Our study retrospectively analyzed the clinical data of patients diagnosed with hyperglycemia, along with a series of laboratory assessments conducted at the Tianjin Third Central Hospital from January 2023 to December 2024. Based on the current or past ultrasound examination, participants were categorized into two groups: T2DM and T2DM + NAFLD. Among the participants, 234 patients had T2DM accompanied by NAFLD, while 197 had T2DM without NAFLD. This cross-sectional included 193 healthy individuals who underwent health check-ups at the Tianjin Third Central Hospital.

The diagnosis of T2DM was determined based on the criteria, which define fasting plasma glucose (FPG) > 7 mmol/L or 2 h postprandial plasma glucose (2hPG) > 11.1 mmol/L (18). The diagnostic criterion for NAFLD was based on the guidelines for prevention and treatment of NAFLD developed by the Chinese Society of Liver Diseases (19). NAFLD was diagnosed in patients presenting with at least two of the following three findings: (a) diffuse enhancement of the liver near-field echo that is stronger than that of the kidney, (b) poorly defined intrahepatic bile duct structures, and (c) gradual attenuation of the far-field echo of the liver. Participants may also exhibit nonspecific symptoms, such as vague pain in the liver area, fatigue, and hepatosplenomegaly.

To ensure the effectiveness of the results, comprehensive exclusion criteria were implemented as follows: (a) individuals with chronic excessive alcohol intake (defined as ≥210 g/week for males and ≥140 g/week for females); (b) individuals with acute or chronic infections; (c) patients with hematologic disorders including anemia, hemolytic diseases, bleeding, and other disorders which resulted in abnormal hemolytic status; (d) patients with lipid metabolism dysfunction; (e) patients with severe spinal curvature. Following rigorous screening, 197 patients with T2DM and 234 patients with T2DM comorbid with NAFLD were included in the final cohort.

The clinical data were retrospectively extracted from the electronic medical record system and categorized into three domains: (a) demographic characteristics: age and gender (b) laboratory parameters: complete blood count parameters [including white blood cells (WBC), neutrophils (NEU), lymphocytes (LYM), monocytes (MONO) and platelets (PLT)], FPG, albumin (ALB), total protein (TP), total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C), alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyl transferase (GGT), C reaction protein (CRP), ferritin (FER) and procalcitonin (PCT). Venous blood samples were collected following a standardized 12 h overnight fast. (c) In the process of data statistics, stratified sampling was employed to conduct random selection among the participants. The criteria for stratified sampling were the patients’ glycosylated hemoglobin index and blood glucose values. Whole blood samples were centrifuged at 1,500 rpm for 20 min. Serum cytokines (IL-2, IL-4, IL-6, IL-10, IL-17, IFN-γ, and TNF-α) were quantified using a flow cytometer (FACS Canto II, BD Biosciences, Inc.).

The values of triglyceride-glucose index (TyG), MHR, NHR, NLR, PWR and Systemic immune-inflammatory index (SII) were derived from the laboratory measurements using the following Equations 1–6

Statistical analyses were conducted using SPSS software (V 27; IBM Corp., Armonk, NY, USA) and GraphPad Prism 8 software (Graph Pad Software, Inc., La Jolla, CA, USA). Continuous variables with normal distribution were expressed as mean ± standard deviation (SD), while non-normally distributed variables were expressed as the median of the interquartile range (IQR) (25%, 75%).

Descriptive analyses were performed using Student’s t-tests or Mann–Whitney U test based on distribution characteristics. Spearman correlation analysis was constructed to assess the association between each biomarker. Additionally, receiver operating characteristic (ROC) curve analysis evaluated predictive capacity and diagnostic accuracy. Optimal cut-off values were determined using Youden’s index maximization criteria. Odds ratios (OR) and their corresponding 95% confidence intervals of combined diagnostic prediction are calculated through binary logistic regression analysis. Use the Bonferroni corrected p-value and set a p-value <0.05 as statistically significant.

