Exploring the gut microbiome in children and young adults with myotonic dystrophy type 1

Abstract

Myotonic Dystrophy type 1 (DM1) is a complex multisystem disorder, and affected children commonly present with gastrointestinal and neurodevelopmental challenges. Despite the high prevalence of these symptoms, their underlying mechanisms remain poorly understood. The gut microbiome is known to be involved in gastrointestinal and neurodevelopmental disorders, and emerging research suggests that it could also play a contributory role in various neuromuscular diseases. However, the gut microbiota of children with DM1 remains unexplored. This pilot study compared gut microbiota profiles of 10 children and one young adult with DM1, to that of their unaffected siblings (n=11). Three-day food records and faecal samples were collected, and bacterial taxa were analysed through 16S rRNA marker-gene sequencing. A within-family paired-sample design accounted for potential confounders, such as diet and lifestyle. Results revealed that overall microbial diversity and community structure were similar between groups. Moreover, participants with DM1 showed a significant lower relative abundance of the bacterial genus Faecalibacterium (mean 4.9% ± 2.1%) compared to their unaffected siblings (mean 9.1% ± 4.4%)(p=0.005). This finding should be considered exploratory, as its univariate analysis was uncorrected for multiple testing. Faecalibacterium is a key gut bacterium, and reduced abundance has been reported in several DM1-associated gastrointestinal and neurodevelopmental disorders. Although causality cannot be inferred from this exploratory study the observed reduction may be relevant to gastrointestinal and neurodevelopmental challenges frequently seen in DM1. If larger further studies confirm its role, targeting Faecalibacterium relative abundance could become a potential therapeutic strategy for alleviating gastrointestinal and neurodevelopmental symptoms in DM1.

1. Introduction

Myotonic Dystrophy type 1 (DM1) is an autosomal dominant inherited disease caused by an expansion of CTG-trinucleotide repeats within the untranslated region of the DMPK (Dystrophia Myotonica Protein Kinase) gene on chromosome 19q13.3.^1,2 DM1 is classified into the congenital, childhood, adult, and late-onset forms, distinguished by the age of symptom onset and correlated with the CTG repeat expansion length.^3,4

DM1 leads to a wide variety of symptoms and affects multiple body systems as the cardiac, pulmonary, ocular, gastrointestinal, endocrine, immune, reproductive, and central nervous systems,^3,5 making DM1 one of the most phenotypically diverse diseases. Clinical presentations differ greatly between patients and are often distinct between adults and children. In adults, primary symptoms include muscle weakness, myotonia, and fatigue, whereas children typically do not show neuromuscular symptoms at first. Instead, they commonly present with cognitive and behavioural problems, fatigue, and gastrointestinal (GI) disturbances, which are reported to have the highest impact on their daily lives.^6–9 The most commonly observed GI disturbances include dysphagia, bloating, constipation, diarrhoea, gastroesophageal reflux, encopresis, and abdominal pain.^9,10 Notably, high variability in GI symptoms and other clinical manifestations is observed among children with DM1 despite comparable CTG repeat lengths, suggesting that factors beyond genetics may play a role in disease modification.

One potential factor may be the gut microbiome, which is increasingly recognized as an important contributor to both, health and disease. ¹¹ The gut microbiome consists of a diverse community of microorganisms, including bacteria, viruses and eukaryotes, and plays essential roles in nutrient absorption, energy metabolism, neurological signalling, gut endocrine function, and immune function.^12,13 Alterations in gut microbial composition have been associated with several symptoms of DM1 and disorders that commonly present with DM1, including muscle atrophy,^14,15 autism spectrum disorder (ASD),^16,17 diabetes mellitus, ¹⁸ cardiac arrhythmia, ¹⁹ attention deficit hyperactivity disorder (ADHD),^20,21 and GI disturbances.^22–25 In addition, alterations in gut microbial composition have also been reported in irritable bowel syndrome (IBS).^26–28 Although DM1 should not be considered an established risk condition for IBS, the overlap between IBS symptoms and the GI phenotype of DM1 raises the possibility that microbiome-related mechanisms may contribute to the GI manifestations observed in DM1.

Furthermore, beyond gastrointestinal health, recent research on the gut-muscle-axis suggests an interdependence between muscle function, metabolism, and gut microbiome composition.^29,30 This understanding positions the gut microbiome as a potential biological marker or therapeutic target in neuromuscular and central nervous system diseases.^31–33 Consistently, gut microbiome alterations have been reported in several neuromuscular diseases, including myasthenia gravis ³⁴ and Duchenne muscular dystrophy.^29,35,36 Furthermore, Mahdavi et al. reported structural changes in the gut microbiome of adult DM1 patients and suggested a potential association between the gut microbiome and the gastrointestinal symptoms observed in adult DM1 participants. ³⁷

Despite accumulating evidence linking the gut microbiome to gastrointestinal, central nervous system and neuromuscular health, its potential role in paediatric DM1 remains unexplored. Therefore, we aimed to explore differences in gut microbiota diversity and composition in children and young adults with DM1 and provide insights into the gut microbiome’s role in paediatric DM1. To address this, we conducted a case-control study using 16S ribosomal RNA (rRNA) marker-gene sequencing-based metataxonomic analysis ³⁸ to characterize and compare the faecal microbiota composition, as a proxy of the gut, of participants with DM1 and their unaffected siblings.

