NAFLD frequently co-occurs with a spectrum of metabolic comorbidities, including type 2 diabetes (T2DM), obesity, insulin resistance, dyslipidaemia, and hypertension. These conditions form a network with shared pathophysiological mechanisms and overlapping gut microbiota features (Quek et al., 2023). This overlap complicates the identification of NAFLD-specific microbial changes (Jarvis et al., 2020). For example, T2DM is associated with increased abundance of lipopolysaccharide (LPS)-producing genera (e.g., Enterobacter) and reduction in butyric acid-producing bacteria (e.g., Faecalibacterium); this pattern is also seen in patients with NAFLD, making it difficult to isolate microbiota signatures unique to NAFLD (Ojo et al., 2021; Yan et al., 2022). The synergistic role of T2DM in accelerating NAFLD progression was reported in large multicentre cohort studies, where patients with T2DM and NAFLD experienced faster progression of hepatic fibrosis and showed a higher risk of severe hepatic events than their non-T2DM counterparts (Huang et al., 2023b; Jarvis et al., 2020).

The Swedish Malmö cohort study (n = 12,548) identified low high-density lipoprotein cholesterol (< 1.0 mmol/L) and high triglycerides (TG≥1.7 mmol/L) levels as independent risk factors for severe hepatic events, with hazard ratios of 1.28 and 1.30, respectively (Nderitu et al., 2017). These metabolic disturbances may partly act through the gut microbiota (Li et al., 2023). For instance, Bifidobacterium lactis BL-99 supplementation improves lipid profiles and modulates bile acid metabolism by increasing the production of short-chain fatty acids (SCFAs) (Bajaj et al., 2018; Manaer et al., 2024).

Most NAFLD microbiome studies have insufficiently controlled for metabolic comorbidities. Few studies have adjusted for T2DM, and even fewer have accounted for other metabolic factors, such as dyslipidaemia (Loomba et al., 2019). This oversight can result in the misclassification of microbial markers, such as incorrectly attributing T2DM-related depletion in Akkermansia to NAFLD (Wong et al., 2013; Sanna et al., 2019; Dao et al., 2016).

Most previous studies have had a cross-sectional design, making it difficult to determine whether the microbial changes are a driver of NAFLD or a secondary outcome of metabolic comorbidities. Additionally, the majority of available cohort studies have focused on middle-aged and elderly populations, limiting the generalizability of the results. Future investigations should prioritize the implementation of large, cross-regional study cohorts to encompass a wider range of patients with NAFLD, particularly those with multiple comorbid metabolic disorders (Neeland et al., 2024) as well as adopt large, geographically diverse cohorts encompassing a broader demographic and comorbidity spectrum (Prado et al., 2024).

Personalized microbiota-based interventions, such as tailored probiotic or prebiotic blends (e.g., Bifidobacterium animalis subsp. lactis BL-99) for patients with T2DM-NAFLD, could improve lipid profiles and inhibit the progression of liver fibrosis. These strategies should be validated through randomized controlled trials to confirm their efficacy and safety.

Obesity is a major risk factor for NAFLD; however, approximately 20% of the patients are lean, highlighting the phenotypic heterogeneity that complicates the elucidation of gut microbiota characteristics. Obesity plays a central role in driving disease heterogeneity through a distinct ‘lipid-flora-liver axis' mechanism (Nogacka et al., 2021; Leong et al., 2020; Zou et al., 2020; Chambers et al., 2019; Depommier et al., 2019; Mokkala et al., 2021). Gut microbiota profiles differ between lean and patients with obesity and NAFLD. Lean individuals exhibit a 1.5–2.3-fold increase in the abundance of Akkermansia muciniphila and depletion of Firmicutes/Bacteroidetes ratio (p < 0.05) (Kivimäki et al., 2022; Raman et al., 2013). In contrast, patients with obesity often show an increase in the relative abundance of LPS-producing bacteria genera, such as Enterobacter, and a depletion of butyric acid-producing bacteria, such as Faecalibacterium prausnitzii (Mocanu et al., 2021; Ghorbani et al., 2023; Zhang Z. et al., 2024). Of note, the abundance of Firmicutes/Bacteroidetes ratio peaks at a BMI of 33 kg/m² and then declines (Haro et al., 2016). The non-linear relationship varies by sex; for instance, Mycobacterium avium decreases significantly with increasing BMI in men but not women (Kasai et al., 2015).

