Our systematic review adheres to the methodological standards outlined in the Cochrane Handbook for Systematic Reviews of Interventions10 and PRISMA guidelines.11 The study protocol was registered in PROSPERO (CRD42024543269) prior to implementation. A comprehensive search was conducted in major databases, including Medline (via PubMed), Web of Science (WoS), Scopus, Embase, and Science Direct. Our search strategy utilized controlled vocabulary (MeSH terms) and free-text terms combined with Boolean operators to address our PECOS research question: Population (adults with acute pancreatitis), Exposure (elevated TyG index), Comparator (normal TyG index), Outcome (progression to severe acute pancreatitis), and Study type (observational studies). Keywords were carefully selected to encompass relevant exposures (triglyceride and glucose index) and the primary outcome (“acute pancreatitis” OR “severe acute pancreatitis”). Further details on the search strategy are provided in the Supplementary material (Table 1S).

All articles identified during the primary and secondary searches were compiled using Zotero® 6.0.15. After duplicate removal, manuscripts were transferred to Rayyan, where two authors (GAVT and OYAV) independently and blinded reviewed the titles and abstracts. Articles selected by both authors were included, and discrepancies were resolved by a third author (EDMR) through an additional review. The selected articles underwent a full-text review to determine their eligibility. The selection process is detailed in Fig. 1.

Fig. 1.

PRISMA 2020 flow diagram of the selection process of the primary studies included.

Observational studies (prospective or retrospective cohorts, case-control, and cross-sectional studies) were included, as well as studies involving adult patients of both sexes diagnosed with acute pancreatitis, regardless of etiology, publication date, or language. These studies had to include TyG index as a variable for evaluating severity. Excluded studies included case reports or case series, abstracts, letters to the editor, duplicate studies across two or more databases, and studies involving pediatric patients. The TyG index was calculated in all included articles as: Ln [fasting TG (mg/dL) × fasting plasma glucose (mg/dL)] / 2.8,12

The primary outcome assessed was the progression to SAP in patients initially admitted with AP. All selected studies used the severity definition from the Atlanta 2012 criteria. Severity was defined as the persistence of organ failure beyond 48 hours.3,8,13

Two researchers collected and extracted relevant information from each study using an Excel spreadsheet. Extracted data included author, country, year, population characteristics, exposure, hospital and ICU stay, TyG values, measures of association, confounding variables, and outcomes. For dichotomous variables, odds ratios (OR) with 95% confidence intervals (CI) were recorded based on events in exposed and unexposed groups. For continuous variables, data were recorded as mean and standard deviation (SD) for exposed and unexposed groups or as mean and interquartile range (IQR).14

We utilized the Inverse Variance (IV)15 method in the meta-analysis to combine adjusted odds ratios (ORs) and their corresponding 95% confidence intervals (CIs). The analysis was conducted using R® version 4.2.2 software, leveraging the meta library. Forest plots were generated to summarize the quantitative synthesis, employing the metabin16 function for dichotomous variables and the metacont17 function for continuous variables. The IV method with the DerSimonian-Laird (DL) approach for tau² estimation was applied.

Heterogeneity among studies was assessed using Cochran’s Q test and Higgins’ I² statistic. When substantial heterogeneity was detected (I² > 40%), a random-effects model without Hartung-Knapp (HK) adjustment was applied. To further explore heterogeneity, we performed a sensitivity analysis using the InfluenceAnalysis function and the Graphic Display of Heterogeneity (GOSH)18 through the gosh.diagnostics function.

Each study was individually assessed for risk of bias. The risk of bias in the primary studies was evaluated using the Newcastle-Ottawa Scale (NOS) tool.19

The search was conducted in databases (Medline via PubMed, Scopus, Embase, WoS, and Science Direct) using an advanced search strategy, yielding a total of 66 articles. Twenty duplicates were removed, leaving 46 articles for screening. Titles and abstracts were reviewed, and 35 articles were excluded, resulting in 11 articles selected for full-text review. During this process, four additional articles were excluded—three due to incomplete data and one for having a different study design. Ultimately, 7 studies were included in the systematic review8,13,20–24 (Fig. 1), comprising 5 retrospective cohort studies and two prospective cohort studies.

