In recent years, the prevalence of non-alcoholic fatty liver disease (NAFLD) has increased significantly (1). More worryingly, a data framework predicted that a growing number of NAFLD patients would develop non-alcoholic steatohepatitis (NASH)-related end-stage or malignant liver diseases (2), who have become the main population for liver transplantation. The aggravating incidence of NAFLD and the severe consequences of disease progression makes it necessary to develop effective methods to predict NAFLD response to LSG treatment. In this regard, serological biomarkers and models have attracted significant clinical attention due to their non-invasive, low cost, simple maneuverability and strong repeatability advantages.

Hepatic steatosis, inflammation and fibrosis in NAFLD are closely related to obesity and insulin resistance (IR) (3, 4). Accumulating evidence suggests that the Peroxiredoxin (PRDX) family is a key antioxidant enzyme regulating the balance between glucose and lipid metabolism (5). PRDX members, including Prdx1 (6), Prdx2 (7), Prdx3 (8), Prdx4 (9) and Prdx6 (10, 11), were reported to be overexpressed in obesity, type 2 diabetes mellitus (T2DM) and atherosclerosis (12). Prdx1 and Prdx4 are the secretory members of the PRDX family. Elevated circulating Prdx4 was found to be associated with certain components (e.g., hypertension and triglycerides [TG]) and mature inflammatory markers (e.g., high-sensitivity C-reactive protein [hs-CRP] and procalcitonin) of the metabolic syndrome (13) and has been incorporative in a T2DM DESIR model in men (14).

vPrdx1 is the main PRDX family member in the pancreas (15) (16), liver insulin resistance, inflammation and steatosis (17). An increase in Prdx1 is released through exosomes in response to various environmental stressors (18), such as inflammatory cytokines (19), oxidative stress (20) and streptozotocin (STZ) administration (21). At present, two studies have reported an increase in Prdx1 in T2DM was positively correlated with the homeostasis model assessment of insulin resistance (HOMA-IR) (r = 0.276, p< 0.01) (18, 22). However, the level and change of circulating Prdx1 in the evolution of NAFLD have not yet been reported.

Metabolic surgery is the most effective treatment to achieve substantial and lasting weight loss in individuals with obesity (23). Among patients with severe obesity undergoing metabolic surgery, the prevalence of NAFLD was found to exceed 90%, the prevalence of NASH was reported to approximate 37% (24 to 98%), with up to 5% of the patients at high risk for unanticipated liver cirrhosis (24). After bariatric surgery, visceral fat and insulin resistance are reduced within a few days (25), liver fatty acid uptake is reversed (26), and liver inflammation is improved (27, 28). LSG has replaced the Roux-en-Y gastric bypass surgery (RYGB) as the most common bariatric operation (29). In particular, NAFLD individuals benefit more from LSG than RYGB (30), with a reported -2.5 (95% CI: -3.5 to -1.5) median NAFLD activity score (NAS) improvement at 192 days after LSG (31).

Since circulating Prdx1 is closely related to chronic metabolism diseases (i.e., arteriosclerosis, diabetes, etc.) and inflammation, we hypothesized that increased circulating Prdx1 might be a biomarker of obesity, insulin resistance and NAFLD and therefore played an important role in the clinical response after LSG.

The most common bariatric procedures are sleeve gastrectomy (SG) and Roux-en-Y gastric bypass (RYGB), which are both effective in alleviating hepatic steatosis, while SG was reported to be associated with fewer complications (43). A published study (44) investigating liver fat fraction at one-year follow-up after LSG found a reduction of -19.7% (95% CI: -22.5% to -16.9%). Most studies about NAFLD have focused on the efficacy of LSG after one year or longer. In this present study, we assessed the remission of NAFLD at three months after LSG because it was at that time period that we observed improvements in relevant pathological status (i.e., insulin resistance, obesity) and liver biopsy after surgery. Studies have shown that within three months after bariatric surgery, insulin resistance usually returns to a normal level (45), with significant weight loss achieved in patients with obesity (46). A reduction in hepatic fat content and improvement in hepatic insulin resistance are considered the earliest beneficial effects of bariatric surgery (47). In this study, liver biopsies were performed to confirm that hepatic steatosis, fibrosis and NAFLD activity scores decreased three months after surgery (48). In addition, we also found an early remission of NAFLD after LSG could be accurately predicted using the proposed nomogram based on the patients’ preoperative TyG, HSI and Prdx1 concentration.

