Gut microbiome changes in people with diabetic retinopathy in India. DRMS-India report # 1: operational protocol and trends from first 100 participants

Abstract

Background

Diabetic retinopathy (DR) is a common microvascular complication of diabetes mellitus (DM), and the leading cause of vision impairment and blindness. India is among the top three countries in DM prevalence, and both DM and DR are projected to rise sharply in the future. There is no accepted strategy for the prevention of DR other than DM control. Recent studies suggest that DM is associated with alterations in a core group of gut microbiota, and progression to DR may be influenced by changes within this core group, highlighting a potential link between DR and gut microbiome. We studied these changes in a protocol-driven large case-control study, the Diabetic Retinopathy Microbiome Study-India (DRMS-India: CTRI/2024/02/062511), analysed the results of the first 100 individuals, and evaluated variations in gut microbiome in DR.

Methods

The DRMS is designed to recruit 462 people aged ≥ 30 years into three cohorts: healthy controls (HCs), DM, and DR, at 17 independent sites in India. Shotgun metagenomic sequencing of first-pass morning fecal samples is performed at a centralized laboratory and correlated with disease status, lifestyle, dietary, and systemic factors.

Results

The first 100 participants included 26 HC, 33 DM, and 41 DR. The trends showed the DR group had 1, 6, and 10 unique core phyla, genera, and species, respectively. Alpha diversity was highest in the DR group; Beta diversity plots showed separate clusters of HCs and DR, with DM overlapping both. Firmicutes (highest in DR), Proteobacteria (highest in DM), Bacteroidetes, and Actinobacteria (highest in HC) were common phyla. Segatella was the most common genus, and Segatella copri was the most common species across all groups to date. Most microbial gene families were annotated to Molecular Functions (MF), and the pathways attributed to carbohydrate, amino acid, lipid, and nucleotide metabolism, indicating distinct functional adaptations in their gut microbiome.

Conclusion

Trends from the first 100 individuals indicate that the gut microbiome of Indians with DR exhibits discriminatory features in microbial diversity and abundance, as well as in gene families and pathways that impact host gut metabolism. Data trends from DRMS-India indicate a region-specific non-invasive biomarker that may guide preventive therapy for DR.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13099-026-00821-9.

Introduction

Diabetic retinopathy (DR) is common and the earliest detectable microvascular complication of diabetes mellitus (DM) [1]. India has a large population with DM (currently in the top three countries in the world). The IDF (International Diabetes Federation) estimates a nearly 75% rise in people with diabetes in the coming twenty-five years [1]. With 4% of people developing visually threatening complications or irreversible vision loss due to DR, it is a major health burden today in the working-age group [2]. India is a large (7th largest in the world) and populous (17.8% of the world population) country. The INDIAB study documented grossly varying trends of DM prevalence in different regions of India [3]. Equally alarming is a near-50% higher crude prevalence of DM in previously much less-affected Indian territories [4]. Regional studies on DR (the SPEED, 2016-17) [5] and the National DR Survey (2015-19) [6] in India have reported that over 50% of people with DM had hyperglycemia for at least 5 years, 60% had poorly controlled DM, and 90% had not received DR screening. Additionally, there were gross regional variations in DR and sight-threatening DR (STDR) [5].

While the current poor access to health care and the quality of eye care in India are likely to improve with new health policies (non-communicable disease clinics) [7] and technology (non-mydriatic fundus camera with artificial intelligence) [8], further research is needed to identify the reasons for regional variations in DM and DR in India [9]. DR takes five years or longer to manifest clinically and still longer to reach the treatable stages [9]. Currently, metabolic control is the only proven strategy for preventing DR. Often, the metabolic memory sets in much earlier than the clinical manifestation of DR [10]. It makes DR progression refractory to metabolic control. Thus, we need a new and/or novel treatment(s) to prevent the progression of early DR to its refractory stages [9, 10].

One such approach could be relating DM and DR through the gut microbiome. The association between gut microbiome and DM is not a new concept [11]. Several pre-clinical and clinical trials conducted over the last decade have shown control of DM with probiotics and microbiota transplant-induced changes in gut permeability, inflammation, glucose homeostasis, and alterations in metabolism [12]. In fact, metformin, the most common oral anti-diabetic drug, is said to act through its action on the gut microbiota [13].

