Non-alcoholic fatty liver disease (NAFLD) is a pathological syndrome characterized by excessive fat accumulation in hepatocytes in the absence of excessive alcohol consumption or other definite factors, including non-alcoholic fatty liver, non-alcoholic steatohepatitis (NASH), cirrhosis, and hepatocellular carcinoma (1–3). Previous studies have shown that NAFLD correlates with metabolic disorders related to insulin resistance (IR), such as type 2 diabetes mellitus, obesity, and metabolic syndromes. Additionally, NAFLD predisposes patients to atherosclerosis (4–7). Previous studies have shown that NAFLD correlates with increased cardiovascular disease (CVD) risks,chronic kidney disease and cancer (8–10). Globally, NAFLD is a common cause of chronic liver disease. The incidence of NAFLD increases annually, and the global prevalence of NAFLD is ~ 30% (11). The prevalence of NAFLD, in Asians is expected to increase by 20-35% during the next decade (12). This rapid increase in the incidences of NAFLD will significantly impact healthcare and economic systems worldwide.

Hemoglobin A1c (HbA1c) is a diagnostic tool for diabetes and pre-diabetes, and it reflects the mean plasma glucose levels over the last 2–3 months. Recent studies suggest that blood glucose levels and interindividual variations contribute to HbA1c levels (13, 14). Consequently, patients with similar mean plasma glucose levels may have differing HbA1c levels. mean plasma glucose levels account for approximately 60%–80% of the variance in HbA1c levels (15). Based on these limitations, Hempe et al (16) developed a mathematical method in 2002, known as the hemoglobin glycation index (HGI), to quantify the difference between measured and predicted HbA1c levels. Subsequent studies demonstrated that HGI correlates with the observed variabilities in HbA1c levels (17, 18). An assessment of HGI in the Actions to Control Cardiovascular Risk in Diabetes trial showed that patients with diabetes and a high HGI had increased risks for diabetic retinopathies and nephropathies. Similar findings were reported in the Diabetes Control and Complication Trial (19, 20). Previous studies have also demonstrated an association between HGI and the risk of CVD in type 2 diabetes mellitus patients (21–24). High HGI levels, even in non-diabetic patients, correlate with an increased risk for atherosclerosis (25, 26). Mi et al. (27) observed that HGI is an independent risk factor for the development of hypertension despite HbA1c levels. In another study, HGI positively correlated with an increased risk for CVD, independent of other glycemic control factors (26). In addition, a prospective cohort study in Korea showed that HGI correlates with significant morbidity in CVD in the presence of controlled HbA1c levels (23). These findings strongly suggest that HGI directly impacts CVD compared to other glucose‐controlling indexes.

Nevertheless, previous studies on HGI primarily focused on vascular complications associated with diabetes. As such, studies demonstrating a relationship between HGI and NAFLD are limited. Therefore, this study aimed to elucidate the association between HGI and NAFLD.

In this study, the physical examination population was evaluated. Higher HGI levels were observed in the NAFLD group than in the non-NAFLD group. Furthermore, HGI increased simultaneously with the prevalence of NAFLD. Multivariate logistic regression analysis showed that HGI was an independent risk factor for NAFLD. Further analysis showed that patients in the highest quartile of HGI had significantly higher risks for NAFLD than those in the three lower quartile groups. A study conducted by Fiorentino et al. (29) demonstrated that HGI correlated with hepatic steatosis; this finding was congruent with the findings of this study. Our results are similar to these two studies (30, 31).,elevated HGI levels are independently associated with NAFLD.We also found that BMI partly mediates the association between HGI and NAFLD.Our research demonstrates several new populations that may be at risk for NAFLD including people younger than 40 with high HGI and obese people with high HGI,which were not previously described.

Several mechanisms may explain the association between HGI and NAFLD. First, intracellular glucose levels were higher in patients with higher HGIs than those with lower HGIs. Increased intracellular glucose release a significant number of noxious metabolites. This release ultimately leads to liver injury (18). Chronic inflammation is considered a core pathogenesis of the development of NAFLD (32, 33). Fiorentino et al. (29) also observed that there is a significant increase in the levels of inflammation-related biomarkers in patients with higher HGIs.

