Pregnant women who received perinatal care at the Third Affiliated Hospital of Zhengzhou University, China between November 2021 and May 2022 were recruited. Thirty pregnant women with hypothyroidism who satisfied the inclusion criteria were included in the hypothyroidism group, whereas thirty normal pregnant women were included in the control group during the same time period. From the hospital’s electronic medical records, we collected demographic details and relevant clinical data of the participants.

The inclusion criteria were as follows: (1) Thyroid function level complied with the diagnostic criteria of the Guideline on Diagnosis and Management of Thyroid Diseases during Pregnancy and Postpartum (2nd edition) (Ad Hoc Writing Committee for Guideline on Diagnosis and Management of Thyroid Diseases during Pregnancy and Postpartum et al., 2019) for hypothyroidism during pregnancy and the diagnostic criteria formulated by the Clinical Laboratory of the Third Affiliated Hospital of Zhengzhou University, China [serum thyroid stimulating hormone (TSH)>4.2 mIU/L] and (2) Gestational age less than 14 weeks.

The exclusion criteria were as follows: (1) age<18 years or ≥35 years; (2) multiple pregnancies; (3) artificial conception; (4) severe stress, anxiety, or depression; (5) endocrine or immune system diseases before and during pregnancy, such as diabetes, gestational hypertension and systemic lupus erythematosus; (6) severe gastrointestinal diseases or gastrointestinal surgery; and (7) received antibiotic, probiotic, or immunosuppressant treatment in the previous 2 months. and (8) applied anti-thyroid drugs or thyroid hormone replacement.

The study was approved by the Medical Ethics Committee of the Third Affiliated Hospital of Zhengzhou University, China (NO: 2021-105-01). All enrolled subjects participated voluntarily and signed an informed consent form.

All fecal samples were collected from all subjects within 24 hours of diagnosis of hypothyroidism during pregnancy. Before collecting the samples, the subjects were informed. After natural defecation, a sterile spoon was used to carefully collect the sample from the middle part of the stool, ensuring that the sample did not contact the bedpan to avoid mixing with urine. After collection, the sample was placed in a 2.0 mL sterile tube, transported to the laboratory within 2 hours, and stored at -80°C until further processing.

Laboratory testing was conducted in the laboratory department of the Third Affiliated Hospital of Zhengzhou University, China. All tests were performed by the manufacturer’s instructions.

The levels of serum thyroid-stimulating hormone (TSH), free thyroxine (FT4), TPOAb, and thyroglobulin antibody (TgAb) were measured using the Cobas e 801 electrochemiluminescence immunoassay analyzer (Roche, USA) and its supporting kit (Roche, Jiangxi, China). The normal reference values were established as follows: TSH (0.27-4.2 mIU/L), FT4 (12-22 pmol/L), TPOAb (0-34 IU/mL), and TgAb (0-114 IU/mL).

Serum fasting blood glucose (GLU), total cholesterol (TC), triglyceride (TG), low-density lipoprotein cholesterol (LDL-C) and high-density lipoprotein cholesterol (HDL-C) levels were measured by an AU5800 automatic biochemistry analyzer (Beckman Coulter, USA) and matching kit (Beckman Coulter, Suzhou, China). The normal reference values were established as follows: GLU (3.9-6.1 mmol/L), TC (0-6.2 mmol/L), TG (0.4-1.86 mmol/L), LDL-C (1.9-3.8 mmol/L), and HDL-C (1.29-1.55 mmol/L).

The levels of C-reactive protein (CRP) and hemoglobin (HGB) in peripheral blood were tested by the BC-5700 (Mindray, China) with a corresponding kit (Mindray, Shenzhen, China). The normal reference values were established as follows: CRP (0-3 mg/L) and HGB (115-150 g/L).

The levels of peripheral blood interleukin (IL)-2, 6, 8, 10 and tumor necrosis factor (TNF)-α were determined by a FACSCanto™ flow cytometer (BD, USA) and a human Th1/Th2 subset detection kit (Cellgene biotech, Jiangxi, China). The normal reference values were established as follows: IL-2 (≤ 11.4 pg/ml), IL-4 (≤ 12.9 pg/ml), IL-6 (≤ 20 pg/ml), IL-10 (≤ 5.9 pg/ml), and TNF-α (≤ 5.5 pg/ml).

