Comparison of TyG indices and atherogenic index of plasma with hypertension in the PERSIAN Guilan cohort

Abstract

Hypertension (HTN) is a major global contributor to cardiovascular morbidity and mortality. Insulin resistance is a key mechanistic factor in HTN development, yet its direct measurement is impractical in large population studies. Triglyceride–glucose (TyG) index derivatives and the Atherogenic Index of Plasma (AIP) have emerged as simple surrogate markers of metabolic dysfunction. However, limited evidence compares their associations withHTN across different glycemic statuses. This study aimed to evaluate the associations of TyG-body mass index (TyG-BMI), TyG-waist circumference (TyG-WC), TyG-waist-to-height ratio (TyG-WHtR), TyG-waist-to-hip ratio (TyG-WHR), and AIP with HTN in a large Iranian population and to determine whether these associations differ among normoglycemic, prediabetic, and diabetic subgroups. This cross-sectional analysis included 10,520 adults aged 35–70 years from the PERSIAN Guilan Cohort Study. Logistic regression models were used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for the association between each index and HTN. Discriminative performance was assessed using receiver operating characteristic (ROC) curves and area under the curve (AUC) analysis. After adjusting for confounding factors, all evaluated indices were significantly associated with HTN in the overall population. The strongest association was observed for AIP (OR = 1.66, 95% CI: 1.43–1.93; P < 0.01), followed by TyG-WHtR (OR = 1.42, 95% CI: 1.35–1.50; P < 0.01), TyG-WHR (OR = 1.36, 95% CI: 1.30–1.43; P < 0.01), TyG-BMI (OR = 1.004, 95% CI: 1.004–1.005; P < 0.01), and TyG-WC (OR = 1.002, 95% CI: 1.001–1.002; P < 0.01). In the subgroup analysis based on glycemic status, AIP showed the strongest association among normoglycemic individuals (OR = 1.35, 95% CI: 1.10–1.67; P < 0.01), whereas TyG-WHR demonstrated the strongest association in the prediabetic group (OR = 1.27, 95% CI: 1.12–1.46; P < 0.01). Among individuals with diabetes, AIP again exhibited the strongest association (OR = 1.46, 95% CI: 1.09–1.97; P = 0.01). AIP exhibited the strongest association with HTN overall and within the normoglycemic and diabetic groups, while TyG-WHR was most strongly associated with HTN among prediabetic individuals. Although all TyG-derived indices and AIP were significantly associated with HTN, their relative strengths varied by glycemic status. Using the most relevant index for each metabolic category may improve risk stratification and support more targeted prevention strategies.

Introduction

Hypertension (HTN) remains a major global public health concern and is widely recognized as a leading risk factor for cardiovascular morbidity and mortality^1–3. Despite advancements in prevention and management, the prevalence of HTN continues to rise, contributing significantly to the global burden of cardiovascular disease (CVD), stroke, and chronic kidney disease (CKD)^1,4–6. Given its multifactorial etiology, early identification of individuals at increased risk is essential for implementing effective preventive strategies and reducing associated health consequences^7,8.

Among the metabolic disturbances linked to HTN, insulin resistance (IR) plays a central role in cardiovascular pathophysiology^8–11. IR promotes endothelial dysfunction, increases sympathetic nervous system activity, and disrupts renal sodium handling—mechanisms strongly implicated in the development of elevated blood pressure^12–15. However, direct assessment of IR, such as through the hyperinsulinemic–euglycemic clamp, is impractical for large population-based studies due to its complexity, invasiveness, and high cost^16,17.

To overcome these limitations, several triglyceride–glucose (TyG)-derived indices have been proposed, including the TyG–body mass index (TyG-BMI), TyG–waist circumference (TyG-WC), TyG–waist-to-height ratio (TyG-WHtR), and TyG–waist-to-hip ratio (TyG-WHR)^18–20. These indices integrate fasting triglyceride and glucose levels with anthropometric measures, providing simple, cost-effective, and reliable surrogate markers of IR-related cardiometabolic risk^21–24.

Another emerging marker is the Atherogenic Index of Plasma (AIP), which reflects the balance between atherogenic and protective lipoproteins. AIP has been associated with HTN and other cardiovascular outcomes and may provide additional insight into lipid-related metabolic dysfunction^25–29.

Although TyG-derived indices and AIP have been individually linked to HTN, direct comparisons of their associations within the same population are limited—particularly in Middle Eastern cohorts. The PERSIAN (Prospective Epidemiological Research Studies in Iran) Guilan Cohort Study (PGCS) provides a unique opportunity to address this gap using data from a large, ethnically homogeneous, and well-characterized population in northern Iran^30–32. Therefore, this study aims to examine the associations between TyG-derived indices and AIP with HTN, compare their discriminatory performance in the overall population and across distinct glycemic categories, and identify the most informative index for HTN assessment in this setting.

Methodology

Study design and population

This cross-sectional study was conducted using baseline data from PGCS, a regional branch of the PERSIAN study designed to investigate non-communicable diseases in Iran^33–35. PGCS included 10,520 participants (4,887 men and 5,633 women) aged 35 to 70 years, residing in urban and rural areas of Some’e Sara County in Guilan Province. Participants were recruited through stratified random sampling from October 8, 2014, to January 20, 2017. The selection of this region was based on its demographic stability, high population density, and relative homogeneity in lifestyle and cultural background³⁰. The study was approved by the Ethics Committee of Guilan University of Medical Sciences (IR.GUMS.REC.1403.254), and all participants provided written informed consent prior to enrollment. The study adhered to the ethical standards outlined in the Declaration of Helsinki³⁶.

Data collection process

Data collection involved a combination of face-to-face interviews, clinical evaluations, and laboratory testing, all conducted by trained personnel. Demographic and lifestyle information included age, sex, marital status, educational attainment, occupational status, and place of residence. Socioeconomic status was assessed using a standardized household wealth index³⁰.

Physical activity was assessed using a validated 36-item questionnaire. The duration of each activity was recorded in hours per day and converted to a metabolic equivalent of task (MET) score. MET was calculated using the formula MET = metabolic intensity × hours per day, and the MET score itself represented the ratio of energy expenditure during an activity to energy expenditure at rest³⁷.