Similarly, Hu et al. (30) reported a positive correlation between HGI and inflammatory biomarkers, including WBC and platelet counts. The US National Health and Nutrition Examination Survey results suggest that HGI independently associates with inflammatory biomarkers (such as C-reactive protein, polymorphonuclear leukocytes, and monocytes.) (34) Inflammation might impair insulin signaling and exacerbate fatty liver infiltration. Alternatively, inflammation induces oxidative stress, damages mitochondria, and causes ER stress. Defective mitochondria are associated with incomplete fat oxidation and the generation of toxic lipid intermediates. These toxic lipid intermediates generate many reactive oxygen species, which eventually promote the progression of liver steatosis into more severe forms of the disease (35, 36) and fuel the transition from NAFLD to non-alcoholic steatohepatitis, liver cirrhosis, and even hepatocellular carcinoma (35, 37, 38). Previous studies reported that HGI represents the degree of non-enzymatic hemoglobin glycation (26, 29, 39). Non-enzymatic hemoglobin glycation might play a key role in NAFLD pathogenesis (40, 41). Studies have also shown that HGI reflects the level of advanced glycation end products (AGEs) (42). AGEs may alter the structure of proteins and their functions, resulting in alterations that contribute to histopathological changes in the liver (42). Studies have also demonstrated that AGEs, in combination with the activation of the RAGE downstream pathways, trigger further inflammation and oxidative stress and impair insulin signaling. Therefore, AGEs increase the development and progression of NAFLD (43). IR is recognized as a potential pathogenesis of NAFLD. In addition, HGI may associate with IR. Furthermore, the degree of IR in patients with lower HGIs was less than not those in patients with higher HGIs (26).

As a result of obesity, an imbalance in pro- and anti-inflammatory adipokines is secreted from adipose tissue. This secretion contributes to the development of NAFLD (44). Similarly, a previous investigation indicated that several adipokines, including adiponectin and leptin, play a role in the pathogenesis of NAFLD, including hepatic fat accumulation, chronic inflammation, and IR (45, 46). BMI is an easy-to-implement body composition classification screening indicator. BMI classifies patients as underweight, normal weight, overweight, or obesity. BMI is widely employed as a surrogate indicator for measuring weight status (adjusted for height) and the percentage of fat mass (47–49). Among anthropometric measurements, BMI has the best discriminatory power (50). Therefore, BMI is the most commonly used anthropometric measure of obesity (51, 52). Numerous studies have shown a strong correlation between BMI and NAFLD; the higher the BMI, the higher the risk of developing NAFLD (53–55). Tang Z et al. (56) noted that increases in BMI correlated with NAFLD and that these increases were independent of the patient’s age. The study also revealed that BMI modulates the association between physical examinations and blood biochemistry parameters and NAFLD (56).Obesity-related inflammatory input has a well-established correlation with NAFLD (22, 57). BMI is a poor predictor of NAFLD severity (58). Even so, BMI was confirmed as the most useful predictive factor for NAFLD onset in both sexes (59). BMI had the highest area under the curve in the prediction of NAFLD (60). BMI is superior to WC in predicting the risk for NAFLD (61). A recent study showed a significant correlation between HGI and obesity (26), which was consistent with the previous findings of Yoo et al. (31) Correspondingly, Hu et al. (30) found that HGI levels had positive associations with BMI. Therefore, a mediator analysis was conducted to determine if BMI mediates the correlation between HGI and NAFLD. Our study showed that HGI and BMI were positively associated with NAFLD, and that HGI positively correlated with BMI. The mediation analysis results indicated that the relationship between the HGI and NAFLD was partly mediated by obesity. In addition, the correlation between the HGI and NAFLD was significantly higher in obese patients.

This study identified that increases in HGI correlated with increased incidences in females. Previous studies also showed gender differences in NAFLD related to gender-specific gene expression, sex hormones, and gender-related economic, behavioral, and environmental factors (62, 63)..

This study had limitations. First, this was a cross-sectional study, and thus could not establish a cause-effect relationship. Further prospective studies are required to validate these findings. Second, HGI levels may vary by population (64, 65). Therefore, the results obtained from this study conducted in Japan might not be applicable to other ethnicities. Third, fatty liver diagnosed by ultrasonography may provide an incorrect diagnosis compared to a liver biopsy. Finally, since the dataset was obtained from a public database, it could not be updated. Therefore, we could not investigate the effect of HGI on liver fibrosis. Some known risk factors of demographic data for NAFLD, such as dietary preferences, exercise habits, work types, and so on, were not included, which limits a comprehensive assessment of the correlation between risk factors and NAFLD.

In conclusion, HGI positively correlates with the prevalence of NAFLD, and BMI partly mediates the association between HGI and NAFLD.Our research demonstrates several new populations that may be at risk for NAFLD including people younger than 40 with high HGI and obese people with high HGI.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://doi.org/10.1038/s41366-018-0076-3.

The studies involving human participants were reviewed and approved by Murakami Memorial Hospital. The patients/participants provided their written informed consent to participate in this study.

YX and YZ performed the data analysis and YX prepared the initial draft of the manuscript. LY and LH gave statistical advice throughout the data analyses and contributed to the manuscript. HM was responsible for the study concept, data analysis, and interpretation of the findings, and critically reviewed and revised the manuscript. All authors contributed to the article and approved the submitted version.