Dietary Glycemic Index and Glycemic Load in Relation to Atherosclerotic Stenosis of Carotid and Cardiovascular Risk Factors in Ischemic Stroke Patients

Abstract

Aim: Glycemic index (GI) and glycemic load (GL) influence postprandi al glucose concentrations and insulin responses. This study aims to ascertain the connection between GI, GL, and carotid atherosclerotic stenosis and cardiovascular disease (CVD) risk factors.

Methods: A total of 669 patients with ischemic stroke within 7 days were enrolled. GI and GL were assessed with a validated food frequency questionnaire from patients. Computed tomography angiography (CTA) was used for the evaluation of carotid atherosclerotic stenosis. Traditional risk factors such as total cholesterol, triglycerides, LDL-C, HDL-C, C-reactive protein, homocysteine, neutrophil to lymphocyte ratio (NLR), fasting plasma glucose, and hemoglobin A1c were measured. GI/GL and its association with CVD risk factors and carotid stenosis were explored with Spearman analysis and multivariable logistic regression, respectively.

Results: The prevalence of carotid stenosis was 63.2% of all 669 participants. The mean value of GI/GL was 49.3/137. Spearman test did not detect significant relationships between GI/GL and CVD risk factors. In multivariable regression models, GI (4^th vs. 1^st quartile, OR = 2.11; 95% CI, 1.30–3.42) and GL (4^th vs. 1^st quartile, OR = 1.82; 95% CI, 1.12–2.96) were observed a significant association with carotid stenosis after adjustment for major confounding factors. The association between GL and carotid stenosis became more pronounced among yo ungers (4^th vs. 1^st quartile, OR = 2.42; 95% CI, 1.13–4.76) and women (4^th vs. 1^st quartile, OR = 3.81; 95% CI, 1.45–5.05).

Conclusion: Higher GI and GL were positively associated with a higher degree of carotid stenosis in these Chinese cerebral infarction patients, especially in younger patients and women.

See correspondence vol. 27: 1243–1244

Introduction

The concept of glycemic index (GI) was originally designed for recommendations of diabetics' diets and was proposed in 1981 by Jenkins et al. GI is a quantitation assessment on the ability of accumulated carbohydrate from the diet to increase blood glucose levels¹⁾. Glycemic load (GL) combines the quality and quantity of overall digested carbohydrates, calculating from multiplying GI of food items by the amount of consumed carbohydrates. Higher carbohydrat es consumption leads to insulin secretion from pancreas; in response, the tissues took in more blood glucose. So we can predict the concentration of postprandial plasma insulin/glucose concentrations and the degree of insulin sensitive in the body through calculating GI/GL²⁾.

Atherosclerosis is a predominant factor of stroke (ischemic or hemorrhagic) events. Approximately one third of ischemic stroke is caused by extracranial carotid stenosis. Research showed that carotid stenosis due to atherosclerosis accounted for 90% of the total carotid stenosis³⁾. Therefore, the prevention of carotid atherosclerotic stenosis is of great importance to the occurrence of stroke.

High concentrations of postprandial glucose and plasma insulin will increase the chronic oxidative stress levels, influence endothelial function, and may play a key role in the generation of cardiovascular diseases (CVD). GI/GL was reported to be associated with blood pressure⁴⁾, type 2 diabetes⁵⁾, and cardiovascular disease risks⁶⁾.

In diabetes, poor glycemic control was associated with the severity of intracranial atherosclerosis⁷⁾. But to date, the relationship between dietary GI/GL, one dominant factor of poor glycemic control, and carotid atherosclerotic stenosis has not been explored. This study aimed to investigate these relationships plus whether GI/GL influencing CVD risk factor in a large scale of symptomatic adults with ischemic stroke.

Materials and Methods

Study Population

The patients of this study were screened from 2016 to 2019 via the Nanjing Stroke Registry Program (NSRP) conducted by the Center of Clinical Research for Stroke Screening and Prevention of Jinling Hospital. This study protocol was approved by Jinling Hospital Ethic Review Committee. Informed consent was received from all participants or their proxies.