This cross-sectional study enrolled a cohort of 624 adults (341 males; 283 females) aged 21–84 years. Based on clinical characteristics, the population comprised three distinct groups: 197 individuals diagnosed with T2DM, 234 with T2DM and NAFLD, and 193 healthy controls. Table 1 summarizes the characteristics of age, ALT, AST, GGT, FPG, ALB, and TP across study groups, expressed as the mean (SD)/IQR. The indicators in the control group were 46.0 (11.8), 15 (10, 18), 18 (14, 21), 19 (14, 29.5), 5.12 (4.93, 5.43), 46.55 (2.43), and 72.37 (3.73), the T2DM cohort were 58.4 (12.7), 18 (13, 30), 17 (14, 25.5), 26 (16, 52), 8.76 (7.67, 9.94), 40.28 (4.05), and 64.79 (5.22), and the T2DM and NAFLD group indicators were 59.7 (10.5), 22.5 (15, 36), 30 (20, 38.25), 48 (26.75, 91), 8.84 (7.41, 10.86), 35.75 (5.95), and 64.17 (8.24).

Significant intergroup differences in baseline characteristics were observed between cohorts with and without NAFLD comorbidity in the T2DM population, except for age, ALT, and TP. In comparison, the participants with NAFLD comorbidity tended toward advanced age and marginally elevated ALT levels. Additionally, the T2DM + NAFLD cohort exhibited clinically significant hepatic biomarker alterations, including elevated AST, GGT, and reduced ALB. These pathophysiological changes collectively suggest progressive hepatic metabolic dysregulation in the comorbid population.

Table 2 showed statistically different lipid metabolism characteristics between cohorts with and without T2DM or NAFLD comorbidity, except for LDL-C (control vs. T2DM) and TyG index (T2DM vs. T2DM + NAFLD). Meanwhile the participants with NAFLD comorbidity tended to have an advanced TyG index. It is worth noting that participants with T2DM and NAFLD exhibited statistically increased levels of TC, TG, LDL-C, TyG and reduced level of HDL-C. Moreover, compared with the control group, the levels of TC, TG, LDL-C, and TyG in T2DM individuals were higher, while the HDL-C level was lower. These findings indicate that the participants with T2DM and NAFLD have hyperlipidemia.

The SD and IQR of inflammatory characteristics variables are shown in Table 3. Compared to the control group, the T2DM cohort demonstrated increased levels of NEU, MONO, CRP, FER, NLR, MHR, NHR, and SII, alongside decreased levels of LYM, PLT, PCT, and PWR. Notably, there was no significant difference in WBC between the two groups. Additionally, Compared to the control group, the T2DM and NAFLD cohort exhibited elevated levels of NEU, CRP, FER, NLR, MHR, NHR and SII couple with attenuated levels of WBC, PLT, PCT and PWR. Furthermore, comparative analysis between the T2DM cohort and the T2DM + NAFLD comorbid cohort revealed intensified inflammatory activity in comorbid patients with higher levels of CRP, FER, NLR, and MHR. While the levels of NHR and SII in the comorbid cohort were lower than in the T2DM cohort. We hypothesized that these results may be related to lower level of WBC in participants with T2DM and NAFLD. These hematologic characteristics may reflect compensatory immunosuppression mechanisms in progressive metabolic.

To further delineate the inflammatory spectrum in NAFLD progression, a stratified random sampling approach was employed. The distribution trends within the subsample were similar to those of the overall sample. Subsequently cytokines profiling was systematically conducted. As detailed in Table 4, the T2DM + NAFLD cohort demonstrated statistically elevated concentrations of IL-6, IL-10, and IFN-γ compared to mono disease cohorts, whereas the concentrations of IL-2, IL-4, and IL-17 were reduced. Additionally, the expression levels of differential cytokines across clinical subgroups were exhibited more clearly in Figure 1, which revealed distinct inflammatory signatures associated with disease progression.

Table 5 illustrates the correlation between the TyG index and the NAFLD-related inflammatory indicators (WBC: r = 0.136, p = 0.011; NEU: r = 0.246, p = 0.0011; CRP: r = 0.373, p = 0.0011; FER: r = 0.541, p = 0.0011; MHR: r = 0.366, p = 0.0011; NHR: r = 0.441, p = 0.0011; NLR: r = 0.319, p = 0.0011; SII: r = 0.143, p = 0.011; IFN-γ: r = 0.229, p = 0.088). The findings revealed a statistically positive association between elevated TyG index and an aggravated inflammation of NAFLD.

