Non-Fasting Plasma Triglycerides Are Positively Associated with Diabetes Mortality in a Representative US Adult Population

: This study aimed to investigate whether non-fasting plasma triglycerides were associated with diabetes mortality. It included 7312 US adult participants. Diabetes mortality data were obtained via the linkage to National Death Index (NDI) records. Hazard ratios of non-fasting plasma triglycerides for diabetes mortality were assessed using Cox proportional hazards models, adjusting for age, gender, ethnicity, obesity, poverty–income ratio, education levels, physical activity, alcohol consumption, cigarette smoking status, survey period, hypercholesterolemia, hypertension, diabetes, and family history of diabetes. Among these participants, 1180 had diabetes. A total of 420 diabetes-caused deaths were recorded during a mean follow-up of 16.8 years. A 1-natural-log-unit increase in non-fasting plasma triglycerides was associated with a 41% higher diabetes mortality risk (hazard ratio, 1.41; 95% confidence interval, 1.19–1.67). Participants with non-fasting plasma triglycerides in the highest quintile, versus those in the lowest quintile, had a 141% higher diabetes mortality risk (hazard ratio, 2.41; 95% confidence interval, 1.46–3.97). The positive association of non-fasting plasma triglycerides with diabetes mortality was independent of diabetes status at the baseline. In conclusion, this study demonstrated that non-fasting plasma triglycerides were positively associated with diabetes mortality, independent of diabetes status at baseline. Non-fasting triglycerides may be a therapeutic target for diabetes-related complications.


Introduction
According to the World Health Organization, diabetes affects 422 million individuals worldwide and causes 1.5 million deaths per year [1].The diabetes mortality rate has increased over time.The age-standardized mortality rate from diabetes is increased by 3% from 2000 to 2019 worldwide, whereas the increase is 13% in lower-middle-income countries [1].In the US, diabetes leads to 30.4 deaths per 100,000 population per year and is the eighth leading cause of death [2].Therefore, it is important to investigate the modifiable risk factors for diabetes mortality.
High triglycerides have been linked to cardiovascular events [3,4], cardiovascular mortality [5], and all-cause mortality [6].Recently, baseline fasting plasma triglycerides have been shown to be positively associated with diabetes mortality [7], suggesting that triglycerides may play a crucial role in glycemic control.However, whether non-fasting triglycerides are associated with diabetes mortality is unknown.Some guidelines have started to recommend the use of non-fasting triglycerides for general screening and risk evaluation [8,9].This shift from fasting to non-fasting triglyceride tests is supported by various reasons.Non-fasting tests are more comfortable and convenient than fasting tests, and they may be safer as certain people may experience hypoglycemia when fasting [8,9].In addition, non-fasting triglyceride levels are ~27 mg/dL Targets 2024, 2 94 above their fasting counterpart [8], and this difference is thought to be not clinically significant for most people [8].Most importantly, non-fasting triglycerides seem to a have similar or better prognostic value for general risk screening compared with their fasting counterpart [6,8].Therefore, it is clinically relevant to investigate whether non-fasting triglycerides are associated with diabetes mortality.
This study aimed to investigate the association of non-fasting plasma triglycerides with diabetes mortality using US adults who attended the National Health and Nutrition Examination Survey (NHANES) from 1988 to 2014.

Study Participants
A total of 7490 adults aged ≥ 20 years attended the NHANES from 1988 to 2014 and had their non-fasting (fasting time < 8 h) plasma triglycerides available.The following participants with missing data were excluded from this study: follow-up time (n = 14), blood hemoglobin A1c (HbA1c, n = 38), plasma glucose (n = 31), serum insulin (n = 45), cigarette smoking status (n = 1), or education (n = 49).The remaining 7312 participants were included in the final analysis.The study was conducted following the ethical standards laid down in the Declaration of Helsinki.It was approved by the National Center for Health Statistics Research Ethics Review Board.All procedures were performed following the guidelines of the Declaration of Helsinki.The participants' records were anonymized before being accessed by the author.

Definitions of Comorbidities
Diabetes was defined as HbA1c ≥ 6.5%, fasting plasma glucose ≥ 126 mg/dL, taking hypoglycemic drugs, or self-reported diagnosis [10].Hypercholesterolemia was defined as total cholesterol ≥ 240 mg/dL or self-reported diagnosis of hypercholesterolemia [11].Hypertension was defined as systolic blood pressure ≥ 140 mm Hg or diastolic blood pressure ≥ 90 mm Hg or prior diagnosis or treatment of hypertension [12].

Diabetes Mortality
Diabetes mortality was obtained from NHANES-linked mortality files and was defined as diabetes being listed as underlying cause of death [7].Follow-up time was defined as the time (in months) from the time when the blood was drawn at the Mobile Examination Center until death, or until the end of follow-up (i.e., 31 December 2015), whichever occurred first.

