Internal transcribed spacer sequencing to explore the intrinsic composition of fungal communities in fungal esophagitis

Abstract

BACKGROUND

Fungal esophagitis (FE) is caused by fungal invasion of the esophageal mucosa. Under endoscopy, the esophageal mucosa shows edema, congestion, erosion, and ulceration, and bleeds easily when touched, and the surface of the mucosa is covered with small white spots like "bean curd residue". Clinical cases showing typical FE under endoscopic imaging but negative esophageal mucosal brush (referred to as suspected FE) have increased the difficulty and challenge of clinical diagnosis and treatment. At present, the esophageal fungal flora of suspected case has not been thoroughly studied.

AIM

To characterize the fungal flora in FE, suspected FE, and the esophageal normal controls (NCs), and to identify marker species to improve detection of FE.

METHODS

This was a case-control study. A total of 19 patients with FE, 16 with suspected FE, and 10 NCs were selected by endoscopy. The esophageal cell brush samples of each group were sequenced by internal transcribed spacer (ITS) 1 and analyzed by bioinformatics.

RESULTS

In FE and suspected FE patients, species richness, species diversity and species evenness, as measured by the Chao1 index, Shannon index and Pielou index, were lower than in the NCs, and the comparison between the FE and NCs was the most significant (P < 0.05). Compared with the NCs, the relative abundance of Candida in FE and suspected FE patients was significantly increased (P < 0.001), while the relative abundance of Yarrowia was significantly decreased (P < 0.05). Moreover, Yarrowia abundance in the FE group was significantly lower than in the NCs and suspected FE groups (P < 0.001). The area under the curve for Candida in FE and suspected FE patients was 99.5% (P < 0.05) and 81.3% (P < 0.05), respectively. Finally, compared with FE patients, the relative abundance of Ascomycota and Candida in the esophageal flora of suspected FE patients was decreased, while the relative abundance of Yarrowia, Thermomyces and Pichia was increased.

CONCLUSION

ITS showed that composition of the fungal community was similar in the FE and suspected FE groups. ITS can be used as an auxiliary diagnostic method for FE and provide a theoretical basis for follow-up diagnosis and treatment.

Key Words: Fungal microbiota; Esophagitis; Candida; Yarrowia; Biomarker

Core Tip: Suspected fungal esophagitis (FE) is difficult to diagnose clinically and makes subsequent treatment difficult. In this paper, internal transcribed spacer sequencing was used to investigate the characteristics of fungal microflora in FE, suspected FE and the esophageal normal controls, and to search for its marker strains, to provide evidence for the diagnosis and treatment of suspected FE.

INTRODUCTION

Fungal esophagitis (FE) refers to esophageal inflammation resulting from invasion of fungi into the esophageal mucosa. The primary manifestations of this condition include swallowing pain, dysphagia, and chest pain behind the sternum[1].As the most common cause of infectious esophagitis, the most common pathogen of FE is Candida albicans (C. albicans), followed by Candida glabrata and Candida tropicalis. FE occurs most commonly in patients with low immune function and esophageal movement disorders (cardiac achalasia and scleroderma)[2,3]. With advances in endoscopic diagnostic techniques, the detection rate of FE has gradually increased. Endoscopy is the first choice for the diagnosis of FE, which is usually made by endoscopic images, brushing of esophageal mucosal wall cells, bacterial culture of esophageal mucosa, or histological examination. Endoscopic FE often presents as white plaques or exudates adhering to the esophageal mucosa, which is difficult to rinse, and may be accompanied by mucosal damage or ulcers. Invasion of the epithelium by fungal spores and pseudomycelia can be found after brushing or biopsy of the esophageal mucosa[1,3].

Studies have shown that shortly after birth, mice exhibit Bifidobacterium enrichment in the proximal esophagus and Lactobacillus enrichment in the distal esophagus. The different distribution of colony abundance and metabolic pathways of the colonies may change the histological morphology and gene expression of the esophagus[4]. Similarly, although fungi are less abundant in the human body than bacteria are, fungi also have a range of significant metabolic pathways and can synthesize a variety of metabolites, such as amino acids or antibiotics, which can have an impact on the esophagus[5,6]. Previous studies on the composition and characterization of fungi have mostly relied on fungal culture. With the development of internal transcribed spacer (ITS) sequencing technology, more methods have been applied to the exploration of fungal flora. At present, fungal components in the oral, respiratory, gastrointestinal and urinary systems have been studied[7,8]. However, patients with FE are still diagnosed by endoscopy combined with esophageal smears, and the composition of the esophageal fungal community remains unclear[9]. We found that some patients had typical manifestations of FE under clinical endoscopy, but the histological results of esophageal mucosa brush were negative. Therefore, in order to understand the composition of fungal communities in these special cases, we used ITS sequencing technology to explore the characteristics and differences in fungal flora in the esophageal normal controls (NCs), suspected FE and FE groups.