The including criteria were patients with ischemic stroke in 7 days and aged over 18 years old; the excluding criteria were patients who had cancer, severe liver/kidney insufficiency, chronic gastrointestinal diseases, and carotid endarterectomy history.

Dietary Intake Assessment

A validated food frequency questionnaire (FFQ) derived from the National Health and Nutrition Examination Surveys (NHNES)⁸⁾ was utilized to assess patients' eating habits during the previous 12 months. The NHNES questionnaire was first translated into Chinese. Then according to China National Nutrition and Health Surveys (which reported the food consumption frequency, but not consumption quantity), the 20–30 most frequently consumed foods in each classification were added if they were not included in NHNES questionnaire. A total of 298 food items classified into 8 groups constituted this Chinese FFQ including staple foods (grains, potatoes, and beans), vegetables, fruits, nuts, meats and fishes, beverages, dietary supplements, and condiments. Nutrients intake was estimated through a Food Composition Table⁹⁾ from the Chinese Center for Disease Control and Prevention and Institute of Nutrition and Food Safety. Participants with incomplete FFQ and reported extremely energy intake were excluded.

GI/GL Calculation

GI values were expressed as a percent of the glycemic response elicited by pure glucose as a standard food (GI for pure glucose = 100). GI and GL calculation in this study was using Chinese Table of Glycemic Index Values⁹⁾ and International Tables of Glycemic Index Values 2008 ¹⁰⁾. GL was calculated by multiplying the content of carbohydrate of food parameters by its GI values and then by its daily average consumption derived from FFQ. The overall GL was the summation of all food items. GL of each unit represents the equivalent of 1 g carbohydrate from pure glucose. The overall GI was then estimated by dividing the overall GL by the participants' daily average intake of carbohydrate.

Anthropometrics, Sociodemographic, and Physical Activity Measurement

Body mass index (BMI) was calculated by dividing the weight by the height squared. Body weight was measured using a digital electronic scale. Body height was measured with a wall-mounted rangefinder. According to annual family incomes, so cioeconomic status was classified as high (200000 yuan or more), moderate (100000–199999 yuan), and low (less than 100000 yuan). Patients with at least 20 minutes physical activities a day for at least 3 days a week were deemed as “physically active”¹¹⁾.

Biochemical Analysis of CVD Risk Factors

Subjects were fasting 8 hours for blood samples. Blood samples were usually collected at the first morning after hospitalization. Most patients (about 90%) were hospitalized w ithin 24 hours of stroke onset; therefore blood samples were usually obtained within 48 hours of stroke onset. Fasting plasma glucose (FPG), hemoglobin A1c (HbA1c), serum total cholesterol (TC), triglycerides (TG), HDL cholesterol (HDL-C), LDL cholesterol (LDL-C), C-reactive protein (CRP), homocysteine, and white blood cells were measured using fresh blood samples by Jinling Hospital Biochemistry Laboratory.

Stenosis Measurement

The degree of carotid stenosis measurement was conducted with a dual-source multidetector CT scanner (Somaton Definition Flash, Siemens, Germany) using the standard scanning protocol: 120-kV tube voltage, 300-mA tube current, 0.75-mm slice thickness, 64 × 0.6-mm collimator width, and 160 mm-FOV. Two neurologists were in charge of the evaluation of carotid enhanced imagines with a carotid reconstruction software (Neurovascular, Siemens, Germany). The presence and extents of stenosis on carotid bifurcations were assessed by NASCET criteria that described as the maximal loss rate percentage of lumen diameter in the lesion region ([normal lumen diameter at distal point – minimal lumen diameter of obstruction lesion)/normal lumen diameter at distal point] × 100%)¹²⁾. The degrees of carotid stenosis were classified according to NASCET criteria: no or mild (0–29%), moderate (30%-69%), and severe (70%–100%). Patients with poor imaging quality or who had carotid lesions other than atherosclerosis (e.g., artery dissection) in bifurcations were excluded.

