Metabolomics-based analysis of the correlation between lipid metabolites and thyroid carcinoma

Abstract

Backgrounds

The pathogenesis of thyroid carcinoma (TC) involves various factors, with the interplay between hormonal and environmental influences being particularly critical. The gut microbiome, a long-overlooked risk factor, may play a significant role in TC development. Studies indicate that gut dysbiosis, bacterial outer membrane components such as lipopolysaccharide(LPS), and metabolites like Short-Chain Fatty Acids (SCFAs)and Trimethylamine N-Oxide (TMAO) are pivotal in mediating or influencing both gastrointestinal and extra-gastrointestinal tumors. Gut dysbiosis influences TC via microbial metabolites (e.g., LPS, SCFAs, TMAO) that modulate host immunity and steroid metabolism. However, reports on the multi-omics analysis of the correlation between gut microbiota, metabolites, and thyroid carcinoma remain scarce.

Methods

This study employs a case-control and cohort design, integrating 16 S rDNA sequencing, and metabolomics for multi-omics joint analysis of fecal samples from thyroid cancer patients before and after surgery. The aim was to identify associations between the gut microbiome, functional genes, and metabolites, as well as potential disease biomarkers.

Results

Postoperative samples showed significant lipid metabolite alterations (283 upregulated, 269 downregulated; P < 0.05).

Conclusions

This paper investigates the correlation between lipid metabolites and thyroid carcinoma through metabolomics-based analysis. Lipid metabolites, particularly cholesterol sulfate, are dysregulated in TC and linked to steroid/arachidonic acid pathways, suggesting potential diagnostic utility.

Introduction

Thyroid carcinoma (TC), one of the most prevalent malignancies of the head and neck, originates pathologically from thyroid follicular epithelial or parafollicular cells [1]. Histologically, 89.8% of cases are classified as papillary thyroid carcinoma (PTC), 4.5% as follicular thyroid carcinoma, 1.8% as Hürthle cell carcinoma, while the remaining cases comprise medullary thyroid carcinoma (1.6%) and anaplastic thyroid carcinoma (0.8%) [2]. While diagnostic methods such as fine-needle aspiration biopsy (FNAB) are well-established, a substantial proportion of cases remain indeterminate, necessitating improved diagnostic and prognostic biomarkers [3]. For indeterminate cases, repeat FNAB is mandatory; when diagnostic uncertainty persists, lobectomy or total thyroidectomy becomes necessary. Therefore, diagnosing thyroid cancer and searching for new markers is a very important task.

Globally, TC incidence has exhibited rapid escalation in recent decades. National cancer registry data reveal that TC now ranks as the fourth most common malignancy among urban Chinese women [4]with an alarming annual growth rate of 20%, imposing substantial societal burdens and healthcare resource strain [5]. Emerging evidence implicates gut microbiota composition and metabolic activity as critical environmental determinants in both gastrointestinal and extra-intestinal tumorigenesis [6, 7]. Dysbiosis of gut microbiota modulates host metabolic and immune homeostasis, influences drug pharmacokinetics (absorption, metabolism, and biotransformation), and consequently impacts carcinogenesis, progression, and therapeutic response. These findings position gut microbiota as a promising diagnostic and therapeutic target in oncology [8].

Despite growing interest in the microbiome-cancer axis, the role of gut microbiota and their metabolic products in thyroid carcinogenesis remains poorly understood. Recent studies have begun to explore links between microbial composition and thyroid dysfunction, but integrative multi-omics analyses—particularly those combining 16 S rDNA sequencing with metabolomic profiling—are still scarce. Such approaches could reveal novel biomarkers and elucidate mechanistic pathways connecting gut microbial activity to thyroid cancer.

