The gut-liver axis links the associations between serum carotenoids and non-alcoholic fatty liver in a 7.8-year prospective study

Introduction

Non-alcoholic fatty liver disease (NAFLD) is the most prevalent chronic liver disease globally, affects approximately 30% of the adult population and over half of the obese individuals (1). NAFLD can process to severe liver conditions, such as non-alcoholic steatohepatitis, fibrosis, and cirrhosis (2-4). Currently, there is no established pharmacological treatment for NAFLD, making dietary modification an effective approach for its prevention and management.

Carotenoids, a group of fat-soluble phytochemicals known for their antioxidant and anti-inflammatory properties, are primarily found in fruits and vegetables (5). Lutein, zeaxanthin, β-cryptoxanthin, lycopene, α-carotene, and β-carotene represent the six major carotenoids in human blood. Although a number of observational studies and small randomized control trials (RCTs) have indicated beneficial associations between dietary or circulating carotenoids and NAFLD (6-11), evidence from prospective studies remains limited, particularly regarding different types of carotenoids (6-8). Furthermore, the underlying mechanisms of these associations are not yet fully understood.

Accumulating evidence suggests that the crosstalk within the gut-liver axis is crucial in the pathogenesis of NAFLD (12-14). Prior research has identified distinct gut microbiota compositions in NAFLD patients compared to healthy individuals (15-17), and interventions targeting the gut microbiome have been linked to improvements in hepatic steatosis (18). Nutritional strategies, particularly those involving carotenoids and polyphenols, have been recognized as influential factors on the gut-liver axis, showing potential in preventing and treating NAFLD (19). Several studies have noted that dietary or supplemental carotenoids could alter the profile of gut microbiota and their metabolites (20-24). While an animal study highlighted the possibility of the gut microbiome mediating the effects of carotenoids on NAFLD (25), this concept has not yet been explored in human subjects through prospective studies.

In this study, we primarily examined the 7.8-year prospective associations between serum carotenoids and the incidence or presence of NAFLD. Secondly, we sought to identify biomarkers within the gut microbiota and metabolites in feces and serum that could elucidate the link between serum carotenoids and NAFLD in a middle-aged and elderly Chinese cohort. We present this article in accordance with the STROBE reporting checklist (available at https://hbsn.amegroups.com/article/view/10.21037/hbsn-23-526/rc).

Methods

Study design and population

The study population was drawn from the Guangzhou Nutrition and Health Study (GNHS), a community-based prospective cohort. This cohort included 4,048 healthy adults aged 40–75 years, recruited between 2008–2013 in Guangzhou, China (trial registered at www.clinicaltrials.gov, ID: NCT03179657) (26). Participants were followed up approximately every three years. At the first follow-up, serum carotenoids were quantified in 3,023 participants. Of these, 38 participants were subsequently excluded due to their excessive ethanol consumption [male: >140 g/week; female: >70 g/week (27)] or the presence of other liver diseases. The remaining participants were evaluated for NAFLD using abdominal ultrasonography (ULS-NAFLD). These evaluations occurred across four follow-up visits, involving 2,921, 2,380, 2,032, and 1,598 participants, respectively, with the last group being assessed as of July 2023. Additionally, magnetic resonance imaging (MRI) was conducted in 828 subjects at 4^th follow-up to confirm NAFLD diagnoses. This analysis included 2,522 subjects who had baseline serum carotenoids data, attended at least two NAFLD evaluations by ultrasound. Among these, the gut microbiome (n=1,661) and targeted fecal (n=893) and serum metabolome (n=896) were further profiled at 2^nd or 3^rd follow-ups (Figure 1). The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013) and was approved by the Ethics Committee of the School of Public Health at Sun Yat-sen University (2018048). All participants provided written informed consent.

Figure 1 Flowchart of study participants. F1–F4 refer to the 1^st, 2^nd, 3^rd, and 4^th follow-up visits, respectively. The time marked above is the time point at the beginning of each follow-up visit. NAFLD, non-alcoholic fatty liver disease; MRI, magnetic resonance imaging.

Data collection

Questionnaire interview and anthropometric measurements

At each visit, detailed information on sociodemographic characteristics, lifestyle factors, and medical history was collected through face-to-face interviews conducted by well-trained interviewers. Habitual dietary intakes were estimated using a 79-item food frequency questionnaire (27). Physical activity level was assessed with a validated 19-item questionnaire (28). Height, weight, waist and hip circumference were measured for participants with light clothing and no shoes.

Laboratory assays

Overnight fasting serum levels of triglycerides (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), alanine aminotransferase (ALT), aspartate aminotransferase (AST), and fasting blood glucose (FBG) were measured using a Hitachi 7600-010 automated analyzer. Serum carotenoids, including lutein & zeaxanthin, β-cryptoxanthin, lycopene, α-carotene, and β-carotene were quantified at baseline through a modified reverse-phase high-performance liquid chromatography (HPLC) method by a Waters 2998 diode array detector (Waters) as detailed previously (29). Due to closely migrating peaks, lutein and zeaxanthin were not quantified separately. The day-to-day coefficients of variation ranged from approximately 7.8% to 10.6%. Total carotenoid levels were defined as the sum of these five carotenoids.

Ascertainment of participants with NAFLD

ULS-NAFLD diagnosis followed the Chinese Liver Disease Association guideline (30), and was performed using a Doppler SSI-5500 sonography system (SonoScape Medical Corp., Shenzhen, China) by experienced physicians. Participants with excessive ethanol intake (male: >140 g/week; female: >70 g/week) were excluded at baseline. Liver steatosis was evaluated based on criteria including diffuse enhancement of near field echo in the hepatic region and any one item of unclear display of intra-hepatic lacuna structure and diaphragm, mild to moderate hepatomegaly, a reduction of the vascular sharpness. The reliability between different operators was tested in the 100 participants (Spearman r=0.911, kappa =0.875, P<0.001). Additionally, a Siemens 3.0T MRI system and LiverLab software were used to quantitatively analyze liver fat percentage, with a cut-off value of 5% for NAFLD evaluation (31). Moderate agreement between ultrasound and MRI in diagnosing NAFLD was observed among 828 participants, with a consistency rate of 71.5% (Spearman r=0.46, kappa =0.44, P<0.001, Table S1). The sensitivity and specificity of the ultrasound method were 82.5% and 63.8%, respectively, compared to MRI.

