Intestinal fungal flora as a biomarker for type 1 diabetes and uses thereof

Abstract

The present application relates to a kind of intestinal fungal flora as type 1 diabetes biomarker and its application, the biomarker includes 19 key fungal genera.The present application is screened from the discovery of 19 key fungal genera in multicenter new T1D child cohort by fungal ITS sequencing and multiomics analysis.Classifier is constructed by random forest model, and the combination of the fungal marker reaches 0.872 in AUC in discovery cohort, and 0.778 in AUC in verification cohort, which shows excellent diagnostic ability.

Description

Technical Field

[0001] This invention relates to the fields of medical testing and microbiome technology, and in particular to an intestinal fungal community as a biomarker for type 1 diabetes and its application. Background Technology

[0002] Type 1 diabetes (T1D) is an autoimmune disease that primarily affects children and adolescents, with a continuously rising global incidence. Currently, T1D diagnosis mainly relies on glucose metabolism indicators (such as fasting blood glucose and HbA1c) and autoantibody detection, lacking early, non-invasive biomarkers. The role of the gut microbiome in the pathogenesis of T1D is receiving increasing attention, but existing research largely focuses on the bacterial community, with limited understanding of the mycobiome. Although the mycobiome comprises only 0.1% of the gut microbiota, it plays a crucial role in immune regulation and metabolism. In recent years, the potential role of the gut mycobiome in immune regulation and metabolic control has been gradually revealed.

[0003] For some patients with type 1 diabetes mellitus (T1DM), who lack the ability to stably control their blood sugar and experience severe hypoglycemic episodes and hypoglycemic coma, clinical islet transplantation can effectively control diabetes. More than 1500 diabetic patients who underwent clinical islet transplantation showed lower risks and longer-lasting effectiveness compared to pancreas transplantation. Furthermore, pancreas transplantation carries significant surgical risks, potential serious complications, and immunosuppression. Therefore, early risk warnings for diabetic patients can significantly reduce these risks.

[0004] The role of the gut microbiome in the pathogenesis of type 1 diabetes (T1D) is receiving increasing attention, but existing research largely focuses on the bacteriome (such as the bacterial and metabolite biomarkers disclosed in CN114317671A), while little is known about the mycobiome. Although the mycobiome constitutes only a small portion of the gut microbiota, it plays a crucial role in immune regulation and metabolism. However, current technologies lack research on T1D-related fungal biomarkers, and there is a lack of systematically screened and validated combinations of fungal biomarkers. There is an urgent need to develop microecological biomarkers for early identification and intervention. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a gut fungal community as a biomarker for type 1 diabetes and its application. This invention utilizes fungal ITS2 sequencing technology and machine learning methods to screen a group of key gut fungal biomarkers from a multicenter cohort of newly diagnosed T1D children, and constructs a high-precision T1D risk prediction model, providing a new tool for non-invasive early screening and diagnosis.

[0006] To achieve this objective, the present invention adopts the following technical solution: In a first aspect, the present invention provides an intestinal fungal community as a biomarker for type 1 diabetes, said biomarker including any one or a combination of at least two of Mortierella, Falciphoriella, Conocybe, Phaeomyces, Hygrocybe, Podosphaera, Retiboletus, Nigrospora, Acaulospora, Calcisporiella, Neocosmospora, Schizothoriium, Neofusicoccum, Cryptococcus f. Tremellaceae, Lecythophora, Digitatispora, Latorua, Rhodotorula, or Dictyosporella.

[0007] Among them, Mortierella belongs to the genus *Mortierella*, Falciphoriella to the genus *Falciphoriella*, Conocybe to the genus *Conocybe*, Phaeomyces to the genus *Phaeomyces*, Hygrocybe to the genus *Hygrocybe*, Podosphaera to the genus *Podosphaera*, Retiboletus to the genus *Retiboletus*, Nigrospora to the genus *Nigrospora*, Acaulospora to the genus *Acaulospora*, Calcisporiella to the genus *Calcarisporiella*, Neocosmospora to the genus *Neocosmospora*, Schizothecium to the genus *Schizothecium*, Neofusicoccum to the genus *Neofusicoccum*, Cryptococcus f. Tremellaceae to the genus *Cryptococcus*, Lecythophora to the genus *Lecythophora*, Digitatispora to the genus *Digitatispora*, Latorua to the genus *Latorua*, Rhodotorula to the genus *Rhodotorula*, and Dictyosporella to the genus *Dictyosporella*.