Additionally, we analyzed the correlation between the inflammatory indexes (MHR, NHR, NLR, SII) and generation laboratory indicators. Tables 6–9 shows that MHR has a positive association with WBC (r = 0.373, p < 0.001), ALT (r = 0.227, p < 0.001), AST (r = 0.227, p < 0.001), GGT (r = 0.295, p < 0.001), FPG (r = 0.244, p = 0.036) and TG (r = 0.303, p < 0.001). NHR positively associates with WBC (r = 0.699, p < 0.001), ALT (r = 0.185, p < 0.001), GGT (r = 0.229, p < 0.001), FPG (r = 0.198, p < 0.001), TG (r = 0.408, p < 0.001), and LDL-C (r = 0.133, p = 0.036). NLR positively correlates with age (r = 0.304, p < 0.001), ALT (r = 0.25, p < 0.001), AST (r = 0.221, p < 0.001), GGT (r = 0.263, p < 0.001), FPG (r = 0.191, p < 0.001), and TG (r = 0.191, p < 0.001). SII has a positive association with WBC (r = 0.462, p < 0.001) and LDL-C (r = 0.156, p < 0.001) and a negative association with AST (r = −0.152, p = 0.108). Among them, NHR presented a superior positive association with generation laboratory indicators.

To further determine the relationship between NAFLD and inflammation, we chose IL-6, IL-10 and IFN-γ to participate in this analysis, which had higher expression levels in the participants with T2DM and NAFLD. We found a positive statistical correlation between IL-6 and AST (r = 0.248, p = 0.027), GGT (r = 0.339, p = 0.0027), and a negative correlation between IL-6 and WBC (r = −0.323, p = 0.0027). Furthermore, IL-10 revealed a positive correlation with AST (r = 0.246, p = 0.027) and GGT (r = 0.316, p = 0.0027). In addition, IL-10 and IFN-γ also have a negative correlation with WBC and HDL-C (Tables 10–12). These results suggested that elevated IL-6 concentrations significantly correlated with aggravated inflammation and lipid deposition.

Figure 2 shows the comparative diagnostic performance of inflammatory indices (MHR, NHR, NLR and SII) and cytokines (IL-6, IL-10 and IFN-γ) through ROC analysis. Moreover, the data distribution of inflammatory indices and cytokines were shown in Supplementary Figure S1. NLR demonstrated superior diagnostic capacity [AUC = 0.868 (0.832–0.904); best cut-off point: 3.28; sensitivity: 72.6%; specificity: 92.4%] relative to MHR (AUC = 0.649), NHR (AUC = 0.568), and SII (AUC = 0.520), with detailed comparisons tabulated in Table 13. Moreover, IL-6 exhibited maximal predictive accuracy among cytokines (AUC = 0.777; best cut-off point: 4.02; sensitivity: 76.8%; specificity: 73.3%) (Table 14). Taken together, NLR and IL-6 were chosen as optimal candidates for subsequent multi parametric diagnostic modeling.

Based on the results presented in Tables 13, 14 and Figure 2, we selected the most effective inflammatory indicator (NLR) and the cytokines indicator (IL-6) and combined them with the glycohepatic index (TyG) to construct a combined diagnostic model. Subsequently, OR and their corresponding 95% confidence intervals of combined diagnostic prediction are calculated through binary logistic regression analysis. Figure 3 illustrates the performance comparison between combined diagnostic prediction and single diagnosis in identifying NAFLD, as determined through ROC curve analysis. Supplementary Figure S2 shows the data distribution for Figure 3 and Tables 15–17 for details. Table 15 demonstrated that combined diagnostic prediction [AUC = 0.956 (0.945–0.978); best cut-off point: 0.44; sensitivity: 90.1%; specificity: 89.1%] performs best in identifying participants with T2DM-NAFLD comorbidity from healthy individuals compared to a single diagnosis. The combined diagnostic prediction [AUC = 0.802 (0.757–0.848); best cut-off point: 0.5; sensitivity: 72.1%; specificity: 86.9%] also exhibited superior diagnostic capacity relative to a single diagnosis in identifying participants with T2DM-NAFLD comorbidity from T2DM participants (Table 16). Additionally, the combined diagnostic prediction demonstrated better predictive accuracy [AUC = 0.891 (0.859–0.922); best cut-off point: 0.57; sensitivity: 76.4%; specificity: 88.0%] in identifying participants with T2DM from healthy individuals than a single diagnosis (Table 17). In summary, the integrative diagnostic model combined metabolic-inflammatory axes (TyG + NHR + IL-6) and demonstrated superior accuracy in NAFLD detection.

