Baseline characteristics
The study design and recruiting strategy for the subjects are shown in Fig. 1. The study included 1750 inpatients with DFUs who were continually followed up from January 2013 to December 2023. In the end, a total of 960 individuals were eligible for the current analysis. The baseline characteristics of the subjects are presented in Table 1. The average age of the participants was 63.23 ± 12.08 years, and 67.5% of the participants were males. Two groups of subjects were established based on the median TyG index level: low (< 9.12, n = 480) and high (≥ 9.12, n = 480). All participants were followed up every 6 months for 11 years, for a median of 83 months. Age, all-cause mortality, DBP, WBC, TC, TG, LDL-C, HDL-C, FBG, HbA1c, and the TyG index were significantly different (all P < 0.05) between the two groups. However, none of the following factors were significantly different (all P > 0.05): sex, previous CHD, PAD, DR, DPN, DN, alcohol consumption, infection severity, duration of diabetes, Hb, ALB, ALT, AST, or Cr.
Table 1
Baseline clinical characteristics of the subjects
Clinical parameters
|
All participants (n = 960)
|
Low-level (< 9.12)(n = 480)
|
High-level (≥ 9.12)(n = 480)
|
p value
|
Gender, n (%)
|
|
|
|
0.098
|
Female
|
312 (32.5)
|
144 (30.0)
|
168 (35.0)
|
|
Male
|
648 (67.5)
|
336 (70.0)
|
312 (65.0)
|
|
Age, years
|
63 (54,72)
|
64 (55,74)
|
62 (53,70)
|
0.003*
|
Hypertension, n (%)
|
|
|
|
0.014*
|
No
|
424 (44.2)
|
231 (48.1)
|
193 (40.2)
|
|
Yes
|
536 (55.8)
|
249 (51.9)
|
287 (59.8)
|
|
Previous CHD, n (%)
|
|
|
|
0.834
|
No
|
85 8(89.4)
|
430 (89.6)
|
428 (89.2)
|
|
Yes
|
102 (10.6)
|
50 (10.4)
|
52 (10.8)
|
|
Previous CVD, n (%)
|
|
|
|
0.008*
|
No
|
843 (87.8)
|
435 (90.6)
|
408 (85.0)
|
|
Yes
|
117 (12.2)
|
45 (9.4)
|
72 (15.0)
|
|
PAD, n (%)
|
|
|
|
1
|
No
|
478 (49.8)
|
239 (49.8)
|
239 (49.8)
|
|
Yes
|
482 (50.2)
|
241 (50.2)
|
241 (50.2)
|
|
DR, n (%)
|
|
|
|
0.096
|
No
|
697 (72.6)
|
360 (75.0)
|
337 (70.2)
|
|
Yes
|
263 (27.4)
|
120 (25.0)
|
143 (29.8)
|
|
DPN, n (%)
|
|
|
|
0.721
|
No
|
273 (28.4)
|
139 (29.0)
|
134 (27.9)
|
|
Yes
|
687 (71.6)
|
341 (71.0)
|
346 (72.1)
|
|
DN, n (%)
|
|
|
|
0.358
|
No
|
566 (59.0)
|
290 (60.4)
|
276 (57.5)
|
|
Yes
|
394 (41.0)
|
190 (39.6)
|
204 (42.5)
|
|
Smoking, n (%)
|
|
|
|
0.047*
|
No
|
709 (73.9)
|
341 (71.0)
|
368 (76.7)
|
|
Yes
|
251 (26.1)
|
139 (29.0)
|
112 (23.3)
|
|
Drinking, n (%)
|
|
|
|
0.495
|
No
|
794 (82.7)
|
393 (81.9)
|
401 (83.5)
|
|
Yes
|
166 (17.3)
|
87 (18.1)
|
79 (16.5)
|
|
Infection severity, n (%)
|
|
|
|
0.945
|
PEDIS 1 grade/uninfected
|
90 (9.4)
|
45 (9.4)
|
45 (9.4)
|
|
PEDIS 2 grade/mild
|
313 (32.6)
|
160 (33.3)
|
153 (31.9)
|
|
PEDIS 3 grade/moderate
|
391 (40.7)
|
195 (40.6)
|
196 (40.8)
|
|
PEDIS 4 grade/Severe
|
166 (17.3)
|
80 (16.7)
|
86 (17.9)
|
|
Duration of diabetes, years
|
9 (4,13)
|
8 (3,12)
|
10 (4,14.75)
|
0.074