[0008] The above 19 key fungal genera were selected by comparing the differences in gut fungal communities between children with T1D and healthy controls, and were ranked as the most important fungal genera using a random forest model.

[0009] This invention collects fecal samples from newly diagnosed T1D children and healthy controls, uses ITS2 sequencing technology combined with a random forest machine learning model to screen for fungal biomarker combinations with excellent discriminative ability, and constructs a T1D risk prediction model. The model has an AUC of 0.872 in the discovery cohort and an AUC of 0.778 in the validation cohort.

[0010] In one specific embodiment, the biomarkers are Mortierella, Falciphoriella, Conocybe, Phaeomyces, Hygrocybe, Podosphaera, Retiboletus, Nigrospora, Acaulospora, Calcisporiella, Neocosmospora, Schizothecium, Neofusicoccum, Cryptococcus_f__Tremellaceae, Lecythophora, Digitatispora, Latorua, Rhodotorula, and Dictyosporella.

[0011] Secondly, the present invention provides a method for constructing a model for predicting the risk of type 1 diabetes. The method includes: sequencing fungi in the feces of healthy individuals and patients with type 1 diabetes, obtaining the relative abundance of gut fungi as biomarkers of type 1 diabetes as described in the first aspect, inputting the data into a random forest model, training the model, storing the trained model, and obtaining a model for predicting the risk of type 1 diabetes.

[0012] The method described in this invention includes the following steps: collecting fecal samples from healthy individuals and patients with type 1 diabetes, respectively; extracting total DNA from the fecal samples; performing PCR amplification and library construction sequencing using specific primers targeting the internal transcribed spacer region 2 (ITS2) of fungi; performing bioinformatics analysis on the sequencing data to obtain the relative abundance of the key gut fungal community; using the relative abundance data as input variables, with health / disease as labels, training a classification model using a random forest algorithm; optimizing the model parameters and evaluating the model performance using cross-validation; and storing the trained model to obtain a model for predicting the risk of type 1 diabetes.

[0013] In one specific embodiment, the bioinformatics analysis includes quality control, noise reduction, and generation of an amplicon sequence variant (ASV) table using the DADA2 workflow, and species annotation based on the UNITE database using QIIME2.

[0014] In one specific embodiment, the sequenced sequence is the ITS2 region.

[0015] In one specific embodiment, the method further includes optimizing the parameters of the random forest model using 10-fold cross-validation.

[0016] Thirdly, the present invention provides a type 1 diabetes risk model constructed according to the method for constructing a type 1 diabetes risk model as described in the second aspect.

[0017] In one specific embodiment, the input variable of the model is the relative abundance of gut fungi as a biomarker for type 1 diabetes, as described in the first aspect.

[0018] This invention constructs a type 1 diabetes risk model by inputting the relative abundance of 19 key fungal genera in the fecal samples of individuals to be tested. The model outputs a risk score (0–1), which is scored using the Youden Index. When the score is ≥0.65, the individual is identified as high-risk and further autoantibody testing or OGTT validation is recommended. The Youden Index is calculated as follows: Youden Index = Sensitivity + Specificity - 1 The type 1 diabetes biomarkers of this invention can be used for early warning of type 1 diabetes risk, prediction of an individual's likelihood of developing type 1 diabetes in the future, and guidance for adjusting the gut microbiota environment. By determining whether one, two, or more of the above biomarkers are present in the gut microbiota of a subject, it is possible to effectively determine whether the subject is susceptible to type 1 diabetes.

[0019] The fungal biomarkers for type 1 diabetes of this invention can be used for early warning, diagnosis, and risk stratification of T1D risk. By detecting the relative abundance of the aforementioned fungal genera in fecal samples, high-risk individuals for T1D can be effectively identified.

[0020] Fourthly, the present invention provides a kit comprising reagents for detecting the gut fungi community described in the first aspect as a biomarker for type 1 diabetes.

