Association of triglyceride-glucose index with atherosclerotic cardiovascular disease and mortality among familial hypercholesterolemia patients

NHANES is a two-year-cycle cross-sectional survey conducted by the Centers for Disease Control and Prevention (CDC) of America, involving a home interview and a medical examination, offering demographics, socioeconomic status, dietary and health information as well as physical and physiological measurements of U.S. population [17]. NHANES study protocol was approved by The National Center for Health Statistics (NCHS) ethics committee (Protocol #2011–17, https://​www.​cdc.​gov/​nchs/​nhanes/​irba98.​htm). Informed consent from all the participants was obtained before participating.

There were 1456 out of 116876 participants in the survey diagnosed as FH in accordance with a Dutch Lipid Clinical Network (DLCN) index that was higher than or equal to 3 points, as described previously [18]. Among them, TyG index was available in 1120 individuals (n = 336 excluded). Participants (n = 179) were excluded due to loss of follow-up. Therefore, a total of 941 individuals (530 females and 411 males), is the final study population (Fig. 1). The study population was divided into 3 groups: < 8.5, 8.5–9.0, > 9.0 to make the participants number in the groups comparable, as reported previously [19].

Information on socioeconomic conditions, behavior and history of diseases was obtained through questionnaires by experienced interviewers. Drinking was defined as having at least 12 alcohol drinks per year; smoking was defined as having smoked ≥ 100 cigarettes in life [20]. History of ASCVD, hypertension or diabetes was defined as self-reported physician diagnosis [21]. History of ASCVD was defined as self-reported physician diagnosis. “Has a doctor or other health professional ever told {you/SP} that {you/s/he} had a coronary heart disease/angina, also called angina pectoris/heart attack (also called myocardial infarction)/stroke?” was a question on the medical conditions section of the household questionnaires via home interview, and those who answered “yes” were deemed to have a history of ASCVD. Body mass index (BMI) was calculated using weight (kg)/height (m2). Income status of the family was described with poverty-income ratio, which was the ratio of family income to the poverty threshold. TyG index was calculated as Ln (fasting triglycerides [TG, mg/dl] × fasting blood glucose [mg/dl]/2) [10]. Laboratory results were obtained from serum specimens when they visited the mobile examination center and vials were stored under appropriate frozen (− 30 °C) conditions until they were shipped to National Center for Environmental Health for testing [22]. The original homeostatic model assessment insulin resistance index (HOMA-IR) was calculated as fasting insulin × fasting blood glucose/22.5 [23]. Homeostatic model assessment insulin sensitivity index (HOMA-IS) was calculated as 1/HOMA-IR. Details for each variable measurement is published on the NHANES website [24].

The period of follow-up lasted from the date of the interview through the last follow-up time, Dec 31 2019, or the date of death, whichever came first. Records from the NDI provided information on these including participants' causes of death. The endpoints for this study included all-cause mortality and cardiovascular death, which encompassed cardiac death (e.g., sudden cardiac death and myocardial infarction) and vascular death (e.g., stroke) [1]. The median follow-up duration is 114 months (interquartile range, 57–159 months). The maximum follow-up duration is 248 months.

Data were analyzed using SPSS complex sample module version 22.0 (IBM Corp, Armonk, NY) and R (version 4.2.2). The Kolmogorov–Smirnov normality test was adopted to test the normality of continuous variables. Normally distributed variables were described with mean ± standard deviation (SD). Variance analysis was adopted to compare the mean levels while Chi-square tests were chosen to compare the percentages of categorical variables across the different groups.

Spearman correlation analysis was used to test the association of TyG index and various established glucose metabolism-related indicators (fasting blood glucose, fasting insulin, HbA1c, HOMA-IR and HOMA-IS index. Both univariable and multivariable-adjusted logistic regression were used to calculate the odds ratio (OR) with 95% confidence interval (CI) for the relationship between TyG index and ASCVD. The possible nonlinear relationships between TyG index and the all-cause or cardiovascular death were further evaluated on a continuous scale with restricted cubic spline (RCS) curves based on the multivariable Cox proportional hazards models, with four nodes at the fixed percentiles of 5%, 35%, 65% and 95% of the distribution of TyG index. The event-free survival rates among the groups were estimated by the Kaplan–Meier method and compared by the log-rank test. Cox regression analysis was used to assess the association between TyG index and mortality. The following variables were utilized as covariates in the study population: age, gender, BMI, smoke, drink, HOMA-IR, low-density lipoprotein cholesterol (LDL-C), creatinine and hypertension. To assess the added prognostic value of TyG index beyond the original model, C-index was calculated, using predict.coxph function to predict Cox model with predict value type = "survival". Furthermore, a sensitivity analysis was applied to further investigate the association of TyG index with ASCVD and mortality in FH patients with IR (HOMA-IR ≥ 2.69) [25]. A two-sided p < 0.05 was considered statistically significant.