|
SBP, mmHg
|
140.5 (124,157)
|
138 (124,154.75)
|
143 (125,159)
|
0.017*
|
DBP, mmHg
|
74 (65,81)
|
72 (64,79)
|
75 (66,83)
|
0.003*
|
WBC, ×10*9/L
|
8.39 (6.74,11.2)
|
8.37 (6.63,10.63)
|
8.39 (6.9,11.73)
|
0.028*
|
Hb, g/L
|
112 (96,127)
|
112 (95.25,126.75)
|
112 (97,127.75)
|
0.737
|
TC, mmol/L
|
4.11 (3.34,4.95)
|
3.88 (3.16,4.70)
|
4.29 (3.51,5.14)
|
< 0.001*
|
TG, mmol/L
|
1.21 (0.95,1.74)
|
1 (0.8,1.22)
|
1.66 (1.28,2.30)
|
< 0.001*
|
HDL-C, mmol/L
|
0.93 (0.74,1.12)
|
0.94 (0.76,1.16)
|
0.92 (0.7,1.1)
|
0.013*
|
LDL-C, mmol/L
|
2.58 (2.01,3.17)
|
2.42 (1.94,3.01)
|
2.71 (2.16,3.36)
|
< 0.001*
|
ALB, g/L
|
33.4 (28.53,37.1)
|
33.4 (28.93,36.8)
|
33.4 (28.1,37.5)
|
0.888
|
ALT, U/L
|
15 (11,22)
|
15 (11,22)
|
15 (11,22)
|
0.84
|
AST, U/L
|
18 (14,24)
|
19 (15,25)
|
18 (14,24)
|
0.069
|
Cr, µmol/L
|
87 (65,118)
|
85.5 (65.25,109)
|
88.5 (64.25,130.75)
|
0.073
|
FBG, mmol/L
|
8.96 (6.37,11.55)
|
6.71 (5.41,9.10)
|
10.82 (8.80,14.36)
|
< 0.001*
|
HbA1c, %
|
8.75 (7.2,11)
|
8.2 (6.7,10.6)
|
9.4 (7.7,11.4)
|
< 0.001*
|
TyG index, mg2/dl2
|
9.12 (8.68,9.53)
|
8.68 (8.35,8.93)
|
9.53 (9.32,9.89)
|
< 0.001*
|
Notes: Median (interquartile range) for continuous variables. Percentages (%) for categorical variables.
Abbreviations: CHD, coronary heart disease; CVD, cerebrovascular disease; PAD, peripheral arterial disease; DR, diabetic retinopathy; DPN, diabetic peripheral neuropathy; DN, diabetic nephropathy; SBP, systolic blood pressure; DBP, diastolic blood pressure; WBC, leukocyte; Hb, hemoglobin; TC, total cholesterol; TG, triglyceride; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; ALB, serum albumin; ALT, alaninetransaminase; AST, alaninetransaminase; Cr, creatinine; FBG, fasting blood glucose; HbA1c, glycosylated hemoglobin; TyG, triglyceride–glucose.
* P < 0.05.
ROC curve analysis of the diagnostic efficacy of the TyG index for predicting endpoint events
The results of the ROC curve analysis are shown in Fig. 2. ROC curve analysis revealed that the TyG index had a significant potential predictive value for MACCE risk and all-cause mortality in DFU patients, with ideal critical values of 8.97 and 9.01, AUCs of 0.617 and 0.603, sensitivities of 72.7% and 72.2%, and specificities of 46.4% and 44.6%, respectively.
Cox regression analysis
The Cox regression analysis results are shown in Table 2. Variables with P < 0.1 or those that were clinically significant according to univariate Cox regression analysis were included in multivariate Cox regression analysis. Three multivariate regression models were built and used to adjust for potential confounding factors for endpoint events gradually. Model I was not adjusted for. Model II was adjusted for sex and age with Model I. Model III was adjusted for hypertension, previous CHD, previous CVD, PAD, DPN, DN, smoking, drinking, duration of diabetes, SBP, and total cholesterol (TC) with Model II.
