Introduction

Peripheral artery disease (PAD) is a central clinical problem affecting about 1.52% of adults older than 40 years while reports estimate > 20% in adults aged 80 or more [1,2,3]. This athero-occlusive condition leads to hind limb ischemia, and despite progress in drug therapies and revascularization, the prognosis remains poor, with almost half of the patients requiring limb amputation [4]. In addition, PAD is often underdiagnosed, with most patients not receiving optimal treatments that are proven to ameliorate prognosis and reduce mortality [5]. To overcome this problem, a diagnostic test such as a sufficiently sensitive and specific blood test would be expected to improve the recognition and treatment of these individuals. However, since PAD shares common risk factors and pathogenetic mechanisms with other cardiovascular diseases, no specific biomarker has yet been identified and implemented in clinical practice [5].

However, in the last decade, particular attention has been focused on the association between diabetes and PAD. Indeed, as demonstrated by clinical and preclinical studies, diabetes in human and animal models of PAD correlates with worsened outcomes [6, 7]. Notably, insulin resistance (IR) is undoubtedly one of the most prominent pathophysiological factors in this association. IR is a typical hallmark of type 2 diabetes mellitus (T2DM), exerting direct effects on the vasculature [8, 9], and as demonstrated by a cross-sectional study by Pande et al. [10], IR is strongly and independently associated with PAD. Similar results were observed by Britton et al. [11] in a population-based study of older American adults (Cardiovascular Health Study).

Interestingly, the studies above assessed IR using the homeostatic model of insulin resistance (HOMA-IR). However, using this method in clinical practice presents several limitations, including costs, time of execution, and invasiveness, making it very challenging, especially for undeveloped countries [12]. Therefore, a more straightforward and low-cost estimating tool, like the triglyceride–glucose (TyG) index, has also been used to identify a relationship between IR and PAD, with results that are not conclusive yet.

A previous systematic review and meta-analysis have shown that a higher TyG index is associated with the incidence of arterial stiffness [13]. Several studies have assessed the correlation between TyG and PAD in terms of diagnosis and prognosis. However, these results have not been pooled and there has been no systematic review that investigated these studies. Based on this premise, as the TyG index, this study aimed to consolidate current evidence regarding the relationship between the TyG index and PAD and test its diagnostic and predictive value for adverse outcomes.

Methods

Protocol and guidelines

The study was registered in PROSPERO (the International Prospective Register for Systematic Reviews) under the registration number (CRD42024548468) and was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [14].

Search strategy and study selection

A thorough systematic search was conducted in international databases including PubMed, Scopus, Web of Science, and Embase using a combination of Medical Subject Headings (MeSH) terms and other keywords to identify all studies published up until the 24th of April 2024 (Supplementary Table 1).

We included studies that (1) evaluated the relationship of TyG index and PAD through quartiles and tertiles and/or by 1-unit increase, (2) measured TyG index in patients with PAD (and a control group), and (3) compared TyG levels based on PAD severity. The following studies were excluded: (1) Review studies, (2) conference abstracts, (3) case reports, (4) non-English language publications, (5) studies not reporting TyG levels, and (6) studies with non-PAD population.

TyG index, as a measure of IR, is calculated from TG and FPG as follows:

$${TyG}=\mathrm{ln}\left({TG}\left(\frac{{mg}}{{dL}}\right)\,\times \,\frac{{FPG}\left(\frac{{mg}}{{dL}}\right)}{2}\right)$$

The records were merged and duplicates were removed using the EndNote software. Subsequently, two reviewers (AS and AK) separately screened the titles and abstracts of all identified studies before full-text screening relevant records to determine their eligibility. Disagreements between the reviewers were resolved through a third reviewer (AHB).

Data extraction and quality assessment

Data extraction was performed by two independent authors (AS and AK) and any disagreements were resolved through mutual consensus. The following information was extracted and charted: first author’s name, publication year, country of origin, study design, study population, sample size, mean age, male percentage, main findings of each study, and the odds ratios (ORs) with their 95% confidence (CI).

The Newcastle-Ottawa Scale (NOS) was used to assess the quality of the included studies [15]. The three main domains assessed by the NOS include selection, comparability, and outcome. On this scale, a score of ≥ 7 is considered high quality.