[0021] This invention can also assess the risk of developing type 1 diabetes by specifically detecting the relative levels of intestinal biomarkers, providing a new detection method for the screening and diagnosis of type 1 diabetes, enabling early clinical intervention and preventing disease progression. Simultaneously, these biomarkers can also serve as intervention targets, continuously monitoring an individual's health status, thereby allowing for early intervention when certain abnormal characteristics associated with type 1 diabetes are detected.

[0022] Fifthly, the present invention provides a drug combination comprising drugs that regulate the gut fungi community as a biomarker of type 1 diabetes as described in the first aspect.

[0023] The biomarkers of the present invention can also be used as intervention targets to prepare drug compositions for regulating the gut microbiota, so as to balance the relative content of various genera. The methods for regulating the microbiota include: (1) dietary intervention: increasing dietary fiber intake to promote the proliferation of beneficial fungi; (2) probiotic supplementation: such as oral administration of yeast-containing probiotics; (3) avoiding the abuse of broad-spectrum antibiotics to prevent fungal imbalance; in order to reduce the risk of type 1 diabetes, the drug composition can regulate the content of metabolites in the intestine, reduce the amount of adverse metabolites that can be absorbed into the enterohepatic circulation and enter the host's blood, and improve the individual's internal environment.

[0024] In one specific embodiment, the drug combination further includes pharmaceutically acceptable excipients.

[0025] In one specific embodiment, the pharmaceutically acceptable excipients include any one or a combination of at least two of the following: carrier, diluent, binder, wetting agent, disintegrant, emulsifier, cosolvent, solubilizer, osmotic pressure regulator, surfactant, coating material, colorant, pH adjuster, antioxidant, antibacterial agent, or buffer.

[0026] In a sixth aspect, the present invention provides the application of the first aspect as a biomarker of type 1 diabetes and / or its detection reagent in the preparation of diagnostic reagents for type 1 diabetes.

[0027] Compared with the prior art, the present invention has at least the following beneficial effects: 1. This invention is the first to systematically screen and validate a group of gut fungal biomarkers that are significantly associated with type 1 diabetes in children, filling a research gap in this field.

[0028] 2. The random forest prediction model constructed based on the fungal biomarker combination of the present invention showed excellent diagnostic performance in both the discovery cohort and the independent validation cohort (AUC values ​​of 0.872 and 0.778, respectively), providing a highly sensitive tool for non-invasive early screening.

[0029] 3. The detection method of the present invention is based on fecal samples, is simple to operate, non-invasive, and is particularly suitable for large-scale screening and early warning of children and adolescents.

[0030] 4. The fungal biomarkers discovered in this invention can not only be used for diagnosis, but also serve as potential intervention targets, providing new ideas and directions for the prevention or adjuvant treatment of T1D by regulating the gut microbiota. Attached Figure Description

[0031] Figure 1 A heatmap showing the relative abundance of 19 key fungal genera in children with type 1 diabetes and healthy controls.

[0032] Figure 2 The ROC curves for a random forest classification model based on 19 key fungal biomarkers are shown in Figure a. Figure a shows the ROC curve for the discovery cohort; Figure b shows the ROC curve for the validation cohort.

[0033] Figure 3 Figure 1 shows the correlation analysis between key fungal biomarkers and clinical indicators. Figure 2a is a Spearman correlation heatmap of the relative abundance of key fungal genera and clinical indicators; Figure 3b is a ranking chart of gut fungal community composition and its relationship with clinical indicators.

[0034] Figure 4 Figure 1 shows the correlation analysis between key fungal markers and gut bacteria and fecal metabolites. Figure 2a is a Spearman correlation heatmap between key fungal genera and gut bacterial taxonomic units; Figure 3b is a Spearman correlation heatmap between key fungal genera and fecal metabolites. Detailed Implementation

[0035] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments. However, the following examples are merely simplified examples of the present invention and do not represent or limit the scope of protection of the present invention. The scope of protection of the present invention is determined by the claims.

[0036] Example 1 This embodiment screens key gut fungal biomarkers for type 1 diabetes. 1. Research subjects and grouping.