2. Methods

2.1. Participants and data collection

This case-control study included children and young adults with DM1 who received treatment and follow-up at Amalia Children’s Hospital, Radboud University Medical Center (Nijmegen, the Netherlands). Unaffected siblings living in the same household were recruited as matched controls. The study protocol was reviewed and approved by the by the institutional ethical review board committee of the Radboud university medical center, Nijmegen. Informed consent was obtained from all participants or their legal guardians after providing oral and written information. The informed consent was signed by the parents, the participant, or both, depending on the participant’s age, in accordance with Dutch legal requirements.

Participants were eligible for inclusion if they were between 2.5 and 21 years of age. Additionally, both, participants with DM1 and their unaffected siblings were genetically tested on the presence of DM1. One unaffected sibling aged 23 years was additionally included to preserve matched sibling pairs. Participants were excluded if they had used antibiotics or probiotics within 3 months prior to the study or if they had made major changes to their medication use or diet within the past month. Additional exclusion criteria included major chronic bowel, metabolic, inflammatory or respiratory diseases, diabetes mellitus, and malignancies. Participants who had been breastfed in the past year were also excluded.

Each family was visited at home to explain the study procedures including, a 3-day dietary record and faecal sample collection. During the home visit, demographic and clinical characteristic were recorded, including age, gender, weight, height, age of symptom onset, and medication usage. Additionally, questionnaires were employed to assess various aspects of general health, including comorbidities, gastrointestinal symptoms, birth mode, allergic conditions, dietary patterns, and physical activity. Gastrointestinal symptoms were assessed using questions based on the Gastrointestinal Symptom Rating Scale ³⁹ and GI symptom frequency was categorized as never (<1 time per month), sometimes (1-4 times per month), often (2-4 times per week), or always (≥5 times per week).

2.2. Faecal sample collection, DNA extraction, and sequencing

Participants were provided with a sterilized container and detailed oral instructions on faecal sample collection. In addition to the sample collection, participants were asked to complete the Bristol Stool Form Scale (Types 1–7) based on the collected sample. ⁴⁰ Stool types 1-2 were classified as hard stool, types 3-4 as normal stool, and types 5-7 as soft stool. After collection, faecal samples were immediately returned by post to the hospital laboratory. Upon arrival, the faecal material was stored at -80°C until subsequent processing. DNA isolation, quality control, and sequencing of microbial DNA were conducted by BaseClear B.V. (Leiden, the Netherlands). For bacterial DNA extraction, 150 µl or mg of faeces was aliquoted from the Eppendorf tube. Microbial DNA was isolated including a bead-beating procedure and purified. The V4 region of 16S ribosomal RNA (rRNA) gene was amplified in duplicate PCR reactions for each sample. The V4 region was targeted using primers 515F (5′-GTGYCAGCMGCCGCGGTAA-3′) and 806R (5′-GGACTACNVGGGTWTCTAAT-3′). ⁴¹ Subsequently, Illumina NovaSeq 6000 was used to sequence the pooled libraries with paired-end 250 base pairs (bp) reads. FASTQ read sequence files were generated using bcl2fastq version 2.20 (Illumina, USA). The initial quality assessment was based on data passing the Illumina Chastity filtering. Subsequently, reads containing PhiX control signal were removed using an inhouse filtering protocol. In addition, reads containing (partial) adapters were clipped (up to a minimum read length of 50 bp). The second quality assessment was based on the remaining reads using the FASTQC quality control tool version 0.11.8 (Bioinformatics, Babraham Institute, Cambridge, UK). Subsequent data analyses were conducted using QIIME2 (version 2020.2). ⁴² The sequenced 16S rRNA marker genes were denoised into amplicon sequence variants (ASV) using the DADA2 ⁴³ denoising package within QIIME2. Default QIIME2 settings were applied, which include truncating reads at the first instance of a Phred quality score of 2 or lower and applying a maximum expected errors threshold of 3 (instead of the default 2). No additional fixed trimming or truncation of bases was performed. Chimera filtering was carried out within the DADA2 workflow using the consensus method, with a minimum fold-abundance difference set to 2 between a potential chimera and its parents for the sequence to be flagged as chimeric. ASVs represents high-resolution taxonomic units corresponding to the microbial entities present in the sample. Classification of representative ASV sequences was accomplished by “qiime2 feature-classifier classify-consensus-vsearch” (default settings) via VSEARCH alignment against the SILVA-138 database (Ref NR 99; i.e. non-redundant 99% identity) as a 16S reference database for taxonomic classification.^42,44