Statistical Analysis

Continuous data were showed as means ± SD, while categorical data were showed as frequency number (rate). Least significant difference test was used for comparing the homogeneity of variance between groups. Carotid stenosis degree was divided into three groups. Baseline data were analyzed with one-way ANOVA, Kruskal–Wallis H test, or Chi-square test. Spearman test was used for investigating the association of CVD risk factors and GI/GL. In multivariable logistic analysis, GI and GL were categorized into quartiles. Ordinal logistic regression models were utilized for calculating odds ratios (OR) and 95% confidence intervals (CI) between GI/GL quartiles and carotid stenosis with (model 1) or without (model 0) controlling for confounding factors which P < 0.1. Stratified analyses among patients were additionally conducted by age (< 65 years or ≥ 65 years) and sex (women or men). Statistical significance in logistic regression models was determined by the Wald test. Statistical analysis was conducted with SPSS Statistics version 22.0 (IBM, New York, USA). P value < 0.05 was deemed as statistical significance.

Results

After excluding 7 patients from all 676 enrolled, 669 patients entered data analysis eventually. The screening process was presented in Fig. 1.

Fig. 1.

Flowchart of the included patients in this study 7 patients were excluded and 669 were eventually enrolled.

The mean age was 65.0 years (SD 10.8 years) and the percentage of men was 66.2%. The average value of GI/GL was 49.3 (SD 24.5)/137 (SD 81.6). The prevalence of carotid stenosis was 63.2% (n = 423) in 669 participants, and 4.6% patients (n = 31) had at least 1 carotid occlu sion. The stenosis rate of 27.1% patients (n = 181) were at 30%–69%, while 16.4% patients (n = 128) were at 70 %–100%.

Of all the 423 patients who had carotid stenosis, the infarction lesions of 277 patients resulted from internal carotid system, while 146 were caused by vertebral artery disease. In those 277 patients with anterior circulation infarction, 198 (71.5%) patients had cerebral infarction (cause of the index stroke) in the same side of dominant carotid stenosis, and 79 (28.5%) patients had cerebral infarction in the other side of dominant carotid stenosis.

Patients' baseline characteristics were shown in Table 1. Comparing to patients with no or mild stenosis, the severe stenosis patients were older (66.0 ± 9.87 vs. 62.9 ± 11.4, p = 0.01) and more likely to be men (71.8% vs. 63.8%, p = 0.04). Those in the severe stenosis participants had a higher prevalence of hypertension (78.2% vs. 64.3%, p = 0.001), diabetes mellitus (40.0% vs. 27.5%, p < 0.001), and heart disease (22.7% vs. 10.1%, p = 0.002). TOAST subtypes were statistically significant between groups, and the proportion of large artery atherosclerosis (LAA) was higher in the most severe carotid stenosis group than the group with no or mild stenosis (56.4% vs. 16.1%, p < 0.001). The severe stenosis patients had an obviously higher serum level of hemoglobin A1c (6.81 ± 2.09 vs. 6.19 ± 1.34, p = 0.001) and higher glycemic index (60.2 ± 36.6 vs. 45.2 ± 15.9, p = 0.001) and higher glycemic load (171 ± 104 vs. 120 ± 59.9, p < 0.001) than participants with no or mild stenosis.

Table 1. Baseline data stratified by carotid stenosis in ischemic stroke patients^a.