Tobacco use was quantified by calculating pack-years, which incorporated both the number of years an individual smoked and the average number of cigarettes smoked per day. Participants were also asked about their use of other tobacco products, including hookah and cigars, as well as alcohol consumption³⁷.

Anthropometric measurements were conducted according to standardized protocols. Height was measured using a Seca wall-mounted stadiometer, and weight was measured using a Seca 755 mechanical scale. BMI was calculated by dividing weight in kilograms by height in meters squared and categorized based on WHO definitions: underweight (< 18.5 kg/m²), normal weight (18.5–24.9 kg/m²), overweight (25.0–29.9 kg/m²), and obese (≥ 30.0 kg/m²)^38,39. Waist and hip circumferences were measured using a non-stretchable tape to calculate WHR and WHtR, which were also used in derivative index calculations.

Blood pressure measurements were obtained using Richter auscultatory mercury sphygmomanometers (MTM Munich, Germany). Two readings were taken from each arm at ten-minute intervals in a quiet room while the participant was seated with back support, feet flat on the floor, legs uncrossed, and arms supported at heart level. An appropriately sized cuff was used for all participants. The mean of the recorded values was used for analysis. HTN was defined as systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg, self-reported prior diagnosis by a healthcare provider, or current use of antihypertensive medications⁴⁰.

Fasting blood samples were collected after a minimum of 12 h of overnight fasting. Fasting plasma glucose (FPG) was measured using the glucose oxidase method. , high-density lipoprotein cholesterol (HDL-C), and total cholesterol (TC) levels were determined using enzymatic colorimetric assays (Pars Azmun, Tehran, Iran) and processed on an automated analyzer (BT 1500, Biotecnica, Italy). Low-density lipoprotein cholesterol (LDL-C) was calculated using the Friedewald formula for participants with TG < 400 mg/dL.

Calculation of indices

In this study, several indices were calculated based on fasting biochemical and anthropometric data to evaluate their associations with HTN. The TyG-derived indices and the AIP were calculated using the following formulas.

TyG-BMI

TyG-BMI = TyG index × Body Mass Index (BMI in kg/m²)⁴¹.

TyG-WC

TyG-WC = TyG index × Waist Circumference (cm)²¹.

TyG-WHtR

TyG-WHtR = TyG index × (Waist Circumference [cm] ÷ Height [cm])⁴².

TyG-WHR

TyG-WHR = TyG index × (Waist Circumference [cm] ÷ Hip Circumference [cm])⁴³.

AIP

AIP = log (Fasting TG ÷ HDL-Cholesterol) in molar concentrations⁴⁴.

Statistical analysis

Statistical analyses were performed using R (version 4.4.1). A p-value < 0.05 was considered statistically significant. Descriptive statistics were used to summarize demographic, clinical, and biochemical characteristics, with continuous variables reported as means (SD) and categorical variables as frequencies (%). The normality of continuous variables was assessed using the Kolmogorov–Smirnov test. Group comparisons were conducted using the student’s t-test for normally distributed variables, the Wilcoxon rank-sum test for non-normally distributed variables, and the Chi-square or Fisher’s exact test for categorical variables.

In this study, all calculated indices were treated as continuous variables. Logistic regression models were used to estimate the association between each index and HTN, and odds ratios (ORs) with 95% confidence intervals (CIs) were reported per one-unit increase in each index. No standardization (e.g., z-scores) or categorization (e.g., quartiles) was applied, in order to preserve the original scale of the indices and ensure consistency with the analytical framework of the study.

Associations were examined in the overall study population and within glycemic status subgroups (normoglycemic, prediabetic, and diabetic). Three models were constructed: a crude model; model 1 adjusted for age and sex; and model 2 fully adjusted for age, physical activity, smoking status, hookah use, alcohol consumption, and opium use.

The discriminatory ability of each index for predicting HTN was assessed using receiver operating characteristic (ROC) curve analysis. Indices were evaluated as continuous variables, and the area under the curve (AUC) was calculated. Optimal cutoff points for each biomarker were determined using Youden’s Index for both the overall population and each glycemic subgroup⁴⁵.

Results

Characteristics of the study population

This cross-sectional analysis included a total of 10,520 participants, comprising 4,887 men (46.45%) and 5,633 women (53.55%). Participants were categorized into three glycemic groups: normoglycemic (n = 6,152), prediabetic (n = 1,837), and diabetic (n = 2,531). The overall mean age of the total study population was 51.51 ± 8.90 years. Specifically, the normoglycemic group had a younger mean age of 49.98 ± 8.67 years, the prediabetic group had a mean age of 52.32 ± 8.74 years, and the diabetic group was older with a mean age of 54.64 ± 8.64 years. Demographic and socioeconomic, anthropometric measurements, and clinical and laboratory measurements are presented in Tables 1 and 2, and Table 3, respectively.

Table 1.

Demographic and socioeconomic characteristics of participants in the PERSIAN Guilan cohort Study, stratified by sex and glycemic status (normoglycemic, prediabetic, and diabetic groups). Data are presented as mean ± SD for continuous variables and Raw numbers for categorical variables.