Methods

Study rationale

Our earlier metataxonomic studies, limited to individuals in South India, showed qualitative and quantitative differences in the gut microbiome (bacteria and fungi) among individuals with DM, DR, and healthy controls [14, 15]. We also demonstrated differences in the intraocular (aqueous humor) microbiome of people with DM and DR, which were more apparent than those in the gut microbiome. These data provide the rationale to extend the study to the entire India to identify regional differences in the gut microbiota of people with DM with and without DR. The Diabetic Retinopathy Microbiome Study- India (DRMS- India) is designed to recruit people from the country’s four regions (North, East, West, and South) and conduct a metagenomic and functional analysis of the faecal samples.

Hypothesis and objectives

Hypothesis

DM is related to changes in a core group of microbiota, regardless of the background microbiome diversity. A healthy individual will progress to DR if further changes develop in the core group microbiota predisposing to DM.

Objectives

The two objectives of DRMS-India are (1) Understanding variations in gut microbiome in people with DM and DR living in four major regions of India and (2) Developing a clinic-friendly gut microbiome index for people with DM and DR in India.

Study design

DRMS-India uses a case-control design. The ‘cases’ are people with DM with/without DR, and the ‘controls’ are healthy people.

Study criteria

Inclusion: (1) People with type 2 DM 30 years or older, presenting to the outpatient service of the study sites and using insulin and/or oral hypoglycaemic agents; (2) with or without systemic comorbidities of DM and DR such as hypertension [16], dyslipidemia [17], diabetic nephropathy [18]. (3) with or without coronary artery and cerebrovascular disease identified from the current medical history and treatment; (4) without major ocular disorders. DM was diagnosed per the WHO (World Health Organization) recommended classification system [19]. DR was diagnosed and quantified using the International Clinical Classification of DR Scale [20].

Exclusion. (1) People with other chronic systemic disorders (including gastrointestinal disorders) not mentioned in the inclusion criteria; (2) Consumption of commercial pre-or pro-biotics or dietary supplements in the last 1 month; (3) Use of any alternate medicine for DM, DR or any other disease in the last 3 months; (4) Use of systemic antibiotic in the last 3 months.

Healthy controls (HCs) were defined as people without definitive DM or any other chronic systemic disease or ocular disorders while matching the other study criteria.

Study sites and population

Only resident Indians across various zones were recruited. This study required stool samples from all sites, stored under optimum conditions (described later), and a detailed ophthalmic workup. Hence, we identified key eye care facilities with proven capabilities that match the study criteria. Seventeen sites across 15 Indian states and 2 union territories agreed to participate in the study. This spread and selection of participating sites were considered a cumulative representation of four zones in India: North (4 sites), East (4 sites), West (4 sites), and South (5 sites). Considering the large sample frame, a consecutive sampling approach was adopted.

Sample size

The ideal sample size depends on the number of people with discriminatory genotypes, as these can serve as a basis for developing potential biomarkers or therapies. Segatella spp. (earlier called Prevotella) is known to be enriched in the gut of non-Western compared to Western populations [21]. It was also the most common organism in our earlier case-control study in South India using 16 S RNA sequencing [14]. We also detected the highest abundance of Segatella spp. in the early results of the current DRMS-India across all the 3 groups. Hence, we compared the proportional abundance of Segatella spp. in the HCs and people with DR groups (described in detail later) to determine the sample size for the DRMS-India. Using online calculators (https://select-statistics.co.uk/calculators/sample-size-calculator-two-proportions/), the minimum required sample was 154 in each group at a 95% confidence interval and a power of 80%. The total sample need is 462 (154 each HC, DM, and DR at a 1:1:1 ratio).

Protocol

Study procedures

Multiple training sessions were conducted for 17 sites. These included clinical evaluation sessions, fundus photography, and the morning stool collection method (using an abridged pre-recorded open-source video from DNAGenotek, Stittsville, Canada). A standardized protocol was developed and shared with the principal investigators at the 17 sites.