Statistical Analyses
Data were presented as mean and standard deviation for continuous variables or percentages for categorical variables.The difference in the baseline variables between participants with and without diabetes was analyzed by T-test (for continuous variables) or chi-squared test (for categorical variables) [17].Associations of non-fasting plasma triglycerides with diabetes markers (glucose, HbA1c, and insulin) were analyzed using scatter plots and linear regression.The association of plasma triglycerides with diabetes diagnosis was analyzed by binary logistic regression [18].
The performance of non-fasting triglycerides for classifying diabetes mortality was analyzed using receiver operating characteristic (ROC) curve analysis [19,20].The optimal cutoff was determined by the Youden Index [21].Hazard ratios (HRs) and 95% confidence intervals (CIs) of non-fasting plasma triglycerides for diabetes mortality were analyzed using Cox proportional hazards models [22], with or without adjustment for the following confounders: age, gender, ethnicity, obesity, poverty-income ratio, education levels, physical activity, alcohol consumption, cigarette smoking status, survey period, hypercholesterolemia, hypertension, diabetes, and family history of diabetes.Triglycerides, glucose, HbA1c, and insulin were natural-log-transformed to improve the data distribution in all the regression analyses.Sensitivity analyses were conducted by further adjustment for total cholesterol and HDL cholesterol.All the analyses were conducted using SPSS (version 27.0).A two-sided p value of <0.05 was considered as statistically significant.

Association of Non-Fasting Triglycerides with Plasma Glucose, Blood Hemoglobin A1c, Serum Insulin, and Diabetes
Scatter plots showed that non-fasting plasma triglycerides were positively associated with glucose, HbA1c, and insulin, independent of diabetes diagnosis at baseline (Figure 1).The positive associations remained after adjustment for all the tested confounders (Table 2).A 1-natural-log-unit increase in non-fasting plasma triglycerides (e.g., from 80 to 217 mg/dL) was associated with a 130% higher diabetes diagnosis risk (adjusted odds ratio, 2.30; 95% CI, 2.01-2.63;p < 0.001).    1 Triglycerides, glucose, hemoglobin A1c, and serum insulin were natural-log-transformed. 2 The analysis was adjusted for age, sex, ethnicity, obesity, poverty-income ratio, education, survey period, physical activity, alcohol consumption, smoking status, hypercholesterolemia, hypertension, diabetes, and family history of diabetes.

Association of Non-Fasting Triglycerides with Diabetes Mortality
During 122,940 person-years of follow-up (mean follow-up 16.8 years), 420 diabetescaused deaths were documented.A 1-natural-log-unit increase in non-fasting plasma triglycerides was associated with a 41% higher risk of diabetes mortality, which was independent of diabetes status at baseline (Table 3).An ROC curve analysis showed that non-fasting triglycerides classified diabetes mortality with an optimal cutoff of 135.5 mg/dL (p < 0.001) (Figure 2).The optimal cutoff was lower in participants without diabetes (103.5 mg/dL) than in participants with diabetes (135.5 mg/dL, Figure 2).
A Kaplan-Meier analysis was conducted to assess the association of triglycerides with diabetes mortality.In this analysis, triglyceride was treated as a dichotomous categorical variable using the optimal cutoffs determined in Figure 2. The results showed that participants with triglycerides above the cutoff had worse survival compared with those with triglycerides below the cutoff in the whole cohort (Figure 3).Similar results were obtained in the sub-cohorts of participants with or without diabetes (Figure 3).A Cox proportional hazards models analysis confirmed that triglycerides above the cutoff remained an independent risk factor for diabetes mortality after adjusting for all the tested confounders (Table 4).An ROC curve analysis showed that non-fasting triglycerides classified diabetes mortality with an optimal cutoff of 135.5 mg/dL (p < 0.001) (Figure 2).The optimal cutoff was lower in participants without diabetes (103.5 mg/dL) than in participants with diabetes (135.5 mg/dL, Figure 2).The optimal cutoff was 135.5 mg/dL in the whole cohort, with a sensitivity of 69%, a specificity of 57%, and an area under the curve (AUC) of 0.674.The optimal cutoff was 103.5 mg/dL for participants without diabetes, with a sensitivity of 80%, a specificity of 42%, and an AUC of 0.644.The optimal cutoff was 135.5 mg/dL for participants with diabetes, with a sensitivity of 73%, a specificity of 36%, and an AUC of 0.547.The blue line represented the ROC curve, and the red line represented the reference line.
A Kaplan-Meier analysis was conducted to assess the association of triglycerides with diabetes mortality.In this analysis, triglyceride was treated as a dichotomous categorical variable using the optimal cutoffs determined in Figure 2. The results showed that participants with triglycerides above the cutoff had worse survival compared with those with triglycerides below the cutoff in the whole cohort (Figure 3).Similar results were obtained in the sub-cohorts of participants with or without diabetes (Figure 3).A Cox proportional hazards models analysis confirmed that triglycerides above the cutoff remained an independent risk factor for diabetes mortality after adjusting for all the tested confounders (Table 4).The optimal cutoff was 135.5 mg/dL in the whole cohort, with a sensitivity of 69%, a specificity of 57%, and an area under the curve (AUC) of 0.674.The optimal cutoff was 103.5 mg/dL for participants without diabetes, with a sensitivity of 80%, a specificity of 42%, and an AUC of 0.644.The optimal cutoff was 135.5 mg/dL for participants with diabetes, with a sensitivity of 73%, a specificity of 36%, and an AUC of 0.547.The blue line represented the ROC curve, and the red line represented the reference line.