A previous study reported that preoperative steatosis (p = 0.006) and insulin resistance index (p = 0.010) were independent predictors of postsurgical persistent severe steatosis one year after surgery via liver biopsy (49). However, the wide application of serial liver biopsy is limited due to its invasiveness, variability in sample collection, and inconsistent evaluation. In terms of imaging exams, MRI is not routinely performed in NAFLD as it is time-consuming and expensive. A meta-analysis indicated that B-ultrasound and transient elastography failed to identify and stage NAFLD in individuals with high BMI (50). Comparatively, a serological model could demonstrate high accuracy in diagnosing hepatic steatosis. HSI is composed of BMI, DM and ALT/AST ratio. Encouragingly, a previous study showed that its AUC for diagnosing NAFLD was 0.81 in the training cohort and 0.82 in the validation cohort (51). More importantly, our study suggested that HSI not only had a good performance in diagnosing NAFLD but was also an important independent predictor of early NAFLD remission after LSG (OR: 0.861, 95% CI: 0.765-0.969, p=0.013). Our correlation analysis of preoperative indicators showed that preoperative TyG was significantly associated with insulin resistance and obesity, consistent with a previous report (52). Interestingly, we also found that HSI was positively correlated with circulating Prdx1 (r=0.383, p =0.002). Several studies have observed increased inflammatory proteins, such as C-reactive protein, which declined after bariatric surgery in T2DM patients with obesity (53).

Circulating PRDX1 was recently identified as a damage-associated molecular pattern (DAMP) that acted as a proinflammatory factor by activating the Toll-like receptor 4-mediated signaling (54). In chronic inflammation, more than a quarter of Prdx1 was secreted actively to the outside of cells as exosomes (55, 56) or through a mechanism of redox-dependent secretion (57). Cross-sectional studies showed that Prdx1 increased in chronic metabolic diseases (i.e., atherosclerosis, diabetes, etc.). However, there is limited research on Prdx1 in the evolution of NAFLD. Our study showed for the first time a high circulating Prdx1 concentration in NAFLD patients with obesity and that its level decreased one month after LSG. Moreover, high baseline Prdx1 levels were associated with poor remission of NAFLD three months after LSG (OR: 0.887, 95% CI: 0.816-0.963, p=0.004). We also observed that Prdx1 plasma concentration positively correlated with BMI, HSI, AST and ALT. Given these observations, combined with recent evidence which showed that the overexpression of Prdx1 induced by oxidative stress enhanced insulin resistance and the accumulation of lipid droplets in mouse’s HepG2 cells (6), we speculated that Prdx1 might be involved in inflammation, oxidative stress and liver fat deposition in NAFLD patients with obesity. However, this hypothesis could not be confirmed in this present study, and additional studies are needed for further clarification.

Of the complex factors in NAFLD (obesity, lipid metabolism disorders, genes, etc.), insulin resistance is the one that has changed most dramatically within one week after LSG. After one week following SG, HOMA-IR reduced significantly from 6.7 ± 11 to 3.0 ± 1.6 (p = 0.019) (3). In addition, a study that reported improved NAFLD with a low-calorie ketogenic diet (VLCKD) found that baseline insulin resistance was a positive independent predictor (r=0.414, p=0.009) of HSI reduction (58). Comparatively, a study of 4784 Chinese adults reported that TyG was a better predictor for NAFLD when compared with HOMA-IR (AUC=0.761) (59). Furthermore, we found that higher TyG was more likely to lead to NAFLD remission, suggesting that excessive insulin resistance might be strongly recommended as an indication of LSG intervention in NAFLD. The most likely reason was that insulin resistance was more significantly reversed in patients with higher baseline TyG, based on our correlation analysis, in which pre-surgical TyG showed a strong positive correlation with delta TyG (r=0.78, p< 0.001). Similarly, two five-year prospective studies found that patients with refractory insulin resistance had persistent obesity and steatosis after surgery (60, 61).

According to our findings, baseline TyG was an independent predictor of HSI remission at three months after LSG in NAFLD patients with obesity, and it is the first time that circulating Prdx1 is proposed as a promising biomarker to assess postoperative improvement of obesity and NAFLD. Significant insulin resistance remissions are more likely to occur in patients with higher TyG.

Despite the interesting and novel findings reported, our study had some limitations. First, the sample size was limited, even though we internally validated the predictive ability of the nomogram via bootstrapping ( Supplementary Figure 1 ), and compared it with another prediction model which was built ( Supplementary Figures 2 , 3A ) and validated ( Supplementary Figures 3B–D ) in all samples (n=93). Second, we used B-ultrasound results and surrogate parameters as a standard instead of histology in this study; thus, our prediction results might not perfectly match NAFLD remission assessed by pathological criteria. In addition, NASH is a severe complication of NAFLD, which might also be associated with fibrosis, cirrhosis and hepatocellular carcinoma. We could not assess the different pathological prognoses of NAFLD (non-alcoholic fatty liver and NASH) due to a lack of liver histology data in this study, even though HSI and ultrasonography were externally validated against liver biopsy or magnetic resonance spectroscopy (62).

In summary, TyG and Prdx1 were identified as important predictors of NAFLD remission, which could be used as an early non-invasive marker or as adjuncts to liver biopsy to confirm NAFLD remission. A future clinical follow-up study would provide further evidence for assessing the changes in TyG and Prdx1 over time in patients with NAFLD.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Ethics Committee of the Third Xiangya Hospital (Project identification code: R20047). The patients/participants provided their written informed consent to participate in this study.

All authors contributed to the conception and design of the study. XC, JM and ZF organized the database. XC, ZF, WX and LZ performed the statistical analysis. All authors wrote the first draft of the manuscript. All authors wrote sections of the manuscript. All authors contributed to the article and approved the submitted version.