Participant de-identification

To maintain confidentiality and avoid errors, each site was assigned a unique site code indicating the zone (N, E, W, S) and the site number (1–5). The enrolled participants were assigned a local site number (e.g., S4-01, S4-02), indicating the South zone, Site 4, and Patients 1 and 2, respectively. A data management committee periodically reviewed the data.

Clinical Protocol consisted of collecting the recruited people’s clinical features, diet history, and morning stool sample. In brief, these were:

1. Clinical Documentation included personal demography (age, sex, spoken language, area in India where most lived, diet, and body mass index [BMI]), diabetes profile (onset and duration of DM measured in years, glucose measurement at time of diagnosis, and antidiabetic treatment), and systemic/ocular morbidities. The preferred language and zone of India where the participants lived for most of their lives were considered surrogates for the geographical influences.

2. Ophthalmic examinations included measuring Snellen’s visual acuity on an illuminated chart placed at 4 m, intraocular pressure by Goldmann applanation tonometer, indirect ophthalmoscopic fundus evaluation, a widefield (60⁰ or more) colour fundus image, and DR grading.

3. Dietary information was documented in two steps. Step (1) The third and Southeast Asia version of the UK Diet and Diabetes Questionnaire documented the impact of diet on DM (https://policystudies.blogs.bristol.ac.uk/2016/09/19/the-uk-diet-and-diabetes-questionnaire-a-new-tool-for-assessing-dietary-habits/); Step (2) A self-gradable microbiome questionnaire with 10 questions (each graded on a scale of 1–10) was used to study the effect of diet on the microbiome.

Quality checks

1. Blood workup: Serum creatinine (mg/dl), urine protein (optional), Hemoglobin (g/dl), fasting or random blood sugars (mg/dl), and HbA1c (%) were obtained in each participant. Each site restricted the blood workup to one NABL (National Accreditation Board for Testing and Calibration Laboratories)-accredited laboratory within their geographical area. A mechanism was put in place to randomly transfer pre-measured blood samples from the central site to the relevant laboratory for quality control of HbA1c reporting.

2. Diagnosis and grouping: The site principal investigator grouped the participants as HC, DM, or DR. The fundus images were transferred to the central site using protected online forms. The central image reading site confirmed the initial grouping.

3. Stool samples. The participant collected the first passage of the morning stool in OMNIgene•GUT | OMR-200 kits (DNAGenotek, Stittsville, Canada), and the timing was recorded. The collected sample was stored at ambient temperature (20–25 °C) per the protocol within 6 h of collection. These kits could preserve DNA for up to 60 days in their buffered solutions and were chosen based on the scope of the study and the need to transfer the samples in batches to the laboratory on a monthly basis. One site in the North, East, and West zones, and two in the South zone, stored the samples at −80 °C to allow additional laboratory analysis. The ambient temperature collected samples were transported to the central laboratory every 4–6 weeks, and the cold-storage samples were transported in a cold chain with a maintained temperature log. The enrolment was completed between January and April 2024 to avoid India’s high summer temperatures.

4. Data transfer: De-identified data from each site was transferred using exclusive online forms. The data were password-protected and accessible only to the site and the central principal investigator (PI). Revisions of the form, if any, were limited to the site PI for one week of sample collection. The site PIs were required to match the site code to the participant identity twice in the online form to confirm the accuracy of the data before the transfers.

Laboratory Protocol consisted of DNA extraction, metagenomic sequencing, and bioinformatics analysis. In brief, these were:

1. DNA Extraction: Microbial genomic DNA was isolated using the QIASymphony PowerFecal Pro DNA kit and a semi-automated method of microbial DNA purification using the QIAsymphony SP instrument. Samples were thawed at room temperature and vortexed for 1 min at maximum speed. Approximately 250–300 mg of the stool sample was placed in the PowerBead Pro Tube, and 800 µL of CD1 solution was added. The PowerBead Pro Tube was secured horizontally on the PowerLyzer24 for bead beating at 2200 rpm for two cycles of 2 min each, with a 2-minute pause between cycles. The PowerBead Pro Tube was centrifuged at 15,000 x g for 1 min. The supernatant was transferred to a new microcentrifuge tube, and 300 µl of CD2 solution was added to remove all contaminating materials like non-DNA organic and inorganic material, including polysaccharides, heme compounds, bile salts, humic substances, cell debris, and proteins, which may reduce DNA purity and inhibit the downstream DNA applications. The mixture was then centrifuged at 15,000 x g for 1 min at room temperature.