16S rRNA amplicon sequencing was used to analyze the fecal samples from all participants. A MagPure Stool DNA KF Kit B (cat. no.MD511, Magen, China) was used to extract total DNA from the microbial community in the stool specimens (0.20 g) (Fadrosh et al., 2014). Polymerase chain reaction (PCR) was used to amplify the highly variable V3-V4 region of the bacterial 16S rRNA gene. The PCR primer sequences were 338F (5’-ACTCCTACGGGAGGCAGCAG-3’) and 806R (5’-GGACTACHVGGGTWTCTAAT-3’). Magnetic beads were used for purification, and library construction was completed. Sequencing was performed using the Illumina MiSeq platform (BGI, Shenzhen, China) to generate 300 bp paired-end reads. After obtaining the Illumina raw data in fastq format, quality control and filtering were performed to obtain CleanData. The 60 samples generated 8,171,576 16S rRNA reads (mean reads per sample = 68,096). After sequence splicing, tags for the hypervariable region were obtained using the FLASH software (version 1.2.1, https://github.com/dstreett/FLASH2) (Magoc and Salzberg, 2011). The software USEARCH (version 7.0.1090, http://www.drive5.com/usearch/) (Edgar, 2013) was used to cluster tags according to 97% sequence similarity to generate operational taxonomic units (OTUs). The software RDP classifier (version 2.2, https://github.com/rdpstaff/classifier) (Wang and Cole, 2024) was used to compare representative OTUs sequences with the database for species annotation. The software Mothur (version 1.31.2, http://www.mothur.org) (Chappidi et al., 2019) was used for alpha diversity (Chao, ACE, Shannon and Simpson indices) analysis, and software QIIME(version 1.80, http://qiime.org/1.8.0/) (Hall and Beiko, 2018) was used for beta diversity analysis via principal coordinate analysis (PCoA) and nonmetric multidimensional scaling (NMDS). The Wilcoxon rank-sum test and linear discriminant analysis effect size (LEfSe) analysis were used to identify species with significant differences in microbiome abundance between the two groups.

Fecal samples from six hypothyroid patients and six control group participants were chosen by simple random sampling and evaluated by metagenomics sequencing. A MagPure Stool DNA KF Kit B (cat. no.MD511, Magen, China) was used to extract total DNA from the microbial community in the stool specimens (0.20 g). The quality and quantity of the extracted DNA were determined using a microplate reader, and the DNA fragment size was assessed by agarose gel electrophoresis. One microgram of genomic DNA was used, and a Covaris instrument was used to ultrasonically interrupt it to obtain a 300 bp fragment. Fastp and the software MEGAHIT (version 1.1.2, https://github.com/voutcn/megahit) (Li et al., 2015) were used for quality control and assembly. The 12 samples generated 73,022,905,500 metagenomic reads, and the average assembly length was 123.16 M.

The software Prodigal (version 2.6.3, https://github.com/hyattpd/Prodigal) (Hyatt et al., 2010) was used for metagenomic gene prediction. The software Diamond (version 2.0.13, https://github.com/bbuchfink/diamond) (Gautam et al., 2022) was used to annotate the gene set in the Kyoto Encyclopedia of Genes and Genomes (KEGG) Database (https://www.genome.jp/kegg/) to obtain KEGG orthologous (KO) abundance tables. The software CD-HIT (version 4.8.1, https://github.com/weizhongli/cdhit) (Kondratenko et al., 2020) was used for clustering to obtain nonredundant gene sets, and the software BLAST (version 2.2.28, http://blast.ncbi.nlm.nih.gov/Blast.cgi) (Schmid et al., 2023) was subsequently used for comparison with the Non-Redundant Protein Sequence (NR) Database (https://ftp.ncbi.nlm.nih.gov/blast/db/FASTA/). The Wilcoxon rank-sum test was used to evaluate differences.

All statistical analyses of 16S rRNA amplicon sequencing and metagenomic sequencing data were performed in R (version 3.4.1). SPSS software (version 26.0, IBM, USA) was used for analysis. Normally distributed measurement data are described as the mean ± standard deviation (SD) according to the t test. Nonnormally distributed measurement data are expressed as medians and quartiles by Wilcoxon rank-sum tests. Categorical variables are reported as frequencies [n (%)] according to the Chi-square test. Correlations were examined by Spearman rank correlation analysis. All the statistical analyses were two-tailed tests, with p < 0.05 indicating statistical significance.