Fecal sample 16S rRNA gene sequencing, and targeted metabolomics profiling

During the 2^nd and 3^rd follow-up visits, 1,661 fecal samples were collected in an ice box and stored at −80 ℃ within 1–4 hours until analysis. Information including the Bristol stool score, usage of antibiotics or antifungals in the past 2 weeks was recorded.

Fecal microbial DNA was extracted using the QIAamp^® DNA Stool Mini Kit (Qiagen, Hilden, Germany). The V3–V4 hypervariable region of the 16S rRNA gene was amplified, purified and quantified. Amplicon sequencing was performed on the Illumina MiSeq System (Illumina, San Diego, USA). FASTQ files were demultiplexed, merge paired, and quality filtered by Quantitative Insights into Microbial Ecology (QIIME) software (version 1.9.0) (32). Sequences were clustered into operational taxonomic units (OTUs) with 99% similarity and annotated based on the Greengenes database (version 13.8) (33).

For metabolomics assays, detailed methods and quality control measures have been described previously (34). In summary, targeted metabolomics profiling of each fecal sample (n=893) or serum sample (n=896) was performed with an ultrahigh-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) system (ACQUITY UPLC-Xevo TQ-S, Waters Corp., Milford, MA, USA). A total of 199 fecal metabolites and 199 serum metabolites were quantified.

Statistical analysis

Participants who attended at least two follow-up visits were categorized into four NAFLD groups (NAFLD-free, improved NALFD, new-onset NAFLD, and persistent NAFLD) based on changes in ULS-NAFLD status over approximately.

Associations between serum carotenoids and NAFLD

Logistic regression was used to analyze the association between sex-specific quartiles (Q) of baseline serum carotenoids and the presence of persistent NAFLD (vs. NAFLD-free). The initial model (model 1) adjusted for baseline age and sex, while a more comprehensive model (model 2) included adjustments for household income, physical activity, multivitamin use, smoking, alcohol drinking, education level, dietary intakes of energy, saturated fat, and fiber. The associations between carotenoid quartiles and incidence of new-onset NAFLD in participants without baseline NAFLD and recovery rate of improved NAFLD in NAFLD patients at baseline were examined using Cox proportional hazards regression, adjusted for the same covariates as in model 1 and 2. We also examined the associations between baseline serum carotenoids quartiles of and the presence of MRI-identified NAFLD (vs. non-NAFLD) by logistic regression for confirmatory analysis. Restricted cubic splines (with 3 knots) were used to assess the dose-response relationship.

Identification of microbiota and metabolite biomarkers link carotenoids and NAFLD

Multivariable linear regression was used to test the association of baseline carotenoids and α-diversity indices (observed OTUs, Shannon index, and Chao1 index), adjusted for the same covariates in the above model 2 plus Bristol stool score, sequencing depth and sequencing run. Analysis of Covariance was used to compare α-diversity among the NAFLD groups. β-diversity at the genus level between carotenoids quartiles and NAFLD groups were analyzed via a principal coordinate analysis (PCoA) and permutational multivariate analysis of covariance (PERMANOVA, 999 permutations) based on Bray-Curtis distance. Multivariate Analysis by Linear Models (MaAsLin) was used to identify overlapping genera associated both with serum carotenoids and NAFLD after arcsine square root transformation of gut microbes’ relative abundance, adjusted for the same covariates as the α-diversity analysis. The false discovery rate (FDR) was controlled using the Benjamini-Hochberg method, an FDR <0.25 was considered statistically significant.

Metabolite concentrations were log-transformed and normalized prior to analyses. We used orthogonal partial least squares discrimination analysis (OPLS-DA) and linear or logistic regression to identify overlapping fecal and serum metabolites related to serum carotenoids and NAFLD, adjusted for the same covariates in model 2.

Mediation and path analyses

For overall effect analysis of selected markers, carotenoids-related microbiome score (MS), carotenoids-related fecal metabolite score (FMS) and carotenoids-related serum metabolites score (SMS) were calculated using the formula: Score=∑ [(concentration of the overlapping genus or metabolites related to carotenoids) × (direction of coefficient with NAFLD)]. Mediation analyses through the R package “mediation” (bootstrap method: Quasi-Bayesian, simulations =2,000) (35) evaluated whether MS, FMS, and SMS mediated the carotenoids-NAFLD association, adjusting for same covariates in model 2. Path analysis was conducted to explore relationships between three types of biomarkers and potential mechanism. A co-occurrence network visualized in Cytoscape software (version 3.9.1) demonstrated the interaction of carotenoids, genera, and metabolites biomarkers and NAFLD based on the regression coefficients. All statistical analyses were performed using R (version 4.1.3) and unless otherwise specified, a P value or FDR <0.05 was considered significant.

Results

Characteristics of the study participants

At baseline, 1,472 subjects were diagnosed with NAFLD, while 1,449 subjects were not. No statistically significant differences were observed in most characteristics between participants included and excluded (Table S2). Among the 2,522 participants followed up a mean time of 7.8 years, 770 were NAFLD-free, 301 had improved-NAFLD, 474 had new-onset-NAFLD, and 977 had persistent NAFLD, as evaluated by ultrasonography. Additionally, 342 out of 828 participants were classified as MRI-identified NAFLD. Compared to the NAFLD-free group, the other three groups generally exhibited higher waist circumference, BMI, FBG, ALT, and poorer physical activity performance, along with lower levels of serum carotenoids. No significant difference was observed in age, education level, household income, alcohol drinking, and other traits (Table 1).