A total of 1,624 patients with AP were included in this meta-analysis, of whom 18% (176 patients) progressed to SAP. The primary studies analyzed were conducted in Asia and Africa, specifically in countries such as Korea,8 China,13,20,23,24 India,21 and Egypt.22 The mean age across all studies and patient groups who developed SAP exceeded 50 years. The main etiology of AP was gallstones, followed by hypertriglyceridemia, and lastly, alcohol-related causes (Table 1).

The risk of bias was assessed using the Newcastle-Ottawa Scale (NOS),19 which evaluates the validity and quality of observational studies, particularly non-randomized studies (NRS). All 7 studies demonstrated a low risk of bias, indicating high methodological quality in terms of selection, adjustment for confounding variables, and reliable outcomes (Table 2).

Of the seven studies included in the meta-analysis, only four8,13,20,23 reported the TyG index with measures of association (OR and 95% CI) for SAP. The meta-analysis of these studies revealed a 48% increased risk of developing SAP in patients with elevated TyG (OR 2.48; 95% CI 1.50–4.08; I² = 80%) (Fig. 2). Due to the high heterogeneity, a sensitivity analysis was conducted using InfluenceAnalysis, which identified Park et al.8 as an outlier. After excluding this study, the analysis showed a consistent association between elevated TyG index and SAP risk, with moderate heterogeneity (OR 1.83; 95% CI 1.27–2.65; I² = 67%) (Fig. 3).

Fig. 2.

Forest plot del TyG as a predictor of SAP.

Fig. 3.

Forest Plot of TyG index as a predictor of SAP after excluding Park et al., showing a moderate reduction in heterogeneity.

When TyG was analyzed as a numerical variable ((8,13,20–24), higher mean values were consistently observed in the SAP group compared to the non-SAP group. The meta-analysis demonstrated that the effect size of TyG values, expressed as the standardized mean difference (SMD) between the non-SAP and SAP groups, was 0.90 (95% CI: 0.70–1.11; I² = 28%), highlighting a significant and substantial difference between the two groups (Fig. 4).

Fig. 4.

Forest plot of TyG as a continuous numerical variable (standardized mean difference, SMD): higher values observed in the SAP group compared to the non-SAP group.

Our analysis of the seven primary studies included in this meta-analysis demonstrates that the TyG index is a significant predictor of progression to SAP (OR 2.48; 95% CI 1.50–4.08; I² = 80%). Patients with SAP showed higher mean TyG index values compared to those who did not progress to greater severity, highlighting its potential as an early risk predictor. This finding supports the integration of the TyG index into traditional severity prediction models for SAP, regardless of etiology.

Currently, there are no previous SR-Ms to compare with our findings. However, the role of the TyG index has been implicated and demonstrated in various diseases beyond AP, including diabetes, hypertension, NAFLD, and cardiovascular events. The TyG index is primarily considered a surrogate marker of insulin resistance (IR) and metabolic abnormalities, given that IR is associated with both hyperglycemia and hypertriglyceridemia.25,26

In the context of AP, peripancreatic fat inflammation plays a crucial role in its pathophysiology. Metabolic disturbances linked to IR, such as hyperglycemia and hypertriglyceridemia, can exacerbate both immune and non-immune responses, contributing to the onset or worsening of SAP severity.27 Specifically, hypertriglyceridemia has been identified as a critical factor in the progression to more severe disease forms. Elevated triglycerides may be hydrolyzed into free fatty acids by pancreatic lipases in damaged tissues, inhibiting mitochondrial function in pancreatic acinar cells. This cascade results in elevated intracellular calcium, inflammatory cytokine release, and tissue injury, clinically manifesting as organ damage or SAP.25–27

Observational studies like a Park et al.8 enrolled 373 patients from three hospitals and found that the TyG index was significantly higher in those who developed SAP, with a 48% increased risk of progression (OR 2.48; 95% CI 2.80–18.19). Additionally, the study showed that incorporating the TyG index into predictive models improved the area under the curve (AUC) from 0.74 to 0.83 when combined with clinical and laboratory variables such as CRP, procalcitonin, age, sex, comorbidities (e.g., diabetes), BMI, and etiology.