	Carotid stenosis
Characteristicsb	No or mild (0–29%) n = 378, 56.5%	Moderate (30∼69%) n = 181, 27.1%	Severe (70%∼100%) n = 110, 16.4%	p-value
Age (years)	62.9 ± 11.4	68.7 ± 8.87	66.0 ± 9.87	0.01*
Sex (males, %)				0.04*
Men	241 (63.8)	133 (73.5)	79 (71.8)
Women	137 (36.2)	48 (26.5)	31 (28.2)
BMI (Kg/m2)	24.8 ± 3.16	25.0 ± 3.40	24.8 ± 3.09	0.93
Socioeconomic states (%)				0.83
Low	109 (28.8)	49 (27.1)	37 (33.6)
Medium	130 (34.4)	64 (35.4)	36 (32.7)
High	139 (36.8)	68 (37.6)	37 (33.6)
Physical activities (%)				0.22
Inactive	193 (51.1)	92 (50.8)	66 (60.0)
Active	185 (48.9)	89 (49.2)	44 (40.0)
Smoking history (%)	171 (45.2)	95 (52.5)	62 (56.4)	0.07
Hypertension (%)	243 (64.3)	140 (77.3)	86 (78.2)	0.001**
Diabetes mellitus (%)	104 (27.5)	83 (45.9)	44 (40.0)	< 0.001***
Dyslipidemia (%)	77 (20.4)	36 (19.9)	15 (13.6)	0.27
Atrial fibrillation (%)	25 (6.6)	14 (7.7)	6 (5.5)	0.75
Heart disease (%)	38 (10.1)	28 (15.5)	25 (22.7)	0.002**
Stroke history (%)	72 (19.0)	46 (25.4)	30 (27.3)	0.09
TOAST subtypes (%)				< 0.001***
LAA	61 (16.1)	68 (37.6)	62 (56.4)
SVD	132 (34.9)	57 (31.5)	27 (24.5)
CE	31 (8.2)	10 (5.5)	3 (2.7)
Others	154 (40.7)	46 (25.4)	18 (16.4)
NLR	2.36 ± 1.28	2.62 ± 1.91	2.65 ± 1.43	0.05
CRP (mg/L)	3.27 ± 7.67	4.98 ± 15.9	4.66 ± 11.6	0.24
Hcy (µmol/L)	14.1 ± 7.60	13.9 ± 6.63	14.3 ± 8.03	0.81
HbA1c (mmol/L)	6.19 ± 1.34	6.63 ± 1.61	6.81 ± 2.09	0.001**
FPG (mmol/L)	5.61 ± 1.73	6.02 ± 1.94	5.80 ± 2.03	0.39
Glycemic index	45.2 ± 15.9	51.2 ± 31.0	60.2 ± 36.6	0.001**
Glycemic load	120 ± 59.9	152 ± 95.5	171 ± 104	< 0.001***

^aData were presented as m ± SD or frequency (rate)

^bLAA, large artery atherosclerosis; SVD, small vascular disease; CE, cardioembolism; NLR, neutrophil to lymphocyte ratio; CRP, C-reactive protein; Hcy, homocysteine; HbA1c, hemoglobin A 1c; FPG, fasting plasma glucose

In multiple regression models (Table 2), the odds were at 2.21 (4^th vs. 1^st quartile (Q), 95% CI, 1.46–3.35; p = 0.001) of GI and 1.92 (4^th vs. 1^st Q, 95% CI, 1.27–2.92; p = 0.002) of GL in the crude model. The OR of GI and GL tended to 2.11 (4^th vs. 1^st Q, 95% CI, 1.30–3.42; p = 0.02) and 1.82 (4^th vs. 1^st Q, 95% CI, 1.12–2.96; p = 0.02), respectively, after adjustment for age, sex, smoking history, TOAST subtypes, hypertension, diabetes mellitus, heart disease, stroke history, NLR, and HbA1c levels.

Table 2. Multiple logistics regression between carotid stenosis and quartile of glycemic index and glycemic load.

	Patients	Crude Model		Model 1
		OR (95% CI)	p-value	OR (95% CI)	p-value
Glycemic index	Number
Q1	168	1 (REF)		1 (REF)
Q2	167	1.05 (0.68–1.95)	0.83	1.11 (0.68–1.83)	0.81
Q3	167	0.97 (0.63–2.60)	0.89	0.92 (0.56–1.53)	0.53
Q4	167	2.21 (1.46–3.35)	0.001**	2.11 (1.30–3.42)	0.02*
Glycemic load	Number
Q1	167	1 (REF)		1 (REF)
Q2	168	1.00 (0.65–1.54)	0.99	1.02 (0.62–1.70)	0.81
Q3	167	1.33 (0.87–2.03)	0.18	1.31 (0.80–2.14)	0.13
Q4	167	1.92 (1.27–2.92)	0.002**	1.82 (1.12–2.96)	0.02*