Association of serum carotenoids with NAFLD persistence and incidence

Longitudinal analyses indicated that higher levels of serum carotenoids were associated with a lower 7.8-year incidence of ULS-NAFLD (P-trend <0.05) and reduced odds of 7.8-year persistent ULS-NAFLD (P-trend <0.001) and presence of MRI-NAFLD (all P-trend <0.05 except β-cryptoxanthin) (Figure 2, Figure S1 and Tables S3-S6). The odds ratios/hazard ratios (ORs/HRs) [95% confidence intervals (CIs)] for NAFLD in quartile (Q) 4 (vs. Q1) of total carotenoids were 0.20 (0.15, 0.27) for persistent NAFLD, 0.63 (0.50, 0.80) for incident NAFLD, 1.53 (1.10, 2.12) for improved-NAFLD, and 0.58 (0.39, 0.87) for MRI-NAFLD (Figure 2). Among these NAFLD types, the association was most pronounced for persistent NAFLD with a statistical power greater than 0.96. Among individual carotenoids, β-carotene exhibited the strongest associations with all types of NAFLD (Figure 2, Figure S1 and Tables S3-S6).

Figure 2 Associations between serum carotenoids and odds/risks of persistent NAFLD, new-onset NAFLD, improved NAFLD and MRI-identified NAFLD. (A) The odds ratio/hazard ratio (95% confidence interval) of persistent NAFLD, new-onset NAFLD, improved NAFLD, and MRI-identified NAFLD by quartile of serum carotenoids. (B-E) Restricted cubic splines for the association between total serum carotenoids and the odds/risks of persistent NAFLD, new-onset NAFLD, improved NAFLD, and MRI-identified NAFLD, respectively. The blue lines and blue dots represent the estimated effects, while the light blue area and the error bars represent the 95% CIs. Covariates adjusted: age, sex, education, income, physical activity, multivitamin use, smoking, alcohol drinking, and dietary intake of energy, saturated fatty acids, and fiber. NAFLD, non-alcoholic fatty liver disease; OR, odds ratio; CI, confidence interval; HR, hazard ratio; MRI, magnetic resonance imaging.

Gut microbial and metabolites biomarkers of serum carotenoids and NAFLD

The α-diversity indices (observed OTUs, Shannon index, and Chao1 index), were higher in participants with higher serum carotenoid levels (P-trend<0.05) compared to those with NAFLD (Figure 3A and Figure S2). Significant differences were observed in microbial structure (β-diversity) among the quartiles of total carotenoids and between the persistent NAFLD and NAFLD-free groups (P<0.05) (Figure 3B,3C, Figure S2 and Tables S7-S10).

Figure 3 Comparison of gut microbial α-/β-diversity by categories of carotenoids and NAFLD and associations of microbial genera with serum carotenoids and odds/risks of NAFLD. (A) Compares standardized gut microbial α-diversity (observed OTUs, Shannon index, Chao1 index) across different levels of serum total carotenoids and NAFLD groups, with P values derived from multivariable linear or logistic regression. The P value was calculated by multivariable linear or logistic regression. (B,C) β-diversity of gut microbiota by categories of serum total carotenoids (B) and NAFLD (C), analyzed by PERMANOVA (999 permutations) based on Bray-Curtis distance. (D) The associations between the 16 overlapping gut microbial genera and the levels of carotenoids and odds/risk of NAFLD. The left heat map in (D) showed the results of MaAsLin, and the forest plots showed the estimated effects from multivariable logistic regression. Covariates adjusted: age, sex, education, income, physical activity, multivitamin use, smoking, alcohol drinking, dietary intake of energy, saturated fatty acids, fiber, Bristol score, and sequencing parameters. *, P<0.05; **, P<0.01; ***, P<0.001. NAFLD, non-alcoholic fatty liver disease; Q1, quartile 1; Q2, quartile 2; Q3, quartile 3; Q4, quartile 4; SD, standard deviation; OR, odds ratio; CI, confidence interval; HR, hazard ratio; OTU, operational taxonomic unit.

To enhance the sample size, the new-onset-NAFLD and persistent NAFLD groups were combined into the incident/persistent NAFLD group, the statistical power is approximately 0.7 [Supplementary file (Appendix 1)]. Sixteen overlapping differential genera biomarkers, distinguishable between extreme quartiles (Q4 vs. Q1) of serum carotenoids and incident/persistent NAFLD (vs. the NAFLD-free group), were identified within in the Firmicutes, Fusobacteria, and Bacteroidetes phyla (Figure 3D and Table S11). Notably, genera such as Megasphaera, Megamonas, Allisonella, Fusobacterium, and Lachnoclostridium, were positively associated with the odds of incident/persistent NAFLD but inversely associated with serum carotenoids. Conversely, Alistipes, Holdemania, Ruminiclostridium 6, various Ruminococcaceae (uncultured, UCG-002, UCG-003, UCG-005, and UCG-014), Christensenellaceae R-7 group, Faecalibacterium, and Intestinibacter had opposite associations (FDR <0.25). Similar patterns were observed between these genera biomarkers with the risk/odds of new-onset-NAFLD and MRI-identified NAFLD (Figure 3D and Table S12).

Multivariable linear or logistic regression and OPLS-DA were used to identify potential fecal and serum metabolites related to both carotenoids and NAFLD (Figure 4A-4C and Tables S13-S16). A total of 19 fecal metabolites differed significantly between incident/persistent NAFLD and NAFLD-free groups, and between the extreme quartiles of serum total carotenoids, mainly comprising fatty acids, bile acids, and carnitines. Carnitines (oleylcarnitine C18:1 and palmitoylcarnitine), along with the majority of the bile acids (e.g., chenodeoxycholic acid, 7-keto LCA, a-muricholic acid and b-muricholic acid), enriched in participants with lower carotenoids, were associated with higher odds/risk of incident/persistent NAFLD, new-onset-NAFLD, and MRI-identified NAFLD. Conversely, MCFAs (e.g., adipic acid, sebacic acid, suberic acid) exhibited opposite associations (Tables S13,S14).