Statistical analysis

The statistical analyses were performed in R version 4.3.0 (R Core Team [2021], Vienna, Austria). For comparison of the TyG index in patients with PAD and healthy controls, standardized mean difference (SMD) and its 95% CI were calculated through random-effect meta-analysis. Odds ratios (ORs) and 95% CIs for a 1-unit increase in TyG index association with PAD were also assessed by random-effect meta-analysis. Similarly, the highest vs. lowest quartile/tertile of TyG comparison was made using meta-analysis.

When the median and interquartile range were reported, the mean and standard deviation were calculated using methods developed by Luo and Wan [16, 17]. The bias-corrected Hedges’ g SMD was used to compare the mean TyG index in patients with PAD and controls and groups with different severity levels. We used the I2 statistic to evaluate heterogeneity. I2 > 50% reflected significant heterogeneity. A sensitivity analysis using the leave-one-out method was performed to assess each study’s effects on the overall pooled effect size. For investigation of publication bias, Egger’s test [18] and visual inspection of the funnel plot by trim-and-fill method were used. P < 0.05 was considered statistically significant throughout the analyses.

Results

Literature search and included study characteristics

The initial search led to a total of 247 records (51 results from PubMed, 62 from Embase, 53 from the Web of Science, and 81 from Scopus). After removing duplicates (n = 141), 65 studies were excluded for title and abstract reasons. Other 19 articles were excluded after full-text screening. A complete PRISMA flowchart and description of the search process and screening are shown in Fig. 1.

Fig. 1
figure 1

PRISMA diagram flowchart representing the search and study selection process.

Full size image

At the end of this screening process, 22 studies were finally included and analyzed [19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40]. Table 1 illustrates the features of these studies as well as their populations’ baseline characteristics. Studies were conducted between 2018 and 2024, and the leading country was China, with 13 studies [23, 24, 27, 30, 32,33,34,35,36,37,38,39,40], followed by Turkey [19, 21, 25, 31], the United States [22, 28], Taiwan [20], Japan [29], and South Korea [26] with a total population composed of 73,168 cases. Among the studies, three [30, 31] examined patients with PAD and compared them with healthy controls [19], two compared different severities of PAD in terms of the TyG index [21, 25], and the others were population-based studies that included four diabetic populations [20, 30, 32, 34] and two hypertensive ones [27, 35]. Based on the NOS criteria for quality assessment of observational studies, all studies had high quality, as shown in Supplementary Table 2.

Table 1 Baseline and population characteristics of included studies.
Full size table

TyG levels in patients with PAD vs control

Three studies assessed TyG levels in patients with PAD and compared them with healthy controls [19, 30, 31]. Caliskan et al. [19], in a single-center, observational, retrospective study, analyzed 440 patients (211 PAD and 229 healthy controls) evaluated with color Doppler ultrasonography. Interestingly, the presence of T2DM was higher in the PAD group than in the control group (p < 0.001). In line with this observation, the TyG index was significantly higher in the PAD group than in healthy controls (9.19 ± 0.57 vs. 8.80 ± 0.59; P < 0.001). In addition, these authors demonstrated that the TyG index was an independent parameter of PAD (OR = 1.111, 95% CI = 1.083–1.139; p < 0.001).

For their part, Ning et al. [30] assessed the TyG index in 1040 patients with diabetes, of which 168 were diagnosed with lower extremity arterial disease (LEAD). These authors observed that the TyG index was higher in the LEAD group compared to the control group (9.94 ± 0.78 vs. 9.36 ± 0.70, P < 0.001). In addition, the TyG index was independently associated with LEAD risk in patients with diabetes (OR = 3.92, 95% CI = 2.92–5.26; P < 0.001).

Finally, Pala et al. [31], in a retrospective study, analyzed a population of 296 patients diagnosed with PAD divided into two groups according to the Rutherford Classification (Group 1 (category 0 to 3) and Group 2 (category 4 to 6). Those in group 2 who developed chronic limb-threatening ischemia (CLTI) and had a higher TyG index compared to those in group 1 (9.27 ± 0.31 vs. 9.00 ± 0.34, P < 0.001).

Notably, the meta-analysis performed on these studies showed that patients with PAD had statistically significantly higher levels of the TyG index (SMD 0.76, 95% CI 0.65 to 0.88, P < 0.0001). There was no heterogeneity in this analysis (I2: 0%). The forest plot for this analysis is shown in Fig. 2.

Fig. 2
figure 2

Forest plot for meta-analysis of TyG levels in patients with PAD vs. controls.