Model 1, adjusted for age, sex, smoking history, hypertension, diabetes mellitus, heart disease, stroke history, TOAST subtypes, neutrophil to lymphocyte ratio level and hemoglobin A 1c levels

Stratified logistic regression by age and sex was performed (Table 3). The results showed that GI was a risk factor for carotid stenosis among the youngers (4^th vs. 1^st Q; OR = 2.17; 95% CI, 1.02–4.61; p = 0.04) and women (4^th vs. 1^st Q; OR = 3.36; 95% CI, 1.22–4.69; p = 0.02). Higher GL had a pronounced relationship with carotid stenosis for younger patients compared to the old aged (4^th vs. 1^st Q; younger patients, OR = 2.42; 95% CI, 1.13–4.76; p = 0.02; elders, OR = 1.45; 95% CI, 0.76–2.77; p = 0.26). Higher GL had also shown a more significant connection with higher carotid stenosis among women compared to men (4^th vs. 1^st Q; women, OR = 3.81; 95% CI, 1.45–5.05; p = 0.007; men, OR = 1.39; 95% CI, 0.04–6.26; p = 0.67). However, no significant relationship between GI and carotid stenosis was observed in the elderly and men.

Table 3. Stratified regression analysis between carotid stenosis and quartile of glycemic index and glycemic load according to age and sex^a.

Modelsb	Age		Sex
	< 65 years OR (95% CI)	≥ 65 years OR (95% CI)	Women OR (95% CI)	Men OR (95% CI)
GI
Q1	1 (REF)	1 (REF)	1 (REF)	1 (REF)
Q2	0.85 (0.37–1.94)	1.31 (0.64–2.45)	1.60 (0.58–3.32)	0.76 (0.13–3.49)
Q3	0.64 (0.28–1.45)	1.18 (0.53–2.37)	1.62 (0.56–4.39)	0.23 (0.02–2.31)
Q4	2.17 (1.02–4.61)	1.88 (0.76–3.58)	3.36 (1.22–4.69)	2.36 (0.58–4.61)
p-value	0.04*	0.06	0.02*	0.23
GL
Q1	1 (REF)	1 (REF)	1 (REF)	1 (REF)
Q2	1.03 (0.44–2.39)	1.12 (0.58–2.17)	1.98 (0.74–3.70)	1.15 (0.26–5.06)
Q3	1.26 (0.56–2.84)	1.34 (0.70–2.54)	2.89 (1.06–4.17)	0.36 (0.03–3.59)
Q4	2.42 (1.13–4.76)	1.45 (0.76–2.77)	3.81 (1.45–5.05)	1.39 (0.04–6.26)
p-value	0.02*	0.26	0.007**	0.67

^aGI, glycemic index; GL, glycemic load

^bAdjusted for age, sex, smoking history, TOAST subtypes, hypertension, diabetes mellitus, heart disease, stroke history, neutrophil to lymphocyte ratio and hemoglobin A 1c levels

Spearman test has not shown statistical significance between GI/GL and total cholesterol, triglycerides, LDL-C, HDL-C, CRP, Hcy, NLR, FPG, and HbA1c (Table 4).

Table 4. Spearman analysis between cardiovascular risk factors and glycemic index and glycemic load.

CVD risk factors	Glycemic index		Glycemic load
	r	p-value	r	p-value
Total cholesterol (mmol/L)	−0.01	0.86	−0.02	0.62
Triglycerides (mmol/L)	0.05	0.23	0.06	0.13
LDL-C (mmol/L)	0.01	0.80	−0.001	0.97
HDL-C (mmol/L)	0.02	0.59	−0.05	0.20
NLR	0.04	0.27	0.03	0.40
CRP (mg/L)	0.03	0.40	0.04	0.32
Hcy (µmol/L)	0.05	0.31	0.07	0.15
HbA1c (mmol/L)	0.01	0.87	0.03	0.42
FPG (mmol/L)	0.01	0.74	0.05	0.25

CVD, cardiovascular disease; NLR, neutrophil to lymphocyte ratio; CRP, C-reactive protein; Hcy, homocysteine; HbA1c, hemoglobin A 1c; FPG, fasting plasma glucose

This study detected 930 plaques in total. Of all 669 patients, 285 (42.6%) had calcified plaques and 124 (18.5%) had ulcerated plaques (Supplementary Table 1).