[0037] This was a multicenter case-control study that included 74 newly diagnosed children with type 1 diabetes mellitus (T1D) and 84 age- and sex-matched healthy controls. All samples were randomly assigned to a discovery cohort (n=120) and a validation cohort (n=38) to ensure that there were no significant differences between the two groups in baseline characteristics such as age, sex, and BMI (P>0.05).

[0038] 2. Sample collection and preservation.

[0039] Fasting stool samples were collected from all subjects and immediately placed in liquid nitrogen for flash freezing. They were then transferred to a -80°C freezer for long-term storage to avoid repeated freeze-thaw cycles.

[0040] 3. Fecal DNA extraction.

[0041] Total fecal DNA was extracted using the QIAamp PowerFecal Pro DNA Kit (Qiagen). DNA integrity was assessed by 1% agarose gel electrophoresis after extraction, and quantification was performed using the Qubit dsDNA HS Assay Kit (Thermo Fisher).

[0042] 4. PCR amplification and library construction.

[0043] The fungal ITS2 region was selected as the target, and PCR amplification was performed using specific primers.

[0044] Forward primer ITS1F: 5′-CTTGGTCATTTAGAGGAAGTAA-3′ (SEQ ID NO:1).

[0045] Reverse primer ITS2:5′-GCTGCGTTCTTCATCGATGC-3′ (SEQ ID NO:2).

[0046] PCR reaction system (25 μL): Template DNA: 10 ng; Takara Ex Taq HS (5 U / μL): 0.25 μL; Primers: 0.5 μM each; dNTP Mixture: 200 μM each. PCR program: 95℃ for 5 min; 33 cycles (95℃ for 1 min, 60℃ for 1 min, 56℃ for 1 min); 72℃ for 7 min.

[0047] The PCR products were detected by 1.5% agarose gel electrophoresis, and those that passed the test were purified using AMPure XP magnetic beads (Beckman).

[0048] 5. Sequencing and data preprocessing.

[0049] The purified library was sequenced using PE250 sequencing on an Illumina NovaSeq 6000 platform. After sequencing, primer sequences were removed using cutadapt, and then the data were processed by DADA2 (QIIME2 plugin) for quality control, noise reduction, assembly, and dechimeric removal to obtain the ASV (Amplicon Sequence Variant) table.

[0050] 6. Species annotation and abundance calculation.

[0051] Species annotation of ASVs was performed using the q2-feature-classifier module of QIIME2, based on the UNITE v9.0 database. Annotation results were summarized at the genus level, and the relative abundance of fungal genera in each sample was calculated.

[0052] 7. Screening of key fungal genera.

[0053] Using the `randomForest` package in R, a random forest classification model was constructed with relative abundance of fungal genera as a feature and T1D / health as a label. 10-fold cross-validation was used to optimize the parameters (ntree=1000, mtry=5). Fungal genera were ranked by importance based on MeanDecreaseGini values, and the top 19 genera were selected as key markers, as shown in Table 1. The top 19 key fungal genera selected based on the random forest model are shown in Table 1. Among them, Mortierella belongs to the genus *Mortierella*, Falciphoriella to the genus *Falciphoriella*, Conocybe to the genus *Conocybe*, Phaeomyces to the genus *Phaeomyces*, Hygrocybe to the genus *Hygrocybe*, Podosphaera to the genus *Podosphaera*, Retiboletus to the genus *Retiboletus*, Nigrospora to the genus *Nigrospora*, Acaulospora to the genus *Acaulospora*, Calcisporiella to the genus *Calcarisporiella*, Neocosmospora to the genus *Neocosmospora*, Schizothecium to the genus *Schizothecium*, Neofusicoccum to the genus *Neofusicoccum*, Cryptococcus f. Tremellaceae to the genus *Cryptococcus*, Lecythophora to the genus *Lecythophora*, Digitatispora to the genus *Digitatispora*, Latorua to the genus *Latorua*, Rhodotorula to the genus *Rhodotorula*, and Dictyosporella to the genus *Dictyosporella*.