MATERIALS AND METHODS

Participants and sample collection

This was a case-control study. The NCs comprised 10 subjects who reported no esophageal discomfort and no esophageal problems were observed under endoscopy. The suspected FE group had 16 patients and the FE group 19 patients who underwent routine upper gastrointestinal endoscopy between February and September 2023 at the Second Hospital of Hebei Medical University. The inclusion criteria were: (1) Signed informed consent; (2) Complete clinical information; (3) Diagnosis as esophagus normal, FE or suspected FE by endoscopy and histology; and (4) ≥ 18 years old. Exclusion criteria included: (1) Taking antibiotics, antifungals, and chemotherapeutic agents within 1 month of the study; and (2) Neoplastic disease.

This research was approved by the Research Ethics Committee of the Second Hospital of Hebei Medical University (No. 2023-R-112) and performed according to the Declaration of Helsinki. With the informed written consent of each study participant, an esophageal brush was used to scan the lesion site of patients with esophagoscopic FE and patients with suspected FE, and the middle esophageal mucosa (25 cm from the incisor) of healthy subjects with normal esophagus. The samples were partially applied for microscopic examination, while the remaining were collected in cryogenic vials and stored at -80 °C. In addition, we have gathered the subjects’ clinical data, which included gender, age, body mass index, smoking or drinking history, and the Kodsi Grade was determined by the same experienced endoscopist[10].

DNA extraction, amplification, and sequencing

By the CTAB method, the fungal genomic DNA was extracted from esophageal brush specimens using the OMEGA kit (Omega Bio-Tek, Norcross, GA, United States). The NanoDrop NC2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, United States) and agarose gel electrophoresis were used to measure the quantity and quality of the extracted DNA. PCR was used to amplify the V1 region of the fungal ITS1 using the specific fungal forward primer GGAAGTAAAAGTCGTAACAAGG and reverse primer GCTGCGTTCTTCATCGATGC. After a single quantitative step, the same amount of amplicons was combined, and the pair-end 2 × 250 bp sequencing was performed at Shanghai Personal Biotechnology Co., LTD. using the Illumina NovaSeq platform and the NovaSeq 6000 SP Reagent Kit (500 cycles). Microbiome bioinformatics was adopted by QIIME2 2019.4[11]. Then primers were removed using cutadapt[12]. The Divisive Amplicon Denoising Algorithm 2 (dada2) plugin, an open source software (R) package for modeling and correcting sequencing amplicon errors on multiple sequencing platforms, was used for quality filtering, denoising, merging, and chimera removal of sequences[13]. Fungal amplicon sequence variants (ASVs) were annotated using sequence alignment based on the UNITE database[14].

Statistical and bioinformatics analyses

The baseline data were analyzed using IBM SPSS statistics 21. The normality test and homogeneity of variances test were performed on the quantitative data. The quantitative data with a normal distribution were expressed as mean ± SD. Depending on the results of the normality test and homogeneity test of variance, we chose the analysis of variance or nonparametric Kruskal-Wallis test to compare the differences between the three groups. The Bonferroni method was used for pairwise comparisons between each group. Qualitative data were described as percentages and analyzed by the χ² test or Fisher's test as appropriate. Grade data were tested by the two-sample rank sum test between each group. P < 0.05 represented a significant difference. In addition, receiver operating characteristic (ROC) analyses were performed for the relative abundance of specific fungal genera, and the area under the ROC curve (AUC) was calculated.

QIIME2 and R packages (v3.2.0) were mainly used for sequence data analysis. The species composition and relative abundance at the phylum and genus levels were analyzed using QIIME2 and R software and plotted as bar graphs. To ensure that sequencing depth met the analytical requirement, ASV rarefaction curves were plotted. We used the ASV table in QIIME2 to calculate ASV level alpha diversity indices, such as Chao1, Shannon diversity index, and Pielou’s evenness, and visualized them as box plots. We then performed a beta diversity analysis using the Bray-Curtis metric to study structural changes in the microbial communities in the sample and visualized them on a non-metric multidimensional scale. Analysis of Similarities was performed using QIIME2 to evaluate the significance of differences in microbial community structure among the groups. A Venn diagram was generated to visualize the shared and unique ASVs among the groups. Linear discriminant analysis Effect Size (LEfSe), and random forest analysis were performed to detect the differences of fungal communities between groups and search for specific fungal genera using QIIME2.