Supplementary Table 1. Plaque characteristics at the maximal stenosis of patients.

	Carotid Stenosis
Plaque characteristics	No or mild (0∼29%) n = 378, 56.5%	Moderate (30∼69%) n = 181, 27.1%	Severe (70%∼99%) n = 79, 11.8%	Occlusion (> 99%) n = 31, 4.6%
Total plaque numbers	212	445	273
Calcification (%)	86 (22.8)	143 (79.0)	56 (67.0)
Ulceration (%)	25 (6.6)	66 (36.5)	33 (41.8)

Discussion

This is the first study that revealed that GI and GL had an independent effect on the development of atherosclerotic stenosis. This cross-sectional study observed increased 2.1-fold odds for higher carotid stenosis of GI and an increased 1.8-fold risk of GL. The stratified results showed that the effect of GL was more pronounced in younger patients and women. The risk of GL increased by about 2.4-fold for younger patients and 3.8-fold for women.

Some previous literatures reported the association between GI/GL and cardiovascular diseases. Among a large sample of asymptomatic adults, one study showed that higher GI and GL were linked to a severe coronary artery calcium, and the association for GL was stronger¹³⁾. The results showed that combining the quality and the quantity of carbohydrates intake was a more important determinant of atherosclerosis. One Chinese study in 2016 and three metaanalyses all concluded that higher glycemic load and higher glycemic index diets were associated with an increased risk of CVD in women^{6, 14)}. Those results were consistent with this study which also showed that women had a greater risk for carotid stenosis. With regard to cardiovascular risk factors, the results were mixed. A previous study showed that dietary GL have been related with increased diabetes risk and lower LDL-C levels⁶⁾. This study did not find that GI/GL had a positive relationship with CVD risk factors which supported the conclusion of one randomized trial that included 163 overweight adults¹⁵⁾.

The mechanisms of GI/GL effect on pathological process of chronic diseases are not completely understood. Long-term high GI/GL diet leads to post-prandial hyperglycemia and serum insulinemia¹⁶⁾. Hyperglycemia decreases the microcirculation perfusion in macrovascular and microvascular by changing the physiological properties of red blood cells and increasing the plasma viscosity and fibrinogen levels. Insulinemia causes the decrease of insulin sensitivity for organs or tissues and produces a chain of endocrine disorders. Habitual high GI/GL diet also causes an imbalance of antioxidation-oxidation resulting in endothelial dysfunction and vascular inflammation which is the initial pathophysiological reaction in atherosclerosis¹⁷⁾. The above consists of the most important determinants for the development and progression of atherosclerotic stenosis.

This study had some limitations. A cross-sectional design limited the ability to deduce causality. Considering the wide differences in dietary structures and eating habits, this scale may be less applicable in populations other than Chinese. In addition, a recall bias inevitably produced during the collection of the past 1-year FFQ data, this may be a cause for the high standard deviation of GI and GL. Atherosclerosis belongs to a systemic disease, but this study assessed carotid bifurcations only for analysis. However, no methods have access to evaluate atherosclerosis in the whole body yet.

Conclusion

The quantity and quality of carbohydrate may take part in the progression of atherosclerotic stenosis of carotid in conjunction with other factors. High-risk stroke population especially youngers and women should reduce their GI and GL for decreasing the unfavorable influence of carbohydrates and preventing cardiovascular diseases. Further research is still needed to carry out for exploring their temporal causality and exact underlying mechanism for the relationship of dietary GI/GL and carotid atherosclerotic stenosis risk.

Disclosure

Min Peng, Xiang Li, Yujing Liu, Min Zou, Yaqia n Xia and Gelin Xu declare that they have no conflicts of interests.