Figure 4 The associations between metabolites in feces and serum and the levels of serum carotenoids and odds/risks of NAFLD. (A) The heat map indicated the beta coefficients of metabolites in feces and serum by each quartile of carotenoids, analyzed using multivariable linear regression. The forest plots showed the odds/hazard ratios (95% CI) of NAFLD by each SD increase in the metabolites using multivariable logistic regression analyses. *, P<0.05; **, P<0.01; ***, P<0.001. (B,C) The VIP of fecal/serum metabolites biomarkers calculated by OPLS-DA. The orange bars represent metabolites enriched in the persistent/incident NAFLD group, and the green bars represent metabolites enriched in the NAFLD-free group. Covariates adjusted: age, sex, education, income, physical activity, multivitamin use, smoking, alcohol drinking, energy, saturated fatty acids, and fiber intake. DHCA, dehydrocholic acid; NAFLD, non-alcoholic fatty liver disease; VIP, variable importance in the projection; CI, confidence interval; OPLS-DA, orthogonal partial least squares discrimination analysis.

Similarly, sixteen overlapping differential serum metabolites were identified, primarily in amino acids and organic acids. Organic acids (e.g., citramalic acid, oxoglutaric acid, pyruvic acid) and three amino acids (L-glutamic acid, pyroglutamic acid, and L-phenylalanine) showed an inverse correlation to serum carotenoids and a positive association with odds/risk of NAFLD. In contrast, other amino acids, such as glycine, N-acetyl-L-aspartic acid, L-glutamine, exhibited opposite associations (Figure 4A,4C and Tables S15,S16).

Mediation and path analyses results

Carotenoids-related microbial and metabolite scores were constructed from the overlapping differential microbes and metabolites detailed in Tables S12,S14,S16. These scores were utilized in mediation analyses. The analyses revealed that the total carotenoids related scores (TC-MS, TC-FMS, and TC-SMS) mediated the carotenoids-NFALD association by 8.72%, 12.30%, and 16.83% (all P<0.05, Figure 5A-5C) for persistent NAFLD, and by 9.46%, 8.74%, and 15.7% for new-onset-NAFLD (Figure 5D-5F). Notably, the mediation effect of β-cryptoxanthin related microbiome score (BY-MS) and β-carotene related serum metabolite score (BC-SMS) on the association between individual carotenoids and persistent NAFLD were particularly prominent, accounting for 29.68% and 11.49% (Figures S3,S4), respectively.

Figure 5 The mediating effects of TC-MS (A,D), TC-FMS (B,E) and TC-SMS (C,F) on the associations between serum total carotenoids and persistent (A-C)/incident (D-F) NAFLD. The standardized beta coefficients on the lines were calculated by multivariable linear or logistic regression, adjusted for age, sex, education, income, physical activity, multivitamin use, smoking, alcohol drinking, energy, saturated fatty acids, and fiber intake. The number in the triangle is the proportion of mediation effect of the score in the association between carotenoids and NAFLD. TC-MS, total carotenoids-related microbial score; TC-FMS, total carotenoids-related fecal metabolite score; TC-SMS, total carotenoids-related serum metabolite score; NAFLD, non-alcoholic fatty liver disease.

Considering the correlation between gut microbiome, fecal metabolites and serum metabolites (Figure 6A), we further performed the path analysis (Figure 6B,6C and Figures S5,S6). The results indicated that the inverse association between serum total carotenoids and persistent NAFLD was mediated through two pathways. One is directly through serum metabolites, while the other links total carotenoids with microbes, fecal metabolites, and subsequently serum metabolites, each with standardized coefficients of 0.143, 0.45, and 0.13, respectively (Figure 6B,6C). The co-occurrence network based on these regression coefficients illustrated the complex interactions of carotenoids, genera, and metabolites biomarkers and NAFLD (Figure 7).

Figure 6 The heatmap of the correlation between gut microbiome, fecal metabolites and serum metabolites and path analysis results of total carotenoids. (A) The heat map indicated the Spearman coefficients of the gut microbiome, fecal metabolites with serum metabolites. *, P<0.05; **, P<0.01; ***, P<0.001. (B,C) The pathways between the TC-MS, TC-FMS and TC-SMS on the association between serum total carotenoids and persistent/incident NAFLD. The standardized beta coefficients were shown on the lines and the red arrow represents the direction of paths. DHCA, dehydrocholic acid; TC-MS, total carotenoids-related microbial score; TC-FMS, total carotenoids-related fecal metabolite score; TC-SMS, total carotenoids-related serum metabolite score; NAFLD, non-alcoholic fatty liver disease.

Figure 7 The co-occurrence network of carotenoids, gut microbial genera, fecal/serum metabolites biomarkers, and NAFLD. Positive associations are denoted by orange lines and borders, while negative associations are indicated by blue lines. The network elements—microbes (round rectangles), fecal metabolites (ellipse), carotenoids (green diamonds), NAFLD groups (red diamonds), and serum metabolites (triangles)—are differentiated by shape. Border width represents regression coefficient magnitude, while the size of the ellipse or triangle denotes the VIP values. The coefficients were calculated by multivariable linear or logistic regression. Covariates adjusted: age, sex, education, income, physical activity, multivitamin use, smoking, alcohol drinking, energy, saturated fatty acids, and fiber intake. NAFLD, non-alcoholic fatty liver disease; VIP, variable importance in the projection.

Discussion

This 7.8-year prospective study elucidated an inverse association between serum carotenoids and both the incidence and presence of NAFLD. We identified sixteen microbial genera, mainly in the Ruminococcaceae and Veillonellaceae family, nineteen fecal metabolites comprising MCFAs, bile acids, and carnitines, and sixteen serum metabolites belonging to organic acids and amino acids. These findings potentially bridge the link between serum carotenoids and NAFLD.

Carotenoids and NAFLD

Several cross-sectional studies have highlighted beneficial associations between serum carotenoids and NAFLD (6-8). The Japanese Mikkabi cohort also observed inverse associations between baseline serum β-carotene and β-cryptoxanthin with serum ALT over a 7.4-year follow-up period (36). Additionally, a 12-week RCT involving 92 Iranian NAFLD outpatients demonstrated that β-cryptoxanthin supplementation significantly improved hepatic steatosis and ALT levels (10). Consistent with the previous findings, this study supported the inverse associations of serum carotenoids with ULS-NAFLD incidence and persistence over a 7.8-year follow-up and replicated the association using MRI-NAFLD. The protective effects of carotenoids to NAFLD may be attributed to their role in enhancing insulin sensitivity (37-39) and ameliorating hepatic oxidative stress status (39-41), reducing inflammatory biomarkers (10,39), and regulating macrophage polarization (42). Our study demonstrated a positive correlation between higher serum carotenoid levels and the reduction of new-onset NAFLD as well as improvements in existing cases. Significantly, β-carotene and α-carotene showed a more marked beneficial impact than lycopene and β-cryptoxanthin, highlighting their crucial role in NAFLD prevention and management.