Variable		All participant				Normoglycemic				Prediabetic				Diabetic
		All (n = 10520)	Men (n = 4887)	Women (n = 5633)	P value	All (n = 6152)	Men (n = 2991)	Women (n = 3161)	P value	All (n = 1837)	Men (n = 907)	Women (n = 930)	P value	All (n = 2531)	Men (n = 989)	Women (n = 1542)	P value
Age		51.51 ± 8.90	51.47 ± 8.86	51.55 ± 8.94	0.64	49.98 ± 8.67	50.30 ± 8.73	49.68 ± 8.61	< 0.01	52.32 ± 8.74	52.12 ± 8.79	52.51 ± 8.70	0.33	54.64 ± 8.64	54.41 ± 8.55	54.79 ± 8.69	0.26
Marital Status	Single	305	78	227	< 0.01	204	55	149	< 0.01	55	15	40	< 0.01	46	8	38	< 0.01
	Married	9527	4733	4794	< 0.01	5617	2889	2728	0.03	1665	877	788	0.02	2245	967	1278	< 0.01
	Widow/ Widower	566	48	518	< 0.01	254	27	227	< 0.01	94	9	85	< 0.01	218	12	206	< 0.01
	Divorced	122	28	94	< 0.01	77	20	57	< 0.01	23	6	17	0.04	22	2	20	< 0.01
Residence Type	Urban	4613	2062	2551	< 0.01	2629	1244	1385	0.08	817	374	443	< 0.01	1167	444	723	0.34
	Rural	5907	2825	3082	< 0.01	3523	1747	1776	0.08	1020	533	487	< 0.01	1364	545	819	0.34
Education	0–5 years of schooling	5050	1841	3209	< 0.01	2688	1061	1627	< 0.01	918	365	553	< 0.01	1444	415	1029	< 0.01
	6–12 years of schooling	4832	2590	2242	< 0.01	3043	1639	1404	< 0.01	815	464	351	< 0.01	974	487	487	< 0.01
	University/college	638	456	182	< 0.01	421	291	130	< 0.01	104	78	26	< 0.01	113	87	26	< 0.01
Occupation	Unemployed	4781	562	4219	< 0.01	2561	277	2284	< 0.01	811	113	698	< 0.01	1409	172	1237	< 0.01
	Employed	5739	4325	1414	< 0.01	3591	2714	877	< 0.01	1026	794	232	< 0.01	1122	817	305	< 0.01
Physical activity		41.25 ± 8.88	43.53 ± 10.10	39.28 ± 7.10	< 0.01	42.00 ± 9.18	44.33 ± 10.34	39.79 ± 7.25	< 0.01	41.27 ± 8.76	43.02 ± 9.75	39.6 ± 7.27	< 0.01	39.45 ± 7.95	41.59 ± 9.36	38.1 ± 6.54	< 0.01
Cigarette Smoking	No	7929	2358	5571	< 0.01	4568	1436	3132	< 0.01	1381	462	919	< 0.01	1980	460	1520	< 0.01
	Yes	2584	2522	62	< 0.01	1580	1551	29	< 0.01	456	445	11	< 0.01	548	526	22	< 0.01

Table 2.

Anthropometric measurements characteristics of participants in the PERSIAN Guilan cohort Study, stratified by sex and glycemic status (normoglycemic, prediabetic, and diabetic groups). Data are presented as mean ± SD for continuous variables.

Variable	All participant				Normoglycemic				Prediabetic				Diabetic
	All (n = 10520)	Men (n = 4887)	Women (n = 5633)	P value	All (n = 6152)	Men (n = 2991)	Women (n = 3161)	P value	All (n = 1837)	Men (n = 907)	Women (n = 930)	P value	All (n = 2531)	Men (n = 989)	Women (n = 1542)	P value
Weight	73.97 ± 13.60	75.32 ± 13.77	72.80 ± 13.35	< 0.01	73.4 ± 13.3	74.47 ± 13.33	72.41 ± 13.14	< 0.01	75.6 ± 14.3	77.38 ± 14.81	73.8 ± 13.5	< 0.01	74.2 ± 13.8	76.02 ± 13.89	72.98 ± 13.64	< 0.01
Height	162.32 9.46	169.74 6.84	155.89 6.13	< 0.01	163.00 ± 9.28	170.00 ± 6.75	156.45 ± 6.03	< 0.01	163.00 ± 9.56	169.82 ± 6.89	155.78 ± 6.08	< 0.01	160.40 ± 9.59	169.13 ± 7.05	154.8 ± 6.24	< 0.01
Body mass index	28.14 ± 5.08	26.08 ± 4.20	29.92 ± 5.11	< 0.01	27.70 ± 4.90	25.70 ± 4.06	29.54 ± 4.90	< 0.01	28.60 ± 5.15	26.77 ± 4.53	30.39 ± 5.10	< 0.01	28.90 ± 5.37	26.50 ± 4.21	30.45 ± 5.46	< 0.01
Waist circumference	98.81 ± 12.41	93.60 ± 10.88	103.34 ± 11.87	< 0.01	97.40 ± 12.17	92.60 ± 10.53	102.00 ± 11.86	< 0.01	100.00 ± 12.58	95.30 ± 11.76	104.46 ± 11.7	< 0.01	101.44 ± 12.36	95.13 ± 10.71	105.49 ± 11.63	< 0.01
Hip circumference	104.19 ± 103.32	99.52 ± 16.28	108.25 ± 140.26	< 0.01	104.54 ± 135.00	99.30 ± 19.86	109.00 ± 187.10	< 0.01	103.76 ± 9.99	100.29 ± 7.93	107.12 ± 10.62	< 0.01	103.69 ± 10.30	99.40 ± 7.78	106.43 ± 10.84	< 0.01
Wrist circumference	16.69 ± 1.33	17.27 ± 1.14	16.19 ± 1.27	< 0.01	16.60 ± 1.32	17.20 ± 1.12	16.10 ± 1.26	< 0.01	16.85 ± 1.40	17.45 ± 1.26	16.26 ± 1.28	< 0.01	16.71.31	17.30 ± 1.11	16.33 ± 1.30	< 0.01
Waist-to-Hip Ratio	0.95 ± 0.06	0.94 ± 0.06	0.97 ± 0.06	< 0.01	0.94 ± 0.06	0.93 ± 0.06	0.96 ± 0.04	< 0.01	0.96 ± 0.06	0.94 ± 0.06	0.97 ± 0.05	< 0.01	0.97 ± 0.05	0.95 ± 0.05	0.99 ± 0.05	< 0.01
Waist-to-height ratio	0.61 ± 0.09	0.55 ± 0.06	0.66 ± 0.07	< 0.01	0.60 ± 0.08	0.54 ± 0.06	0.65 ± 0.07	< 0.01	0.62 ± 0.09	0.56 ± 0.06	0.67 ± 0.07	< 0.01	0.63 ± 0.09	0.56 ± 0.06	0.68 ± 0.07	< 0.01
Systolic blood pressure	118.24 ± 16.75	118.84 ± 16.53	117.72 ± 16.92	< 0.01	115.75 ± 15.89	116.57 ± 15.60	114.98 ± 16.10	< 0.01	120.37 ± 16.90	121.54 ± 16.42	119.24 17.30	< 0.01	122.76 ± 17.53	123.29 ± 18.10	122.43 ± 17.17	0.23
Diastolic blood pressure	76.97 ± 11.01	77.91 ± 11.08	76.17 ± 10.88	< 0.01	75.63 ± 11.10	76.60 ± 11.02	74.74 ± 11.00	< 0.01	78.80 ± 10.82	80.10 ± 10.84	77.58 ± 10.70	< 0.01	78.90 ± 10.50	79.92 ± 10.90	78.24 ± 10.20	< 0.01

Table 3.