2.3. Dietary record, PDHD-index and score allocation

Each participant completed a three-day food record, which included two weekdays and one weekend day, to capture habitual dietary intake. To evaluate the dietary reports, the Paediatric Dutch Healthy Diet Index (PDHD-index) was developed, adapted from the Dutch Healthy Diet Index (DHD-15). ⁴⁵ Adjustments for the paediatric population were done in collaboration with dietitians of the Radboudumc and aimed to represent adherence to the Dutch Guidelines for a Healthy Diet ⁴⁶ . The PDHD-index consists of five components, forming a continuous score. These components include the consumption of vegetables, fruits, fibres, saturated fatty acids (SFA), and products containing easily fermentable sugars (ADF). Each component can receive a maximum score of ten points, yielding a total PDHD-index score ranging from zero to 50 points. Age-specific cut-off and threshold values were applied to ensure appropriate scoring across different age groups (1-3, 4-8, 9-13, and 14-18 years). For participants older than 18 years, the cut-off and threshold values applied were identical to those for the 14-18 years age group. Cut-off values indicate the required amount of intake for the adequacy components, whereas a threshold value indicate the intake level that is considered zero points for the moderation components. The dietary components and their corresponding cut-off and threshold values are presented in Supplementary Table 1.

The first component assesses vegetable intake and is based on the recommendations of the Netherlands Nutrition Centre, the higher of the two recommendations was selected as cut-off value (Supplementary Table 1). ⁴⁶ The second component, fruit consumption, is also based on the guidelines of the Netherlands Nutrition Centre of 150 or 200 grams of fruit per day. ⁴⁶ The third component assesses the fibre intake per day and was calculated from the recommendation of the grams fibre per megajoule per day. ⁴⁷ Moreover, the fourth component scored the consumption of saturated fatty acids based on the recommendation that children under the 4 years of age and those over 4 years should consume less than 15 and 10 energy percent of SFA, respectively. ⁴⁸ The upper limits were derived from those applied in the DHD-15-index. ⁴⁵ The number of ADF consumptions is assessed as the fifth component. Including the three main meals, it is recommended to not consume more than 7 ADF consumptions per day. To reflect the dietary guidelines that advise limiting foods not included in ‘The Wheel of Five’, ⁴⁶ a cut-off value of two ADF consumptions was established.

All scores were derived from the average of the three-day dietary record. For adequacy components, a minimum score of zero points was assigned in cases of non-consumption. Scores for intake are within the range of zero and the cut-off value were calculated using equation (1):

$$\text{Score}\hspace{0.17em}\text{for}\hspace{0.17em}\text{adequacy}\hspace{0.17em}\text{components}=\frac{\text{Intake}}{\text{Cut}‐\text{off}\hspace{0.17em}\text{value}}\times 10$$ (1)

For moderation components, a score of zero points was allotted when intake was above the threshold value. Conversely, a maximum score of ten points was assigned when the intake was below the cut-off value. Scores for intake between the threshold and cut-off value were determined using equation (2):

$$\text{Score}\hspace{0.17em}\text{for}\hspace{0.17em}\text{moderation}\hspace{0.17em}\text{components}=\frac{\left(\text{Cut}‐\text{off}\hspace{0.17em}\text{value}-\text{Intake}\right)}{\left(\text{Cut}‐\text{off}\hspace{0.17em}\text{value}-\text{Threshold}\hspace{0.17em}\text{value}\right)}\times 10$$ (2)

2.4. Bioinformatics and statistical analysis

Python version 3.10, ⁴⁹ R version 4.3.3, ⁵⁰ and QIIME2 version 2020.2 ⁴² were used for the data analysis of the sequenced reads. A customized Python workflow based on QIIME2 was primarily adopted for the gene sequencing analysis. Relative abundances of taxa at phylum, class, order, family, genus, and species level were calculated. Values were not corrected for zero-inflation. The hierarchical clustering of samples was performed using a weighted UniFrac distance measure as implemented in QIIME2 to compute distances between samples. ⁵¹ Alpha diversity was measured in order to examine the diversity of species within the gut microbiota, considering the number of species and their relative abundance (Shannon index ⁵² ) and the evolutionary relationships among taxa (Faith index ⁵³ ). The ASV table within QIIME2 was used to determine the alpha diversity indices, which were visualized using boxplots. Next, an analysis of microbial diversity across different sample groups, β-diversity, was evaluated by performing a Principal Coordinate Analysis (PCoA). Moreover, Principal Component Analysis (PCA) and Redundancy Analysis (RDA) were conducted using the R package vegan version 2.6. ⁵⁴ RDA is tested for significance with the ANOVA like permutation test with 999 permutations. ⁵⁵