Bariatric surgery studies offer further insights into microbiota-host interactions. Postoperative increment in A. muciniphila (2.1–3.4-fold) is correlated with reduced hepatic adiposity (r = −0.47, p = 0.002) (Jian et al., 2022). However, higher baseline microbiota alpha diversity is associated with reduced postoperative fat loss (p = 0.016), indicating that diversity does not always confer metabolic benefits (Foster et al., 2017). This complexity reflects the influence of obesity-related factors such as insulin resistance and inflammation on microbiota function (Li et al., 2021). Moreover, Dong et al. (2023) reported that sleeve gastrectomy resulted in reduced glucose-dependent insulinotropic polypeptide signaling, conferring resistance to NAFLD. This effect was linked to elevated levels of A. muciniphila and changes in indolepropionic acid, a bacterial tryptophan metabolite. Bariatric surgery also alters ileal physiology, affecting bile acid composition independently of weight loss (Talavera-Urquijo et al., 2020).

Many previous studies have failed to adequately control for obesity-related confounders. BMI is a unidimensional indicator and may mask the effects of body fat distribution on the microbiome. Visceral fat area, for example, shows a stronger correlation with LPS-producing species than BMI (Nie et al., 2020; Yan et al., 2021). Future studies should incorporate precise indicators of body fat distribution and combine multidimensional data, such as computed tomography-measured visceral fat and metabolic markers (e.g., homeostasis model assessment of insulin resistance), to better define obesity-related microbiome changes in NAFLD.

Most microbiome studies have focused on bacteria, overlooking fungi and phages. Longitudinal studies are needed to assess microbiome remodeling after weight loss surgeries or interventions in different obesity phenotypes, emphasizing key taxa such as A. muciniphila. Additionally, the sex hormone-microbiota-hepatic steatosis regulatory axis should be examined separately for male and female patients with obesity to identify sex-specific therapeutic targets.

The progression of NAFLD, from simple steatosis to NASH to liver fibrosis, cirrhosis, and hepatocellular carcinoma (HCC), significantly alters gut microbiota composition and function (Ponziani et al., 2019). Studies have shown that NAFLD/NASH is associated with an increased abundance of Firmicutes/Bacteroidetes ratio (1.5–2.3-fold) and reduced microbiome diversity (Shannon's index, decrease of 0.8–1.2, p < 0.05), along with shifts in key functional genera (Boursier et al., 2016; Vallianou et al., 2021).

In the Rotterdam cohort study, 472 out of 1,355 participants were diagnosed with hepatic steatosis, and their microbiotas were characterized by elevated levels of Coprococcus (fecal cocci) and Ruminococcus, with reduced microbial diversity (Alferink et al., 2021). Zhang et al. (2021) reported that a high-cholesterol diet promoted NASH progression, with an increase in Mucispirillum, Desulfovibrio, and Anaerotruncus and depletion of Bifidobacterium and Bacteroides. Hoyles et al. (2018) reported an increase in Acidaminococcus, Escherichia coli, and Bacteroides and depletion of Faecalibacterium and Anaerobacterium in patients with NASH.

Shifts in the fungal profile also mark disease progression. A study of 69 patients with NAFLD showed that non-obese patients had a significantly increased proportion of Trichoderma/Saccharomyces cerevisiae (wood mold/sterile yeasts). This feature holds potential value for distinguishing between mild and advanced liver disease (Demir et al., 2022). As NAFLD progresses to cirrhosis and HCC, microbiota profiles evolve. Loomba et al. (2021) reported the predominance of Firmicutes in patients with NAFLD/NASH, and an increase in Tenericutes abundance in patients with advanced fibrosis. Ponziani et al. (2019) reported increments in Enterobacteriaceae and Streptococcus spp. and depletion of A. muciniphila in patients with NASH-cirrhosis. Behary et al. (2021) identified enriched species composition in patients with NAFLD-HCC, including Bacteroides cecum, Ruminococcus gnavus, Veillonella parvula, Bacteroides xylanisolvens, and Clostridium bolteae, that may promote tumorigenesis via the Toll-like receptor 4 (TLR4)/nuclear factor kappa-B (NF-κB) pathway. These changes are summarized in Figure 2.