Age differences, body mass index (BMI) at enrollment, gestational age, TPOAb, TgAb, TC, TG, LDL-C, HDL-C, GLU, and HGB were not significant between the hypothyroidism and control groups (P > 0.05).Higher serum CRP and lower FT4 levels were observed in the hypothyroidism group than in the control group ( Table 1 ). The clinical characteristics of the 12 subjects who received shotgun metagenomics are displayed in Supplementary Table 1 of the Supplementary Materials .

There was no significant difference in the alpha diversity between the hypothyroidism group and the control group ( Figure 1A ). Based on the unweighted UniFrac distance, beta diversity was significantly different between the two groups ( Figure 1B ). PCoA showed that the difference between the two groups was statistically significant ( Figure 1C ). A segregation trend between the two groups was observed by NMDS analysis ( Figure 1D ).

At the genus level ( Figure 2A ), the top 5 dominant bacterial genera were Faecalibacterium, Gemmiger, Prevotella, Bacteroides and Bifidobacterium in the hypothyroidism group and Gemmiger, Faecalibacterium, Bifidobacterium, Bacteroides and Blautia in the control group. The relative abundances of the phyla Bacteroidetes and Firmicutes and the corresponding Firmicutes/Bacteroidetes ratios are illustrated in Figure 2B , indicating that the differences were not statistically significant.

The Wilcoxon rank-sum test was performed on the top 10 genera ( Figure 2C ). Faecalibacterium, Paraprevotella, and Prevotella were significantly enriched in the hypothyroidism group, and Blautia was enriched in the control group. LEfSe analysis revealed that at the genus level, the abundances of Faecalibacterium, Paraprevotella, and Prevotella were strikingly increased in the hypothyroidism group, whereas those of Anaerococcus, Blautia, Corynebacterium, Eggerthella, etc. were significantly increased in the control group( Figure 2D ).

At the species level, Prevotella_copri, Faecalibacterium_prausnitzii, Phocaeicola_vulgatus and Eubacterium_sp. were the most abundant in the two groups ( Figure 3A ). The Wilcoxon rank-sum test showed that Paraprevotella_clara, Prevotella_hominis, Prevotella_sp_AM34_19LB, Prevotella_sp_TF12_30, and Prevotella_stercorea_CAG_629 were enriched in the hypothyroidism group, and Phascolarctobacterium_faecium was enriched in the control group ( Figure 3B ). LEfSe analysis revealed that the hypothyroidism group had greater relative abundances of Prevotella_hominis, Prevotella_sp_AM34_19LB, Prevotel-la_sp_TF12_30, Prevotella_stercorea_CAG_629, and Paraprevotella_clara ( Figure 3C ).

Microbial KOs were identified in gut metagenomes. As shown, the metabolism was active, and dominated by global and overview maps, carbohydrate metabolism, amino acid metabolism, and metabolism of cofactors and vitamins ( Figures 4A, B ). Differences in the number of KOs between the two groups were calculated by the rank-sum test. Different functions were initially screened according to “P < 0.05”, and then, through a literature review, relevant functions were selected and displayed visually in a box plot. The hypothyroidism group had higher K08681 (pyridoxal 5’-phosphate synthase pdxT subunit [EC: 4.3.3.6]) and lower K02106 (short-chain fatty acid transporter) and K06999 (phospholipase/carboxylesterase) ( Figure 4C ).

Serum inflammation-related factors were compared between the two groups. The serum CRP, IL-2, IL-4, IL-10 and TNF-α levels in the hypothyroidism group were greater than those in the control group (P < 0.05). There was no statistically significant difference in IL-6 between the two groups ( Table 2 ).

At the genus level ( Figure 5A ), TSH was positively correlated with Faecalibacterium, Paraprevotella, Prevotella, etc., and negatively correlated with Blautia. Prevotella was positively correlated with IL-4 and IL-10, Paraprevotella was positively correlated with IL-2, IL-10 and TNF-α, and Blautia was negatively correlated with CRP, IL-2, IL-4 and TNF-α. At the species level ( Figure 5C ), Prevotella_stercorea_CAG_629 was positively correlated with IL-4, IL-10, and TNF-α. Paraprevotella_clara was positively correlated with IL-10. Correlation analysis between different microbial species and functions revealed that these three functions were associated with different species of the Prevotella genus ( Figure 5B ).