Full size image

TyG levels in different severities of PAD

Next, we assessed two studies to compare PAD severities regarding the TyG index [21, 25]. In a retrospective observational comparative study, Cora and colleagues [21] analyzed 200 patients divided into two groups (group 1=moderate and group 2=severe) with PAD grouped based on Global Limb Anatomic Staging System (GLASS) criteria. Patients in group 2 had significantly higher TyG index than in group 1 (9.21 ± 0.61 vs. 8.96 ± 0.54, P = 0.040).

Similarly, Karaduman et al. [25] examined the TyG index in 71 patients with PAD. The patients were divided into two groups according to the angiographically detected lesions and the TransAtlantic InterSociety Consensus-II (TASC-II) classification (Group 1 = TASC-II grade A–B and Group 2 = TASC-II grade C–D). Importantly, these authors observed that patients in Group 2 had a higher TyG index than those in Group 1. Finally, a meta-analysis of these studies revealed that the TyG index was associated with the severity of PAD (SMD 0.48, 95% CI 0.22 to 0.74, P = 0.0003, Fig. 3).

Fig. 3
figure 3

Forest plot for meta-analysis of TyG levels in patients with severe PAD vs. moderate PAD.

Full size image

Change in risk of PAD by one unit increase in TyG

Seven studies evaluated the association between one unit increase in the TyG index and the risk of PAD using brachial-ankle pulse wave velocity (baPWV) [23, 24, 27, 32, 34, 38, 39]. Of note, the meta-analysis of these studies revealed that one unit increase in the TyG index was associated with a 60% higher risk of PAD (OR 1.60, 95% CI 1.41 to 1.80, P < 0.0001, Fig. 4). This analysis had 69% heterogeneity (P < 0.01). None of the studies had a significant effect on the overall effect size (Supplementary Fig. 1). Finally, Egger’s test did not indicate the presence of publication bias (P = 0.55). However, the funnel plot showed asymmetry and the addition of two studies led to an OR of 1.49 (95% CI 1.29 to 1.71) (Supplementary Fig. 2).

Fig. 4
figure 4

Forest plot for meta-analysis of one unit increase in TyG level association with high baPWV.

Full size image

Change in risk of PAD by TyG tertiles

Three studies compared the risk of PAD (high baPWV) in the TyG tertiles [24, 34, 38]. We performed a meta-analysis comparing the risk of PAD between the third and first tertiles of the TyG index in the study population. Our pooled analysis demonstrated that patients in the third tertile of the TyG index had a 72% higher risk of PAD compared to the reference tertile (OR 1.72, 95% CI 1.42 to 2.07, P < 0.0001, Fig. 5). This analysis was not associated with heterogeneity between studies (I2 = 0, P = 0.53).

Fig. 5
figure 5

Forest plot for meta-analysis of tertile 3 vs. tertile 1 of TyG association with high baPWV.

Full size image

Change in risk of PAD by TyG quartiles

Six studies compared the risk of PAD (high baPWV) in the TyG quartiles [26, 27, 29, 33, 39, 40]. In our meta-analysis, we compared the risk of PAD between the fourth and first quartiles of the TyG index in the study population. In the pooled analysis, we observed that patients in the fourth quartile of the TyG index had a 94% higher PAD risk than the reference quartile (OR 1.94, 95% CI 1.49 to 2.54, P < 0.0001, Fig. 6). This analysis was associated with high heterogeneity between studies (I2 = 85%, P < 0.01). While there was no significant effect of each individual study on the overall effect size by leave-one-out analysis (Supplementary Fig. 3), the funnel plot showed asymmetry for publication bias (Supplementary Fig. 4). The addition of four studies led to higher odds of PAD in Q4 vs. Q1 (OR 1.41, 95% CI 1.00 to 1.97, P = 0.049). However, Egger’s test was not significant for this analysis (P = 0.107).

Fig. 6
figure 6

Forest plot for meta-analysis of quartile 4 vs. quartile 1 of TyG association with high baPWV.

Full size image

Diagnostic ability of TyG index for PAD

Two studies explored the diagnostic ability of the TyG index in PAD [19, 23]. In their research, Caliskan et al. [1] found that the TyG index had acceptable predictive ability for PAD (sensitivity: 57.8%, specificity: 70%, AUC: 0.689). In addition, Guo and colleagues’ study found that AUC for predicting the TyG index was 0.708 in women and 0.580 in men [23].