Virome studies further reveal changes in enterovirus composition in patients with NAFLD. A cross-sectional study of 73 patients with NAFLD found reduced numbers of Lactococcus and Lactobacillus phages and an increased number of phiAT3 phages (Lang et al., 2020). In addition, Chen et al. (2016) reported that viral diversity differed significantly between patients with NAFLD, different disease stages, and study and control groups and that the presence of enterovirus was correlated with NAFLD severity. Figure 2 highlights the unique microbial profiles associated with the different stages of liver disease, emphasizing the potential role of the gut microbiome in disease progression.

Currently available non-invasive diagnostic tools struggle to distinguish NAFLD stages, leading to ambiguous links between microbiota and pathology. This ambiguity is compounded by the fact that most studies have focused on bacterial groups, with limited attention to fungi and viruses. Small sample sizes and lack of multicentre validation further limit generalizability.

Future research should pursue a multimodal diagnostic platform integrating liver spatial transcriptomics with fecal virome sequencing. Developing new biomarker combinations could improve diagnostic accuracy. Understanding the proinflammatory synergy between fungi and bacteria in patients with NASH is also essential. Macrogenomic co-occurrence analysis may clarify these interactions. Monitoring phage dynamics in high-risk patients with NAFLD-HCC may offer early warning markers for timely intervention.

Diet strongly influences gut microbiota. High-fat and highly processed foods can induce dysbiosis, which is closely linked to NAFLD onset and progression (Pi et al., 2024). Variability in dietary habits across populations contributes to heterogeneity in microbiota results. This section examines the effects of alcohol, nicotine, substance use, and circadian rhythms on the NAFLD-microbiota relationship.

The relationship between alcohol and NAFLD is complex. Although NAFLD excludes excessive alcohol consumption by definition, the impact of light-to-moderate alcohol consumption on NAFLD remains a focus of research (Hashimoto et al., 2015; Wongtrakul et al., 2021). Most studies suggest that low-to-moderate alcohol consumption may reduce NAFLD risk and progression (Hamaguchi and Kojima, 2013). For example, Ajmera et al. (2018) reported less improvement in steatosis and NASH among moderate drinkers than in abstainers. A meta-analysis of 8,936 participants showed a lower risk of advanced liver fibrosis in moderate drinkers with NAFLD (Wijarnpreecha et al., 2021). However, the findings are inconsistent. A study of 132 bariatric surgery patients found no association between alcohol intake and histological severity (Cotrim et al., 2009). This inconsistency may be related to small sample size and unadjusted confounders such as genetics, diet, and lifestyle.

Alcohol also affects gut microbiota. Low-to-moderate alcohol consumption can shift gut microbial communities, including those that produce SCFAS (Morrison and Preston, 2016). Caslin et al. (2019) reported higher A. muciniphila levels among moderate alcohol consumers, reducing the severity of experimental autoimmune encephalomyelitis. Alcohol may also disrupt gut barrier integrity and gut–brain signaling (de Timary et al., 2015). Moreover, alcohol influences the gut virome. Hsu et al. (2023) reported depletion of Lactobacillus phages and increased bIL67 in patients with NAFLD with moderate alcohol consumption compared to those with no or low consumption of alcohol.

These findings suggest that alcohol influences NAFLD via multiple microbiota-related pathways, although the mechanisms remain unclear. Current studies have focused on low-to-moderate alcohol intake, with limited data on heavy drinking. Future large-scale longitudinal studies should quantify alcohol consumption gradients and correlate them with microbial ethanol production through metabolomics to define thresholds for alcohol-induced hepatotoxicity.