Sensitivtiy Analyses
Further analysis showed that similar results were obtained when triglycerides were stratified as quintiles: participants with non-fasting plasma triglycerides in the highest quintile, versus those in the lowest quintile, had a 141% higher risk of diabetes mortality (adjusted HR, 2.41; 95% CI, 1.46-3.97,Table 5).The positive association remained in those

Sensitivtiy Analyses
Further analysis showed that similar results were obtained when triglycerides were stratified as quintiles: participants with non-fasting plasma triglycerides in the highest quintile, versus those in the lowest quintile, had a 141% higher risk of diabetes mortality (adjusted HR, 2.41; 95% CI, 1.46-3.97,Table 5).The positive association remained in those with or without diabetes at baseline (Tables 6 and 7).Sensitivity analyses showed that the positive association remained after a further adjustment for total cholesterol and HDL cholesterol (Tables 5-7).

Discussion
Using a representative cohort of US adults, this study found, for the first time, that nonfasting plasma triglycerides were positively associated with diabetes mortality.This study extended the previous finding that triglycerides are positively associated with diabetes mortality from the fasting state [7,23] to the non-fasting state.Similar to the previous finding [7], such an association was independent of diabetes status at the baseline.Therefore, non-fasting triglycerides might be used to detect those with a high risk of diabetes mortality.
The mechanism underlying the positive association between triglycerides and diabetes mortality is unclear.A few hypotheses have been put forward.For example, triglycerides promote inflammation [24].Additionally, a higher number of triglycerides may co-exist with other morbidities, such as hypercholesterolemia and hypertension, and diabetes [25]; however, after adjusting for these co-morbidities, our results showed that triglycerides remained positively associated with diabetes mortality.Some guidelines have started to recommend the use of non-fasting triglycerides for general screening and risk evaluation [8,9].The results of the study also support the use of non-fasting triglycerides for risk prediction, which is consistent with previous reports [6,8].
This study found that the optimal cutoff of non-fasting triglycerides for diabetes mortality was 135.5 mg/dL.This cutoff is supported by the mortality data: those with non-fasting triglycerides above the cutoff had a 37% higher multivariable-adjusted risk of diabetes mortality compared with those below the cutoff.However, this cutoff is lower than the current hypertriglyceridemia threshold of 150 mg/dL [40,41].The lower threshold of triglycerides for cardiovascular incidence [42,43], all-cause mortality [6,44], and diabetes mortality [23] have been reported previously.Therefore, whether the hypertriglyceridemia cutoff of 150 mg/dL should be lowered needs to be further investigated in the future.
The present study has a number of strengths, e.g., a large sample size (n = 7312) derived from a nationally representative adult sample, prospective study design, and adjustment for many confounders.
This study has some limitations.Firstly, it lacked multiple triglyceride measurements throughout the study [45].Secondly, mortality outcomes were ascertained by the linkage to National Death Index (NDI) records with a probabilistic match, which might result in misclassification, although this matching method had a high accuracy of 98.5% [46,47].

5 Figure 1 .
Figure 1.Scatter plots of non-fasting triglycerides with plasma glucose, blood hemoglobin A1c, and serum insulin.The plots on the left were from all the participants, the plots in the middle were from participants without diabetes, and the plots on the right were from participants with diabetes.B values represented correlation coefficients.p < 0.05 for all the associations.

Figure 1 .
Figure 1.Scatter plots of non-fasting triglycerides with plasma glucose, blood hemoglobin A1c, and serum insulin.The plots on the left were from all the participants, the plots in the middle were from participants without diabetes, and the plots on the right were from participants with diabetes.B values represented correlation coefficients.p < 0.05 for all the associations.
family history of diabetes.