Associations between carotenoids, gut microbiota, and NAFLD

The gut-liver axis plays a pivotal role in the pathogenesis of NAFLD (13,14). Earlier studies have noted that NAFLD is associated with an increased abundance of Lachnospiraceae, Veillonellaceae, Porphyromonadaceae families and Clostridium, Dorea genera, but a decreased abundance of Christensenellale, Bifidobacteriacea, Ruminococcaceae families and Faecalibacterium, Bifidobacterium genera (15-17,43-45). Observational studies and RCTs have reported that higher serum carotenoids are associated with an increased abundance of beneficial bacteria (i.e., Ruminococcaceae UCG-002, Lachnospiraceae NC2004 group, Ruminococcus 1 genera) (22-24). In line with most previous research, our study found that higher serum carotenoids were associated with a higher abundance of 11 genera (e.g., from the Ruminococcaceae family, Alistipes, Ruminiclostridium 6) but with a lower abundance of 5 genera (Megasphaera, Megamonas, Allisonella, Fusobacterium, and Lachnoclostridium) that overall were linked to a decreased incidence and persistence of NAFLD. Ruminococcaceae is (46,47) a well-known butyrate-producing bacterium, and butyrate can down-regulate pro-inflammatory factors and alleviate adipose tissue inflammation in NAFLD (46,48). Our previous analyses also suggested that Ruminococcaceae UCG-002 and UCG-003 might protect against metabolic syndrome and diabetes by reducing muro-cholic acid (49). In addition, three opportunistic pathogens (Megamonas, Megasphaera and Allisonella genera) belonging to the Veillonellaceae family, found to be enriched in NAFLD patients in this study, have been previously associated with cardio-metabolic conditions such as T2DM (50) and obesity (51).

Associations between carotenoids, metabolites, and NAFLD

Metabolome analyses in our study revealed significant associations between various metabolites and NAFLD. Specifically, higher levels of fecal bile acids (predominantly secondary bile acids) and carnitines, along with serum organic acids (e.g., citramalic acid, oxoglutaric acid) and certain amino acids (e.g., L-glutamic acid, pyroglutamic acid) correlated with lower serum carotenoids and increased odds/risk of NAFLD, including MRI-NAFLD. Conversely, lower levels of fecal MCFAs and some serum amino acids (e.g., N-acetyl-L-aspartic acid, glycine) were associated with higher carotenoid levels and reduced NAFLD risk.

These metabolic associations align with findings from numerous studies linking similar fecal and circulating metabolites with cardio-metabolic conditions. MCFAs (C₈-C₁₀) known as peroxisome proliferator-activated receptor-γ activators, have been shown to play beneficial roles in promoting insulin sensitization, glucose-stimulated insulin secretion, and adipogenesis (52,53). Secondary bile acids such as muro-cholic acid (an antagonist) and chenodeoxycholic acid (an agonist), act as ligands for the nuclear receptor farnesoid X receptor and Takeda-G-protein-receptor-5 (54,55). The enrichment of genera like Fusobacterium and Lachnoclostridium in NAFLD patients (56) is linked to increased levels of secondary bile acids, potentially leading to dysregulated hepatic lipid and carbohydrate metabolism.

In terms of serum amino acids, glycine, L-glutamate, and L-serine are precursors for glutathione synthesis. L-glutamic acid, a precursor of L-glutamate, and pyroglutamic acid may accumulate in response to oxidative stress (57). Moreover, various organic acids, produced through metabolic pathways such as glycolysis, the citric acid cycle, fatty acids oxidation, are implicated in NAFLD. Intermediates like L-Lactic acid, oxoglutaric acid, and pyruvic acid are essential in gluconeogenesis, a key pathway in hepatic glucose production. Impaired hepatic glucose production is indicative of hepatic insulin resistance, potentially explaining their strong positive association with NAFLD (58). Furthermore, the enrichment of carnitine like oleylcarnitine C18:1 and palmitoylcarnitine in NAFLD patients might suggest a link to reduced hepatic mitochondrial fatty acid β-oxidation, thereby increasing NAFLD risk (58,59). Interestingly, our findings observed that serum metabolites exhibited a stronger association with health outcomes than fecal metabolites. This could be attributed to the direct involvement of serum metabolites in systemic circulation and metabolic processes. In contrast, fecal metabolites, being excretory products, primarily reflect gut activity, impact the gut barrier, immunity, and gut metabolites (14) but have a less direct role in systemic metabolism. Existing literature suggests a link (60), and occasionally causality (61), between gut microbiota and blood metabolites. Our pathway analysis results also provide the possibility that intestinal flora affects fecal metabolites and then affects serum metabolites, which warrants further investigation.

Taken together, serum carotenoids appear to influence gut microbiota composition, promoting beneficial bacteria like the Ruminococcaceae family and reducing pathogenic bacteria such as Megamonas, Megasphaera, and Allisonella from the Veillonellaceae family. This alteration leads to an increase in MCFAs and a decrease in secondary bile acids and carnitine in fecal metabolites. Concurrently, changes occur in serum metabolites, including amino acids and organic acids, triggering a cascade of insulin sensitization promotion, improved hepatic lipid and carbohydrate metabolism, reducing mitochondrial fatty acid β-oxidation and oxidative stress, ultimately lowering the risk of developing NAFLD.

Strengths and limitations

This study has several strengths. To our best knowledge, it is the first prospective study to explore the influence of the gut-liver axis in interpreting the associations between serum carotenoids and NAFLD in a larger longitudinal cohort. The robustness of our findings is underscored by the consistent associations of the identified biomarkers with both the odds of 7.8-year persistent NAFLD and the risk of 7.8-year incident ULS-NAFLD. Furthermore, the associations between carotenoids and biomarkers with NAFLD were also substantiated in MRI-NAFLD within a sub-sample of 828 participants.