Two principal mechanisms underlie microbial contributions to carcinogenesis. The first involves direct genotoxicity, where enteric bacteria (e.g., Escherichia coli) compromise genomic integrity through DNA damage, mutagenesis, and apoptotic dysregulation, ultimately promoting colorectal carcinogenesis [9]. The second mechanism centers on inflammation modulation, wherein oncogenic microbiota activate pattern recognition receptors (e.g., Toll-like receptors, TLRs), perpetuating NF-κB signaling cascades within the tumor microenvironment [10]. Substantial evidence confirms that gut microbiota collectively influence tumorigenesis through metabolic products [11–13]. Investigating gut microbial and metabolic profiles in cancer patients may unveil novel biomarkers and elucidate pathways linking microbiota to malignant transformation.

Current Methodologies for Gut Microbiota Profiling and Metabolomics Applications Multiple approaches are available for analyzing the overall composition of gut microbiota, with 16 S rDNA sequencing and metagenomic sequencing being the most widely adopted gold-standard method [14]. The 16 S rDNA gene sequence consists of variable and conserved regions with sufficient interspecies polymorphism of genes, whose conserved sequence regions reflect the interspecies affinities of bacteria, while the variable regions show the interspecies differences that can provide statistically valid measurements for distinguishing different bacteria. With the advancement of PCR and DNA sequencing technologies, 16 S rDNA sequencing of enteric flora by existing technologies can obtain information on the overall composition of the microbial system of the enteric flora, functional genes and their abundance, microbe-host interactions, and the effects of different treatments on species and genes.

Qunye Zhang et al. used 16 S, qPCR, high-resolution metabolome and qRT-PCR to investigate the mechanism of gut flora mediating the production of high salt diet-induced hypertension [15]; Shun Lu et al. found that the diversity of gut flora in non-small-cell lung cancer patients was correlated to the good results of anti-PD-1 immunotherapy through 16 S gene sequencing and FACS detection of peripheral blood immune cells correlation. Yu et al. shows that thyroid carcinoma patients demonstrate significant changes in gut microbiota [16]. This all indicates the wide application of metabolomics sequencing.

In this study, we hypothesize that thyroid cancer is associated with specific alterations in gut microbial metabolites, particularly lipid species, which may reflect or influence disease status. To test this, we conducted a case-control study involving pre- and post-operative fecal samples from TC patients, integrating 16 S rDNA sequencing and metabolomics to identify differential metabolites and enriched metabolic pathways. Our aim is to uncover potential diagnostic biomarkers and shed light on the functional role of gut microbiota in thyroid cancer, ultimately contributing to more precise diagnostic and therapeutic strategies.

Materials and methods

Experimental materials

Experimental objects

This study strictly adhered to the inclusion criteria to recruit and screen 12 patients with thyroid cancer. Samples and baseline data were collected from thyroid cancer volunteers before and after surgery while taking levothyroxine sodium as prescribed, including age, gender, and various physiological indicators, and the volunteers were followed up. The obtained stool samples were subjected to 16 S rDNA sequencing, Metagenomic sequencing and metabolomic sequencing, and the sequencing results were analyzed by multi-omics analysis to obtain the association of intestinal flora-functional genes-metabolites and potential disease markers, which were combined with the existing clinical diagnostic indexes of thyroid cancer, to provide a new disease marker for assisting in the diagnosis of thyroid cancer, and to validate it clinically.

Subjects should complete the collection of relevant medical history information within 24 h. Comprehensive and detailed collection of basic information, general physical examination, neurological examination, measurement of temperature, blood pressure, respiration and pulse rate, and recording of age, gender, history of hypertension, diabetes mellitus, smoking, alcohol consumption, white blood cell counts, blood glucose levels, homocysteine levels, total cholesterol levels, triglyceride levels, LDL cholesterol levels, HDL cholesterol levels, and other clinical indicators for clinical diagnosis of thyroid cancer. cholesterol level, HDL cholesterol level, cerebral white matter lesion score and other relevant indicators were fully documented on the case report form for backup.