   After initial cell lysis and DNA extraction, the supernatant was transferred to a clean microcentrifuge tube and proceeded with the QIAsymphony DNASoilStool_600_V1 Protocol on the QIAsymphony SP instrument for washing and elution. After washing and eluting the DNA in the QIAsymphony SP instrument, the concentration and quality of the isolated DNA were checked, and the DNA was stored at −20 °C.

2. DNA Quality Assessment and Microbial DNA Confirmation: DNA concentrations of the samples were checked using a Nanodrop spectrophotometer. To confirm the presence of microbial DNA, the selected samples were amplified by polymerase chain reaction (PCR) using primers targeting the V3-V5 region of the 16 S rRNA gene. The PCR products were then subjected to gel electrophoresis to visualize the 560 bp band generated after amplification. Samples with a concentration greater than 2 ng/µl and an absorbance ratio of 260/280 between 1.86 and 2 were selected for sequencing.

3. Metagenomics sequencing: Shotgun metagenomic sequencing was performed at the BRIC-NIBMG core genomic facility (Kalyani, India). Initially, DNA quantification was carried out using the Qubit fluorometer to ensure accurate input amounts for library preparation. Libraries were prepared using the Nextera XT library preparation kit, which enables efficient fragmentation and segmentation of DNA samples with sequencing adapters. The prepared libraries were sequenced on the Novaseq 6000 platform using 2 × 250 bp read-length chemistry.

Bioinformatics statistical analysis

The raw fastq files obtained by shotgun sequencing were initially checked for quality parameters using the FastQC. In the preprocessing step, Kneaddata was used to trim adapter sequences and remove low-quality reads, short reads, and human reads from the raw data. After the data preprocessing, MetaPhlAn4 was used to profile the taxonomic features, and HUMAnN3 was used to profile functional features such as the gene families and the metabolic pathways. Further statistical analyses, such as alpha- and beta-diversity, ordination analysis of beta-diversity distances, differential abundance of features, and association of metagenomic features with metadata, were performed using built-in scripts in MetaPhlAn and HUMAnN, along with various R packages, including MaAsLin and LEfSe. To test for differences in alpha diversity and bacterial taxa among the three groups, a Kruskal-Wallis test was performed. For comparing the beta diversity estimate, i.e., Bray-Curtis dissimilarity index between groups, permutational MANOVA (PERMANOVA) was performed, and a Non-metric Multidimensional Scaling (NMDS) plot was generated.

Results

Clinical

This communication reports data from the first 100 participants, who have been completely analysed. The participant distribution across the 3 groups and the 4 zones, along with the basic demographics, is presented in Table 1. Among the 100 samples, the mean age of HCs was 10 years lower than that of people with DM or DR. The mean duration of DM was higher in people with DR. In the cohort, 10 people were insulin-dependent. Hypertension was the most common comorbidity (n = 23). The collected stool sample was the first stool pass in nearly all participants (98%), with a similar time of collection.

Table 1.

Demographic and Clinical parameters of the 100 participants distributed in the 4 zones