The notable metabolic-inflammatory features exhibited by patients with the comorbidity of T2DM and NAFLD are underpinned by a profound mechanism. This mechanism is embedded in a vicious cycle instigated by IR and intricately intertwined with multiple signaling pathways. The elevation of liver enzymes, dyslipidemia, and the upregulation of systemic inflammation markers, including CRP, FER, NLR, and MHR, as observed in this study, can all be accounted for within this framework.

The core initiating factor is IR. In the setting of T2DM, the insulin signaling pathway, primarily the phosphatidylinositol 3-kinase (PI3K)-Akt axis, is compromised. This leads to unregulated lipolysis in white adipose tissue. A substantial quantity of free fatty acids (FFAs) then inundates the liver, forming the material basis for hepatic steatosis (18–21). This finding aligns with our observation that the TyG index is significantly associated with lipid parameters (Table 2).

The FFAs that enter hepatocytes not only function as lipotoxicity agents but also serve as danger signals to activate the innate immune system. They initiate the downstream inhibitor of nuclear factor kappa B kinase subunit beta (IKKβ)/nuclear factor kappa B (NF-κB) and c-Jun N-terminal kinase (JNK) signaling pathways via pattern recognition receptors such as Toll-like receptor 4 (TLR4) (22, 23).

The nuclear translocation of NF-κB directly promotes the transcription and release of crucial pro-inflammatory factors, including TNF-α, IL-1β, and IL-6 (24, 25). This offers a precise mechanistic account for the significantly elevated IL-6 level and its positive correlation with liver injury markers observed in our study (Table 10).

Simultaneously, FFA-induced ERS and mitochondrial dysfunction further intensify the inflammatory state. ERS indirectly enhances NF-κB activity by activating the inositol-requiring enzyme 1 alpha (IRE1α) and protein kinase RNA-like endoplasmic reticulum kinase (PERK) branches of the unfolded protein response (UPR). Dysfunctional mitochondria generate reactive oxygen species (ROS), which promotes oxidative stress and activates the NLR family pyrin domain-containing 3 (NLRP3) inflammasome. This, in turn, leads to the maturation and secretion of IL-1β and interleukin-18 (IL-18), thereby continuously amplifying the inflammatory cascade (26–29).

Notably, the compensatory increase in IL-10 that we detected might originate from a feedback regulatory mechanism designed to limit excessive inflammatory damage. Nevertheless, under persistent metabolic stress, this protective mechanism is evidently inadequate.

Furthermore, lipid accumulation within hepatocytes and the consequent inflammatory microenvironment exacerbate hepatic insulin resistance via the Janus kinase-signal transducer and activator of transcription (JAK-STAT) pathway, particularly the IL-6-activated JAK2-STAT3 axis, and the feedback of suppressor of cytokine signaling (SOCS) proteins. This process forms a self-reinforcing “metabolic-inflammatory” positive feedback loop (30, 31). Significantly, this loop is not confined to the liver alone; it also impacts whole-body metabolic homeostasis by releasing inflammatory mediators (such as IL-6 and IFN-γ, as detected by our research) and pathogenic lipid species. This phenomenon elucidates the more profound metabolic derangements observed in patients with comorbidities.