Subsequently, to compare the composition of the gut microbiome of participants with DM1 to that of their unaffected siblings, a paired-analysis was conducted. Each DM1 participant was paired with their unaffected sibling. One family consisted of two children with DM1 and two unaffected siblings. Within this family, sibling pairs were matched by age to minimize the age gap within each pair. Two of the included families consisted of an unequal number of participants with DM1 and unaffected siblings. One family included two participants with DM1 and one unaffected sibling, while another family included two unaffected siblings and one participant with DM1. As a result, two participants were paired twice. This approach was preferred over excluding subjects from the analysis which already contained a relatively small number of participants. The Wilcoxon signed-rank test was employed as statistical evaluation of the paired microbial analysis. The output of this paired-analysis was screened for significance, fold-change, and relative abundance of bacterial genera (≥ 1% on average over all samples). Given the exploratory nature of this study and the small sample size relative to the number of taxa tested, significance was not adjusted for multiple testing correction. Therefore, any observed differences are considered suggestive. All statistical analyses were performed using R version 4.3.3 or SPSS 29.0 software. Data normality was assessed using the Shapiro-Wilk test and by visually inspecting Q-Q plots. Descriptive statistics are presented as mean ± standard deviation for continuous variables and percentages for categorical variables. Baseline characteristics were compared between participants with DM1 and their unaffected siblings using paired analyses. Continuous variables were analysed using the non-parametric Wilcoxon signed-rank test or the parametric paired t-tests. Binary categorical variables were compared using McNemar’s test, whereas nominal and ordinal variables were analysed using the Wilcoxon signed-rank test. Statistical significance is defined by a p-value of ≤ 0.05. Graphs were generated using GraphPad Prism (version 10.1.2).

3. Results

3.1. General characteristics of the study population

The study included a total of 22 participants, consisting of 11 participants with DM1 (10 children and 1 young adult) and 11 unaffected siblings serving as controls. The DM1 cohort consisted of six females and five males, with a mean age of 11.0 ± 5.2 years, average height of 141.4 ± 27.5 cm, and mean weight of 41.8 ± 22.5 kg. Of these participants, eight presented with the childhood-onset form and three presented with the congenital-onset form of DM1. The unaffected sibling group included seven females and four males, with a mean age of 10.1 ± 6.1 years, average height of 135.8 ± 28.6 cm, and mean weight of 35.3 ± 19.2 kg. Detailed questioning of the study population revealed no significant differences in age, height, weight, BMI, medication use, presence of comorbidities and birth mode between the participants with DM1 and their unaffected siblings. However, significant within-pair differences were observed for gestational age, stool form, and frequency of abdominal pain. Participants with DM1 were born at a lower gestational age and reported abdominal pain more frequently than their unaffected siblings. Comprehensive demographic data for both groups are summarised in Table 1.

Table 1.

Population demographics of children and a young adult with DM1 and their unaffected siblings.

Parameter		Participants with DM1 (n = 11)	Unaffected siblings (n = 11)	p-value
Female, n (%)		6 (54.5%)	7 (63.6%)	1.000 ^
Classification DM1, n (%)	Congenital	3 (27.3%)	NA	NA
	Childhood	8 (72.7%)
Age (years)		11.0 ± 5.2 (2.6 – 21.2)	10.1 ± 6.1 (2.7 – 23.9)	0.879 +
Height (cm)		141.4 ± 27.5 (94 – 183)	135.8 ± 29 (94 – 178)	0.566 +
Weight (kg)		41.8 ± 22.5 (15.2 – 84.4)	35.3 ± 19.2 (15.1 – 75.5)	0.263 +
BMI (kg/m2)		19.5 ± 5.5 (14.7 – 32.6)	17.7 ± 2.7 (15.0 – 23.8)	0.367 +
Stool form, n (%)	Soft stool	5 (45.5%)	2 (18.2%)	0.048 ∼
	Normal stool	6 (54.5%)	6 (54.5%)
	Hard stool	0 (0%)	3 (27.3%)
Preterm, n (%)		2 (18.2%)	0 (0%)	0.500 ^
Birth delivery, n (%)	Vaginal	6 (54.5%)	6 (54.5%)	1.000 ^
	C-section	5 (45.5%)	5 (45.5%)
Gestational age (weeks)		37.4 ± 4.5 (27.7 – 41.7)	40.0 ± 1.8 (37.0 – 42.1)	0.016 ∼
Presence of comorbidity, n (%)		4 (36.4%)	4 (36.4%)	1.000 ^
Comorbidity type, n (%)	ASS	4 (36.4%)	1 (9.1%)	NA
	ADHD	2 (18.2%)	2 (18.2%)
	Chronic bronchitis	0 (0%)	2 (18.2%)
	Sleep apnoea	1 (9.1%)	0 (0%)
	VACTERL syndrome	1 (9.1%)	0 (0%)
	Solitary kidney	1 (9.1%)	0 (0%)
Medication usage, n (%)		5 (45.5%)	2 (18.2%)	0.218 ^
Medication type, n (%)	Methylfenidaat	2 (18.2%)	2 (18.2%)	NA
	Macrogol	4 (36.4%)	1 (9.1%)
	Domperidon	1 (9.1%)	0 (0%)
	Levocetrizine	1 (9.1%)	0 (0%)
	Mebeverine	1 (9.1%)	0 (0%)
	Melatonine	1 (9.1%)	0 (0%)
	Omeprazol	1 (9.1%)	0 (0%)
	Lisonopril	1 (9.1%)	0 (0%)
	Desmopressine	0 (0%)	1 (9.1%)
	Risperidon	0 (0%)	1 (9.1%)
	Sertraline	0 (0%)	1 (9.1%)
Reported frequency of abdominal pain, n (%)	Never	3 (27.3%)	9 (81.8%)	0.013 ∼
	Sometimes	2 (18.2%)	1 (9.1%)
	Often	5 (45.5%)	1 (9.1%)
	Always	1 (9.1%)	0 (0%)