RESULTS

Characteristics of patients and controls

The demographic and clinical data of the 10 NCs who had no endoscopic evidence of esophageal, gastric or duodenal disease or gastrointestinal symptom, 16 suspected FE patients and 19 FE patients, are shown in Table 1. There were no significant differences in age, gender, BMI, smoking status, alcohol consumption, or Kodsi Grade among the three groups. Among patients with FE and suspected FE (n = 35), 21 were Kodsi grade I, including 10 with FE (47.6%) and 11 with suspected FE (52.4%); 12 were Kodsi grade II, including seven with FE (58%) and five with suspected FE (41.7%); and two with FE were Kodsi grade III.

Table 1 The baseline clinical characteristics of the participants.

Variables	NC, n = 10	NFE, n = 16	FE, n = 19	P value
Age (years, mean ± SD)	41.80 ± 9.56	54.06 ± 16.62	56.84 ± 18.05	0.06
Gender, n (%)				0.29
Male	6 (60)	5 (31.3)	10 (52.6)
Female	4 (40)	11 (68.7)	9 (47.4)
BMI (kg/m2, mean ± SD)	24.61 ± 2.53	23.97 ± 4.28	23.77 ± 2.48	0.80
Smoker, n (%)	1 (10)	4 (25)	3 (15.8)	0.69
Drinker, n (%)	5 (50)	3 (18.8)	4 (21.1)	0.21
Kodsi grade, n (%)				0.26
I		11 (68.7)	10 (52.6)
II		5 (31.3)	7 (36.9)
III		0 (0)	2 (10.5)
IV		0 (0)	0 (0)

BMI: Body mass index; NC: The esophageal normal control; NFE: Suspected fungal esophagitis; FE: Fungal esophagitis.

Endoscopic images of patients with FE were similar to those of patients with suspected FE (Figure 1). Compared with healthy esophagus (Figure 1A), Kodsi grade I esophagus had a small number of raised white spots, < 2 mm in diameter, accompanied by edema, with no congestion or ulceration (Figure 1B). Kodsi grade II presented with multiple raised white spots, > 2 mm in diameter, accompanied by hyperemia and edema, but no ulcers (Figure 1C). Kodsi grade III presented as fused linear or patchy white plaques with edema and ulcers (Figure 1D).

Figure 1 Endoscopic manifestations of fungal esophagitis and suspected fungal esophagitis. A: Endoscopic esophageal manifestations in esophageal health patients; B: Kodsi grade I manifestations of fungal esophagitis (FE) and suspected FE; C: Kodsi grade II manifestations of FE and suspected FE; D: Kodsi grade III manifestations of FE.

Diversity of fungal communities in FE was similar to that in suspected FE

Sequencing depth was checked by drawing a richness sparse curve (Figure 2). All curves reached saturation, indicating sufficient sequencing depth. Chao1 index, Shannon index and Pielou's evenness index were used to represent the richness, diversity and uniformity of fungal microbial communities in the three groups, respectively. Compared with the NCs group, the richness, diversity and uniformity of the fungal microbial community in the FE group were significantly decreased (Chao1 index, Shannon index and Pielou's evenness were all P < 0.05 as determined by nonparametric Kruskal-Wallis test). Only the fungal richness in the suspected FE group was significantly lower than in the NCs group (Chao1 index P < 0.05 as determined by nonparametric Kruskal-Wallis test), and there was no significant difference in diversity and evenness. Alpha diversity subgroup analysis of Kodsi grade in the FE and suspected FE groups found no significant difference in Chao1 index, Shannon index and Pielou's evenness index between Kodsi grades I and II of the two groups (P > 0.05 as determined by nonparametric Kruskal-Wallis test; Figure 2C and D). However, the richness, diversity and evenness of fungal microbial communities showed a decreasing trend with increase of Kodsi grade for both FE and suspected FE. Based on Bray-Curtis distance matrix analysis, the differences in fungal community distribution among the three groups were shown. As the distance between the two points decreased, the similarity increased, and the difference in fungal microbial community structure between the groups decreased (Figure 2E). The results suggested that there were some differences in fungal microbial communities between the NCs, FE and suspected FE groups.