All TC patients underwent total thyroidectomy or lobectomy by the same surgical team. Standard perioperative care was followed, including general anesthesia and prophylactic antibiotics (Cefazolin). Fecal samples from the TC group were collected at two time points: (1) pre-operatively (FF group, 3 days before surgery, prior to any bowel preparation or fasting) and (2) post-operatively (BF group, within 72 h after surgery, before discharge).

Inclusion exclusion criteria

(1) Inclusion criteria Thyroid cancer volunteers:

① Adults (age ≥ 18 years) with a first-time, histologically confirmed diagnosis of primary papillary thyroid carcinoma;

② No previous treatment for thyroid cancer (including surgery, radiofrequency ablation, radiation therapy, or chemotherapy); No history of antibiotic, probiotic, or prebiotic use within the 3 months preceding sample collection;

③ Patients gave informed consent, volunteered to participate and signed an informed consent form.

(2) Exclusion criteria.

① Volunteers have cancers other than primary thyroid cancer;

② Volunteers have a history of alcohol or narcotic abuse, drug abuse, or a history of psychiatric disorders (e.g., schizophrenia, obsessive-compulsive disorder, depression), antagonistic personality, poor motivation, paranoia, or other emotional or intellectual problems that may affect the informed validity of participation in this study;

③ Volunteers have other endocrine-related disorders;

④ Use of medications known to significantly alter gut microbiota or metabolism (e.g., antibiotics, immunosuppressants, metformin, statins) within the 3 months prior to the study.

This study has been reviewed and approved by the Clinical Research Ethics Review Committee of the Affiliated Hospital of Guangdong Medical University, with ethics approval number: PJ2021-079. All participants have signed informed consent forms.

Methods of fecal sample collection

(1) Distribute one sterilized fecal cup to the subject in advance;

(2) Use the spoon in the sampling tube to intercept one spoonful of the inside of the middle section of the feces;

(3) Place the spoon in the fecal cup and tighten the lid;

(4) Hand the sample to the person in charge, who registers the subject’s information on the wall of the cup and quickly places the sample in a -80 °C refrigerator for storage.

The preoperative group (FF) consisted of fecal samples collected from 12 patients 3 days prior to surgery, while the postoperative group (BF) consisted of fecal samples collected from the same 12 patients within 72 h after surgery. Rationale for Stool Sampling: The gut-thyroid axis is a burgeoning field of research. The premise is that gut microbiota and their metabolites can influence systemic inflammation, immune function, and host metabolism, including the regulation of hormones and signaling molecules that may impact thyroid pathophysiology. Stool samples provide a non-invasive window into this complex interplay, capturing the functional output (metabolomes) of the gut microbiome.

Experimental methods

DNA extraction from fecal samples

We perform fecal DNA extraction using the QIAMEM DNA Stool Mini Kit 51,604 from QIAGEN. Take 0.15 g of feces and vortex with Bead Solution; add C1 lysis, heating at 65℃ + vortex; centrifugation to get the supernatant; C2/C3 precipitate impurities, centrifugation to retain the supernatant; C4 binding DNA, divided into 3 times through the column; C5 washing; C6 elution, detection of concentration (≥ 10ng/µl, total amount of ≥ 1 µg) and purity (AGE / Qubit); qualified samples were sent to sequencing.

Amplification, sequencing and analysis of 16Sr DNAV3-V4 of fecal DNA

Sample extraction, amplification, sequencing and analysis were done by Shanghai Zhongke New Life Technology Co. The basic experimental flow of sequencing was as follows:

(1) Primer design and synthesis:

The amplification region was selected as V3-V4 region. The primers 341 F and 806R were used to synthesize the sequences applicable to Illumina Miseq.

The sequences of Illumina Miseq were added to the 5’ end of 341 F and 806R to make them into specific primers with barcode.

Primer sequences: Forward primer (338 F, 5’-3’): ACTCCTACGGGGAGGCAGCA.

Reverse primer (806R, 5’-3’): GGACTACHVGGGGTWTCTAAT.

