A retrospective study utilized MIMIC-IV database to explore the potential association between triglyceride-glucose index and mortality in critically ill patients with sepsis

Background

Sepsis is a life-threatening organ dysfunction caused by an imbalance in the body’s response to infection. The incidence of blood glucose metabolism disorder is relatively high in sepsis patients, and its pathophysiology is that when the body is infected, a large amount of pro-inflammatory mediators is released, leading to peripheral insulin resistance¹. This is typically manifested as hyperglycemia and increased glycemic variability (GV)². Despite plenty of studies^3–5 elucidating the mechanisms of glucose metabolism disorders, the interplay between overall blood glucose and GV remains unclear. Early studies^6,7 demonstrated that infection and inflammation can lead to significant changes in lipids and lipoproteins, resulting in a decrease in lipid and lipoprotein levels in sepsis patients. Lower baseline lipid levels are correlated with an increased risk of sepsis and poorer prognosis⁸. It has been proved that⁹ decreased plasma triglyceride (TG) levels are linked to higher mortality in sepsis patients, though its use as a prognostic marker remains debated. At present, the pathological and physiological mechanisms by which blood lipids affect the prognosis of sepsis patients are not well understood, and whether various lipoproteins can serve as predictive biomarkers and therapeutic targets for sepsis and renal dysfunction has not been investigated.

With the extensive research on insulin sensitivity parameters, several simple insulin sensitivity parameters that can be applied clinically and accurately reflect patients’ insulin sensitivity have gradually been validated, and the triglyceride-glucose (TyG) index is one of them¹⁰. The TyG index is positively correlated with contrast-induced nephropathy, risk of ischemic stroke¹¹, and severity and prognosis of coronary artery disease¹², which is supported by moderate-quality evidence. Although prior studies^13,14 have highlighted the potential prognostic value of TyG in sepsis patients, they have not investigated the survival time of the population or the presence of a nonlinear association between the TyG index and short-term mortality risk in sepsis patients. Therefore, the relationship between the TyG index and mortality rate in sepsis patients remains unraveled.

In the Surviving Sepsis Campaign (SSC), early identification and improved prognosis were two major goals of sepsis treatment¹⁵, and the application of improvement plans for sepsis patients, including sepsis screening, intervention effectiveness assessment, and patient prognosis evaluation are recommended¹⁶. Therefore, it is urgent to identify novel biomarkers that can predict the prognosis of sepsis. This study employed the open-source critical care database, multiple statistical models, survival analysis, and subgroup analysis to shed light on the association between the TyG index and hospital/ICU mortality rate in sepsis patients.

Methods

Results

Among the adult patients in the MIMIC-IV database, a total of 22,518 subjects met the eligibility criteria. From the database, 135 prognostic factors were initially extracted. Following data cleaning, 50 predictors with over 30% missing data were excluded. In the end, 85 forecast factors were included in the model.

Discussion

Our study retrospectively delved into the correlation between the TyG index and mortality rate in critically ill sepsis patients based on the latest MIMIC database. RCS was also employed to explore the nonlinear relationship between the TyG index and mortality rate. To account for potential confounding variables, Cox regression analysis, merging multiple models, and appropriately classified subgroup analysis were carried out. Our findings indicate that a higher TyG index is associated with increased in-hospital and ICU mortality rates among critically ill sepsis patients. Moreover, there was a non-linear correlation between the TyG index and in-hospital mortality and ICU mortality of critically ill sepsis patients. Furthermore, subgroup analyses were performed to compare hazard risks across different patient demographics and conditions, including age, sex, BMI, hypertension, type 2 diabetes, heart failure, continuous renal replacement therapy, and ventilation. The results showed consistency across these subgroups, with no significant differences, suggesting that these factors did not substantially impact the reliability of the TyG index within our study population. This provides valuable insights into the relationship between glucose and lipid metabolism parameters and clinical outcomes in severe sepsis.

Recent studies from China, South Korea, and Spain have confirmed the predictive value of the TyG index for type 2 diabetes^24–26, and it has also been linked to hypertension²⁷, coronary artery stenosis²⁸, hyperuricemia²⁹, and other conditions. Most research has focused on the association between the TyG index and major adverse cardiovascular events. Our study, however, found a significant correlation between a higher TyG index and increased risk of in-hospital and ICU mortality in severe sepsis patients. These results are consistent with findings from Fang et al.¹⁵ and Zheng et al.¹⁶. By further examining the cumulative survival time of patients with severe sepsis and the nonlinear relationship between the TyG index and outcomes, our study offers additional evidence supporting our hypothesis and provides a more comprehensive understanding of the association between the TyG index and clinical outcomes.