Clinical and laboratory measurements characteristics of participants in the PERSIAN Guilan cohort Study, stratified by sex and glycemic status (normoglycemic, prediabetic, and diabetic groups). Data are presented as mean ± SD for continuous variables.

Variable	All participant				Normoglycemic				Prediabetic				Diabetic
	All (n = 10520)	Men (n = 4887)	Women (n = 5633)	P value	All (n = 6152)	Men (n = 2991)	Women (n = 3161)	P value	All (n = 1837)	Men (n = 907)	Women (n = 930)	P value	All (n = 2531)	Men (n = 989)	Women (n = 1542)	P value
Systolic blood pressure	118.24 ± 16.75	118.84 ± 16.53	117.72 ± 16.92	< 0.01	115.75 ± 15.89	116.57 ± 15.60	114.98 ± 16.10	< 0.01	120.37 ± 16.90	121.54 ± 16.42	119.24 17.30	< 0.01	122.76 ± 17.53	123.29 ± 18.10	122.43 ± 17.17	0.23
Diastolic blood pressure	76.97 ± 11.01	77.91 ± 11.08	76.17 ± 10.88	< 0.01	75.63 ± 11.10	76.60 ± 11.02	74.74 ± 11.00	< 0.01	78.80 ± 10.82	80.10 ± 10.84	77.58 ± 10.70	< 0.01	78.90 ± 10.50	79.92 ± 10.90	78.24 ± 10.20	< 0.01
Fasting blood sugar	104.56 ± 37.17	103.63 ± 35.27	105.36 ± 38.72	0.01	88.75 ± 6.40	88.89 ± 6.38	88.60 ± 6.41	0.07	107.00 ± 6.31	107.02 ± 6.32	106.85 ± 6.31	0.56	141.28 ± 60.45	145.13 ± 60.01	138.82 ± 60.63	0.01
Total cholesterol	192.79 ± 38.97	191.19 ± 38.17	194.18 ± 39.60	< 0.01	192.00 ± 36.00	190.05 ± 35.70	193.53 ± 36.20	< 0.01	199.00 ± 43.60	196.97 ± 40.60	201.00 ± 46.30	0.05	190.66 ± 41.90	189.34 ± 42.40	191.51 ± 41.50	0.20
Triglycerides	160.25 ± 103.27	165.90 ± 111.80	155.36 ± 94.99	< 0.01	148.00 ± 84.00	153.50 ± 91.60	142.64 ± 75.70	< 0.01	174.00 ± 130.12	182.36 ± 140.00	165.55 ± 119.48	< 0.01	180.39 ± 118.00	188.34 ± 132.00	175.29 ± 109.00	< 0.01
High density lipoprotein	48.37 ± 10.97	46.52 ± 10.52	49.99 ± 11.09	< 0.01	48.70 ± 11.10	46.72 ± 10.50	50.59 ± 11.30	< 0.01	48.44 ± 10.80	46.64 ± 10.39	50.19 ± 10.81	< 0.01	47.60 ± 10.80	45.80 ± 10.60	48.67 ± 10.80	< 0.01
Low density lipoprotein	112.84 ± 32.05	112.15 ± 31.73	113.45 ± 32.31	0.03	114.00 ± 30.50	113.06 ± 30.30	114.59 ± 30.60	0.04	116.35 ± 33.40	115.08 ± 33.20	117.60 ± 33.50	0.10	107.88 ± 34.14	106.71 ± 33.90	108.63 ± 34.30	0.16

The mean values of TyG-derived indices and AIP are presented in Table 4. For the total study population, the mean ± SD values were as follows: TyG-BMI: 249.64 ± 50.51, TyG-WC: 876.20 ± 133.06, TyG-WHR: 8.48 ± 0.88, TyG-WHtR: 5.42 ± 0.93, and AIP: 0.11 ± 0.28. When stratified by glycemic status, corresponding values for normoglycemic participants were: 240.21 ± 46.60, 844.55 ± 122.65, 8.20 ± 0.78, 5.20 ± 0.85, and 0.08 ± 0.27; for those with prediabetes: 257.28 ± 49.65, 899.01 ± 129.03, 8.64 ± 0.78, 5.54 ± 0.88, and 0.14 ± 0.28; and for those with diabetes: 267.02 ± 54.65, 936.56 ± 136.01, 9.02 ± 0.90, 5.86 ± 0.96, and 0.16 ± 0.28, respectively.

Table 4.

Metabolic and atherogenic indices measurements characteristics of participants in the PERSIAN Guilan cohort Study, stratified by sex and glycemic status (normoglycemic, prediabetic, and diabetic groups). Data are presented as mean ± SD for continuous variables.