Discussion

This study was the first to systematically examine the potential association between the TyG index and PAD. Accordingly, we performed a comprehensive analysis of 22 out of 247 studies that were exhaustively searched and identified on four leading online databases (Pubmed, Embase, Scopus, and Web of Science), ensuring the inclusion of the most relevant and reliable data. Notably, among the essential findings resulting from our analysis, we report three main results. The first is that a higher TyG index is associated with PAD. Secondly, we revealed that the TyG index was considerably higher in patients with severe PAD compared with those with moderate PAD. Hence, this suggests that the TyG index is a valuable marker of PAD severity. Finally, our analysis revealed that an increase in the TyG index was associated with a more increased risk of PAD, as higher TyG quartiles had elevated rates of PAD.

Diabetes is a worldwide health problem characterized by elevated blood glucose levels induced by defects in insulin production, IR, or both [37, 38]. Consequently, diabetes and IR contribute to developing vascular disorders, including PAD, impairing vasodilation, elasticity of the arterial walls (arterial stiffness), and increasing intima-media thickness and vascular calcification [39]. Importantly, these IR-related vascular dysfunctions are also present in asymptomatic PAD subjects and remain highly undetected, significantly limiting adequate treatments and contributing to overall mortality [40]. Therefore, assessing IR may positively affect PAD patients in terms of therapeutic choice, survival, and morbidity. Among the tools used to evaluate IR, the TyG index remains the best choice because of its low costs, execution time, and safety. In addition, compared to other methods, like HOMA-IR, the TyG index had higher sensitivity in predicting prediabetes/diabetes diagnosis [41] and is more strongly associated with arterial stiffness and atherosclerosis diseases [30]. Therefore, based on this premise, different studies have assessed the association between the TyG index and PAD, and the results from these studies still need to be unified.

Among several methods of IR assessment, the traditional HOMA-IR has been used as a valuable and reliable marker [41]. However, due to difficulties in measuring insulin levels, especially in developing countries with limited resources, the availability of this method of IR assessment has been questioned [41]. On the other hand, TyG index calculation is more financially reasonable and easily available through common laboratory measurements, as it has shown promising results in the identification of IR among normal populations [42, 43]. In line, this index has outperformed HOMA-IR in several studies [43,44,45]. Hence, given the importance of PAD and its assessment, and considering the availability of TyG measurement, its use is suggested in these conditions.

Collectively, our study comprehensively analyzed all these studies and corroborated the importance of assessing the TyG index in the clinical assessment of PAD. In addition, it supports the valuable predictive role in patients at risk of PAD development. In clinical settings, since TG and FPG are routinely measured for patients with high cardiovascular risk, including those with PAD, the calculation of TyG and considering it in the assessment of prognosis in patients with PAD could be easy and reliable if confirmed in future studies.

Strengths and limitations

The fundamental strength of the present study lies in the large number of included studies and cases that can provide substantial evidence and insights for further research. Meta-analysis of continuous increase in TyG and TyG quartiles is another strength of the current review. However, five main limitations need to be acknowledged in this study. First, results may be influenced by study populations and sample sizes. Second, our research did not consider all metabolic conditions that could affect the results. Third, using different TyG index cutoffs and quartiles in various studies may introduce bias. Fourth, most existing studies are cross-sectional, which limits their ability to establish causality. Finally, the few studies included in the meta-analyses of PAD vs. control and PAD severities limit our findings and warrant additional studies. These limitations contribute to the mixed findings and emphasize the necessity for further research to clarify the relationship between the TyG index and PAD.

Conclusion

To date, there is no ideal specific marker for PAD, neither for risk stratification nor for diagnosis. In the last decades, even though the direct role of IR in PAD has not been fully established, a linkage between these conditions has been proven. In this context, we reported the clinical utility of assessing the TyG index, an easy-to-dose IR marker and a valuable index of several IR-related disorders. For instance, we have previously documented the association of this index with several IR-related CV and non-CV disorders, supporting its valuable diagnostic and prognostic role [12, 46,47,48,49,50]. In line with this notion, the present study’s findings provide evidence for a significant association between this index and PAD and demonstrate the usefulness of this index as a predictive marker of PAD incidence. Therefore, our study supports the importance of including the TyG index assessment in the clinical setting to identify individuals at risk for PAD and CVDs in general, in which IR plays a pathogenic role.