(2) PCR amplification and purification:

All samples were performed according to the standard procedure, with three replicates for each group of samples. The PCR products of the same sample were mixed and detected by 2% agarose electrophoresis, followed by cutting and recovery of PCR products using AxyPrepDNA Gel Recovery Kit (AXYGEN), and the recovered products were detected by 2% agarose gel electrophoresis.

(3) Miseq library construction:

① PCR amplification was used to introduce the Illumina official specific junction sequence into the outer end of the target region; ② the target bands were separated by agarose gel electrophoresis, and the specific fragment was gel recovered; ③ the purity of the target sequences was verified by 2% agarose gel electrophoresis after elution with Tris-HCl buffer; ④ single-stranded DNA templates were obtained by denaturing with sodium hydroxide for subsequent sequencing. template for subsequent sequencing.

(4) Miseq sequencing: Bridge PCR amplification of DNA clusters, sequencing while synthesizing (fluorescent labeled dNTP), and reading the sequence.

(5) Bioinformatic analysis of 16 S rDNA data.

The basic principle of targeted analysis of lipid-related metabolites

Lipidomics is a science that studies the composition, structure, and function of lipid molecules in organisms. Lipids are the main components of biological membranes and are involved in important physiological processes such as cell signaling and energy storage. Lipidomics systematically analyzes and characterizes lipid molecules in biological samples through high-throughput technologies such as mass spectrometry and liquid chromatography, thereby revealing their functions and regulatory mechanisms in organisms.

Metabolomic profiling of lipid metabolites

Metabolite extraction

Lipid metabolites were extracted from frozen fecal samples using a methanol: methyl-tert-butyl ether (MTBE) solvent system. Briefly, 20 mg of feces was homogenized in 400 µL of methanol and 1,360 µL of MTBE. After phase separation with the addition of water, the upper organic layer (containing lipids) was collected, dried under nitrogen, and reconstituted in isopropanol for LC-MS analysis.

LC-MS analysis

Targeted lipidomic analysis was performed using an ultra-high-performance liquid chromatography (UHPLC) system (e.g., Thermo Vanquish) coupled to a triple quadrupole mass spectrometer (e.g., Thermo Quantis). Separation was achieved on a reversed-phase C18 column (e.g., ACQUITY UPLC BEH C18, 1.7 μm, 2.1 × 100 mm) using a gradient of mobile phase A (water: acetonitrile, 40:60, 10 mM ammonium formate) and mobile phase B (isopropanol: acetonitrile, 90:10, 10 mM ammonium formate). The mass spectrometer operated in multiple reaction monitoring (MRM) mode for specific, sensitive quantification of a predefined panel of lipid metabolites, including fatty acyls, glycerolipids, glycerophospholipids, and sterols.

Data processing and analysis

Raw MS data were processed using vendor-specific software (e.g., Thermo TraceFinder) for peak integration, calibration, and quantification. Metabolite levels were normalized to the total protein content of the sample (determined by BCA assay) to account for variations in sample mass. Quality control (QC) was maintained by analyzing pooled quality control samples intermittently throughout the analytical run to monitor instrument stability.

Statistical analysis

Statistical analyses were performed using R software (v4.1.0). Clinical characteristics were compared using Student’s t-test or Mann-Whitney U test for continuous variables and Chi-square test for categorical variables. 16 S rRNA sequencing data were analyzed for alpha and beta diversity. For metabolomic data, multivariate statistical analyses, including Principal Component Analysis (PCA) and Partial Least Squares-Discriminant Analysis (PLS-DA), were performed using the ropls package. Differential metabolites between groups were identified based on a combination of Variable Importance in Projection (VIP) score > 1.0 from the PLS-DA model and a p-value < 0.05 from a paired or unpaired t-test (with FDR correction for multiple comparisons, where appropriate). Pathway enrichment analysis was performed using the MetaboAnalyst 5.0 platform based on KEGG pathways.