Variable	All participant				Normoglycemic				Prediabetic				Diabetic
	All (n = 10520)	Men (n = 4887)	Women (n = 5633)	P value	All (n = 6152)	Men (n = 2991)	Women (n = 3161)	P value	All (n = 1837)	Men (n = 907)	Women (n = 930)	P value	All (n = 2531)	Men (n = 989)	Women (n = 1542)	P value
Triglyceride-glucose–body mass index	249.64 ± 50.51	232.06 ± 44.20	264.89 ± 50.70	< 0.01	240.21 ± 46.60	224.19 ± 41.34	255.37± 46.17	< 0.01	257.28 ± 49.65	242.03 ± 46.22	272.15 ± 48.38	< 0.01	267.02 ± 54.65	246.73± 45.14	280.03± 56.25	< 0.01
Triglyceride-glucose–waist circumference	876.20 ± 133.06	832.15 ± 124.34	914.41 ± 128.49	< 0.01	844.55 ± 122.65	805.79 ± 114.83	881.23 ± 118.49	< 0.01	899.01 ± 129.03	861.52 ± 128.14	935.57 ± 119.08	< 0.01	936.56 ± 136.01	884.95 ± 125.76	969.67 ± 132.00	< 0.01
Triglyceride-glucose–waist-to-hip ratio	8.48 ± 0.88	8.35 ± 0.88	8.59 ± 0.87	< 0.01	8.20 ± 0.78	8.11 ± 0.79	8.29 ± 0.76	< 0.01	8.64 ± 0.78	8.56 ± 0.83	8.73 ± 0.72	< 0.01	9.02 ± 0.90	8.88 ± 0.91	9.11 ± 0.89	< 0.01
Triglyceride-glucose–waist-to-height ratio	5.42 ± 0.93	4.90 ± 0.73	5.87 ± 0.84	< 0.01	5.20 ± 0.85	4.74 ± 0.67	5.63 ± 0.77	< 0.01	5.54 ± 0.88	5.07 ± 0.73	6.01 ± 0.77	< 0.01	5.86 ± 0.96	5.23 ± 0.74	6.27 ± 0.87	< 0.01
Atherogenic index of plasma	0.11 ± 0.28	0.13 ± 0.28	0.09 ± 0.26	< 0.01	0.08 ± 0.27	0.11 ± 0.28	0.05 ± 0.25	< 0.01	0.14 ± 0.28	0.18 ± 0.29	0.11 ± 0.27	< 0.01	0.16 ± 0.28	0.20 ± 0.28	0.14 ± 0.28	< 0.01

Status of hypertension in the population

Among the total 10,520 participants, 4,543 were identified as having HTN. Of these, 1,904 were men and 2,639 were women. In the normoglycemic group, 2,105 participants had HTN, comprising 952 men and 1,153 women. In the prediabetic group, 843 individuals had HTN, including 378 men and 465 women. Within the diabetic group, 1,595 participants were hypertensive, consisting of 574 men and 1,021 women (Fig. 1).

Fig. 1.

Hypertension status in men and women in the overall study population, presented across different glycemic groups (normoglycemic, prediabetic, diabetic) from the PERSIAN Guilan Cohort Study.

Association of TyG index, TyG index related parameters, and AIP with hypertension

In overall population, the analysis demonstrated significant associations between HTN and all evaluated indices in the crude model. The strongest association was observed for AIP (OR = 1.59, 95% CI: 1.38–1.83, P < 0.01), followed by TyG-WHR (OR = 1.55, 95% CI: 1.48–1.63, P < 0.01) and TyG-WHtR (OR = 1.51, 95% CI: 1.45–1.58, P < 0.01). After adjusting for potential confounding variables (including age, sex, physical activity, smoking status, alcohol consumption, hookah use, and opium use), all associations remained statistically significant. In the adjusted model, the strongest association was again observed for AIP (OR = 1.66, 95% CI: 1.43–1.93, P < 0.01), followed by TyG-WHtR (OR = 1.42, 95% CI: 1.35–1.50, P < 0.01) and TyG-WHR (OR = 1.36, 95% CI: 1.30–1.43, P < 0.01) (Table 5).

Table 5.

Table 5. Logistic regression analysis evaluating the association of TyG index, its derivatives, and AIP with hypertension in the overall population and across glycemic subgroups (normoglycemic, prediabetic, and diabetic). Odds ratios (ORs) and 95% confidence intervals (CIs) are presented for the crude model and three adjusted models: model 1 is adjusted for age and sex; and model 2 is further adjusted for physical activity, smoking status, Hookah use, alcohol consumption, and opium use (Data are presented as OR with 95%).

Population		TyG-BMI	TyG -WC	TyG-WHR	TyG-WHtR	AIP
Overall population	Crude	1.005(95%CI: 1.004–1.006, P < 0.01)	1.002(95%CI: 1.022–1.028, P < 0.01)	1.55(95%CI: 1.48–1.63, P < 0.01)	1.51(95%CI:1.45–1.58, P < 0.01)	1.59 (95%CI:1.38–1.83, P < 0.01)
	Model 1	1.006 (95% CI: 1.005–1.006, P < 0.01)	1.002 (95% CI: 1.002–1.003, P < 0.01)	1.39(95% CI: 1.32–1.46, P < 0.01)	1.45(95%CI:1.37–1.53, P < 0.01)	1.80 (95%CI:1.56–2.09, P < 0.01)
	Model 2	1.004 (95% CI: 1.004–1.005, P < 0.01)	1.002 (95% CI: 1.001–1.002, P < 0.01)	1.36 (95% CI: 1.30–1.43, P < 0.01)	1.42 (95% CI: 1.35–1.50, P < 0.01)	1.66 (95% CI: 1.43–1.93, P < 0.01)
Normoglycemic population	Crude	1.004(95%CI:1.003–1.005, P < 0.01)	1.002(95%CI:1.001- 1.002, P < 0.01)	1.36(95%CI:1.27–1.46, P < 0.01)	1.35(95%CI:1.27–1.44, P < 0.01)	1.27 (95%CI:1.04–1.54, P < 0.01)
	Model 1	1.005(95% CI: 1.003–1.006, P < 0.01)	1.002(95% CI: 1.001–1.002, P < 0.01)	1.25(95% CI: 1.16–1.34, P < 0.01)	1.31(95% CI: 1.22–1.42, P < 0.01)	1.47 (95% CI: 1.20–1.80, P < 0.01)
	Model 2	1.004 (95% CI: 1.003–1.005, P < 0.01)	1.001 (95% CI: 1.001–1.002, P < 0.01)	1.22 (95% CI: 1.14–1.32, P < 0.01)	1.29 (95% CI: 1.19–1.39, P < 0.01)	1.35 (95% CI: 1.10–1.67, P < 0.01)
Prediabetic population	Crude	1.002 (95%CI:1.0005–1.0040, P = 0.01)	1.001(95%CI:1.001–1.003, P < 0.01)	1.37(95%CI:1.21–1.55 P < 0.01)	1.30(95%CI:1.17–1.44 P < 0.01)	1.21 (95%CI:0.87–1.68, P = 0.26)
	Model 1	1.003(95% CI: 1.001–1.005, P < 0.01)	1.001(95% CI: 1.001–1.002, P < 0.01)	1.29 (95%CI:1.14–1.46, P < 0.01)	1.22(95% CI:1.08–1.39, P < 0.01)	1.44 (95% CI: 1.03–2.03, P = 0.03)
	Model 2	1.002 (95% CI: 1.001–1.004, P = 0.02)	1.001 (95% CI: 1.000–1.002, P < 0.01)	1.27 (95% CI: 1.12–1.46, P < 0.01)	1.21 (95% CI: 1.06–1.37, P = 0.02)	1.39 (95% CI: 0.97–1.97, P = 0.06)
Diabetic population	Crude	1.004(95%CI: 1.002–1.005, P < 0.01)	1.002(95%CI: 1.001–1.002, P < 0.01)	1.17(95%CI: 1.07–1.28, P < 0.01)	1.34(95%CI: 1.23–1.46, P < 0.01)	1.35 (95%CI: 1.01–1.81, P = 0.04)
	Model 1	1.004(95%CI: 1.002–1.005, P < 0.01)	1.001(95%CI: 1.001–1.002, P < 0.01)	1.09(95%CI: 1.00–1.20, P = 0.04)	1.28 (95% CI: 1.15–1.41, P < 0.01)	1.53 (95% CI: 1.14–2.06, P < 0.01)
	Model 2	1.003 (95% CI: 1.001–1.005, P < 0.01)	1.001 (95% CI: 1.001–1.002, P < 0.01)	1.09 (95% CI: 1.00–1.20, P = 0.04)	1.27 (95% CI: 1.14–1.41, P < 0.01)	1.46 (95% CI: 1.09–1.97, P = 0.01)