Data are presented as mean ± standard deviation (SD), median with interquartile range (IQR) or n (%). Abbreviations: DM1 = Myotonic Dystrophy type 1; BMI = Body Mass Index; C-section = Caesarean section.

^{^}McNemar test.

⁺Paired t-test.

^∼Wilcoxon signed-rank.

3.2. Dietary records and Paediatric Dutch Healthy Diet (PDHD) index

None of the participants reported any dietary restrictions, including vegetarian, vegan, dairy-free, or other special diet. The dietary intake of each participant was assessed using the PDHD-index, derived from the three-day dietary record completed by each participant. There were no significant differences in the PDHD-index scores between weekdays and weekend days. Minimal intra-family variation in PDHD-index was observed, while differences between families were more pronounced, as illustrated in Figure 1. The overall mean PDHD-index score for the study population was 30.4 ± 9.3, with no significant differences between the DM1 group (30.6 ± 9.0) and unaffected sibling group (30.3 ± 10.5). Moreover, no individual PDHD component differed significantly between the two study groups (Supplementary Table 2). Given the absence of statistical differences in diet, no dietary adjustments were required for the microbial analysis.

Figure 1.

Paediatric Dutch Healthy Diet (PDHD) index scores for each participant are displayed based on the mean intake from the three-day dietary record and categorized by family. Abbreviations: DM1 = Myotonic Dystrophy type 1.

3.3. Sequencing process

Microbial gDNA was extracted from the 22 faecal samples, and the V4 region of the 16S rRNA gene was sequenced, resulting in the identification of a total of 18,254,601 raw sequence reads. After quality filtering, 67.0% of the reads remained, with a mean ± SD read length of 247.9 ± 0.60 bp. On average, each sample retained 552,500 ± 76,660 reads for taxonomic classification, with 95.11% ± 1.66% of the reads classified down to the genus level.

3.4. Composition of the gut microbiome at genus level

The gut microbiome composition at the genus levels of all participants is displayed in Figure 2, where the samples are clustered based on UPGMA of weighted UniFrac values (pairwise distance matrix).

Figure 2.

Gut microbiota composition of all participants at genus level, clustering is based on UPGMA of weighted UniFrac values (pairwise distance matrix). Each bar represents a participant, with the first column’s colours indicating the family, the second column showing the relative abundance of all bacterial genera, and the last column distinguishing between participants with DM1 (orange) and healthy siblings (blue). Abbreviations: fam = family, DM1 = Myotonic Dystrophy type 1.

Bifidobacterium and Bacteroides were the most relatively abundant genera in both groups, with mean relative abundances of 18.84% and 11.32% in the DM1 cohort, and 17.37% and 10.55% in unaffected siblings, respectively (Figure 3). Among participants with DM1, the third most abundant genus was Prevotella (9.84%) followed by Faecalibacterium (4.95%), Subdoligranulum (4.06%), Blautia (3.31%), Akkermansia (2.40%) and Ruminococcus (2.67%). The third to eighth most relatively abundant genera in the unaffected siblings were Faecalibacterium (9.08%), Prevotella (8.84%), Akkermansia (4.12%), Blautia (3.49%), and Subdoligranulum (2.76%). Notably, the composition of the top eight most relatively abundant bacterial genera differed between the DM1 group and the unaffected sibling group. The relative abundances of these genera in both groups are presented in Figure 3. Between-group comparison (Wilcoxon signed rank test) revealed a statistically significant lower relative abundance of the genus Faecalibacterium in participants with DM1 compared to the unaffected siblings (p = 0.022; Figure 3). Moreover, the bacterial genera Bifidobacterium and Prevotella exhibited substantial variation across the samples, yet the extent of this variation was comparable for both study groups.