Figure 2 The mycobiota diversity in three groups. A: A rarefaction curve based on species richness in the Chao1 index. The X-axis shows the drawing depth, and the Y-axis shows the median value of the alpha diversity index calculated 10 times vs the boxplot. The flatter the curve, indicating that the sequencing result is sufficient to reflect the diversity contained in the current sample; B: Comparison of alpha diversity analyses between groups in the esophageal normal controls, suspected fungal esophagitis (NFE), and FE groups; C: Comparison of alpha diversity analysis between Kodsi I and Kodsi II within the FE group; D: Comparison of alpha diversity analysis between Kodsi grade I and Kodsi grade II within the suspected FE group. The alpha-diversity was evaluated by the Chao1 index, Shannon, and Pielou’s evenness, Chao1 index, Shannon index and Pielou's evenness index represent richness, diversity and evenness within the community, respectively; E: Analysis of NMDS using the Bray-Curtis distance matrix (stress = 0.14). ^aP < 0.05, ^bP < 0.001. NC: The esophageal normal control; NFE: Suspected fungal esophagitis; FE: Fungal esophagitis.

Composition of fungal communities FE and suspected FE was similar

The relative abundance of microbial composition in the NCs, FE and suspected FE groups at phylum and genus levels was compared. At the gate level, Ascomycota had the highest abundance, followed by Basidiomycota (Figure 3A). At the genus level, the relative abundances of the three groups of fungi are shown in Figure 3B. The top three dominant genera in the NCs group were Yarrowia (28.5%), Candida (16.8%) and Thermoascus (11.1%). The suspected FE group was dominated by Candida (70.9%), Yarrowia (10.0%) and Pichia (3.1%). The top three genera in the FE group were Candida (94.4%), Thermoascus (3.1%) and Yarrowia (0.8%). We can intuitively see that although relative abundance of fungi in the FE group differed from that in the suspected FE group, the species composition was roughly the same. In the order of NCs group, suspected FE group and FE group, Candida showed a gradual increase, while Yarrowia showed a gradual decrease. The same trend was found in subgroup analysis of endoscopic Kosdi grade in both the suspected FE and FE groups (Figure 3C). We performed separate analyses of Candida and Yarrowia and found that an increase in Candida colonization was associated with a decrease in Yarrowia abundance compared with the NCs group, suggesting an antagonistic effect (Figure 3D).

Figure 3 Taxonomic characteristics of esophageal fungi microflora in the esophageal normal control group, suspected fungal esophagitis group, and fungal esophagitis group. A: Relative abundance at the phylum level in three groups; B: Relative abundance at the genus level in three groups; C: Relative abundance at the genus level of the Kodsi subgroup in the suspected fungal esophagitis (NFE) and FE groups; D: Relative abundance of Candida genera and Yarrowia genera in three groups. Diverse letters specify statistically significant differences among groups. ^aP < 0.05, ^bP < 0.001. NC: The esophageal normal control; NFE: Suspected fungal esophagitis; FE: Fungal esophagitis.

Analysis of fungal community species in esophageal mucosa can improve histological diagnosis

To assess the similarities and differences in the fungal communities of the three groups, we compared the fungal ASVs and presented the results in a Venn diagram. There were 290, 307 and 282 ASVs in the NCs group, suspected FE group and FE group, respectively, and the three groups shared 30 ASVs (Figure 4A). LEfSe analysis and random forest analysis were used to find the dominant species. We used the intersection of the two results as a marker. Comparison between the NCs and FE groups showed that the fungi genera endemic to the FE group included Candida and Pleurotus (Figure 4B and C). The specific fungal genus in the NCs group was Yarrowia. In the comparison between the suspected FE and FE groups, the specific fungal genus in the FE group was Candida, compared with Yarrowia, Thermomyces and Pichia in the suspected FE group (Figure 4D and E). In comparison of the suspected FE and NCs groups, the specific fungal genera in the suspected FE group were Candida and Aspergillus. In the NCs group there were Yarrowia, Pleurotus and Cladosporium (Figure 4F and G). To search for biomarkers, ROC analysis was performed on Candida, the most specific genus in the suspected FE and FE groups (Figure 4H and I). The AUC of Candida in the FE and suspected FE groups was 99.5% (95%CI: 97.8%-100%, P < 0.05, cutoff = 0.65) and 81.3% (95%CI: 63.7%-98.8%, P < 0.05, cutoff = 0.59), respectively. In order to further explore the diagnostic performance of histological brush and ITS sequencing, we compared the positive diagnostic rates of FE. The positive rate of endoscopy + microscope group and the positive rate of endoscopy + ITS are shown in Table 2. The positive rate of the two groups was statistically different (P < 0.05). It was proved that Candida was the marker pathogen in the FE and suspected FE groups.