In the normoglycemic population, logistic regression analysis showed significant associations between HTN and all evaluated indices in the crude model. The strongest association was observed for TyG-WHR (OR = 1.36, 95% CI: 1.27–1.46, P < 0.01), followed by TyG-WHtR (OR = 1.35, 95% CI: 1.27–1.44, P < 0.01) and AIP (OR = 1.27, 95% CI: 1.04–1.54, P < 0.01). After adjusting for potential confounding variables, all associations remained statistically significant. In the adjusted model, the strongest association was observed for AIP (OR = 1.35, 95% CI: 1.10-1.67, P  < 0.01), followed by TyG-WHtR (OR = 1.29, 95% CI: 1.19–1.39, P < 0.01) and TyG-WHR (OR = 1.22, 95% CI: 1.14–1.32, P < 0.01) (Table 5).

In the prediabetic population, logistic regression analysis showed significant associations between HTN and most evaluated indices in the crude model. The strongest association was observed for TyG-WHR (OR = 1.37, 95% CI: 1.21–1.55, P < 0.01), followed by TyG-WHtR (OR = 1.30, 95% CI: 1.17–1.44, P < 0.01). After adjusting for potential confounding variables, TyG-WHR remained the strongest significant predictor (OR = 1.27, 95% CI: 1.12–1.46, P < 0.01), followed by TyG-WHtR (OR = 1.21, 95% CI: 1.06–1.37, P < 0.01) (Table 5).

In the diabetic population, logistic regression analysis demonstrated significant associations between HTN and all evaluated indices in the crude model. The strongest association was observed for AIP (OR = 1.35, 95% CI: 1.01–1.81, P = 0.04), followed by TyG-WHtR (OR = 1.34, 95% CI: 1.23–1.46, P < 0.01) and TyG-WHR (OR = 1.17, 95% CI: 1.07–1.28, P < 0.01). After adjusting for potential confounding variables, AIP remained the strongest significant predictor (OR = 1.46, 95% CI: 1.09–1.97, P =0.01), followed by TyG-WHtR (OR = 1.27, 95% CI: 1.14–1.41, P < 0.01) and TyG-WHR 1.09 (95% CI: 1.00–1.20, P = 0.04) (Table 5).

Discriminative performance of indices based on ROC curve analysis

The discriminative performance of TyG-related indices for HTN varied substantially. TyG-WHR exhibited the highest discriminative ability for HTN (AUC = 0.61, 95% CI: 0.59–0.62), followed closely by TyG-WHtR (AUC = 0.60, 95% CI: 0.59–0.61). TyG-WC demonstrated moderate discriminative performance (AUC = 0.59, 95% CI: 0.58–0.60), while TyG-BMI showed lower discriminative capacity (AUC = 0.57, 95% CI: 0.56–0.58). AIP exhibited the weakest discrimination (AUC = 0.53, 95% CI: 0.52–0.54) (Fig. 2).

Fig. 2.

Receiver Operating Characteristic curves of TyG index derivatives and AIP for hypertension in the overall population.

Discussion

In this study, we examined the association between TyG index derivatives, and AIP with HTN across different glycemic categories. Our findings underscore the importance of these indices, particularly TyG-WHR, TyG-WHtR and AIP, as valuable predictive markers for HTN, with distinct patterns emerging based on glycemic status.

In our study, the overall prevalence of HTN was 43.2%, with the highest rate observed in the diabetic group (62.9%), followed by the prediabetic group (45.9%) and the normoglycemic group (34.2%). These figures align with global reports, where individuals with diabetes or prediabetes are at significantly higher risk for developing HTN⁴⁶. The high predictive role of AIP, followed by TyG-WHtR and TyG-WHR, highlights the importance of lipid-related markers in HTN risk stratification. AIP, which reflects the ratio of TG to HDL-C, has long been associated with cardiovascular risk, including HTN, due to its capacity to capture lipid imbalances that contribute to endothelial dysfunction and arterial stiffness^47,48.

Our findings align with previous research that has evaluated TyG and its derivative indices in relation to HTN and cardiometabolic outcomes. Huang et al., using NHANES data from American adults, similarly reported positive correlations between TyG, TyG-BMI, TyG-WHtR, and TyG-WC with HTN. Their identification of non-linear threshold effects further supports the relevance of TyG-related markers across the metabolic spectrum⁴⁹.

Yang et al. also demonstrated that TyG and its related anthropometric derivatives were significant predictors of new-onset HTN in a large Chinese cohort, with TyG-WHtR emerging as the strongest indicator. Our results similarly identified TyG-WHtR and TyG-WHR as highly informative indices. However, our results uniquely showed that AIP surpassed TyG-based markers in some analyses, suggesting that lipid-driven mechanisms—including atherogenic dyslipidemia—may play a more prominent role in HTN risk in our population compared to East Asian cohorts⁵⁰.