Nicotine, the primary component of cigarettes, has a strong influence on gut microbiota composition and NAFLD progression. Smoking is strongly associated with oral and gut dysbiosis, marked by increased Prevotella abundance and reduced microbial diversity (Wu et al., 2016). Mechanistically, nicotine synergizes with a high-fat diet to exacerbate hepatic and muscular steatosis, potentially via enhanced abdominal lipolysis and altered lipid metabolism (Sinha-Hikim et al., 2014). In smokers, the gut microbiota shifts include elevated Prevotella, decreased Bacteroidetes, and lower Shannon diversity—partially reversible after cessation (Stewart et al., 2018). Smoking also increases intestinal permeability and alters luminal pH, promoting pathogenic colonization (Chen J. et al., 2024; Su et al., 2025) and disrupting the GLA, leading to hepatic lipid buildup and insulin resistance (Miao et al., 2023; Sun and Debosch, 2023).

Emerging evidence links electronic cigarettes to microbiota changes, with increased Prevotella abundance and depletion of Anaerobacter, despite stable overall diversity (Stewart et al., 2018). A 31-year study found a 99% higher NAFLD risk in individuals who were persistently exposed to second-hand smoke; further, heavy smoking (>10 pack-years) has been correlated with higher all-cause mortality in women (Liu et al., 2013; Charatcharoenwitthaya et al., 2020; Wu F. et al., 2021). Bacteroides xylanisolvens may mitigate NAFLD by reducing intestinal nicotine bioavailability (Chen et al., 2022).

These findings highlight the pathogenic link between nicotine exposure, microbiota dysregulation, and NAFLD, although the mechanisms remain unclear. Preclinical studies suggest the therapeutic potential of probiotics such as B. xylanisolvens in countering nicotine-associated NAFLD (Lee et al., 2018), but randomized controlled trials and mechanistic animal studies are needed. Current research is limited by poor quantification of nicotine exposure, confounding by other tobacco toxins, and reliance on cross-sectional designs, which hinder causal inference.

Emerging nicotine delivery systems, particularly electronic cigarettes, are underexplored. Most studies on vaping involve small, homogeneous cohorts (n < 100) and short follow-up periods (< 6 months), yielding inconsistent results. Future research should prioritize longitudinal studies assessing the effects of smoking cessation or nicotine replacement therapies on the gut microbiota and metabolic status of patients with NAFLD. Additionally, screening and validating engineered strains that degrade nicotine in animal models could offer new therapeutic avenues.

Pharmacotherapy is a major source of interindividual heterogeneity in gut microbiota composition among patients with NAFLD (Li et al., 2025), driven by bidirectional drug-microbiota interactions that affect both therapeutic outcomes and microbial ecology (Chen L. P. et al., 2024; Liu et al., 2024). Metformin illustrates this complexity: approximately 30% of users experience gastrointestinal side effects linked to Escherichia coli overgrowth (Sun et al., 2018; Mueller et al., 2021), and its strain-specific effects on gut microbes have varied across studies (Rosell-Díaz and Fernández-Real, 2024; Szymczak-Pajor et al., 2025). For example, Bacteroidetes abundance shows conflicting trends depending on host metabolic status (Ejtahed et al., 2019); meanwhile, increments in Blautia and Butyrivibrio coupled with a depletion in Faecalibacterium is a signature pattern associated with insulin sensitivity (de la Cuesta-Zuluaga et al., 2017; Barengolts et al., 2018; Pastor-Villaescusa et al., 2021). Microbiota responses to metformin differ markedly among patients with NAFLD, reflecting differences in study design, participant characteristics, analytical methods, and individualized nature of drug–microbiota interactions (Jang et al., 2024).

Cardiovascular drugs also modulate the microbiota in NAFLD. SGLT2 inhibitors promote hepatoprotection through Akkermansia-enriched remodeling, while thiazolidinediones and DPP-4 inhibitors have neutral or harmful effects on hepatic steatosis (Zhang S. et al., 2024). Statins alter gut microbial communities by inhibiting 3-hydroxy-3-methylglutaryl coenzyme A reductase, potentially explaining the lower T2DM incidence in obese users (5.9% vs. 17.7%) (Zhou H. et al., 2023; Vieira-Silva et al., 2020). Proton pump inhibitors cause long-lasting dysbiosis, marked by Enterobacteriaceae/Lactobacillaceae abundance and Ruminococcaceae/Bifidobacteriaceae depletion (Jackson et al., 2016). These changes are linked to NAFLD progression and persist for over 2 years post-treatment (Imhann et al., 2016). Even brief antibiotic exposure (< 7 days) causes long-term microbial diversity loss and functional gene depletion, promoting pathobiont overgrowth and hepatic inflammation (Jernberg et al., 2010). In Finnish children, initial macrolides use led to depletion of Actinobacteriaceae and an increase of Anaplasma and Aspergillus phyla (Korpela et al., 2016).