Parameters	Entire cohort (n = 100)	HCs (n = 26)	DM (n = 33)	DR (n = 41)
Age (yrs)	52.83 ± 12.10	45.53 ± 11.95	55.48 + 12.55	55.40 + 9.94
Zone as per site (N/E/W/S)	(40/17/24/19)	(10/7/3/6)	(13/4/12/4)	(17/6/9/9)
Duration of DM (years)	-	-	8.75 + 7.16	11.55 ± 6.61
Initial HbA1c (%) at diagnosis of DM	-	-	7.73 (n = 6)	9.1 (n = 9)
Height (cm)	160.44 + 8.67	160.82 + 9.94	160.81 + 7.93	160.38 + 8.59
Weight (Kg)	67.01 + 12.32	67.69 + 13.22	68.51 + 14.20	65.33 + 9.94
BMI*	20.79 ± 3.30	20.94 ± 3.28	21.14 + 3.98	20.33 ± 2.65
Creatinine (mg/dl)	0.91 ± 0.25	0.85 + 0.14	0.91 ± 0.16	0.95 ± 0.35
Haemoglobin (gm/dl)	12.12 + 5.70 (n = 73)	12.52 + 5.42 (n = 21)	11.98 + 5.95 (n = 22)	11.95 + 5.67 (n = 30)
Current HbA1c (%)	7.69 ± 2.31	5.73 ± 0.42	8.86 ± 2.30	8.2 ± 2.40

E=East, N=North, W=West, S=South. BMI*=Body mass index, measured as weight (Kg)/height² (m)

Microbiome

The 100 samples (HC − 26, DM − 33, and DR − 41) generated 915.2 million (average: 9.15 M) paired-end reads. After the initial quality check and removal of human reads, 778.2 million (average: 7.8 M) quality-filtered microbial paired-end reads were retained for phylogenetic analysis, gene family profiling, and metabolic pathway identification. There were more phyla, genera, and species in the DR group (Table 2). We used a subset of taxa > 0.5% average relative abundance in any group as the core microbiome and checked for common and unique core taxa (Table 3). The distribution of common and unique phyla, genera, and species in the HC, DM, and DR groups is shown in Fig. 1. Information on all bacterial taxa observed is shown in Supplementary Fig. 2.

Table 2.

Distribution of microbial taxa among different participant groups

Microbial features	HC (n = 26)	DM (n = 33)	DR (n = 41)
Phyla	11	15	17
Genera	422	487	690
Species	669	801	1064
Metabolic pathways	462	485	486

HC: healthy controls, DM: diabetes mellitus, DR: diabetic retinopathy

Table 3.

Distribution of Core microbial taxa among different participant groups with average relative abundance > 0.5% in any of the groups

Microbial features	HC (n = 26)	DM (n = 33)	DR (n = 41)
Phyla	4	4	5
Genera	31	29	34
Species	33	34	40

HC: healthy controls, DM: diabetes mellitus, DR: diabetic retinopathy

Fig. 1.

Numbers of different taxa in HC, DM and DR with an average relative abundance of more than 0.5% a) common and unique phyla in different participant groups, b) common and unique genera in different participant groups, c) common and unique species in different participant groups

Diversity analysis

The Shannon (Fig. 2a), Chao1 (Fig. 2b), and Gini (Fig. 2c) α diversity indices estimated the intra-individual diversity in DR, DM, and HC groups. It showed the highest α diversity in the DR, followed by the DM and HC groups. Non-metric Multidimensional Scaling (NMDS) plots were prepared using Bray-Curtis dissimilarity indices to compare the inter-individual variability (ß-diversity). The NMDS plots showed that the DR group clustered separately from HC, but DM overlapped with HC and DR (Fig. 3).

Fig. 2.

Alpha diversity estimation of each sample from three groups (Healthy controls, Diabetes mellitus, and Diabetic retinopathy) based on their a) Species richness and relative abundance [Shannon index], b) Total number of different species [Chao1 index], and c) Evenness of species abundances [Gini index]

Fig. 3.

NMDS plot based on Bray-Curti’s dissimilarity index of the microbial communities of each group (Healthy controls, Diabetes mellitus, and Diabetic retinopathy)

Distribution of Taxa

Firmicutes was the most abundant phylum in all groups, followed by Bacteroidetes, Actinobacteria, and Proteobacteria. The abundance of Firmicutes was higher in DR than in DM and HC (DR > DM> HC); it was the opposite for Bacteroidetes, which was higher in HC than in DM and DR (HC > DM> DR) (Fig. 4). At the genus level, Segatella was the most abundant genus in all groups, followed by Bifidobacterium. The abundance of Segatella, Rosburia, and Ligilactobacillus was highest in HC, followed by DM and then DR (HC > DM> DR) (Fig. 5). Segatella copri was the most abundant species in all groups, with the highest abundance in HC, followed by DM and DR (HC > DM > DR) (Supplementary Table 1).