[0054] Table 1 Example 2 This embodiment constructs a random forest model. A T1D classification model was constructed using the R package `randomForest` with the abundance of 19 fungal genera as input variables. Model training employed 10-fold cross-validation with parameters: `ntree=1000`, `mtry=5`, and minimum number of samples per node = 3. Model performance (AUC) was evaluated. The importance ranking of the 19 fungal genera was based on MeanDecreaseGini values, as detailed in Table 1. Specific results are as follows: Figure 1 and Figure 2 As shown, 19 key fungal genera were selected. The random forest model achieved an AUC of 0.872 in the discovery queue and an AUC of 0.778 in the validation queue.

[0055] 1. Model Building A random forest classification model was constructed using the relative abundance of the aforementioned 19 key fungal genera as input variables. Training was performed using a discovery queue, with the following parameter settings: ntree=1000; mtry=5; Minimum number of samples per node = 3; The stability of the model was evaluated using 10-fold cross-validation, and AUC curves were plotted for different numbers of variables.

[0056] 2. Model Performance Evaluation Within the discovery cohort, model stability was assessed using 10-fold cross-validation, and the area under the receiver operating characteristic (ROC) curve (AUC) was calculated. Figure 2 As shown, the model achieved an AUC of 0.872 in the discovery queue. Subsequently, the model performance was externally validated using an independent validation queue (n=38), with an AUC of 0.778, indicating that the model has good discriminative ability and generalization ability.

[0057] 3. Risk scoring and threshold setting The model outputs a risk score between 0 and 1 for each sample. To determine the optimal diagnostic cutoff value, we plotted receiver operating characteristic (ROC) curves in the discovery cohort and calculated the sensitivity and specificity corresponding to different thresholds. The optimal threshold was determined using the Youden's Index maximization principle, calculated as: Youden's Index = Sensitivity + Specificity - 1.

[0058] In the detection cohort, the Youden index reached its maximum (J=0.632) when the threshold was set to 0.65, corresponding to a sensitivity of 82.1% and a specificity of 81.0%. This threshold also demonstrated good discriminative ability in the validation cohort, indicating that it has good generalization and clinical applicability. The optimal diagnostic cutoff value (0.65) was determined using the Youden index. Individuals with scores higher than this threshold were considered high-risk for T1D and further clinical examination (such as autoantibody testing) was recommended.

[0059] Example 3 This embodiment performs a correlation analysis between biomarkers and clinical indicators. (1) Spearman correlation analysis To explore the biological significance of key fungal biomarkers, Spearman correlation coefficients were calculated between the relative abundance of 19 fungal genera and clinical indicators (such as glycated hemoglobin, white blood cells, fasting blood glucose, and blood lipids).

[0060] HbA1c: Reflects the average blood glucose level over the past 3 months; ≥6.5% is the diagnostic threshold for T1D. FBG: Fasting blood glucose; ≥6.1mmol / L is abnormal. WBC: White blood cell count; elevated levels suggest infection or inflammation. TG: Triglycerides, an indicator of metabolic disorders.

[0061] The results are presented as heatmaps, such as Figure 3 As shown, this reveals that some fungal genera (such as Typhula) are positively correlated with blood glucose levels, while others (such as Rhodotorula) may be negatively correlated with inflammatory markers.

[0062] To explore the biological significance of key fungal biomarkers, we calculated the Spearman correlation coefficients between the relative abundance of 19 fungal genera and major clinical indicators (including age, lipid profile, liver and kidney function, BMI, complete blood count, and blood glucose-related indicators). The results are presented in the form of a heatmap, as shown below. Figure 3 As shown in the figure. Correlation analysis revealed significant associations between different fungal genera and clinical indicators, suggesting their potential role in host metabolism and immune status.

[0063] Examples are given below: Typhula (Corynebacterium) and HbA1c (glycated hemoglobin) showed a positive correlation (correlation coefficient > 0.05, P < 0.05). HbA1c is a key indicator reflecting recent glycemic control, and its elevation is associated with the progression of type 1 diabetes mellitus (T1D). This result suggests... Typhula The accumulation of [a substance] may be related to glucose metabolism disorders, or it may serve as a microecological marker of poor glucose control.

[0064] Rhodotorula (Rhodotorula spp.) showed a negative correlation with WBC (white blood cell count) (correlation coefficient < -0.05, P < 0.05). WBC is a commonly used indicator of systemic inflammation, and its decrease is often associated with immunosuppression or specific infectious states. This association may mean... Rhodotorula It plays a certain role in regulating the intestinal immune microenvironment, and changes in its abundance may reflect the host's immune status.