Findings from Li et al. further support the clinical relevance of TyG-based indices, demonstrating that TyG, and especially TyG-WHtR, predicted both all-cause and cardiovascular mortality among hypertensive adults. Their use of machine learning models highlighted the incremental predictive value of TyG derivatives when added to traditional risk models. While our study did not evaluate mortality outcomes, the strong associations observed between TyG-derived indices, AIP, and HTN reinforce the broader concept that metabolic markers reflecting IR, visceral adiposity, and dyslipidemia contribute meaningfully to long-term cardiovascular risk⁵¹.

Our results also corroborate findings from other studies, such as the Framingham Heart Study, which established that IR markers, including the TyG index, are predictive of HTN, even in individuals without overt metabolic disorders⁵². The consistency of these findings across different populations suggests that TyG based indexes and AIP may be universally applicable markers for early detection of HTN.

Among normoglycemic individuals, AIP and TyG-WHtR exhibited the strongest associations with HTN. These findings are consistent with studies such as that by Mingjuan et al., which demonstrated that the AIP could predict HTN even in normoglycemic individuals⁵³. Notably, TyG-WHtR, a combination of the TyG index and waist-to-height ratio, appears to be particularly effective in capturing the risk of HTN associated with abdominal obesity, which is a key factor in both IR and HTN^54,55. Additionally, Sawaf et al. found that the TyG index is a strong predictor of HTN risk even among non-diabetic patients, further emphasizing the utility of this index in identifying HTN risk in metabolically healthy individuals⁵⁶.

⁵⁷.

In the prediabetic subgroup, AIP did not reach statistical significance in the fully adjusted model. This may be explained by the distinct metabolic profile of prediabetes, which represents an intermediate stage between normal glucose regulation and overt diabetes. In this phase, lipid abnormalities, while present, are often less severe than in diabetes, whereas IR and central obesity tend to play a more prominent role in blood pressure elevation. Because AIP reflects the balance between TG and HDL-C, its predictive strength may be reduced when lipid disturbances are milder. In contrast, indices such as TyG-WHR, which incorporate both –TG-glucose levels and body fat distribution, may better capture the combined metabolic and anthropometric influences on HTN risk in prediabetic individuals. Nevertheless, this interpretation is based on our single large-cohort analysis, and future studies in other populations should specifically examine this finding to confirm whether it holds true and to determine its generalizability across different ethnic, geographic, and clinical contexts.

Among diabetic individuals, AIP outperformed TyG-derived indices, suggesting it as a strong predictor of HTN among this population. 

The associations between the TyG index derivates, AIP, and HTN are largely attributed to IR and lipid dysregulation. IR impairs endothelial function by disrupting the balance between vasodilatory and vasoconstrictory signals, leading to increased arterial stiffness and elevated blood pressure⁵⁹. Elevated TG levels, reflected in both the TyG-derived indices and AIP, are associated with the formation of atherogenic lipoproteins, which increase vascular inflammation and blood pressure⁶⁰. Additionally, abdominal obesity, as reflected by TyG-WHtR, contributes to HTN through increased release of inflammatory cytokines from visceral fat ⁶¹.

The consistent predictive performance of TyG-WHtR across all glycemic subgroups may be explained by its ability to simultaneously capture two key components of HTN risk:IR and central obesity. The TyG index, is a well-recognized surrogate marker of IR ⁶². The waist-to-height ratio component reflects central adiposity, which is linked to HTN^63,64. By integrating these biochemical and anthropometric dimensions, TyG-WHtR may provide a more comprehensive assessment of the metabolic and structural determinants of HTN risk than either parameter alone, which could explain its robust performance across different metabolic states observed in our study.

Among the indices, TyG-WHR and TyG-WHtR demonstrated slightly better discriminatory performance, which may be attributed to their incorporation of central obesity measures, a key contributor to HTN. These findings highlight that while these indices are useful for association analyses, their practical discriminatory value should be interpreted with caution.

The associations observed between TyG-based indices, AIP, and HTN can be explained by several intertwined metabolic, inflammatory, and vascular mechanisms^65,66. TyG and its derivatives serve as reliable surrogate markers of IR, a central abnormality that promotes HTN through multiple biological pathways. IR impairs endothelial nitric oxide synthesis, enhances endothelin-1–mediated vasoconstriction, and increases oxidative stress, collectively leading to reduced vascular compliance and increased peripheral resistance^67,68. It also stimulates sympathetic nervous system activity, elevates heart rate and vascular tone, and augments renal sodium reabsorption, thereby expanding plasma volume and raising blood pressure. Furthermore, IR accelerates hepatic overproduction of triglyceride-rich lipoproteins, increases circulating free fatty acids, and promotes ectopic fat accumulation, especially in visceral depots. These processes activate inflammatory pathways—including TNF-α, IL-6, and CRP—which further damage endothelial cells and intensify vascular stiffness^69–71.

AIP adds another mechanistic dimension by reflecting atherogenic dyslipidemia, characterized by elevated s TG, low HDL-C, and an abundance of small dense LDL particles. These lipid alterations accelerate the formation of foam cells, promote subclinical atherosclerosis, and diminish endothelial repair capacity. Triglyceride-rich remnants penetrate the vascular wall and induce oxidative modification, while low HDL-C reduces reverse cholesterol transport and anti-inflammatory activity^62,72–75. Together, these disturbances enhance arterial stiffness, thicken the vascular intima, and impair vasodilatory responses, all of which increase systolic and diastolic blood pressures. The convergence of these mechanisms explains why TyG-based indices and AIP consistently correlate with HTN: they reflect complementary dimensions of metabolic overload, lipid toxicity, adipose-tissue inflammation, and vascular injury that collectively drive the pathophysiology of elevated blood pressure.