On the other hand, Wei et al.20 studied 353 AP patients in Shanghai, categorizing them by severity. After adjusting for confounders in a multivariate logistic regression analysis, the TyG index was an independent predictor of progression to SAP (OR 1.83; 95% CI 1.38–2.44; p < 0.001). The study also evaluated the TyG index by etiology (hypertriglyceridemia [HTG] vs. non-HTG), demonstrating its predictive value in both groups.

Moreover, Lee et al.28 examined 313 patients and found that those with a higher TyG index were more likely to require ICU admission and had prolonged hospital stays. The analysis reported an OR of 2.59 (95% CI 1.27–5.30) for predicting SAP and an AUC of 0.731, emphasizing its predictive utility.

Although there are other severity prediction scores, such as PANC3,3 BISAP (Bedside Index for Severity in Acute Pancreatitis), the Ranson score, and the Computed Tomography Severity Index (CTSI), comparisons with TyG have only been attempted in primary studies. Park et al.8 evaluated the AUC of these scores and found that BISAP outperformed both the Ranson score and TyG, with AUCs of 0.82, 0.79, and 0.79, respectively. Similarly, Sayiti et al.13 compared their predictive performance for SAP and reported that BISAP (AUC 0.82) was superior to CTSI and TyG, which had AUCs of 0.78 and 0.65, respectively. While both BISAP and CTSI are reliable severity scores, CTSI is an imaging-based score and therefore not an early predictor. In contrast, BISAP, as a clinical score, can be applied within the first 24 hours, similar to TyG. Although TyG has been evaluated in only a few studies, it has shown promising results, is easy to calculate, and may be considered a biomarker based on triglycerides and fasting glucose.

The TyG index offers notable advantages for clinical practice due to its accessibility and ease of calculation. Unlike other severity assessment systems such as APACHE II, BISAP, or Ranson scores, which require multiple physiological and biochemical parameters, the TyG index relies solely on fasting triglyceride and glucose levels. This simplicity makes it a particularly valuable tool in resource-limited settings, allowing clinicians to quickly and effectively identify high-risk patients and tailor their management strategies accordingly.3,6,8

The evidence presented in this study suggests that the TyG index could be integrated as an early and accessible marker for managing AP, aiding not only in severity prediction but also in optimizing hospital resource allocation and treatment planning.

Therefore, we believe that stratifying at-risk patients using the TyG index will enable a more targeted treatment approach for those at higher risk of severe disease progression. However, further research is needed to determine the optimal cutoff point for its application across different populations and assess whether the etiology of pancreatitis influences these thresholds. Additionally, it is essential to compare the TyG index with other validated scores or predictors and evaluate its potential as a prognostic marker for hospitalization and mortality.

Our study has several strengths. First, we employed a comprehensive search strategy across six major databases and clinical trial registries. Second, we implemented a robust methodology, including a thorough quality assessment and statistical analyses addressing heterogeneity. Third, our findings provide robust data supporting the role of the TyG index as a predictor of SAP, with consistent results across individual studies that employed rigorous statistical evaluations, controlled for confounders, and used uniform definitions for calculating the TyG index.

However, some limitations must be acknowledged. First, when assessing the strength of association using ORs, we observed high heterogeneity due to the inclusion of all AP patients without stratifying by etiology or uniformly controlling for confounders. Second, most studies were conducted in Asia, necessitating further research in diverse populations to enhance generalizability across different ethnicities and geographic regions.