Figure 3.

Box plots of the relative abundances of the most abundant bacterial genera in each group. Genera are ordered from highest to lowest based on their relative abundance across the total study population. Boxes represent the interquartile range (IQR), and medians are indicated by horizontal lines. The whiskers extend to values within 1.5 IQR and data points beyond this range are plotted individually as potential outliers. Abbreviations: DM1 = Myotonic Dystrophy type 1.

3.5. Significant differences in relative abundance of Faecalibacterium

Paired-analysis between each participant with DM1 and their respective unaffected sibling revealed a statistically significant decrease (p = 0.0048) in the relative abundance of the bacterial genus Faecalibacterium in children and a young adult with DM1 (non-parametric Wilcoxon signed-rank test; Figure 4). The mean relative abundance was 4.95% ± 2.14% among participants with DM1 compared to 9.08% ± 4.45% in their unaffected siblings, with a mean Log2 fold change of -0.87 ± 0.86. This genus is ranked as the third most abundant bacteria in the unaffected sibling group, whereas it was the fourth most abundant bacteria in the DM1 group (Figure 3). Information about the age differences within sibling pairs is provided in the Supplementary Table 3.

Figure 4.

Relative abundance of the bacterial genus Faecalibacterium was significantly decreased in participants with DM1 compared to their unaffected siblings as determined by paired-analysis (non-parametric Wilcoxon signed-rank test; p = 0.0048). Abbreviations: DM1 = Myotonic Dystrophy type 1; Fam = Family.

3.6. Gut microbial diversity and community structure

Alpha-diversity indices, Shannon and Faith, were used to evaluate species richness and phylogenetic diversity within each group. The Shannon and Faith indices indicated that the gut microbiota diversity of participants with DM1 was not statistically different from that of their unaffected siblings (Mann-Whitney U test; Figure 5).

Figure 5.

Comparison of gut microbiota diversity of the participants with DM1 and unaffected siblings group using alpha-diversity indices Shannon (left) and Faith (right). The Mann-Whitney U test was employed for statistical evaluation. Data are presented as median (5-95 percentile). Abbreviations: DM1 = Myotonic Dystrophy type 1.

Differences in microbial community structure were explored by performing beta-diversity analysis with using Principal Coordinates Analysis (PCoA) based on weighted UniFrac distances. PCoA did not reveal distinct clustering between the DM1 and unaffected sibling group (Supplementary Figure 1). PERMANOVA analysis of the weighted UniFrac distances confirmed no statistically significant differences in microbial community structure between the two groups (p > 0.05). Subsequent Principal Component Analysis (PCA) performed at the genus level indicated that PC1 and PC2 collectively explained 73.9% of the variance in the dataset (Figure 6A). As illustrated in Figure 6B and C, the bacterial genera contributing most to PC1 and PC2 were Prevotella and Bifidobacterium, respectively. Consistent with the PCoA results, PC1 and PC2 did not separate the participants with DM1 and unaffected sibling group.

Figure 6.

Principal component analysis (PCA) of bacterial abundance. (A) Principal component analysis (PCA) performed by using the relative abundance of bacterial genera. The six bacterial genera explaining most of the variance in (B) PC1 and (C) PC2. Abbreviations: DM1 = Myotonic Dystrophy Type 1, PC1 = first principal component, PC2 = second principal component.

Subsequently, Redundancy Analysis (RDA) was performed to quantify the factors which explain the most variance at the genus level of microbiota composition (Figure 7). Family was identified as the main variance explaining factor, accounting for 33.7% of the variance. Although the factors: family, BMI, age, abdominal pain, and stool scale accounted for a substantial proportion of the observed variance, none of the factors achieved statistical significance based on RDA permutation testing as implemented in the Vegan R-package.

Figure 7.

Explained variance between the faecal microbial samples calculated using Redundancy Analysis (RDA) for each separate baseline variables. BMI = Body Mass Index, PDHD = Paediatric Dutch Healthy Diet.

4. Discussion

To the best of our knowledge, this study is the first to examine gut microbiota composition in children and young adults with DM1. Previous research has associated gut microbiome alterations to a variety of DM1-symptoms and disorders that commonly present in DM1, including GI disturbances,^22,24,25 ADHD, ²⁰ diabetes mellitus, ¹⁸ muscle atrophy, ¹⁴ and ASD. ¹⁷ This exploratory study compared the gut microbiome compositions of children and a young adult with DM1 and those of their unaffected siblings. Based on the microbiological analysis of 22 samples, participants with DM1 had a significant lower faecal abundance of the bacterial genus Faecalibacterium compared to their unaffected siblings. However, we observed no differences in overall bacterial diversity and community structure.