[0065] (2) Network analysis To further elucidate the role of fungi in the gut microbiota, bacteriome (16S rRNA gene sequencing) and metabolome data from the same cohort were integrated. A fungal-bacterial-metabolite network diagram was constructed using Cytoscape software, as shown below. Figure 4 As shown, network analysis revealed significant co-occurrence or mutual exclusion relationships between key fungal genera and certain known beneficial bacteria (such as Bifidobacterium) or specific fecal metabolites, suggesting potential microbial community interaction mechanisms.

[0066] In summary, this invention, through systematic microbiome research, has for the first time identified a combination of gut fungal biomarkers comprising 19 fungal genera, which can be used for early risk prediction of type 1 diabetes. The machine learning model constructed based on this biomarker combination exhibits high accuracy. This invention provides a new solution for non-invasive early screening of T1D and lays the foundation for further research on the role of gut fungi in the pathogenesis of T1D and the development of microecological intervention strategies.

[0067] The applicant declares that the above description is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Those skilled in the art should understand that any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention fall within the protection and disclosure scope of the present invention.

Claims

1. A gut fungal community as a biomarker for type 1 diabetes, characterized in that, The biomarkers include any one or a combination of at least two of Mortierella, Falciphoriella, Conocybe, Phaeomyces, Hygrocybe, Podosphaera, Retiboletus, Nigrospora, Acaulospora, Calcisporiella, Neocosmospora, Schizothothecium, Neofusicoccum, Cryptococcus f. Tremellaceae, Lecythophora, Digitatispora, Latorua, Rhodotorula, or Dictyosporella.

2. The gut fungi community as a biomarker for type 1 diabetes according to claim 1, characterized in that, The biomarkers are Mortierella, Falciphoriella, Conocybe, Phaeomyces, Hygrocybe, Podosphaera, Retiboletus, Nigrospora, Acaulospora, Calcarisporiella, Ne ocosmospora, Schizothecium, Neofusicoccum, Cryptococcus_f__Tremellaceae, Lecythophora, Digitatispora, Latorua, Rhodotorula and Dictyosporella.

3. A method for constructing a model to predict the risk of type 1 diabetes, characterized in that, The method includes: sequencing fungi in the feces of healthy individuals and patients with type 1 diabetes, obtaining the relative abundance of gut fungi as biomarkers of type 1 diabetes as described in claim 1 or 2, inputting the data into a random forest model, training the model, storing the trained model, and obtaining a model for predicting the risk of type 1 diabetes.

4. The method for constructing a model for predicting the risk of type 1 diabetes according to claim 3, characterized in that, The sequenced sequence is the ITS2 region.

5. The method for constructing a model for predicting the risk of type 1 diabetes according to claim 3 or 4, characterized in that, The method also includes optimizing the parameters of the random forest model using 10-fold cross-validation.

6. A type 1 diabetes risk model constructed using a method for constructing a type 1 diabetes risk model according to any one of claims 3-5.

7. The type 1 diabetes risk model according to claim 6, characterized in that, The input variable of the model is the relative abundance of gut fungi as a biomarker for type 1 diabetes, as described in claim 1 or 2.

8. A reagent kit, characterized in that, The kit includes a reagent for detecting the gut fungi as a biomarker for type 1 diabetes as described in claim 1 or 2.

9. A drug combination, characterized in that, The drug combination includes a drug that modulates the gut fungi community as a biomarker of type 1 diabetes, as described in claim 1 or 2. Preferably, the drug combination further includes pharmaceutically acceptable excipients; Preferably, the pharmaceutically acceptable excipients include any one or a combination of at least two of the following: carrier, diluent, binder, wetting agent, disintegrant, emulsifier, cosolvent, solubilizer, osmotic pressure regulator, surfactant, coating material, colorant, pH adjuster, antioxidant, antibacterial agent, or buffer.

10. The use of the biomarker of type 1 diabetes and / or its detection reagent as described in claim 1 or 2 in the preparation of diagnostic reagents for type 1 diabetes.

Images