Some limitations warrant attention. Firstly, our study measured serum carotenoids only at baseline. However, most participants (73.2%) in our study exhibited minimal changes (within one quartile) in dietary carotenoid intake over six years. These intake changes showed no significant variation among different NAFLD groups. Previous study (62) has also reported relatively high levels of intraclass correlations (0.63–0.85) of serum carotenoid concentrations over periods of 3 years. Our analyses also suggest that the associations tended to be attenuated, they are unlikely to significantly distort the observed association over time. Secondly, due to constraints on biospecimens and budget, not all variables were measured at the same time point. Despite this, our study employed a prospective cohort design for analyzing the association between baseline carotenoids and NAFLD progression. Additionally, we conducted a confirmatory evaluation of the relationship between baseline carotenoids and MRI-NAFLD odds during the fourth follow-up and incorporated a longitudinal approach in assessing the roles of omics biomarkers as mediators between baseline carotenoids and NAFLD progression, aligning with the temporal relationship from exposure to mediator and then to outcome. Third, since our study was focused on a single city, the microbial and metabolite biomarkers we identified needed to be cautiously generalized to other populations.

Conclusions

Our study’s findings illuminate a significant inverse association between most serum carotenoids and both the persistence and incidence of NAFLD. Key biomarkers, such as gut microbial Ruminococcaceae and Veillonellaceae family, fecal MCFAs and bile acids, as well as serum amino acids and organic acids, appear to mediate this beneficial association. The present study provides robust evidence from a multi-omics perspective, particularly focusing on the gut-liver axis, to elucidate the mechanisms underlying the epidemiological associations between carotenoids and NAFLD. By bridging this gap, our research contributes to a deeper understanding of the potential roles of serum carotenoids in managing and preventing NAFLD.

Acknowledgments

We thank all study participants of the Guangzhou Nutrition and Health Study. We thank all other team members involved in the cohort study and data analyses.

Funding: This work was supported jointly by the National Natural Science Foundation of China (grants 82073546, 81773416 to Y.M.C., and 82073529 to J.S.Z.); the Key Research and Development Program of Guangzhou, China (grant 202007040003 to Y.M.C.); and the 5010 Program for Clinical Researches of the Sun Yat-sen University (Guangzhou, China) (grant 2007032 to Y.M.C.). The funding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://hbsn.amegroups.com/article/view/10.21037/hbsn-23-526/rc

Data Sharing Statement: Available at https://hbsn.amegroups.com/article/view/10.21037/hbsn-23-526/dss

Peer Review File: Available at https://hbsn.amegroups.com/article/view/10.21037/hbsn-23-526/prf