Table 2
Univariate and multivariate Cox proportional hazard models evaluating endpoint events in patients with DFUs based on TyG index levels
|
Hazard ratio, 95% CI, and P value
|
Model I
|
Model II
|
Model III
|
HR
|
95% Cl
|
p value
|
HR
|
95% Cl
|
p value
|
HR
|
95% Cl
|
p value
|
MACCEs
|
Low-level (< 9.12)
|
ref
|
-
|
-
|
ref
|
-
|
-
|
ref
|
-
|
-
|
High-level (≥ 9.12)
|
1.836
|
1.435–2.349
|
< 0.001*
|
1.952
|
1.523–2.503
|
< 0.001*
|
1.807
|
1.402–2.328
|
< 0.001*
|
All-cause mortality
|
Low-level (< 9.12)
|
ref
|
-
|
-
|
ref
|
-
|
-
|
ref
|
-
|
-
|
High-level (≥ 9.12)
|
1.639
|
1.041–2.582
|
0.033*
|
1.661
|
1.051–2.623
|
0.03*
|
1.661
|
1.039–2.656
|
0.034*
|
Nonfatal myocardial infarction
|
Low-level (< 9.12)
|
ref
|
-
|
-
|
ref
|
-
|
-
|
ref
|
-
|
-
|
High-level (≥ 9.12)
|
2.118
|
1.147–3.912
|
0.016*
|
2.323
|
1.252–4.311
|
0.008*
|
2.185
|
1.164-4.1
|
0.015*
|
Revascularization
|
Low-level (< 9.12)
|
ref
|
-
|
-
|
ref
|
-
|
-
|
ref
|
-
|
-
|
High-level (≥ 9.12)
|
2.258
|
1.075–4.744
|
0.032*
|
2.223
|
1.053–4.694
|
0.036*
|
1.96
|
0.911–4.22
|
0.085
|
Nonfatal stroke
|
Low-level (< 9.12)
|
ref
|
-
|
-
|
|
|
|
|
|
|
High-level (≥ 9.12)
|
1.787
|
1.249–2.555
|
0.001*
|
1.951
|
1.361–2.799
|
< 0.001*
|
1.761
|
1.216–2.55
|
0.003*
|
Notes: Model I: adjusted for none. Model II: adjusted for sex and age with Model I. Model III: adjusted for hypertension, previous CHD, previous CVD, PAD, DPN, DN, smoking, drinking, duration of diabetes, SBP, and TC with Model II. Abbreviations: HR, hazard ratio; CI, confidence interval; TyG, triglyceride-glucose; MACCEs, major adverse cardio-cerebral events. * P < 0.05. |
In Model I, there was a positive correlation between the TyG index and the risk of MACCEs (HR = 1.836; 95% CI = 1.435–2.349; P < 0.001) and all-cause mortality (HR = 1.639; 95% CI = 1.041–2.582; P = 0.033). Model II revealed similar positive associations between the TyG index and MACCEs (hazard ratio (HR) = 1.952; 95% CI = 1.523–2.503; P < 0.001) and all-cause mortality (HR = 1.661; 95% CI = 1.051–2.623; P = 0.03). Furthermore, the TyG index was strongly associated with MACCEs (HR = 1.807; 95% CI = 1.402–2.328; P < 0.001) and all-cause mortality (HR = 1.661; 95% CI = 1.039–2.656; P = 0.034) in Model III.
Additionally, the TyG index was also an independent risk factor for each component of MACCEs in Model I for nonfatal myocardial infarction (HR = 2.118; 95% CI = 1.147–3.912; P = 0.016), revascularization (HR = 2.258; 95% CI = 1.075–4.744; P = 0.032), and nonfatal stroke (HR = 1.787; 95% CI = 1.249–2.555; P = 0.001). Model II included nonfatal myocardial infarction (HR = 2.323; 95% CI = 1.252–4.311; P = 0.008), revascularization (HR = 2.223; 95% CI = 1.053–4.694; P = 0.036), nonfatal stroke (HR = 1.951; 95% CI = 1.361–2.799; P < 0.001), and Model III included nonfatal myocardial infarction (HR = 2.185; 95% CI = 1.164-4.1; P = 0.015); revascularization (HR = 1.96; 95% CI = 0.911–4.22; P = 0.085); and nonfatal stroke (HR = 1.761; 95% CI = 1.216–2.55; P = 0.003), revealing that the higher the TyG index was, the greater the risk of MACCEs was.
Restricted cubic spline model analysis
The correlation analysis results are shown in Fig. 3. Restricted cubic spline (RCS) model analysis indicated that there was an S-shaped nonlinear dose‒dependent positive relationship between the TyG index and the risk of MACCEs in patients with DFU (nonlinear P < 0.001). The risk of MACCEs gradually increases with increasing TyG index. This dose-dependent relationship was more pronounced, especially for TyG index ≥ 9.12.
Stratified analysis in subgroups
The forest plots in Fig. 4 present the subgroup-level stratified analysis by age, sex, hypertension, previous CHD, previous CVD, PAD, DPN, and DN. The adjustment variables included sex, age, hypertension, previous CHD, previous CVD, PAD, DPN, DN, smoking, drinking, duration of diabetes, SBP, and total cholesterol (TC). The findings showed that the TyG index did not significantly affect the risk of MACCEs among patients with DFUs (interaction P values > 0.05).
K‒M survival analysis
The data from the Kaplan‒Meier survival curve analysis are shown in Fig. 5. During the follow-up period, out of 960 participants, 271 experienced MACCEs (28.2%), of which 79 (29.15%) died. We analyzed the effect of different TyG indexes on the risk of endpoint events in patients with DFUs using Kaplan‒Meier survival curves and log-rank tests. It was clearly evident from the results that the high-level TyG index group had a greater cumulative incidence of MACCEs (all-cause mortality, nonfatal myocardial infarction, revascularization, and nonfatal stroke) than did the low-level group (all log-rank tests, P < 0.05).