Results

Baseline characteristics of the preoperative and postoperative thyroid cancer groups

The baseline characteristics of the 12 enrolled papillary thyroid carcinoma (PTC) patients are summarized in Table 1. Serum levels of thyroid hormones and related biomarkers were measured pre-operatively (FF group) and within72 hours post-operatively (BF group) using electrochemiluminescence immunoassays. As expected, parathyroid hormone (PTH) levels were significantly decreased after surgery (45.97 ± 19.87 pg/ml vs. 24.9 ± 11.39 pg/ml, p = 0.00629), consistent with transient post-thyroidectomy hypoparathyroidism. Free Thyroxine (FT4) levels were significantly elevated in the BF group (16.98 ± 1.46 pmol/L vs. 20.35 ± 3.63 pmol/L, p = 0.00991), attributable to the initiation of levothyroxine hormone replacement therapy immediately after surgery.

Table 1.

Baseline characteristics and serum biomarker levels of PTC patients pre- and post-operation

Variant	FF (N = 12)	BF (N = 12)	P-value
FT3 (pmol/L, χ ± s)	4.82 ± 0.7	4.65 ± 0.89	0.63236
FT4 (pmol/L, χ ± s)	16.98 ± 1.46	20.35 ± 3.63	0.00991**
TSH (mIU/L, χ ± s)	1.53 ± 1.21	1.79 ± 2.55	0.76213
A-TG (mmol/L,χ ± s)	42.13 ± 77.43	36.59 ± 64.95	0.85766
PTH (pg/ml, χ ± s)	45.97 ± 19.87	24.9 ± 11.39	0.00629**
HTG (ng/ml, χ ± s)	36.03 ± 51.87	4.66 ± 6.49	0.06038

Notably, Free Triiodothyronine (FT3) and Thyroid-Stimulating Hormone (TSH) levels showed no significant change (p > 0.05). The stability of TSH is likely due to its long serum half-life (~ 1 week), meaning a significant rise would not be expected within the 72-hour post-operative sampling window. The stability of FT3, while less intuitive, may be related to the short study timeframe and the complex kinetics of hormone replacement and metabolic clearance in the immediate post-operative state(Data presented as mean ± standard deviation. P-values calculated using a paired Student’s t-test).

Results of 16s rDNA sequencing of fecal samples

PCA and PLS-DA analysis

To assess the global differences in fecal lipid metabolites, unsupervised Principal Component Analysis (PCA) and supervised Partial Least Squares-Discriminant Analysis (PLS-DA) were performed. Both models showed a clear separation between the pre-operative (FF) and post-operative (BF) groups in both positive and negative ionization modes (Fig. 1A-B). The distinct clustering indicates a significant alteration of the fecal metabolome following thyroidectomy, even within this short timeframe.

Fig. 1.

Multivariate analysis of fecal lipid metabolites. (A) PCA and (B) PLS-DA score plots demonstrate separation between pre-operative (FF, blue) and post-operative (BF, red) samples in positive ion mode. Ellipses represent 95% confidence intervals

Between-group difference metabolite analysis

Differential analysis between groups was performed by lipid metabolites in positive and negative ion modes. We found that in positive ion mode (Fig. 2A and C), 112 metabolites were down-regulated and 171 metabolites were up-regulated in the BF group relative to the FF group (|logFC|>0.5, P < 0.05), while in negative ion mode (Fig. 2B and D), 157 metabolites were down-regulated and 111 metabolites were up-regulated in the BF group relative to the FF group (|logFC|>0.5, P < 0.05). Volcano plot analysis (threshold: |log₂FC| >0.5, P < 0.05) identified 283 significantly upregulated and 269 significantly downregulated lipid metabolites in the BF group compared to the FF group. A hierarchical clustering heatmap of the top 30 most variable metabolites showed a consistent pattern of regulation within each group, further validating the distinct metabolic phenotypes (Fig. 2).

Fig. 2.