Figure 4 Specific species differences in the three groups. A: Venn diagram based on amplicon sequence variant (ASV) level. Each color block represents a group, and the overlapping areas between the color blocks indicate the ASVs shared among the corresponding groups. The number of each block indicates the number of ASVs contained in that block; B: Linear discriminant analysis Effect Size (LEfSe) analysis of the esophageal normal controls (NCs) group and fungal esophagitis (FE) group at genus level; C: Random forest analysis of NCs group and FE group at genus level; D: LEfSe analysis of suspected FE group and FE group at genus level; E: Random forest analysis of suspected FE group and FE group at genus level; F: LEfSe analysis of suspected FE group and NCs group at genus level; G: Random forest analysis of suspected FE group and NCs group at genus level; H: Receiver operating characteristic (ROC) curve of Candida species relative abundance in the NCs group and FE group; I: ROC curve of Candida species relative abundance in the NCs group and suspected FE group. NC: The esophageal normal control; NFE: Suspected fungal esophagitis; FE: Fungal esophagitis.

Table 2 Comparison of the positive rate between the two diagnostic methods.

Groups	Positive, n (%)	Negative, n (%)	Total (n)	χ2	P value
Endoscopy + microscope	19 (54.29)	16 (45.71)	35	8.23	0.008
Endoscopy + ITS	30 (85.71)	5 (14.29)	35

ITS: Internal transcribed spacer.

The FE and suspected FE groups were similar in terms of fungal species diversity, richness and evenness, and tended to be similar in terms of fungal species composition. Most importantly, Candida, a marker species of FE, also showed high specificity in suspected FE. Therefore, we have reason to believe that cases diagnosed as suspected FE with negative brush of esophageal mucosa can be confirmed as FE, and the method of fungal community species analysis of esophageal mucosa can improve the rate of histological diagnosis.

DISCUSSION

In recent years, fungal communities have received less attention than bacterial communities, but the conclusions are the same. In other words, whether bacteria or fungi, changes in the species community (including the composition of the species community and the abundance ratio of the microbial community) are always associated with the development of disease. For example, both polysaccharides from alkali extraction and molecular sieve purification (PNS2A) and molecular sieve process and a novel polysaccharide (PRY1-1) isolated from red yeast rice can reduce blood sugar and cholesterol by enhancing glucose and lipid metabolism by changing intestinal microbial composition, thereby alleviating the symptoms of diabetic mice and high-fat diet-induced mice to protect liver and kidney function[15,16]. Sporoderm-removed Ganoderma lucidum spore inhibits esophageal squamous cell carcinoma through monocyte chemoattractant protein 1, which is involved in the regulation of PI3K/AKT/mTOR and Erk signaling pathways[17]. At the same time, studies have also shown that the ratio of Candida to yeast increases with the progression of colon cancer[7]. Moreover, such changes are not limited to the digestive system, but are also found in respiratory diseases. There are also significant differences in the composition of bacterial and fungal microflora in bronchoalveolar lavage fluid between patients with or without chronic obstructive pulmonary disease[18]. Ghannoum et al[8] conducted IST sequencing on oral tissues of healthy people and found that the most common fungi were Candida species, and Cladosporium, Aureobasidium, and Saccharomycetales, the first three fungal genera with high abundance except Candida. In the stomach, Saccharomyces, Malassezia, and Candida were predominant[19]. In our experiments, we found that in the esophageal healthy people, the most common fungal genus in the esophagus is Yarrowia, followed by Candida. However, the most common fungal genus in FE and suspected FE was Candida.

C. albicans is a normal colonizing fungus that exists widely in the human digestive tract. When the microflora is unbalanced due to the low immunity of the human body, the morphological transformation between the mycelia of C. albicans will transform them into virulent pathogenic fungi[20,21]. Gotthardt et al[22] found Candida enrichment in the bile of some patients with cholangitis after orthotopic liver transplantation, which was significantly correlated with the reduced survival rate of retransplantation. Matsuo et al[23] found a significant correlation between endoscopic grade and Candida load in 323 cancer patients with esophageal candidiasis, and Candida invasion of squamous mucosa was also associated with treatment failure. At the same time, studies have shown a significant increase in Candida colonization in the oral microbiota of human immunodeficiency virus (HIV)-infected patients[24].

