increase in alpha diversity, 2 of which had some methodological variance compared to the majority of the studies.^23,38,39 The Chen et al. study was based on metagenomics analysis.²³ Yoon et al. showed an increase of alpha diversity only when measured with the abundance-based coverage estimator and not with the more frequently applied Simpson or Shannon index, but first and foremost the DNA extraction from stool samples did not include a mechanical component (eg beads beating), and thus less than 50% of the samples yielded fungal DNA.³⁸ Notably, an alpha diversity increase in patients with IBD compared to controls was only observed in Asian study populations (3 out of 8 studies^23,38,39), whereas all studies with a decrease originated from Western Europe or North America (5 out of 13 studies).^{22,26,33,42,44} This was underscored by our estimation of the association between alpha diversity and different variables from the included studies, in which geography was strongly associated with both alpha and beta diversity. This reflects the complexity of this variable, which is related to lifestyle, diet, physical activity, and hygiene aspects that may affect the mycobiome. The impact of medication (AB, GCS) on alpha diversity would be best investigated within each group, and not across cases and controls in heterogeneous populations; thus, the low association between medications and alpha diversity is not surprising.

There was a clear trend, with most studies pointing toward differences in beta diversity between patients with IBD and controls (Figure S1). This underscores that there are significant changes in the mycobiome of patients with IBD when compared to non-IBD controls. The studies not finding significant differences in beta diversity had fewer participants compared to the 16 studies that found significant differences in beta diversity, which indicates that sample sizes of around 40 in each group are necessary to analyze beta diversity.

Disease activity, as proportion of participants, ranged from 0% to 100%, and the definition of active disease ranged from biochemical markers to patient-reported outcome measures across the studies. Nevertheless, disease activity is pertinent to the findings, as underscored by the increase of Candida in patients with UC and active disease.^21,24 However, it is unclear how much influence the treatment of active inflammation has on the gut mycobiome, where Candida similarly seems to increase with GCS usage in patients with UC, in line with previous evidence.^16,24 Only 3 studies included treatment-naïve patients (95 pediatric patients with CD in total).^25,29,37 Even as they all applied ITS sequencing of stool samples, the sample size was small, and one study did not account for multiple testing. There was no overlap in the findings they reported, and notably all significant findings of differences in relative abundance were identified among organisms present far less than 1%, increasing the risk of incidental findings in small study populations. Whether longstanding chronic inflammation, disease extent, and disease behavior, with complications such as strictures, fistulas, or abscess, also affect the mycobiome is still uncertain. This could all be brought to light if the mycobiome of new onset and untreated patients with IBD was well characterized, which would make it possible to compare these with studies of patients with longstanding disease, complications, and various treatments.

Based on the knowledge gained by this review, it is difficult to give evidence-based recommendations on how future studies should be conducted. Nevertheless, we will point out 6 considerations that are important to take into account in future studies:

1. Inclusion of treatment naïve participants to minimize treatment effects.

2. Inclusion of symptomatic controls to minimize stool consistency influence.

3. Standardized DNA extraction, including mechanical lysis of cell wall.

4. Amplicon sequencing, as metagenomics-based studies have yet to overcome issues of low abundance of fungi. So far, ITS2-based studies are most prevalent and thus comparison to existing literature favors this approach.

5. Inclusion of stool as well as biopsy or colonic lavage to increase the chance of finding true commensals and not dietary or environmental traces.

6. Validation of results in independent study populations.

Methodological Considerations

This review was limited by the broad inclusion criteria. The interpretations of the findings were challenged by their heterogeneity, but the lack of stringent methodological approaches and the very diverse patient populations necessitated this approach. Based on the collected information, a meta-analysis was not feasible.

Conclusion

The current knowledge of the gut mycobiome documents a change in fungal composition in patients with IBD, but research on this topic lacks greatly behind the bacterial microbiome. The notion that changes in the mycobiome could contribute to IBD pathogenesis is supported by reports that link the mycobiome to both the gut immune response and the bacterial microbiome.

To address the importance of the gut mycobiome, larger multi-center studies that allow detailed investigation of at least genus-level differences among CD, UC, and controls are warranted. We suggest that the collection of samples and the analysis of samples should be standardized to allow for external validation. These studies should include incident patients who have not been treated with IBD medications or undergone IBD-related surgery.

Supplementary Data

Supplementary data is available at Inflammatory Bowel Diseases online.

Funding

This work was supported by the Odense University Hospital PhD Fund [grant number A5031 to J.D.F.]; NordForsk [grant number 90569 NORDTREAT to J.H.]; Innovation Fund Denmark [grant number 8114-00026B to J.K. and V.A.]; the Research Council of Norway [grant number 2988039 to M.L.H.]; the Vinnova [grant number 2019-01185 NORDTREAT to J.H.]; the Region of Southern Denmark’s Pulje for Fri og Strategisk Forskning [grant number A1381 to J.K. and J.D.F.]; the Deutsche Forschungsgemeinschaft Cluster of Excellence “Precision Medicine in Chronic Inflammation” [EXC 2167]; and the Deutsche Forschungsgemeinschaft Research Unit 5042 miTarget.

Conflicts of Interest

J.D.F. reports travel grants from Tillotts Pharma. C.A. served as an advisory board member for Ferring, Takeda, Tillotts, and Janssen. M.L.H. has served as a speaker and/or advisory board member for AbbVie, Ferring, Galapagos, MEDA, MSD, Pfizer, Takeda, and Tillotts Pharma. She has also received grant support from Ferring, Tillotts Pharma, Takeda, and Pfizer. V.A. has served as an adviser for MSD/Merck, Janssen and as a member of an advisory board for MSD/Merck. J.H. served as a speaker and/or advisory board member for AbbVie, Alfasigma, Aqilion, BMS, Celgene, Celltrion, Dr Falk Pharma and the Falk Foundation, Eli Lilly, Ferring, Galapagos, Gilead, Hospira, Index Pharma, Janssen, MEDA, Medivir, Medtronic, Merck, MSD, Novartis, Pfizer, Prometheus Laboratories Inc., Sandoz, Shire, STADA, Takeda, Thermo Fisher Scientific, Tillotts Pharma, Vifor Pharma, and UCB. He received grant support from Janssen, MSD, and Takeda. O.B., K.S., C.B., A.F., and J.K. declare no conflict of interest.

Ethical Considerations

This systematic review of previous published research generated no new data. Ethical approval was not required, and we adhered to the PRISMA statement.