4.1. Significant decrease of Faecalibacterium in participants with DM1

Alterations in microbial composition can affect a number of physiological functions. In this study, a paired-analysis of the microbial composition between participants with DM1 and their unaffected siblings revealed significant differences in relative abundance of taxa at family, genus, and species level. We report a significant decrease in the relative abundance of the bacterial genus Faecalibacterium in participants with DM1, with a Log2 fold-change of -0.87 (Figure 4). This indicates that the relative abundance of Faecalibacterium in children and a young adult with DM1 was reduced to approximately 54% of the levels observed in their unaffected siblings. Interestingly, Faecalibacterium was detected in the microbiome of all participants in the study. However, this bacterial genus was not reported in a previous gut microbiome study in adult DM1 patients from Mahdavi et al. ³⁷ Nevertheless, the decrease in Faecalibacterium was also found in a non-paired statistical analysis between the primary study groups (p=0.0215; Figure 3).

Faecalibacterium, belonging to the phylum Firmicutes, is one of the most abundant commensal bacteria in the human gut microbiota, as shown in Figure 3 and reported by previous studies.^56,57 In the intestine, Faecalibacterium is a major producer of the short-chain fatty acid butyrate, which possesses anti-inflammatory properties. ⁵⁷ Notably, reduced abundance of Faecalibacterium has been reported in IBS.^26–28 IBS is characterized by abdominal pain, constipation, and diarrhea, which overlap with the gastrointestinal symptoms frequently observed in DM1. Although DM1 is not considered an established risk condition for IBS, the overlap in symptomatology may point towards shared microbiome-related mechanisms. The reduction in Faecalibacterium observed in the present study therefore aligns with previous studies linking this bacterial genus to gastrointestinal dysfunction and gut microbiome alterations.^26–28 Furthermore, ASD and ADHD, disorders that are frequently observed in DM1, have also been associated with alterations in Faecalibacterium abundance. However, microbiome studies in children with ASD have shown inconsistent findings, with some reporting increased levels^58,59 and others reporting decreased levels^60,61 of Faecalibacterium. Moreover, microbiome studies in ADHD describe reduced Faecalibacterium abundance.^62–64 Together, the reduced abundance of Faecalibacterium observed in children and a young adult with DM1 may reflect alterations in gut microbial composition that are also observed in disorders and symptoms that are commonly observed in DM1, highlighting the potential significance of this bacterial genus.

4.2. No changes in gut microbial diversity and community structure

The comparison of microbial diversity and community structure between participants with DM1 and unaffected siblings was conducted using alpha diversity (Shannon and Faith indices) and beta diversity (PCoA), respectively. The Shannon and Faith indices, both commonly utilized for assessing alpha diversity, ⁶⁵ revealed no significant differences between the two groups, suggesting that overall microbial diversity was not different (Figure 5). This aligns with findings from a study by Mahdavi et al., that examined the gut microbiota in adults with DM1 using paired analysis and similarly observed no significant differences in alpha diversity between DM1 patients and controls. ³⁷ Notably, although Mahdavi et al. employed a paired study design, their control group consisted of family members living within the same house, which may have included both related and unrelated individuals. Moreover, in our study, beta diversity analysis revealed no distinct clustering, suggesting no differences in microbial community structure between participants with DM1 and their unaffected siblings. However, this result is in contrast with Mahdavi’s findings of differences in beta diversity among possibly unrelated adult DM1 cases and controls, but aligns with studies that utilized sibling comparisons to assess the impact of disease on the microbiome in children.^66,67 These studies did not report significant differences in alpha or beta diversity between the groups. The lack of distinct clustering in the PCoA (Supplementary Figure 1) can be explained by the greater similarity of microbial communities within families, overshadowing the differences attributed to disease status. This was further supported by the Redundancy Analysis (RDA), which showed that family-related factor accounted for most of the variance observed between samples, while disease status explained the least variance (Figure 7). These findings underscore the importance of accounting for familial influences in microbiome studies.

4.3. Gut microbiome composition at genus level

Compositional analysis of the bacterial genera in the whole dataset showed that the predominant gut microbiota genera identified were Bifidobacterium, Bacteriodes, Prevotella, Faecalibacterium, and Subdoligranulum, consistent with previous research on microbiome composition in children.^62,68,69 Notably, the relative abundance of Bifidobacterium, Prevotella, and Bacteriodes exhibited substantial variation compared to other dominant bacteria (Figure 3), as supported by the PCA. The PCA showed that Prevotella and Bacteroides contributed most to the variance in PC1, while Bifidobacterium was the main contributor to PC2. In the study of Mahdavi, the most prevalent genera in the adult DM1 participants were Blautia (24%) and Bacteroides (14%), both of which were also prominent genera in our DM1 cohort. Bacteroides was the second most abundant genus (11.3%), but Blautia (3.31%) was notably less abundant among our DM1 participants, potentially reflecting differences in age, environment, diet, or geography. Additionally, a higher relative abundance of Prevotella was observed in the DM1 group, consistent with Mahdavi’s findings, though this increase was not statistically significant in our study. Therefore, it can be observed that the microbial composition is highly diverse among the samples, with Prevotella, Bacteroides, and Bifidobacterium accounting for the majority of the observed variance.