The association between the TyG index and the risk of poor prognosis in sepsis patients may be attributed to disruptions in glucose and lipid metabolism observed in these individuals. The coexistence of hyperglycemia and dyslipidemia during sepsis, and their significant correlation with disease prognosis, underscore the importance of understanding the metabolic disturbances and their metabolites. This may facilitate the identification of early biomarkers with prognostic potential. In sepsis, the body experiences hypercatabolism driven by stress hormones and pro-inflammatory cytokines. Initially, the stress response leads to the secretion of various hormones and neurotransmitters, stimulating the liver to synthesize and release substantial amounts of glycogen into the bloodstream, thereby increasing blood glucose levels³⁰. As the disease advances, enhanced gluconeogenesis results in elevated levels of anaerobic metabolites in the circulation, exacerbating hyperglycemia³¹. Additionally, the presence of infectious agents may impair pancreatic function, thereby reducing the production of insulin³². Recent studies have highlighted that during sepsis, a substantial release of inflammatory factors occurs. These factors can impair insulin signaling and contribute to insulin resistance through mechanisms involving free fatty acids³³, oxidative stress³⁴, and endoplasmic reticulum stress³⁵. Such acute insulin resistance can occur within a few hours after the body’s acute response and last for several weeks³⁶. Similarly, lipids also play an important role in the pro-inflammatory and anti-inflammatory effects of sepsis microvasculature, and changes in lipid metabolism and activation of lipid signaling pathways have been found to be components of the complex environment underlying the pathophysiology of sepsis^37,38. During the onset and progression of severe sepsis, triglyceride levels rise more significantly than other lipoproteins. This increase is driven by cytokines, leading to the synthesis and secretion of triglyceride-rich low-density lipoproteins by the liver, a condition clinically referred to as the high-fat state of sepsis. It is now also considered a part of the body’s innate immunity, especially in binding and inactivating lipopolysaccharides (LPS)³⁹. The protective effect of lipoproteins in infected patients is primarily due to their ability to bind and neutralize LPS from gram-negative bacteria and lipophilic acids from gram-positive bacteria. This interaction effectively blocks the biological activity of endotoxins, preventing them from activating inflammatory cells^40,41. Animal experiments have shown that after LPS injection or live infection, animals with increased blood lipoprotein concentration have lower levels of cytokines, reduced organ dysfunction, increased LPS clearance, and improved survival rates^42–44. In critical cases, lipolysis of adipose tissue can elevate triglyceride levels, particularly when lipid absorption is impaired⁴⁵. Additionally, in sepsis, increased VLDL synthesis by the liver and decreased VLDL clearance by peripheral tissues exacerbate lipid metabolism disorders⁴⁶. The mechanism is related to the action of some cytokines (TNF-α, IL-1, IL-6, and INF-γ)^47,48 on the liver to increase fatty acid synthesis. Therefore, abnormal glucose and lipid metabolism may be crucial in the progression of sepsis, and the TyG index could potentially serve as a marker for monitoring these metabolic changes and even predicting the prognosis.

A systematic review and meta-analysis by Hofmaenner et al.⁴⁹ found that in critically ill sepsis patients, lower levels of total cholesterol, high-density lipoprotein cholesterol, and low-density lipoprotein were associated with higher mortality, whereas triglyceride levels did not show a similar association. This finding was corroborated by a clinical study⁵⁰. In contrast, Muhammad et al.⁵¹ and Yin et al.⁵² highlighted a link between the TyG index and cardiovascular events, while Liu et al.⁵³ reported no association between the TyG index and cardiovascular or all-cause mortality, suggesting contradictory outcomes. Notably, none of these studies specifically focused on critically ill patients with relevant disease conditions. Previous research^54,55 indicates that the TyG index may be a more accurate measure of insulin resistance compared to alternatives like the Homeostasis Model Assessment of Insulin Resistance (HOMA-IR). In critical care settings, the TyG index has demonstrated potential prognostic value. A Mendelian randomization study identified a causal relationship between a higher TyG index and reduced stroke risk in critically ill patients⁵⁶, and Huang et al.⁵⁷ observed a significant negative correlation between the TyG-BMI index and all-cause mortality in severe stroke cases. These studies suggest a negative correlation between the TyG index and adverse clinical outcomes. In a retrospective study of non-surgical intensive care patients, Nishigoori et al.⁵⁸ found that a high TyG index was correlated with negative outcomes in patients with acute coronary syndrome, but also found that a low TyG index was associated with adverse outcomes in acute heart failure. It seems that the relationship between high or low TyG index and clinical outcomes depends on specific outcome types under specific disease conditions. However, more studies have reported a positive relationship between high TyG index and poor prognosis in critically ill patients. For example, Liu et al.⁵⁹ and Yang et al.⁶⁰ both found that a high TyG index correlates with increased mortality in critically ill patients with cerebral hemorrhage. Although findings are not entirely consistent, they underscore the potential of the TyG index in predicting mortality among critically ill patients. Additionally, a review⁶¹ noted persistent lipid abnormalities in sepsis survivors and highlighted the role of these factors in atherosclerosis, which may contribute to mortality during hospitalization or ICU admission. Thus, the TyG index, integrating indicators of lipid and glucose metabolism, may serve as a valuable marker for insulin resistance and a predictor of metabolic abnormalities in acute sepsis.