Fig. 4.

Distribution of core phylum in Healthy controls, diabetes mellitus and diabetic retinopathy groups

Fig. 5.

Distribution of top 15 genera in Healthy controls, Diabetes mellitus, and Diabetic retinopathy groups

Gene family and pathways diversity with functional annotation

We annotated gene families derived from HUMAnN3 outputs with Gene Ontology (GO) terms and assigned them to their respective functional categories: Biological Process, Molecular Function, and Cellular Component using the GO.db package. A total of 5165 unique GO terms (i.e., gene families) were detected across all samples. The relative abundance of GO terms was compared across the three study groups (DM, DR, and HC). Across all groups, MF (Molecular Function) accounted for the highest proportion of functional activity, followed by BP (Biological Processes) and CC (Cellular Component) (Fig. 6). In the DR, DM, and HC groups, the MF comprised 59.3%, 58.59%, and 58.72%, respectively; the BP comprised 28.86%, 29.18%, and 29.13%, respectively; and the CC comprised 12.12%, 12.22%, and 12.14%, respectively. These proportional trends reflect consistent functional dominance of Molecular Functions across all groups, with possible impact on host gut metabolism.

Fig. 6.

Proportion of gene family categories in 100 samples

The gene families were further reconstituted into 509 microbial pathways at the community level; 448 pathways were common to all groups, whereas 13, 19, and 1 pathways were uniquely present in the DM, DR, and HC groups, respectively (Supplementary Fig. 2). Fifteen pathways were commonly present between DR and DM, 9 between DM and HC, and 4 between DR and HC. Most of the metabolic pathways could be categorized under several functions belonging to carbohydrate metabolism (n = 52), amino acid metabolism (n = 106, 21%-highest), lipid metabolism (n = 49), Cofactor biosynthesis (n = 55), and nucleotide metabolism (n = 59) (Supplementary Fig. 4).

Discussion

Preliminary trends from DRMS-India indicate possible uniqueness within the DR group at all evaluated taxonomic ranks. The participants with DR seem to have richer and more diverse gut microbiomes, while the microbiome was more evenly distributed in the DM group. Inter-participant diversity analysis revealed separate clusters of DR and HCs, while DM overlapped. The number of core taxa at different taxonomic levels was largely similar, indicating that the predominant taxa were shared across the HC, DM, and DR groups (Table 3). However, the relative abundance of the predominant taxa, such as Segatella at the genus level and Segatella copri at the species level, was not similar across the three groups (Supplementary Table 1). Faecalibacterium prausnitzii, Bifidobacterium adolescentis, and Bifidobacterium longum were higher in the DR group (Supplementary Table 1). Thus, the similarity in the number of core taxa indicates that disease is associated with the abundance and functional balance of shared taxa rather than their presence or absence. These alterations in microbial taxa abundance are consistent with the expected progression from metabolic dysfunction to microvascular complications.

This analysis provides valuable insights into gut microbial patterns in DR and lays the groundwork for future large-scale microbiome-based risk stratification models. Our previous work was limited to 83 individuals from a single site in South India, using metataxonomy with 16 S rRNA-based amplicon sequencing [14]. By employing a computational approach to predict the functional composition using marker genes and reference data, we reported a reduction of anti-inflammatory and probiotic bacteria in DM and DR compared to HCs [14]. Khan et al. evaluated fecal swabs from 54subjects from another south Indian territory and, using similar meta-taxonomy sequencing, identified differences between people with different grades of sight-threatening DR (STDR) [22]. The authors also identified potential biomarkers for the gut microbiome in STDR [23]. Independent laboratories from China have also reported the biomarker-related significance of the gut microbiome in DR [24–26].