4.4. Strengths and limitations

The paired-sibling study design is a major strength of the current study, effectively accounting for various confounding factors that could influence gut microbiota. It ensured that genetic, housing, ethnic, environmental, and dietary factors were comparable between groups. Consequently, it is likely that the results of this analysis are attributable to the differentiating factor of disease. Additionally, detailed questioning of the study population ensured no differences in age, BMI, birth mode, and medication usage, within the sibling pairs. Moreover, by excluding participants who had used antibiotics, the accuracy of reflecting the disease-associated effect on gut microbial communities between DM1 and unaffected siblings was further enhanced. A further strength of the study is that all participants were genetically tested for DM1. This ensured that the control group consisted of individuals confidently without DM1, and not of pre-symptomatic individuals.

However, despite its strengths, this study has some limitations. First, the small sample size reduced the statistical power, though this is understandable given the rarity of the patient population. With the current sample size, the study was insufficiently powered to support the development of a multivariable model to adjust for potential confounders. Additionally, multiple testing correction was not applied due to the exploratory nature of this study and the small sample size relative to the number of taxa tested. Moreover, the Illumina NovaSeq platform outputs binned quality scores, which may have affected DADA2 default settings. However, given the high genus-level classification rate (∼95%) and deep sequencing coverage (>550K reads/sample), we expect the impact to be minimal. Future studies should employ the adjusted NovaSeq-specific error rates learning model for DADA2 to address this. In addition, two families consisted unequal numbers of DM1 and unaffected participants, resulting in two participants being paired twice. While statistically unconventional, this approach was deemed acceptable given the small sample size. It was preferred over excluding participants and losing valuable data. Finally, it is essential to note that our findings are based on observational data. Therefore, it cannot be established whether the observed microbial alterations are causative, consequential, or both in relation to DM1.

4.5. Recommendations for future research

Future gut microbiota research should consider prioritizing intra-family matching, preferably through the inclusion of sibling controls, to more effectively distinguish disease-associated microbial alterations from those attributable to shared genetic background and environmental exposures. Additionally, future studies should also consider stricter age matching pairs to further minimize potential age-related confounding. Moreover, future research on the gut microbiota in DM1 should focus on larger and more statistically robust studies to identify potentially more subtle microbiome alterations. Increasing the sample size will enhance the ability to establish a causal relationship between DM1 (symptoms) and the gut microbiome. Furthermore, the relatively high use of macrogol among participants with DM1 suggests that constipation may be clinically relevant in this population. Given the reported associations between constipation and gut microbiome composition, future studies should investigate constipation to better understand its potential contribution to the microbial alterations. In addition, combining 16S sequencing with metagenomic, metatranscriptomic, and proteomic approaches can provide deeper insights into microbiome function in DM1. Importantly, if the decrease in Faecalibacterium is confirmed, modulation of its abundance could potentially serve as a strategy to alleviate gastrointestinal and neurodevelopmental DM1 symptoms.

5. Conclusion

Overall, our findings indicate differences in abundance in the gut microbiome of children and a young adult with DM1, particularly marked by a reduction in one of its most abundant bacterial genera, Faecalibacterium. A log2 fold change of -0.87 suggests that Faecalibacterium abundance was approximately 46% lower in participants with DM1 than in their unaffected siblings. However, these findings should be considered exploratory. Previous studies have linked Faecalibacterium to DM1-symptoms and disorders that commonly present in DM1, but whether its shift plays a direct role in DM1 pathogenesis or is a secondary effect remains unclear. The insights presented here could pave the way for new approaches and therapeutic strategies aimed at alleviating the neurodevelopmental and gastrointestinal symptoms associated with DM1.

ORCID iDs

Anne M. van Uden https://orcid.org/0009-0009-3705-7864

Derick G. Wansink https://orcid.org/0000-0002-6773-8662

Alem Gusinac https://orcid.org/0009-0006-1896-4176

Hilde M. H. Braakman https://orcid.org/0000-0001-8053-5918

Data Availability Statement

All scripts, raw data and generated datasets have been deposited fully anonymized in the Radboud Data Repository (https://data.ru.nl/). Access to the data is controlled and available to registered users of the repository. The data and scripts of this study are available on request from the corresponding author.