Differential metabolite analysis. (A) Volcano plot of metabolites in positive ion mode. Red dots represent significantly upregulated metabolites (n = 171), blue dots represent downregulated metabolites (n = 112) (|log₂FC| >0.5, P < 0.05). Grey dots are non-significant. (B) Heatmap of the top 30 differential metabolites shows distinct clustering between FF and BF groups (rows are metabolites, columns are samples; red indicates high abundance, blue indicates low abundance)

Identification of metabolites differing between groups

After we identified the lipid metabolites that varied significantly between groups, we merged the up- and down-regulated metabolites of the positive and negative ion modes, respectively, and performed de-emphasis to obtain 92 up-regulated metabolites,104 down-regulated metabolites, respectively, and we demonstrated the categorization of up- and down-regulated metabolites of the positive and negative ion modes (Fig. 3A), of which 40 were down-regulated metabolites: Fatty Acyls [FA] 40 kinds of metabolites; Glycerol lipids [GL] 7 species; Glycerophospholipids [GP] 11 species; Polyketides [PR] 19 species; Prenol Lipids [PR] 8 species; Sphingolipids [SP] 9 species; Sterols [ST] 10 species. Up-regulated metabolites: Fatty Acyls [FA] 24; Glycerol lipids [GL] 2; Glycerophospholipids [GP] 9; Polyketides [PR] 10; Prenol Lipids [PR] 12; Sphingolipids [SP] 15 Sterols [ST] 20 species. After identifying the lipid metabolites that changed significantly between groups, we show the top 15 lipid metabolites with VIP scores in the PLS-DA analysis in the positive and negative ion mode, and all of them had scores of 1 or more (Fig. 3B). They may be the key differential lipid metabolites before and after surgery. These differentially expressed metabolites were categorized into eight major lipid classes (Fig. 3A). Sterols (ST) and Fatty Acyls (FA) constituted the largest proportion of dysregulated lipids. The top 15 metabolites with the highest Variable Importance in Projection (VIP) scores from the PLS-DA model (all VIP > 1.5) are listed in Fig. 3B; These are proposed as the key metabolites driving the separation between pre- and post-operative states.

Fig. 3.

Characterization of differential metabolites. (A) Stacked bar chart showing the number of significantly altered metabolites within each lipid class. (B) VIP scores of the top 15 metabolites most responsible for group separation in the PLS-DA model

Metabolic pathway composition and differential analysis

KEGG functional enrichment was performed for lipid metabolites that differed significantly between groups (Fig. 4A). As shown, the lipid metabolites upregulated in the BF group were significantly enriched to the Steroid hormone biosynthesis pathway, while the lipid metabolites downregulated by BF were significantly enriched to the Arachidonic acid metabolism pathway. And we found that the BF group was significantly enriched to the Steroid hormone biosynthesis pathway due to up-regulation of cholesterol sulfate(P < 0.01); to the Arachidonic acid metabolism pathway due to significant down-regulation of the 2,3-dinor-5,6-dihydro-15-F2t-IsoP metabolism pathway.(P < 0.05) (Fig. 4).Among them, cholesterol sulfate is a sulfated derivative of cholesterol, formed by the binding of cholesterol to a sulfate group catalyzed by sulfotransferase (SULT). The role of cholesterol sulfate in cancer has not been fully clarified, with some studies suggesting that cholesterol sulfate may be able to promote cancer. metastasis, but some studies suggest that it exerts anti-tumor effects by regulating T cells.

Fig. 4.

Pathway analysis. Scatter plot of KEGG pathway enrichment. The top enriched pathways are labeled. The y-axis represents the -log10(P-value), and the x-axis represents the Pathway Impact value. The size of the dot corresponds to the number of metabolites mapped to the pathway