It is well known that Candida is a major opportunistic pathogen. If the immune function of the body is low or antibiotics are administered, fungal invasiveness can be enhanced through gene regulation, and the esophageal mucosa can be invaded, resulting in a series of inflammatory and immune responses, and FE[25]. In our study, it was verified that Candida is the main pathogenic strain in FE, with high specificity and sensitivity. The main pathogenic species of Candida is C. albicans, which is a polymorphic fungus. Its growth cycle consists of four forms: Yeast, pseudohyphae, fungal hyphae and chlamydospores, among which, yeast and mycelia are the most distinctive[21,26]. With changes in host conditions, the growth cycle and histological morphology of C. albicans change, resulting in changes in fungal virulence and biofilm formation[27-29]. Therefore, we speculated that although the relative abundance of Candida was the highest in FE and suspected FE with negative brush histology, the relative abundance of C. albicans differed between patients with FE and suspected FE. There may be different forms of C. albicans on the esophageal mucosa in the two groups, which may be one of the reasons for the negative histological detection of esophageal mucosal brush.

We observed that increased colonization of Candida in FE correlates with a decreased abundance of Yarrowia, suggesting an antagonistic relationship between these two fungi. Yarrowia is a member of the ascomycetes, belonging to the yeast family. It is widely found in nature and can be isolated from water, soil and even foods that contain a lot of fat and protein. It has the ability to degrade a variety of hydrophobic substances, and can effectively produce a large number of valuable metabolites, such as proteins, minerals, vitamins and fatty acids and so on. Therefore, it is mainly used in food processing, feed raising and other fields[30,31]. In addition to the above biological functions, it can also inhibit the growth of pathogenic bacteria such as Vibrio parahaemolyticus[32]. When Yarrowia grows on the contaminated substrate, it can form a protective biofilm and play a bioremediation role[33].

Although Yarrowia is relatively safe, in some extreme cases, such as after prolonged intubation or indwelling of an intravenous catheter, it can also cause disease as an opportunistic pathogen in immunocompromised patients[34,35]. In our study, we found that the relative abundance of Yarrowia was higher in esophageal healthy people, but significantly lower in patients with FE and suspected FE, suggesting that Yarrowia may exist as a beneficial bacterium in FE and suspected FE.

We found the presence of Pichia in the esophagus. Pichia is a mature host for protein expression that is widely used in biopharmaceutical and industrial enzyme production[36]. In addition to its good antibacterial properties, it can also inhibit the growth of fungal mycelia and colony size, and play an antifungal role[37]. Mukherjee et al[38] found that there was an obvious antagonistic relationship between Candida and Pichia in the oral fungal community of HIV-infected individuals. Pichia can ameliorate oral candidiasis by blocking the growth of C. albicans, mycelial morphology development, and biofilm formation, and competing with nutritional restriction. Pichia kudriavzevii can block Akt-1, mTOR and JAK-1 pathways, and induce apoptosis of colorectal cancer cells[39]. We also found that, compared with FE, the relative abundance of Pichia pastoris was significantly increased in suspected FE and in the NCs group, which was negatively correlated with Candida. Therefore, we speculate that the above changes in the relative abundance of pathogenic and beneficial fungi may be a possible reason for the negative esophageal brush histology in patients with endoscopic FE.

However, several limitations should be acknowledged in our study. First, the study population was small. Second the microbial communities of esophageal fungi between different groups may be disturbed by factors such as underlying diseases. Third, our current research was limited to the clinical level. In the future, we will carry out basic biological studies and randomized controlled trials.

CONCLUSION

In summary, our study explored the composition and biodiversity of esophageal fungal flora in esophageal health people, and patients with FE and suspected FE. We compared the differences in fungal microbial communities between groups, and used high-throughput sequencing to search for specific marker fungi. Comparing the composition and differences of fungal microbial communities, we have reason to think that suspected FE can be diagnosed as FE. In conclusion, in cases of FE, study of the composition of the esophageal fungal community can reduce the false-negative rate of esophageal mucosal brush histology, improve the diagnosis rate and provide guidance for subsequent treatment.

ACKNOWLEDGEMENTS

We thank all subjects who participated in the study.