This study offers therapeutic insights into the management of critical sepsis and advocates for a comprehensive risk management strategy to mitigate sepsis-related mortality associated with elevated TyG index levels. Such a strategy should include active management of cardiovascular risk factors such as lipid levels, body mass index (BMI), fasting blood glucose, and glycosylated hemoglobin. Regular monitoring and timely intervention for ICU patients with elevated TyG index levels are essential to reduce adverse outcomes. While the “gold standard” for insulin assessment is the hyperinsulinemic-euglycemic clamp test, its extensive duration, complexity, and high cost limit its routine clinical application. In contrast, the TyG index, which relies on simple laboratory biochemical markers, offers a more economical and time-efficient alternative. This suggests that the TyG index has the potential to evolve into a straightforward and effective tool for monitoring sepsis prognosis. Dynamic assessment of the TyG index in clinical practice could facilitate real-time evaluation of ICU sepsis patients’ prognoses, allowing for more tailored and cautious treatment strategies in cases of identified high mortality risk. Additionally, the TyG index may serve as a valuable evaluation tool for assessing the efficacy of treatments targeting glucose and lipid metabolism disorders in sepsis patients. Continuous monitoring of the TyG index and analysis of its trends can aid in the adjustment of therapeutic interventions for glucose and lipid management.

However, this study has several limitations. The study cohort, derived from the MIMIC-IV database, provides a large sample size and reliable results, yet the critically ill population studied may not fully represent a continuous spectrum of metabolic status. Residual confounding remains a possibility within the model, and specific causes of patient mortality are not accessible. Moreover, the exclusion of certain laboratory test data due to missing values may introduce bias. Lastly, this study represents a retrospective and observational investigation; therefore, it cannot definitively establish a causal relationship between the TyG index and mortality in patients with severe sepsis. Future research of a higher evidence level is needed to validate the association between the TyG index and clinical outcomes in septic patients.

Conclusions

In conclusion, an elevated TyG index is associated with a higher risk of in-hospital and ICU mortality in critically ill sepsis patients. Our analysis demonstrated a nonlinear relationship that remained consistent across subgroups, thereby enhancing the stability and robustness of our findings. Moreover, this study, with its comprehensive statistical analyses and larger patient cohort in comparison to previous investigations, clearly quantifies the heightened mortality risk correlated with the TyG index. Our research underscores the utility of the TyG index as a crucial biomarker for clinicians and its role in facilitating early identification and intervention for high-risk sepsis patients in the ICU. Therefore, the foregoing results may lay a foundation for future prospective studies aimed at validating these observations and evaluating potential interventions based on the TyG index to improve sepsis patient outcomes.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

JQ L, ZY X and XY Z contributed equally to this work. JQ L and N H designed and conceptualized the study. JQ L, ZY X, XY Z, JY S, JL L, SY C and GY J completed the record retrieval and data extraction. Mathematical modeling and meta-analysis were conducted with the help of YF F and JY S. The original draft was written by JQ L, ZY X and XY Z. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the HwaMei Research Foundation of Ningbo No.2 Hospital (Grant No.2022HMKY48 and No.2023HMZD07), Medical Scientific Research Foundation of Zhejiang Province (Grant No.2023RC081), the Project of NINGBO Leading Medical & Health Discipline (Project No.: 2022-F17), and the Ningbo Top Medical and Health Research Program (No.2023030615). Funders played no role in the study design, execution, or manuscript writing.

Data availability

Raw data supporting the obtained results are available at the corresponding author.

Declarations