To our knowledge, DRMS-India is the first regional effort at large-scale metagenomic analysis of the gut microbiome to identify enriched gene families and metabolic pathways in DR. Findings from the first 100 participants warrant comparative interpretation of larger datasets collected from different regions of India. We hope that the completion of the DRMS-India will help us develop a clinic-friendly index/biomarker, the ‘gut microbiome Index’ (GMI), to understand diabetic people’s retinal health. GMI could differentiate between healthy individuals and those with DM/DR in Indians. Given the proximity of lifestyle, it could be extrapolated to people in South Asia. DRMS-India is not designed to develop a preventive therapy for DR. However, by obtaining true species, gene families, and metabolic pathways-related data through shotgun sequencing of fecal samples, DRMS-India may guide us to develop the same for delaying the occurrence of DR [27]. Gut microbiome can be altered by use of medication, and the design of the DRMS-India excluded people who used antibiotics, alternative forms of medications, pre/probiotics and allopathic medication other than those for DM and hypertension. The presented data reflect only initial trends, and medication use allowed under the study criteria of DRMS-India shall be adjusted for in the complete analysis.

The gut microbial changes identified in the DM and DR patients will serve as early biomarkers for understanding disease progression. However, establishing functional links between the microbiome alterations and disease progression is crucial for effective translation. The metabolically active species in the gut microbiome of healthy individuals, DM and DR patients will produce the secondary metabolites that can be identified with metabolomics profiles through the gut microbiome-metabolome axis, triggering inflammation in DM and DR patients. These metabolites can potentially modulate the host epigenome which in turn regulates the host transcriptome. The differential transcription of host genes may directly impact on the proteome that has been favored as clinical biomarkers for understanding the pathophysiology of different diseases [28]. Thus, future integration of gut metagenomic data, especially the enriched metabolic pathways from completed DRMS, along with the metabolomics, epigenomics, transcriptomics and proteomics could pave the way for enduring preventive strategies against DR.

Conclusions

The DRMS-India is evaluating the gut microbiome and related enriched metabolism across India. Trends from the first 100 participants suggest a biomarker-related utility of the gut microbiome in discriminating DR. While the current data is not sufficient evidence, the final comprehensive analysis may substantiate these findings and provide further evidence of the relationship between gut metabolism and DR. DRMS-India could lay the groundwork for a public-level preventive therapy and identify the individuals most likely to benefit from it.

Abbreviations

DRMS

Diabetic retinopathy microbiome study

DR

Diabetic retinopathy

DM

Diabetes mellitus

HC

Healthy controls

STDR

Sight-threatening diabetic retinopathy

BMI

Body mass index

PI

Principal investigator

PCR

Polymerase chain reaction

rRNA

Ribosomal ribonucleic acid (Ribosomal RNA)

MetaPhlAn4

Metagenomic phylogenetic analysis version 4

HUMAnN3

HMP unified metabolic analysis network version 3

MaAsLin

Microbiome multivariable association with linear models

LEfSe

Linear discriminant analysis effect size

NMDS

Non-metric Multidimensional Scaling

GO

Gene ontology

BP

Biological process

MF

Molecular function

CC

Cellular component

GMI

Gut microbiome Index

Author contributions

TD, PM, SM and BT participated in designing the study question, sample size calculation and methodology. TD procured the funding. BT and the DRMS group collected the clinical data, while AM, PA, VR and SM processed the samples and analysed the sequencing data. BT and AM wrote the initial drafts later revised by SM and TP. All authors, including the DRMS group, performed revisions to the final version of the manuscript.

Funding

Hyderabad Eye Research Foundation.

Data availability

The data and materials used and/or analysed during the current study are accessible upon reasonable request to the corresponding author.

Declarations

Ethics approval and consent to participate

The DRMS-India is registered with the Clinical Trial Registry of India (CTRI/2024/02/062511). Each participating clinical and laboratory centre obtained ethics committee approval from its respective institute. These were collated at the central driving institute (Central Clinical Coordination Center) for the final ethics committee clearance (LEC-BHR-P-07-23-1064). The study protocol adhered to the Declaration of Helsinki.

Consent for publication

Consent for publication is obtained from all authors.

Competing interests

The authors declare no competing interests.