From a clinical perspective, the consistent associations of TyG-derived indices and AIP with HTN highlight their potential value as practical, low-cost tools for cardiometabolic risk stratification in routine healthcare settings. Because these indices rely on simple laboratory and anthropometric measurements, they can be easily incorporated into primary care workflows, especially in regions where access to advanced metabolic testing is limited. Their ability to capture early metabolic disturbances—such as i IR, visceral adiposity, and atherogenic dyslipidemia—suggests they may help clinicians identify individuals at elevated risk for HTN even before overt metabolic disease develops. Integrating these markers into standard assessments may allow for earlier lifestyle interventions, closer monitoring, and more personalized prevention strategies. Additionally, the stronger associations observed for specific indices such as TyG-WHtR and AIP emphasize that different metabolic pathways may dominate HTN risk in different clinical contexts, supporting a more targeted approach to patient management. Ultimately, these indices could serve as complementary tools alongside traditional risk factors, improving the precision and efficiency of HTN prevention efforts.

This study has several limitations that should be considered when interpreting the findings. First, due to its cross-sectional design, causal associations between TyG-related indices, AIP, and HTN cannot be established. Longitudinal cohort studies are required to clarify temporal relationships and evaluate the predictive value of these indices for incident HTN.

Second, although we adjusted for a wide range of demographic, behavioral, and clinical factors, the possibility of residual confounding cannot be fully eliminated. Important variables such as detailed dietary patterns were not available, despite their known influence on lipid profiles and blood pressure. Additionally, other unmeasured factors—including genetic susceptibility, psychosocial stress, and early-stage or subclinical comorbidities (e.g., kidney dysfunction or cardiovascular abnormalities)—may have influenced both the exposure indices and HTN outcomes.

Third, several lifestyle-related covariates, including smoking, alcohol consumption, hookah use, opium use, and physical activity, were assessed through self-reported questionnaires. These measures are subject to recall bias and misclassification, which may have introduced measurement error and contributed to attenuation or variability in the observed associations.

Fourth, participants using lipid-lowering medications were not excluded because detailed medication data were not consistently available at the precise time of biochemical assessment. Excluding these individuals based on incomplete information could have introduced selection bias and reduced the generalizability of the findings. However, lipid-modifying therapies may lower TG levels and alter HDL-cholesterol concentrations, potentially influencing AIP and TyG-derived indices. While this may have partially attenuated some associations, the main findings remained statistically significant, suggesting robustness despite this limitation.

Finally, the study population was drawn from a specific geographic region in northern Iran. Although the cohort is large and well-characterized, the results may not be fully generalizable to populations with different ethnic, cultural, socioeconomic, or lifestyle characteristics. Future studies should evaluate these associations in diverse populations and explore additional biomarkers, including novel metabolic indices and machine learning–based risk prediction tools, to strengthen the clinical applicability of TyG-derived indices and AIP. Investigating these markers in younger populations or individuals with early metabolic disturbances may also yield valuable insights for earlier prevention strategies.

Future studies should build upon our findings by incorporating longitudinal designs to determine the temporal and causal relationships between TyG-derived indices, AIP, and HTN. Additionally, integrating comprehensive dietary assessments, genetic profiling, and objective measures of lifestyle behaviors (such as accelerometer-based physical activity or biochemical verification of tobacco exposure) could reduce measurement bias and improve analytical precision. Large, multi-center studies across diverse ethnic and geographic populations are also warranted to evaluate the generalizability of these associations. Furthermore, combining TyG-related indices and AIP with emerging biomarkers, imaging modalities, or machine learning–based risk prediction models may enhance early detection of individuals at elevated cardiometabolic risk. Finally, evaluating these indices in younger populations and in those with early metabolic dysfunction may help identify opportunities for earlier intervention and targeted prevention strategies.

Conclusion

This study compared TyG index derivatives and AIP in association with HTN across different glycemic statuses in a large Iranian population. In the fully adjusted analysis, AIP showed the strongest association with HTN in the overall population, as well as among normoglycemic and diabetic individuals, while TyG-WHR demonstrated the strongest association in the prediabetic group. TyG-WHtR consistently showed meaningful associations across all glycemic categories. These findings indicate that although both TyG-derived indices and AIP are valuable markers associated with HTN, their relative strengths vary by glycemic status. Incorporating the most relevant index for each metabolic profile into clinical assessments may enhance the identification of individuals who are more likely to present with HTN and support more targeted prevention strategies.

Abbreviations

AIP

Atherogenic Index of Plasma

TyG

Triglyceride-Glucose

TyG-BMI

Triglyceride-Glucose Body Mass Index

TyG-WC

Triglyceride-Glucose Waist Circumference

TyG-WHtR

Triglyceride-Glucose Waist-to-Height Ratio

TyG-WHR

Triglyceride-Glucose Waist-to-Hip Ratio

BMI

Body Mass Index

HDL-C

High-Density Lipoprotein Cholesterol

LDL-C

Low-Density Lipoprotein Cholesterol

FPG

Fasting Plasma Glucose

MET

Metabolic Equivalent of Task

PGCS

PERSIAN Guilan Cohort Study

PERSIAN

Prospective Epidemiological Research Studies in Iran

CI

Confidence Interval

OR

Odds Ratio

ROC

Receiver Operating Characteristic

AUC

Area Under the Curve

SD

Standard Deviation

WHO

World Health Organization

IR

Insulin Resistance

HTN

Hypertension

TG

Triglyceride

Author contributions

F.J, and F.MG designed and supervised the study. E.AS and N.L wrote the manuscript. S.H, S.M and M.A consulted on the possible associated factors to be taken into account. E.AS, and N.L, analyzed and interpreted the data. F.J, and F.MG performed the technical revision of the manuscript. All authors read and approved the final manuscript.

Funding

The study was funded of Guilan University of Medical Sciences (IR.GUMS.REC.1403.254).

Data availability

The datasets used and/or analyzed during the current study can be provided from the corresponding author on reasonable request.

Declaration

Artificial Intelligence statement

Artificial intelligence tool (ChatGPT by OpenAI) was used exclusively for language editing purposes, including grammar, punctuation, andclarity. AI was not involved in the study design, data analysis or interpretation, or the development of scientific content. All authorsreviewed the manuscript in full and take complete responsibility for the accuracy, integrity, and originality of the work.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The study was approved by the Ethics Committee of Guilan University of Medical Sciences (IR.GUMS.REC.1403.254). All participants provided written informed consent prior to enrollment. The study was conducted in accordance with the ethical principles outlined in the Declaration of Helsink.