Figure 2 .
Figure 2. ROC curves of non-fasting triglycerides to classify diabetes mortality.The optimal cutoff was 135.5 mg/dL in the whole cohort, with a sensitivity of 69%, a specificity of 57%, and an area under the curve (AUC) of 0.674.The optimal cutoff was 103.5 mg/dL for participants without diabetes, with a sensitivity of 80%, a specificity of 42%, and an AUC of 0.644.The optimal cutoff was 135.5 mg/dL for participants with diabetes, with a sensitivity of 73%, a specificity of 36%, and an AUC of 0.547.The blue line represented the ROC curve, and the red line represented the reference line.

Figure 2 .
Figure 2. ROC curves of non-fasting triglycerides to classify diabetes mortality.The optimal cutoff was 135.5 mg/dL in the whole cohort, with a sensitivity of 69%, a specificity of 57%, and an area under the curve (AUC) of 0.674.The optimal cutoff was 103.5 mg/dL for participants without diabetes, with a sensitivity of 80%, a specificity of 42%, and an AUC of 0.644.The optimal cutoff was 135.5 mg/dL for participants with diabetes, with a sensitivity of 73%, a specificity of 36%, and an AUC of 0.547.The blue line represented the ROC curve, and the red line represented the reference line.

Figure 3 .
Figure 3. Kaplan-Meier curves of triglycerides for diabetes mortality.The blue line (the top line) represented participants with triglycerides below the optimal cutoff, and the red line (the bottom line) represented participants with triglycerides above the optimal cutoff.

Figure 3 .
Figure 3. Kaplan-Meier curves of triglycerides for diabetes mortality.The blue line (the top line) represented participants with triglycerides below the optimal cutoff, and the red line (the bottom line) represented participants with triglycerides above the optimal cutoff.

Table 1 .
Baseline characteristics of the study cohort.

Table 2 .
Association of non-fasting plasma triglycerides 1 with plasma glucose, 1 blood hemoglobin A1c, 1 and serum insulin, 1 analyzed by linear regression.

Table 3 .
Non-fasting plasma triglycerides 1 and risk for diabetes mortality.
Abbreviations: CI, confidence interval; HR, hazard ratio. 1 Triglycerides were natural-log-transformed. Model 1: unadjusted; Model 2: adjusted for age, sex, and ethnicity; Model 3: adjusted for all the factors in Model 2 plus obesity, poverty-income ratio, education, and survey period; Model 4: adjusted for all the factors in Model 3 plus physical activity, alcohol consumption, and smoking status; Model 5: adjusted for all the factors in Model 4 plus hypercholesterolemia, hypertension, diabetes, and family history of diabetes.

Table 4 .
Non-fasting plasma triglycerides for diabetes mortality, stratified according to dichotomous triglycerides (above versus below the optimal cutoff).
Abbreviations: CI, confidence interval; HR, hazard ratio.Model 1: unadjusted; Model 2: adjusted for age, sex, and ethnicity; Model 3: adjusted for all the factors in Model 2 plus obesity, poverty-income ratio, education, and survey period; Model 4: adjusted for all the factors in Model 3 plus physical activity, alcohol consumption, and smoking status; Model 5: adjusted for all the factors in Model 4 plus hypercholesterolemia, hypertension, diabetes, and family history of diabetes.

Table 4 .
Non-fasting plasma triglycerides for diabetes mortality, stratified according to dichotomous triglycerides (above versus below the optimal cutoff).
Abbreviations: CI, confidence interval; HR, hazard ratio.Model 1: unadjusted; Model 2: adjusted for age, sex, and ethnicity; Model 3: adjusted for all the factors in Model 2 plus obesity, poverty-income ratio, education, and survey period; Model 4: adjusted for all the factors in Model 3 plus physical activity, alcohol consumption, and smoking status; Model 5: adjusted for all the factors in Model 4 plus hypercholesterolemia, hypertension, diabetes, and family history of diabetes.

Table 5 .
Non-fasting plasma triglycerides in quintiles and adjusted risk for diabetes mortality in the whole cohort 1 .: CI, confidence interval; HR, hazard ratio; Q, quintile (Q1 = lowest quintile and Q5 = highest quintile). 1 out of 7312 participants had missing HDL cholesterol values and were excluded.Therefore, the remaining 7268 participants were included in the analysis.2Totalcholesterol and HDL cholesterol were naturallog-transformed.Model 1: adjusted for age, sex, ethnicity, obesity, poverty-income ratio, education, survey period, physical activity, alcohol consumption, smoking status, hypercholesterolemia, hypertension, diabetes, and family history of diabetes. Abbreviations

Table 6 .
Non-fasting plasma triglycerides in quintiles and adjusted risk for diabetes mortality in the sub-cohort of participants without diabetes 1 .

Table 7 .
Non-fasting plasma triglycerides in quintiles and adjusted risk for diabetes mortality in the sub-cohort of participants with diabetes 1 .