Among the 941 FH participants in this study, 204 (21.68%) had TyG index < 8.5 (low TyG index), 311 (33.05%) had TyG index ≥ 8.5 and ≤ 9.0 (moderate TyG index), and 426 (45.27%) had TyG index > 9.0 (high TyG index; Table 1). As expected, those with higher TyG index had higher levels of fasting blood glucose, fasting total glyceride (TG), fasting insulin, HbA1c and HOMA-IR index (all p < 0.001), in addition to higher occurrence rate of diabetes and impaired fasting blood glucose (IFG; p < 0.001) than those with lower TyG index. Besides, individuals with higher TyG index were also more likely to be older, white and had higher BMI, systolic blood pressure (SBP), fasting total cholesterol (TC) and LDL-C as well as lower high-density lipoprotein cholesterol (HDL-C; all p < 0.05) compared with those with relatively lower TyG index. There was no significant difference among the four groups in DLCN score, poverty-income ratio, drink, smoke, hypertension history, diastolic blood pressure (DBP), alanine aminotransferase (ALT), aspartate aminotransferase (AST) as well as serum creatinine (p > 0.05).

As a novel indicator of IR, the diagnostic value of TyG index in FH population was examined by assessing the association with those well-recognized indicators reflecting glucose metabolism status including fasting blood glucose, HbA1c, fasting insulin, HOMA-IR and HOMA-IS. As shown in Table 2, TyG index was positively associated with fasting glucose, HbA1c, fasting insulin and HOMA-IR index, and negatively associated with HOMA-IS (all p < 0.001) among FH patients, indicating that TyG index was applicable to FH individuals to reflect the glucose metabolism status.

Given the idea that TyG index is associated with the development and prognosis of cardio-cerebrovascular diseases [26, 27], we evaluated the relationship in FH population by conducting a multivariable-adjusted logistic analysis (Table 3). After adjustment with age, gender, BMI, smoke, drink, HOMA-IR, LDL-C, creatinine and hypertension, the OR was 2.38 (95% CI 1.25–4.53, p = 0.01) in the group with high TyG index compared with the group with low TyG index; while no significant difference was observed between the group with moderate and low TyG index (OR, 1.45; 95%CI 0.73–2.89, p = 0.29). For every 1 unit increase of TyG index, the risk of ASCVD increased by 74% after adjustment (95%CI 1.15–2.63, p = 0.01). In the sensitivity analysis, TyG index remained significantly associated with ASCVD in FH patients with IR (Crude OR, 1.52, 95%CI 1.09–2.13, p = 0.02; Adjusted OR, 2.14; 95%CI 1.19–3.86, p = 0.01). These results indicated that increase of TyG index was independently associated with the elevation of ASCVD risk among FH adults.

In this study, the median follow-up duration is 114 months (interquartile range, 57–159 months). In Fig. 2, we used RCS to flexibly model and visualize the relationship with all-cause and cardiovascular mortality in FH adults. The risk of both all-cause and cardiovascular mortality was relatively flat until around 8.6 of TyG index, and then started to increase rapidly afterwards (p for non-linearity = 0.0083 and 0.0046, respectively). The study population was divided into three groups (< 8.5, 8.5–9.0 and > 9.0) due to the strong U/J-shaped relation between TyG index and mortality. Figure 3 depicts the cumulative hazard of all-cause death and cardiovascular death in the groups with different TyG index. FH patients reported > 9.0 had lower survival probability compared with those reported low or moderate TyG index (log-rank p = 0.035). No significant difference was observed in the incidence of cardiovascular mortality (log-rank p = 0.093). Cox analysis was utilized to further assess the association of TyG index with all-cause death and cardiovascular death (Table 4). After multivariable adjustment, TyG index was independently associated with both all-cause death (HR, 1.55; 95%CI 1.10–2.18, p = 0.01 for every 1 unit increase of TyG index) and cardiovascular death (HR, 1.79; 95%CI 1.04–3.09, p = 0.04 for every 1 unit increase of TyG index). In sensitivity analysis (Table 5), TyG index remained significantly associated with all-cause and cardiovascular death in FH patients with IR (HR, 1.91; 95%CI 1.12–3.24, p = 0.02 and HR, 2.73; 95%CI 1.11–6.74, p = 0.03, respectively). The adjusted HR was 1.61 (95%CI 1.07–2.42, p = 0.02) for all-cause death and 2.09 (95%CI 1.02–4.30, p = 0.045) for cardiovascular death in the group with high TyG index comparing with the group with moderate TyG index. However, no significant difference was observed when comparing the two groups reported low and moderate TyG index (HR, 1.19; 95%CI 0.71–2.00, p = 0.51 for all-cause death; HR, 2.19; 95%CI 0.96–4.97, p = 0.06 for cardiovascular death). In Cox prediction models, C-statistic values were 0.489 (95%CI 0.437–0.542) and 0.408 (95%CI 0.324–0.491) for survival from all-cause death and cardiovascular death with traditional risk factors, respectively (Table 6). Addition of TyG index to the original model resulted in significant improvements in discrimination of both survival from all-cause death (C-statistic 0.483; 95%CI 0.432–0.535, p = 0.042) and cardiovascular death (C-statistic 0.397; 95%CI 0.315–0.479, p = 0.042). These results demonstrated that TyG index was an independent risk factor for all-cause death and cardiovascular death among FH adults.