Although some studies considered drug-related variables, most have focused on single drugs, neglecting the impact of polypharmacy effects and their synergistic or antagonistic effects on microbial metabolism. Notably, microbial responses to the same drug varied widely between individuals. Longitudinal data on microbiota resilience and antibiotic resistance gene enrichment in patients with NAFLD are lacking. Future studies should assess the long-term effects of various drugs on the gut microbiome and NAFLD and develop standardized criteria for assessing drug use, including information on the duration of drug use, dosage, and co-administration, to improve the comparability of results. In particular, multicentre, large-sample clinical studies are needed to validate the prevalence of drug-microbiome interactions.

Circadian misalignment is a key modulator of hepatic pathophysiology, with bidirectional links between chronobiological disturbances and metabolic dysregulation in NAFLD (Bolshette et al., 2023). The liver, which governs ~43% of rhythmic transcriptomes (Zhang et al., 2014), depends on circadian regulation of lipid and glucose metabolism via signaling from the suprachiasmatic nucleus to peripheral clocks (Panasiuk et al., 2024; Speksnijder et al., 2024; Petrenko et al., 2020). Modern lifestyles increasingly disrupt circadian rhythms, contributing to metabolic diseases (Cui et al., 2024) and gut microbiota dysbiosis. The microbiota, considered a “second circadian pacemaker”, shows diurnal oscillations in B/F ratios and microbial metabolite production (Lal et al., 2024; Adafer et al., 2020).

Time-restricted feeding, a chrononutrition strategy, has shown promise in restoring microbial rhythmicity (e.g., Lactobacillus, Mucispirillum, and acetate-producing taxa) and reducing hepatic steatosis in preclinical animal models (Schrader et al., 2024; Snijder and Axmann, 2019; Xia et al., 2023). However, current research is largely preclinical, lacks large-scale human data, and offers limited insight into the GLA. To our knowledge, no study has established a framework for personalized circadian-based interventions.

Future studies should employ multi-omics approaches (e.g., transcriptomics and metabolomics) to map circadian-microbial networks and clarify clock-controlled pathways, such as REV-ERBα-mediated bile acid regulation, as potential therapeutic targets. Large prospective cohort studies are needed to build predictive models linking circadian disruption to metabolic disease risk. In summary, circadian rhythm research offers new insights into liver metabolism and may inform targeted prevention and treatment strategies. With further advances, personalized chronotherapeutic approaches are poised to become a key area in future medicine.

Differences in participant age, sex, and geographic region have been found to influence gut microbiota composition in NAFLD research.

In NAFLD research, given the ethical concerns over liver biopsy restricting histological data collection (Di Pasqua et al., 2022; Panday et al., 2022; Wang et al., 2024), many studies rely on non-invasive diagnostics with known variability. Some studies have used non-invasive assays (e.g., MRI-PDFF) to report differences in microbial abundance with nominal P-values, overlooking multiple testing burdens. For example, a study linking Bacteroides abundance to BMI in obese children (Morán-Ramos et al., 2022) lost significance after false discovery rate correction, a mandatory practice in genomics, yet inconsistently applied in microbiome research.

Future studies must adopt false discovery rate correction (e.g., Benjamini–Hochberg procedures) to reduce false positives and ensure statistical rigor. Bayesian hierarchical models with shrinkage priors (e.g., horseshoe priors) can address sparse signals and population stratification. Machine learning methods, such as elastic net regression, are suitable for multicollinear microbial data. Power analysis tools (e.g., micropower R package) should guide sample size planning, and active learning frameworks should be used to prioritize the enrolment of phenotypically extreme cases (e.g., patients with discordant imaging-histologic staging).

Given NAFLD's potential for reclassification and temporal microbiome shifts, hybrid designs using Mendelian randomization (e.g., PNPLA3 rs738409 as an instrument) and longitudinal mixed-effects models are needed to establish causality and translate statistical rigor into biologically meaningful insights.