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://hbsn.amegroups.com/article/view/10.21037/hbsn-23-526/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The trial was registered at https://www.clinicaltrials.gov/ with the identifier: NCT03179657. The study protocol of GNHS was approved by the Ethics Committee of the School of Public Health at Sun Yat-sen University (2018048), and informed consent was obtained from all individual participants.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Younossi ZM, Golabi P, Paik JM, et al. The global epidemiology of nonalcoholic fatty liver disease (NAFLD) and nonalcoholic steatohepatitis (NASH): a systematic review. Hepatology 2023;77:1335-47. [Crossref] [PubMed]
2. Powell EE, Wong VW, Rinella M. Non-alcoholic fatty liver disease. Lancet 2021;397:2212-24. [Crossref] [PubMed]
3. Paik JM, Golabi P, Younossi Y, et al. Changes in the Global Burden of Chronic Liver Diseases From 2012 to 2017: The Growing Impact of NAFLD. Hepatology 2020;72:1605-16. [Crossref] [PubMed]
4. Toh JZK, Pan XH, Tay PWL, et al. A Meta-Analysis on the Global Prevalence, Risk factors and Screening of Coronary Heart Disease in Nonalcoholic Fatty Liver Disease. Clin Gastroenterol Hepatol 2022;20:2462-2473.e10. [Crossref] [PubMed]
5. Rodriguez-Concepcion M, Avalos J, Bonet ML, et al. A global perspective on carotenoids: Metabolism, biotechnology, and benefits for nutrition and health. Prog Lipid Res 2018;70:62-93. [Crossref] [PubMed]
6. Cao Y, Wang C, Liu J, et al. Greater serum carotenoid levels associated with lower prevalence of nonalcoholic fatty liver disease in Chinese adults. Sci Rep 2015;5:12951. [Crossref] [PubMed]
7. Christensen K, Lawler T, Mares J. Dietary Carotenoids and Non-Alcoholic Fatty Liver Disease among US Adults, NHANES 2003⁻2014. Nutrients 2019;11:1101. [Crossref] [PubMed]
8. Liu X, Shen H, Chen M, et al. Clinical Relevance of Vitamins and Carotenoids With Liver Steatosis and Fibrosis Detected by Transient Elastography in Adults. Front Nutr 2021;8:760985. [Crossref] [PubMed]
9. Chai W, Eaton S, Rasmussen HE, et al. Associations of Dietary Lipid-Soluble Micronutrients with Hepatic Steatosis among Adults in the United States. Biomedicines 2021;9:1093. [Crossref] [PubMed]
10. Haidari F, Hojhabrimanesh A, Helli B, et al. An energy-restricted high-protein diet supplemented with β-cryptoxanthin alleviated oxidative stress and inflammation in nonalcoholic fatty liver disease: a randomized controlled trial. Nutr Res 2020;73:15-26. [Crossref] [PubMed]
11. Xiao ML, Chen GD, Zeng FF, et al. Higher serum carotenoids associated with improvement of non-alcoholic fatty liver disease in adults: a prospective study. Eur J Nutr 2019;58:721-30. [Crossref] [PubMed]
12. Tilg H, Adolph TE, Trauner M. Gut-liver axis: Pathophysiological concepts and clinical implications. Cell Metab 2022;34:1700-18. [Crossref] [PubMed]
13. Tilg H, Adolph TE, Moschen AR. Multiple Parallel Hits Hypothesis in Nonalcoholic Fatty Liver Disease: Revisited After a Decade. Hepatology 2021;73:833-42. [Crossref] [PubMed]
14. Sharpton SR, Ajmera V, Loomba R. Emerging Role of the Gut Microbiome in Nonalcoholic Fatty Liver Disease: From Composition to Function. Clin Gastroenterol Hepatol 2019;17:296-306. [Crossref] [PubMed]
15. Ruuskanen MO, Åberg F, Männistö V, et al. Links between gut microbiome composition and fatty liver disease in a large population sample. Gut Microbes 2021;13:1-22. [Crossref] [PubMed]
16. Hoyles L, Fernández-Real JM, Federici M, et al. Molecular phenomics and metagenomics of hepatic steatosis in non-diabetic obese women. Nat Med 2018;24:1070-80. [Crossref] [PubMed]
17. Li F, Sun G, Wang Z, et al. Characteristics of fecal microbiota in non-alcoholic fatty liver disease patients. Sci China Life Sci 2018;61:770-8. [Crossref] [PubMed]
18. Sharpton SR, Maraj B, Harding-Theobald E, et al. Gut microbiome-targeted therapies in nonalcoholic fatty liver disease: a systematic review, meta-analysis, and meta-regression. Am J Clin Nutr 2019;110:139-49. [Crossref] [PubMed]
19. Ji Y, Yin Y, Sun L, et al. The Molecular and Mechanistic Insights Based on Gut-Liver Axis: Nutritional Target for Non-Alcoholic Fatty Liver Disease (NAFLD) Improvement. Int J Mol Sci 2020;21:3066. [Crossref] [PubMed]
20. Lyu Y, Wu L, Wang F, et al. Carotenoid supplementation and retinoic acid in immunoglobulin A regulation of the gut microbiota dysbiosis. Exp Biol Med (Maywood) 2018;243:613-20. [Crossref] [PubMed]
21. Umoh FI, Kato I, Ren J, et al. Markers of systemic exposures to products of intestinal bacteria in a dietary intervention study. Eur J Nutr 2016;55:793-8. [Crossref] [PubMed]
22. Schmidt KM, Haddad EN, Sugino KY, et al. Dietary and plasma carotenoids are positively associated with alpha diversity in the fecal microbiota of pregnant women. J Food Sci 2021;86:602-13. [Crossref] [PubMed]
23. Frankenfeld CL, Hullar MAJ, Maskarinec G, et al. The Gut Microbiome Is Associated with Circulating Dietary Biomarkers of Fruit and Vegetable Intake in a Multiethnic Cohort. J Acad Nutr Diet 2022;122:78-98. [Crossref] [PubMed]
24. Wiese M, Bashmakov Y, Chalyk N, et al. Prebiotic Effect of Lycopene and Dark Chocolate on Gut Microbiome with Systemic Changes in Liver Metabolism, Skeletal Muscles and Skin in Moderately Obese Persons. Biomed Res Int 2019;2019:4625279. [Crossref] [PubMed]
25. Li CC, Liu C, Fu M, et al. Tomato Powder Inhibits Hepatic Steatosis and Inflammation Potentially Through Restoring SIRT1 Activity and Adiponectin Function Independent of Carotenoid Cleavage Enzymes in Mice. Mol Nutr Food Res 2018;62:e1700738. [Crossref] [PubMed]
26. Wang X, Chen X, Zhang H, et al. Circulating retinol-binding protein 4 is associated with the development and regression of non-alcoholic fatty liver disease. Diabetes Metab 2020;46:119-28. [Crossref] [PubMed]
27. Zhang CX, Ho SC. Validity and reproducibility of a food frequency Questionnaire among Chinese women in Guangdong province. Asia Pac J Clin Nutr 2009;18:240-50. [PubMed]
28. Liu B, Woo J, Tang N, et al. Assessment of total energy expenditure in a Chinese population by a physical activity questionnaire: examination of validity. Int J Food Sci Nutr 2001;52:269-82. [Crossref] [PubMed]
29. Burri BJ, Dopler-Nelson M, Neidllinger TR. Measurements of the major isoforms of vitamins A and E and carotenoids in the blood of people with spinal-cord injuries. J Chromatogr A 2003;987:359-66. [Crossref] [PubMed]