Discussion

It has been concluded that thyroid carcinoma (TC) is mainly a disease caused by a combination of hormonal-environmental factors [17], and intestinal microorganisms may be an important environmental risk factor that has been overlooked. Although there have been a large number of reports on multi-omics combined head and neck tumor diagnostic models [18], there are still few studies related to TC, and even fewer studies discussing the changes in intestinal flora, metabolic profiling, and related diagnostic models in preoperative and postoperative TC. Our team took this as an entry point to explore, through metabolomics sequencing, data analysis to obtain gut flora-functional gene-metabolite associations and potential disease markers, combined with the existing clinical diagnostic indicators of thyroid cancer, to provide new disease markers to assist in the diagnosis of thyroid cancer, and its clinical validation. This pilot study utilized an integrated LC-MS-based metabolomics approach to profile fecal lipid metabolites in PTC patients before and after thyroidectomy. Our primary finding is that surgery induces rapid and significant alterations in the gut metabolic landscape. The most notable changes were the upregulation of cholesterol sulfate and the downregulation of a prostaglandin-like metabolite (2,3-dinor-5,6-dihydro-15-F2t-IsoP), which mapped to the steroid hormone biosynthesis and arachidonic acid metabolism pathways, respectively.

In recent years, the investigation of the correlation between the microbiology of the human gut and various diseases has deepened, and our research team was prompted to think about the possible connection between the pathogenesis of thyroid cancer and metabolites. Metabolomics was chosen as the entry point for this study. Of interest is the potential influence of structural changes in the gut microbial community on the development of thyroid cancer [19], which was revealed through the systematic analysis of metabolic pathways and their regulatory networks. Examples have shown that the use of metabolomics approaches not only enables the identification of key differential metabolites, but also provides a deeper understanding of the chain of molecular events during disease progression [20]. Moreover considering that gut flora is an important microbial component of the human body, and its impact on human health has received widespread attention [21]. We conduct the research through metabolomics methods. The significant enrichment of the steroid hormone biosynthesis pathway [22], primarily driven by elevated cholesterol sulfate, is a finding with potential biological relevance to thyroid cancer. Cholesterol sulfate is an enigmatic molecule with dual roles in cellular regulation. Some studies posit a pro-tumorigenic role, suggesting it can promote cell proliferation and survival in the tumor microenvironment [26]. Conversely, other research indicates it can modulate immune function, particularly T-cell activity, potentially exerting anti-tumor effects [26]. In the context of our study, its sharp increase post-operatively could be a stress response to surgery, a direct or indirect effect of altered thyroid hormone levels, or a reflection of changes in gut microbial composition post-antibiotic prophylaxis. This ambiguity highlights the need for functional validation to determine whether this shift represents a bystander effect or a active mechanistic player in post-surgical recovery or cancer biology.

Conversely, the downregulation of metabolites in the arachidonic acid metabolism pathway suggests a post-operative suppression of inflammatory eicosanoid production. Arachidonic acid derivatives are potent mediators of inflammation, a well-known hallmark of cancer progression. The observed downregulation of a prostaglandin metabolite could indicate a reduction in systemic inflammation following tumor resection, potentially representing a beneficial shift in the host environment. This finding aligns with studies linking eicosanoid signaling to cancer pathogenesis [23], though a direct causal link to thyroidectomy remains to be established.

The significant down-regulation of 2,3-dinor-5,6-dihydro-15-F2t-IsoP was found to correlate with the enrichment of the arachidonic acid metabolism pathway in this group. The formation of cholesterol sulfate was elucidated as a derivative of cholesterol catalyzed by sulfotransferase (SULT) and combined with a sulfate group. It is widely distributed in human tissues such as skin, intestine, adrenal glands, liver and brain [24]. It plays an important role in a variety of physiological activities, including T cell signaling regulation, glucose metabolism regulation, and lipid homeostasis [25].At the current stage of research, controversy still exists regarding its biological effects in cancer, with some studies suggesting that it may promote tumor progression and influence tumor value-addition and survival through metabolic regulation in the tumor microenvironment [26].

While studies directly linking gut microbial metabolites to thyroid cancer are scarce, our findings on disrupted steroid and lipid metabolism are consistent with a growing body of literature on the gut-thyroid axis. Recent work by Yu et al. [19] also reported distinct gut microbial communities in TC patients, suggesting a functional link that our metabolomic data now substantiate at the biochemical level. Our results extend these findings by identifying specific metabolic pathways that are perturbed, moving beyond correlation to propose potential functional mechanisms.