In summary, the comorbidity of T2DM and NAFLD is not merely a simple juxtaposition of two distinct diseases. Instead, it represents a persistent, low-grade inflammatory state centered in the liver. This state is initiated by IR and is jointly propelled by multiple signaling pathways, including TLR4/NF-κB, JNK, ERS, and the NLRP3 inflammasome, through intricate cross-talk mechanisms. These findings provide a robust theoretical foundation for considering inflammatory markers as viable diagnostic and therapeutic targets.

In Table 4, the observed cytokine profile-specifically, the notable elevation of IL-6, IL-10, and IFN-γ, accompanied by the reduction in IL-2, IL-4, and IL-17- does not represent an isolated biological occurrence. Rather, it serves as the central manifestation of the imbalance within the immune metabolic regulatory network during the progression of T2DM-NAFLD. This particular cytokine pattern profoundly uncovers the molecular essence of the disease’s transition from simple metabolic derangements to chronic inflammation-induced liver injury.

Among these cytokines, the core driving role of IL-6 is of particular significance. IL-6 directly facilitates the synthesis of acute-phase response proteins [such as CRP and FER, which also exhibited corresponding increases in this study (Table 3)] by activating the JAK2/STAT3 signaling pathway within hepatocytes. Simultaneously, it induces the expression of suppressor of cytokine signaling 3 (SOCS3) (32). The latter competitively inhibits the tyrosine phosphorylation of the insulin receptor substrate (IRS), thereby exacerbating hepatic insulin resistance and establishing a vicious cycle.

Furthermore, the continuous activation of STAT3 can also upregulate the activity of hepatic stellate cells (HSCs), which paves the way for the development of liver fibrosis. This offers a potential mechanistic interpretation for the significant elevation of AST and GGT in patients with comorbidities (Table 1) (33).

IFN-γ acts in concert with IL-6. It is predominantly secreted by activated T cells and NK cells. Through its distinct JAK1/STAT1 signaling axis, IFN-γ drives the polarization of liver macrophages (Kupffer cells) toward the pro-inflammatory M1 phenotype and enhances the antigen presentation ability, thereby further amplifying the inflammatory cascade reaction. The positive correlation between IFN-γ and FPG indicates that it may directly participate in the immune inflammatory pathological process of diabetes (Table 12) (34).

The seemingly contradictory elevation of IL-10 in Table 4 interpreted as a compensatory feedback mechanism aiming to limit tissue damage. IL-10 inhibits macrophages from generating TNF-α and IL-12 and promotes their transformation into the anti-inflammatory M2 phenotype by activating the JAK1/STAT3 pathway (sharing some downstream signals with IL-6 but yielding different outcomes in diverse contexts). Nevertheless, in the setting of continuous metabolic stress (such as lipotoxicity and ERS) in patients with T2DM-NAFLD comorbidity, this anti-inflammatory feedback is overpowered by potent pro-inflammatory signals, resulting in the failure to restore immune homeostasis. On the other hand, the reduction in IL-2 and IL-4 indicates the attenuation of helper T cell (Th) function, particularly that of Th2 cells. The substantial decline in IL-17 might reflect a shift in the Th17/Treg balance toward immunosuppression.

This could potentially represent an adaptive reconstruction of the body under persistent inflammatory stress. However, it might also result in an overall deterioration of anti-inflammatory and repair capacities. This finding aligns with the “immune exhaustion” state inferred from the reduction in WBC count, as observed in the study (Table 3). In addition, we supposed that the clinical manifestation of thrombocytopenia was caused by splenomegaly (Supplementary Table S1) in patients with liver disorder (35, 36).

In summary, the dysregulation of the cytokine network in the comorbidity of T2DM and NAFLD is a complex phenomenon. It is predominantly governed by core pro-inflammatory pathways, such as the IL-6/STAT3 and IFN-γ/STAT1 axes. Although there is compensatory activation of anti-inflammatory mechanisms, the overall process ultimately spirals out of control. This intricate network not only directly facilitates the progression of hepatocyte steatosis, inflammatory injury, and fibrosis but also acts as a crucial link connecting systemic insulin resistance to liver-specific pathological alterations.