30. Zeng MD, Fan JG, Lu LG, et al. Guidelines for the diagnosis and treatment of nonalcoholic fatty liver diseases. J Dig Dis 2008;9:108-12. [Crossref] [PubMed]
31. Schaapman JJ, Tushuizen ME, Coenraad MJ, et al. Multiparametric MRI in Patients With Nonalcoholic Fatty Liver Disease. J Magn Reson Imaging 2021;53:1623-31. [Crossref] [PubMed]
32. Caporaso JG, Kuczynski J, Stombaugh J, et al. QIIME allows analysis of high-throughput community sequencing data. Nat Methods 2010;7:335-6. [Crossref] [PubMed]
33. DeSantis TZ, Hugenholtz P, Larsen N, et al. Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl Environ Microbiol 2006;72:5069-72. [Crossref] [PubMed]
34. Ling CW, Miao Z, Xiao ML, et al. The Association of Gut Microbiota With Osteoporosis Is Mediated by Amino Acid Metabolism: Multiomics in a Large Cohort. J Clin Endocrinol Metab 2021;106:e3852-64. [Crossref] [PubMed]
35. Tingley D, Yamamoto T, Hirose K, et al. mediation: R Package for Causal Mediation Analysis. Journal of Statistical Software 2014;59:1-38. [Crossref]
36. Sugiura M, Nakamura M, Ogawa K, et al. High serum carotenoids are associated with lower risk for developing elevated serum alanine aminotransferase among Japanese subjects: the Mikkabi cohort study. Br J Nutr 2016;115:1462-9. [Crossref] [PubMed]
37. Qiu X, Gao DH, Xiang X, et al. Ameliorative effects of lutein on non-alcoholic fatty liver disease in rats. World J Gastroenterol 2015;21:8061-72. [Crossref] [PubMed]
38. Harari A, Coster ACF, Jenkins A, et al. Obesity and Insulin Resistance Are Inversely Associated with Serum and Adipose Tissue Carotenoid Concentrations in Adults. J Nutr 2020;150:38-46. [Crossref] [PubMed]
39. Kim JE, Clark RM, Park Y, et al. Lutein decreases oxidative stress and inflammation in liver and eyes of guinea pigs fed a hypercholesterolemic diet. Nutr Res Pract 2012;6:113-9. [Crossref] [PubMed]
40. Lee Y, Hu S, Park YK, et al. Health Benefits of Carotenoids: A Role of Carotenoids in the Prevention of Non-Alcoholic Fatty Liver Disease. Prev Nutr Food Sci 2019;24:103-13. [Crossref] [PubMed]
41. Matsuura B, Miyake T, Yamamoto S, et al. Usefulness of beta-cryptoxanthin for nonalcoholic fatty liver diseases. J Food Nutr Disord 2016; [Crossref]
42. Ni Y, Zhuge F, Nagashimada M, et al. Novel Action of Carotenoids on Non-Alcoholic Fatty Liver Disease: Macrophage Polarization and Liver Homeostasis. Nutrients 2016;8:391. [Crossref] [PubMed]
43. Iino C, Endo T, Mikami K, et al. Significant decrease in Faecalibacterium among gut microbiota in nonalcoholic fatty liver disease: a large BMI- and sex-matched population study. Hepatol Int 2019;13:748-56. [Crossref] [PubMed]
44. Hullar MAJ, Jenkins IC, Randolph TW, et al. Associations of the gut microbiome with hepatic adiposity in the Multiethnic Cohort Adiposity Phenotype Study. Gut Microbes 2021;13:1965463. [Crossref] [PubMed]
45. Astbury S, Atallah E, Vijay A, et al. Lower gut microbiome diversity and higher abundance of proinflammatory genus Collinsella are associated with biopsy-proven nonalcoholic steatohepatitis. Gut Microbes 2020;11:569-80. [Crossref] [PubMed]
46. Chang PV, Hao L, Offermanns S, et al. The microbial metabolite butyrate regulates intestinal macrophage function via histone deacetylase inhibition. Proc Natl Acad Sci U S A 2014;111:2247-52. [Crossref] [PubMed]
47. Jiang Z, Zhuo LB, He Y, et al. The gut microbiota-bile acid axis links the positive association between chronic insomnia and cardiometabolic diseases. Nat Commun 2022;13:3002. [Crossref] [PubMed]
48. Hold GL, Schwiertz A, Aminov RI, et al. Oligonucleotide probes that detect quantitatively significant groups of butyrate-producing bacteria in human feces. Appl Environ Microbiol 2003;69:4320-4. [Crossref] [PubMed]
49. Watanabe M, Sianoya A, Mishima R, et al. Gut microbiome status of urban and rural Filipino adults in relation to diet and metabolic disorders. FEMS Microbiol Lett 2021;368:fnab149. [Crossref] [PubMed]
50. Duan M, Wang Y, Zhang Q, et al. Characteristics of gut microbiota in people with obesity. PLoS One 2021;16:e0255446. [Crossref] [PubMed]
51. Huang L, Gao L, Chen C. Role of Medium-Chain Fatty Acids in Healthy Metabolism: A Clinical Perspective. Trends Endocrinol Metab 2021;32:351-66. [Crossref] [PubMed]
52. Muthulakshmi S, Chakrabarti AK, Mukherjee S. Gene expression profile of high-fat diet-fed C57BL/6J mice: in search of potential role of azelaic acid. J Physiol Biochem 2015;71:29-42. [Crossref] [PubMed]
53. Song Z, Cai Y, Lao X, et al. Taxonomic profiling and populational patterns of bacterial bile salt hydrolase (BSH) genes based on worldwide human gut microbiome. Microbiome 2019;7:9. [Crossref] [PubMed]
54. Jiao N, Baker SS, Chapa-Rodriguez A, et al. Suppressed hepatic bile acid signalling despite elevated production of primary and secondary bile acids in NAFLD. Gut 2018;67:1881-91. [Crossref] [PubMed]
55. Masoodi M, Gastaldelli A, Hyötyläinen T, et al. Metabolomics and lipidomics in NAFLD: biomarkers and non-invasive diagnostic tests. Nat Rev Gastroenterol Hepatol 2021;18:835-56. [Crossref] [PubMed]
56. Chávez-Talavera O, Tailleux A, Lefebvre P, et al. Bile Acid Control of Metabolism and Inflammation in Obesity, Type 2 Diabetes, Dyslipidemia, and Nonalcoholic Fatty Liver Disease. Gastroenterology 2017;152:1679-1694.e3. [Crossref] [PubMed]
57. Petersen MC, Vatner DF, Shulman GI. Regulation of hepatic glucose metabolism in health and disease. Nat Rev Endocrinol 2017;13:572-87. [Crossref] [PubMed]
58. Bjørndal B, Alterås EK, Lindquist C, et al. Associations between fatty acid oxidation, hepatic mitochondrial function, and plasma acylcarnitine levels in mice. Nutr Metab (Lond) 2018;15:10. [Crossref] [PubMed]
59. Berge RK, Bjørndal B, Strand E, et al. Tetradecylthiopropionic acid induces hepatic mitochondrial dysfunction and steatosis, accompanied by increased plasma homocysteine in mice. Lipids Health Dis 2016;15:24. [Crossref] [PubMed]
60. Vojinovic D, Radjabzadeh D, Kurilshikov A, et al. Relationship between gut microbiota and circulating metabolites in population-based cohorts. Nat Commun 2019;10:5813. [Crossref] [PubMed]
61. Liu X, Tong X, Zou Y, et al. Mendelian randomization analyses support causal relationships between blood metabolites and the gut microbiome. Nat Genet 2022;54:52-61. [Crossref] [PubMed]
62. Toniolo P, Van Kappel AL, Akhmedkhanov A, et al. Serum carotenoids and breast cancer. Am J Epidemiol 2001;153:1142-7. [Crossref] [PubMed]