This study has several important limitations that must be acknowledged. First, the small sample size (n = 12) limits the statistical power and generalizability of our findings; these results should be considered preliminary and hypothesis-generating. Second, the lack of a healthy control group makes it impossible to discern which metabolic differences are specific to thyroid cancer itself versus non-specific to the pre-operative state. Third, the post-operative sampling timeframe (within 72 h) is a major confounder. The metabolic changes observed are likely a complex amalgamation of the effects of tumor removal, surgical stress, anesthesia, antibiotic administration, and dietary changes, making it difficult to attribute changes solely to the absence of the tumor. Future longitudinal studies with larger cohorts, healthy controls, and multiple post-operative time points (e.g., 1 month, 3 months) are essential to isolate the cancer-specific metabolic signature from the acute effects of surgery.

Conclusions

In summary, this study systematically analyzed the role of lipid metabolites in the gut microbiota on the occurrence and development of thyroid cancer using metabolomics technology. In this pilot study, we observed significant alterations in the fecal lipid metabolome of patients following thyroidectomy for papillary thyroid carcinoma. The most notable changes included the upregulation of cholesterol sulfate, associated with the steroid hormone biosynthesis pathway, and the downregulation of metabolites within the arachidonic acid metabolism pathway.

It is important to emphasize that these findings are preliminary and observational. The study design, particularly the lack of a healthy control group and the acute post-operative sampling timeframe, means that these metabolic shifts cannot be directly attributed to the presence or absence of thyroid carcinoma itself. The changes are likely a complex response to the combined effects of surgical trauma, anesthesia, perioperative medications, and dietary changes.

Therefore, the identified metabolites should not be considered validated “biomarkers” for TC diagnosis. Instead, this work generates a hypothesis: that specific lipid metabolic pathways may be involved in the physiological response to thyroidectomy and could potentially be linked to the disease state. This hypothesis requires rigorous validation in future studies.

Future research must employ a longitudinal design with larger cohorts, include matched healthy controls, and feature multiple post-operative sampling points to disentangle the acute effects of surgery from any underlying cancer-specific metabolic signature. Only through such controlled validation can the true diagnostic or mechanistic significance of these lipid metabolites be determined.

Abbreviations

TC

Thyroid Carcinoma

SCFAs

Short-Chain Fatty Acid

TMAO

Trimethylamine N-Oxides

TLRs

Toll-like receptors

LDL

Low-Density Lipoprotein‌

HDL

High-Density Lipoprotein‌

FA

Fatty Acyls

GL

Glycerol lipids

GP

Glycerophospholipids

PR

Polyketides

SP

Sphingolipids

ST

Sterols

SULT

Sulfotransferase

Author contributions

JL and JH: Writing–original draft, Writing–review & editing, Data curation, Visualization. XX: Data curation. SL: Formal analysis. HL: Project administration. YW, WC and ZC: Validation. CZ, ML, BL, GZ and ZH: Investigation. JL and BL: Resources. ZZ: Writing–review & editing, Supervision, Methodology. WQ: Writing–review & editing, Funding acquisition, Project administration, Conceptualization.

Funding

This study was supported by the following two projects: Clinical Research Project of the Affiliated Hospital of Guangdong Medical University (LCYJ2021B010)) 2024 Unsubsidized Science and Technology Tackling Program of Zhanjiang City, Guangdong Province, China (2024B01154).

Data availability

All data within the literature are available for use. The sequence data supporting the results of this study have been deposited in Metabolights with the primary accession code MTBLS12769.The following is access link https://www.ebi.ac.uk/metabolights/MTBLS12769.

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki, and was approved by the Clinical Research Ethics Committee of the Affiliated Hospital of Guangdong Medical University (approval number: PJ2021-079).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.