Our correlation analysis unveiled an intricate interaction network among inflammatory markers and metabolic parameters. The TyG index, a dependable marker of insulin resistance, exhibited a significant and positive correlation with multiple inflammatory markers, including CRP, FER, MHR, NHR, and NLR (Table 5). This strongly suggests a bidirectional promoting association between insulin resistance and systemic inflammation in the comorbidity of T2DM and NAFLD.

Among diverse inflammatory indexes, NHR demonstrated the most extensive and robust correlations with conventional metabolic parameters. Specifically, it had a high correlation with WBC (r = 0.699) and TG levels (r = 0.408), indicating that NHR might serve as a sensitive indicator integrating inflammation and dyslipidemia (Table 7). This finding builds upon recent research regarding the value of combined indicators, such as the TyGFI index developed by the Yan Miao team. By integrating metabolic stress and physiological vulnerability, it significantly enhances the capacity to predict cardiovascular risk (37–44).

Notably, we discovered that IL-6 was negatively correlated with WBC (r = −0.323), which contradicts the traditional concept of inflammation. We postulate that under the condition of chronic metabolic inflammation, continuous inflammatory stimulation may result in the exhaustion or redistribution of immune cell functions, and this phenomenon merits further investigation.

Given that the majority of patients with NAFLD exhibit splenomegaly (45), chronic inflammation associated with insulin resistance (46), the phagocytic function and anti-inflammatory effects of the spleen are augmented (47–50). From the perspective of anatomical structure, as the spleen and the liver are connected to visceral fat via the portal vein circulation, the substances secreted by visceral fat directly impact these two organs. Consequently, some scholars have put forward the concept of the hepatosplenic axis. As a crucial organ in the immune cycle, the spleen assumes a dual role in the development of chronic inflammation (51). The spleen is directly implicated in the progression of chronic low-grade inflammation, which in turn precipitates insulin resistance. Similar with our results, the increased expression of IL-6 (52) and IFN-γ (47) were observed in high fat diet mice with enlarged spleen. Moreover, The spleen Ki/V(0) (representing the tissue phosphorylation 18F-fluorodeoxyglucose distribution volume) is correlated with plasma glucose, suggesting the level of insulin sensitivity (45). In comparison to MSG-NO rats, splenectomy in MSG-obese animals can effectively mitigate hyperinsulinemia, enhance insulin sensitivity, and reduce the hypertrophy of adipocytes and islets (53).

However, certain studies have also revealed that the spleen might play a protective role in obesity. Experimental findings have demonstrated that in male Sprague–Dawley rats, regardless of whether they are fed a high-fat diet or a normal diet, splenectomy results in an elevation of serum lipid levels, with the exception of triglycerides and high-density lipoproteins. More significantly, splenectomy markedly accelerates the progression of liver steatosis. Through Western blot analysis and real-time polymerase chain reaction assays, it was discovered that splenectomy not only significantly downregulates the expression level of phosphatase and tensing homolog in the liver but also induces an abnormally high ratio of phosphorylated Akt/Akt in the liver (54). Similarly, researchers observed that after splenectomy in obese mice, the serum interleukin-10 level was further diminished, while the levels of pro-inflammatory cytokines did not decline. These data suggest that spleen-derived interleukin-10 plays a crucial role in the diet-induced inflammatory responses of white adipose tissue and the liver (55).

According with these researches, we plan to establish a database for patients with non-alcoholic fatty liver disease in the future, conduct cohort follow-ups, and track their subsequent progress to continue relevant research, such as “the analysis of the current immune-mediated mechanisms underlying the chronic low-grade inflammation and its subsequent triggering of insulin resistance.”

One of the most significant findings of this study was to assess the diagnostic value of multiple inflammatory markers for the comorbidity of T2DM and NAFLD. Receiver operating characteristic (ROC) analysis indicated that the NLR exhibited the highest diagnostic accuracy among blood inflammatory markers [area under the curve (AUC) = 0.868], surpassing the MHR, NHR, and SII (Table 13, Figure 2A). Among cytokines, IL-6 demonstrated the optimal diagnostic performance (AUC = 0.777) (Table 14; Figure 2B). These results suggest that the relatively straightforward and readily obtainable NLR could serve as a valuable tool for primary healthcare institutions to screen the risk of NAFLD in T2DM patients (56).

In line with our findings, numerous studies have corroborated the value of inflammatory markers in the risk assessment of metabolic diseases. The research team led by Dai Dongling discovered that the modified TyG index (incorporating parameters such as waist circumference) had outstanding predictive ability for metabolic-associated fatty liver disease in adolescents (AUC 0.915–0.923). The study by Alam et al. (57) revealed that the fibrosis-4 (FIB-4) index had good performance in the assessment of liver fibrosis in patients with NAFLD complicated by T2DM (AUC = 0.73). Our study further supplements the value of inflammatory markers, especially NLR and IL-6, in the early identification of the simple steatosis stage. Moreover, the comparative analysis of the FIB-4 model have been supplemented. As the results shown in Supplementary Table S2, the AUC for the predictors of combined testing of NLR, IL-6, and TyG is higher than FIB-4 in identifying participants with T2DM-NAFLD comorbidity from T2DM participants (Supplementary Table S2-2) and participants with T2DM from healthy people (Supplementary Table S2-3), further suggesting that the predictors may provide a better predictive capacity in screening patients with T2DM in the early stage or NAFLD in the T2DM patients.

The most groundbreaking discovery of this study lies in the establishment of a composite diagnostic model (TyG + NHR + IL-6) that integrates metabolism, inflammation, and cytokines. This model exhibited outstanding diagnostic performance in discriminating patients with T2DM complicated by NAFLD from those with simple T2DM, with an area under the receiver operating characteristic curve (AUC) of 0.802 (Table 16; Figure 3B). Moreover, it demonstrated even superior performance when differentiating comorbid patients from healthy controls, achieving an AUC of 0.956 (Table 15; Figure 3A).

This multi-dimensional diagnostic approach is highly congruent with the concept of contemporary precision medicine, which aims to enhance disease identification capabilities by integrating biomarkers from diverse pathophysiological pathways. Our method shares similarities with the research concept of the metabolic-inflammatory subtypes described above; however, it is more clinically applicable and feasible. In contrast to subtype classification that necessitates intricate metabolomics analysis, our composite model is founded on routine clinical indicators, rendering it more amenable to promotion in clinical practice.

Furthermore, compared with the TyGFI index developed by the Yan Miao team, our model specifically targets the identification of T2DM-NAFLD comorbidity, potentially offering a more precise screening tool for this particular patient population.

According to our results and in line with the 2024 ADA guidelines, we suggest for the annual implementation of a systematic screening protocol in high-risk populations, such as T2DM patients. This protocol usually begins with the FIB-4 index, followed by confirmatory testing with transient electrography (e.g., Fibro Scan) for those with indeterminate or elevated scores. To address the limitations of the secondary consumption and complex inspection process, our study proposes an alternative method utilizing a novel combination predictive marker of NLR, IL-6 and TyG. Combination predictive marker with a value greater than 0.44 may be considered high risk (Tables 15–17). The management for high-risk patients must be comprehensive, which including grounded in lifestyle modifications (targeted weight loss, dietary control, exercise) and augmented by individualized pharmacological regimens. This integrated strategy coupled with regular 3- to 6-month follow-ups to monitor key parameters is essential for effective clinical management.

This investigation encompasses several limitations. Firstly, the cross-sectional design employed herein precludes the determination of causal relationships among the observed associations. In the future, prospective cohort studies are imperative to validate the predictive utility of these indicators. Secondly, the diagnosis of non-alcoholic fatty liver disease (NAFLD) is predicated on clinical characteristics rather than the gold standard of liver biopsy. This approach may potentially overlook subclinical cases. Thirdly, not all potential confounding factors, such as dietary patterns, physical activity levels, and gut microbiota composition, which could influence the inflammatory state, were comprehensively evaluated. Finally, the study population was sourced from a single center. Consequently, the external validity of the findings necessitates verification through multi-center investigations.

Future research endeavors should center on the following aspects: