Risk factors and a predictive model for left ventricular hypertrophy in young adults with salt‐sensitive hypertension

Abstract

Salt‐sensitive hypertension is common among individuals with essential hypertension, and the prevalence of left ventricular hypertrophy (LVH) has increased. However, data from early identification of the risk of developing LVH in young adults with salt‐sensitive hypertension are lacking. Thus, the present study aimed to design a nomogram for predicting the risk of developing LVH in young adults with salt‐sensitive hypertension. A retrospective analysis of 580 patients with salt‐sensitive hypertension was conducted. The training set consisted of 70% (n = 406) of the patients, while the validation set consisted of the remaining 30% (n = 174). Based on multivariate analysis of the training set, predictors for LVH were extracted to develop a nomogram. Discrimination curves, calibration curves, and clinical utility were employed to assess the predictive performance of the nomogram. The final simplified nomogram model included age, sex, office systolic blood pressure, duration of hypertension, abdominal obesity, triglyceride‐glucose index, and estimated glomerular filtration rate (eGFR). In the training set, the model demonstrated moderate discrimination, as indicated by an area under the receiver operating characteristic (ROC) curve of 0.863 (95% confidence interval: 0.831–0.894). The calibration curve exhibited good agreement between the predicted and actual probabilities of LVH in the training set. Additionally, the validation set further confirmed the reliability of the prediction nomogram. In conclusions, the simplified nomogram, which consists of seven routine clinical variables, has shown good performance and clinical utility in identifying young adults with salt‐sensitive hypertension who are at high risk of LVH at an early stage.

1. INTRODUCTION

Hypertension is a frequent cause of cardiovascular diseases and premature death, and its prevalence is increasing in young patients. ¹ , ² Numerous studies have demonstrated the link between dietary salt intake and hypertension, and a reduction in dietary sodium intake significantly lowers blood pressure. ³ , ⁴ Salt‐sensitive hypertension, characterized by an exaggerated increase in blood pressure in response to salt loading, is a common but often overlooked phenotype of hypertension. ⁵ More than one‐half of individuals with essential hypertension are affected by salt‐sensitive hypertension, the prevalence of which increases with age. ³ , ⁶ In hypertensive patients who exhibit salt sensitivity, there is a greater occurrence of target organ damage. ⁷ Left ventricular hypertrophy (LVH) is a prevalent type of cardiac damage caused by hypertension and frequently leads to heart failure. ⁸ It has been reported that LVH is observed in approximately two‐fifths of hypertensive patients. ⁹ While it typically manifests clinically in midlife and beyond, it can also develop in young adulthood. ¹⁰ Unlike patients with other hypertension phenotypes (masked hypertension, salt‐resistant hypertension, etc.), patients with salt‐sensitive hypertension exhibit an increased incidence of LVH, regardless of blood pressure. ¹¹ There is particular interest in targeted screening programs to identify those at highest risk of developing LVH. However, it is challenging to predict which young patients who meet the criteria for salt‐sensitive hypertension will ultimately progress to LVH. More predictors should be explored to facilitate risk stratification in these patients.

The development of LVH in hypertensive patients is influenced by various factors. Numerous studies have shown strong correlations between LVH risk and traditional risk factors, including age, sex, coexisting medical conditions, lifestyle factors, dietary habits, and lipid profiles. ⁸ , ¹¹ Early screening should be conducted in high‐risk populations to enable timely intervention and delay the onset and progression of LVH. Several clinical models are currently used to assess future risk in patients with hypertension in the general population, ¹² , ¹³ but none have focused on salt‐sensitive hypertension. In addition, considerable evidence has been obtained based on Caucasian populations and neglected to address ethnic disparities in the incidence of LVH, such as in Asian populations. ¹⁴ To develop effective preventive strategies for LVH in patients with salt‐sensitive hypertension, it is crucial to analyze and quantify the influence of key risk factors comprehensively. Nevertheless, there is no nomogram model for assessing the risk of developing LVH in patients with salt‐sensitive hypertension, especially in China.

This study aimed to identify and incorporate vital clinical characteristics related to LVH into a nomogram model for predicting the risk of LVH in young patients with salt‐sensitive hypertension.

2. METHODS

2.1. Patient population

The initial participants were recruited mainly from the Physical Examination Department of the First Affiliated Hospital of Chengdu Medical College and from the annually free community‐based blood pressure screening program in the Xindu and Pengzhou Districts, Chengdu, Sichuan Province, China (Figure S1). Based on the results of blood pressure measurements and salt intake as estimated by a dietary questionnaire, patients who were suspected to have salt‐sensitive hypertension or another hypertension phenotype were highly recommended to be hospitalized for further examinations by local community health centers and the Health Benefits Program through an increase in Medicare reimbursement. Overall, we reviewed records from 2983 hypertensive patients who were hospitalized at the First Affiliated Hospital of Chengdu Medical College, China, between April 2021 and November 2023. The inclusion criteria were as follows: (a) aged 18−45 years, (b) diagnosed with essential hypertension, ³ , ⁵ , ¹⁵ (c) underwent echocardiography and routine medical workup, and (d) fulfilled the salt‐sensitive hypertension criteria. ³ , ⁶ , ¹⁶ The exclusion criteria were as follows: had (a) secondary hypertension, (b) severe cardiovascular and cerebrovascular diseases, (c) serious liver or kidney diseases, (d) autoimmune diseases and malignancies, (e) active inflammatory or infectious diseases, (f) pregnancy, or (g) incomplete records. Finally, complete data from 580 patients with salt‐sensitive hypertension were obtained (Figure 1). This study was performed in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the First Affiliated Hospital of Chengdu Medical College (approval number: LXKY00100). Written informed consent was obtained from the patients (or parents or legal guardians) before the commencement of the study.

FIGURE 1.

The workflow of this study.

2.2. Data collection

Two researchers collected and accessed all eligible patients. Demographic parameters, self‐report questionnaires, previous medical history, laboratory test results, and echocardiography data were recorded for subsequent analysis. The 580 patients with salt‐sensitive hypertension were evaluated by the modified acute saline loading method. ¹⁶ , ¹⁷ Blood pressure parameters of each patient were measured during the first 24 h after admission. Office blood pressure was measured using a validated semiautomatic device (Omron HEM‐1020, Omron Healthcare, Kyoto, Japan) on the same arm while the participant was in a seated position. Three consecutive measurements were taken at 5‐min intervals and averaged for each patient. Ambulatory blood pressure was measured using a validated automatic dynamic blood pressure monitoring device (Oscar 2, SunTech Medical Instruments, USA). Recordings were taken every 20 min during the daytime (7:00 a.m.−10:00 p.m.) and every 30 min during nighttime (10:00 p.m.−7:00 a.m.). A minimum of 20 daytime and seven nighttime recordings were required for a valid ambulatory blood pressure monitor for subsequent analysis. ¹⁸ Echocardiography was performed using a Vivid‐7 Pro Ultrasound machine (GE Healthcare, USA) by at least two experienced sonographers. Body mass index (BMI) was calculated by dividing weight (in kg) by height squared (in meters) (kg/m²). ¹⁹ Waist circumference was measured in the horizontal plane midway between the superior iliac crest and the lower margin of the last rib. ¹⁹ The estimated glomerular filtration rate (eGFR) was calculated as follows: 186 × (serum creatinine) −1.154 × (age)−0.203 × (0.742, if female). ²⁰ The triglyceride‐glucose (TyG) index was calculated as ln (fasting triglycerides (mg/dL) × fasting plasma glucose (mg/dL)/2). ²¹ Each patient was required to have a urine sample collected at any time within a 24‐h period after admission (casual spot urine sample). The method for the estimation of 24‐h urinary sodium excretion using spot urine samples was the Sun_C method. ²²

2.3. Cold pressor test

The cold pressor test (CPT) was administered during the initial assessment. ²³ Following a 20‐min period of rest in a seated position, three blood pressure readings were taken using a standard mercury sphygmomanometer prior to ice water immersion. Subsequently, the participants submerged their left hand in an ice water bath (3−5°C) up to just above the wrist for a duration of 1 min. ²³ Blood pressure readings were then recorded at intervals of 0, 1, 2, and 4 min on the right arm using a standard mercury sphygmomanometer after the left hand had been removed from the ice water bath. ²³ A positive result for the test was determined if the systolic blood pressure, diastolic blood pressure, or mean arterial pressure (MAP) exhibited an increase of 15 mm Hg or more following cold stimulation. ²³

2.4. Assessment of salt sensitivity

Considering the half‐life of antihypertensive drugs, all the patients were asked to pause taking antihypertensive drugs for at least a full day before the assessment of salt sensitivity. ¹⁶ The assessment of salt sensitivity was conducted using a modified acute saline loading method (Figure S2) known as modified Sullivan's method after CPT assessment. ¹⁶ , ¹⁷ Blood pressure was measured three times at 8:00 a.m., and the average value was considered the baseline blood pressure for each fasting patient. Subsequently, a 4‐h intravenous infusion of 2000 mL of 0.9% saline was administered, and 40 mg of furosemide was given orally immediately after the saline infusion. Blood pressure was then measured hourly during the trial, up to 4 h after the administration of furosemide. An increase in the MAP of more than 5 mm Hg after acute salt loading and/or a decrease in the MAP of more than 10 mm Hg after the diuresis shrinkage test were considered to indicate salt‐sensitive. ¹⁶ , ¹⁷

2.5. Definitions

Hypertension was defined as a mean systolic blood pressure ≥140 mm Hg, a diastolic blood pressure ≥90 mm Hg, or the use of antihypertensive medication. ³ , ¹⁵ Nocturnal hypertension was defined as a mean nighttime systolic blood pressure ≥120 mm Hg and/or mean diastolic blood pressure ≥70 mm Hg, irrespective of nighttime blood pressure dipping patterns or daytime blood pressure levels. ¹⁸ LVH was described as a left ventricular muscle mass index (LVMI) of 115 g/m² for men and >95 g/m² for women. ²⁴ All patients were divided into a non‐LVH group and an LVH group based on the results of echocardiography. Abdominal obesity was defined as a waist circumference ≥90 cm for males or ≥85 cm for females. ¹⁹ Regular exercise was described as an aerobic exercise of 90−150 min per week with 65%−75% of heart rate reserve. ²⁵ Current smoking was defined as smoking at least one cigarette per day in the past 6 months. ²⁶ Alcohol intake was defined as drinking alcohol at least weekly during the past year. ²⁷ Renal function was determined by the eGFR. ²⁰ Obstructive sleep apnea hypopnea syndrome (OSAHS) was characterized by daytime sleepiness and obstructive apnea and hypopnea episodes during sleep. ²⁸ Diabetes was defined as a fasting plasma glucose level of 126 mg/dL or higher (≥7.0 mmol/L) and/or a postprandial 2‐h plasma glucose level of 200 mg/dL or higher (≥11.1 mmol/L) and/or a glycated hemoglobin (HbA1c) level of 6.5% or higher (≥48 mmol/mol). ²⁹ A family history of hypertension was identified by the presence of hypertension in at least one parent of the patient being studied. An educational level ≥12 years was defined as tertiary education.

2.6. Development and assessment of the nomogram

A total of 580 patients were randomly divided into two cohorts: 70% (n = 406) formed the training set for constructing the nomogram model, while the remaining 30% (n = 174) formed the validation set. Randomized sequences were generated using the R software random number generator with a permuted block design. ³⁰ Potential predictive variables were screened using least absolute shrinkage and selection operator (LASSO) regression and multivariate logistic regression analysis. The importance of each screened predictor in the multivariate logistic regression model was estimated by the partial chi‐square statistic minus the predicted degrees of freedom. ³¹ Subsequently, independent predictors (p < .05) identified through multivariate logistic regression were included in the nomogram for predicting the risk of LVH. For simplicity, all selected continuous variables were classified as categorical variables based on traditional clinical cutoffs or preliminary data assessment.

Receiver operating characteristic (ROC) curves were performed to evaluate the ability of the nomogram to differentiate between LVH. Calibration curves and nonparametric bootstraps of 1000 resamples were employed to assess the agreement between the actual observed values and the predicted values from the nomogram. Decision curve analysis (DCA) was used to evaluate the clinical utility of the nomogram by quantifying the net benefit at various threshold probabilities. The derivation and validation of the nomogram were performed according to the checklist in the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) guideline (TRIPOD Checklist). ³⁰

2.7. Sample size

The adequate sample size for the prediction research, which included both the development and validation phases, was determined by the number of outcome events. ³² To ensure accuracy, we aimed to have at least 10 outcome events per variable (EPV). ³² Considering the documented prevalence of LVH in the general population, ¹² , ¹³ particularly the incidence of 20.4% in our center, ³³ we anticipated a 25% event rate for LVH in our research. To include 10 or fewer predictors in the final multivariable logistic regression model, we estimated that a sample size of 400 patients or more would be required. Our study had a sample size and number of outcome events that far exceeded those of the EPV method, thus providing reliable estimates.

2.8. Statistical analysis

Descriptive statistics were used to summarize categorical and numerical variables. Categorical data are expressed as numbers and percentages, while continuous variables are summarized as the mean (standard deviation) or median (interquartile range). Univariate tests (such as the chi‐square test, t test, or Mann‒Whitney U test) and multivariate logistic regression analyses were conducted. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. All the statistical analyses were performed using SPSS 24.0 software (IBM Corp., Armonk, NY), and the R 4.2.2 programming language (rms, pROC, ggplot2, or dca packages, as appropriate) was employed to establish the nomogram. Values of p < .05 were considered statistically significant.

3. RESULTS

3.1. Patient characteristics

In total, 580 patients with salt‐sensitive hypertension were enrolled in this study. The clinical characteristics of the patients in the training and validation sets are listed in Tables 1 and S1. The prevalence of LVH was 25.52% (148/580). In the training set, patients with LVH were more likely to be female and to have a history of alcohol intake, abdominal obesity, a longer duration of hypertension, a lower eGFR, and higher office systolic blood pressure, nighttime systolic blood pressure, triglycerides, fasting blood glucose, TyG index, uric acid, and estimated 24‐h urinary sodium excretion than patients without LVH were (Table 1). For the 580 patients included in the training and validation analyses, no significant differences were observed for any of the parameters between the two groups (Table S2). The geometry of the LVH in the training and validation sets is shown in Figure S3.

TABLE 1.

Baseline characteristics of patients in the training and validation sets.

	Training set			Validation set
Variables	LVH (n = 102)	Non‐LVH (n = 304)	p value	LVH (n = 46)	Non‐LVH (n = 128)	p value
Age (years)	38.66 ± 6.09	33.43 ± 8.60	<.001	38.86 ± 5.80	33.44 ± 8.65	<.001
Female, n (%)	58 (56.9)	86 (28.3)	<.001	26 (56.5)	36 (28.1)	.001
Educational level ≥12 years, n (%)	55 (53.9)	182 (59.9)	.299	25 (54.3)	75 (58.6)	.728
Current smoking, n (%)	40 (39.2)	110 (36.2)	.636	17 (37.0)	46 (35.9)	.902
Alcohol intake, n (%)	46 (45.1)	70 (23.0)	<.001	22 (47.8)	29 (22.7)	.002
BMI (kg/m2)	25.66 ± 2.52	25.38 ± 2.63	.348	25.23 ± 2.42	25.62 ± 2.47	.357
Waist circumference (cm)	88.46 ± 4.72	87.74 ± 4.48	.167	88.87 ± 3.91	87.86 ± 4.62	.188
Regular exercise, n (%)	21 (20.6)	76 (25.0)	.422	9 (19.6)	33 (25.8)	.431
Family history of hypertension, n (%)	30 (29.4)	79 (26.0)	.520	14 (30.4)	34 (26.6)	.701
Diabetes mellitus, n (%)	12 (11.8)	27 (8.9)	.437	6 (13.0)	12 (9.4)	.573
OSAHS, n (%)	12 (11.8)	24 (7.9)	.233	5 (10.9)	11 (8.6)	.766
Abdominal obesity, n (%)	42 (41.2)	49 (16.1)	<.001	19 (41.3)	21 (16.4)	.001
Duration of hypertension (years)	5.76 ± 2.99	2.75 ± 2.13	<.001	5.70 ± 2.92	2.61 ± 2.24	<.001
Antihypertensive therapy, n (%)	22 (21.6)	54 (17.8)	.383	10 (21.7)	23 (18.0)	.661
At admission
Office SBP (mm Hg)	183.26 ± 17.74	163.18 ± 15.81	<.001	184.20 ± 18.06	163.73 ± 17.38	<.001
Office DBP (mm Hg)	79.12 ± 6.31	78.82 ± 6.25	.676	78.98 ± 5.97	79.72 ± 6.40	.810
Office heart rate (bpm)	75.23 ± 10.03	74.85 ± 9.70	.735	74.30 ± 9.50	76.24 ± 9.28	.229
24‐h ABPM
24‐h mean SBP (mm Hg)	158.78 ± 21.99	156.79 ± 21.19	.417	158.50 ± 20.59	155.52 ± 20.50	.400
24‐h mean DBP (mm Hg)	78.93 ± 16.70	77.49 ± 16.78	.453	77.02 ± 16.71	76.26 ± 16.55	.790
Daytime SBP (mm Hg)	170.48 ± 16.97	168.58 ± 16.41	.316	169.48 ± 16.63	166.65 ± 16.67	.326
Daytime DBP (mm Hg)	84.25 ± 10.84	83.42 ± 10.57	.496	83.37 ± 10.62	82.82 ± 8.90	.734
Nighttime SBP (mm Hg)	138.66 ± 13.26	134.72 ± 14.19	.014	139.00 ± 13.58	133.33 ± 14.28	.021
Nighttime DBP (mm Hg)	73.41 ± 10.17	72.29 ± 10.99	.365	73.54 ± 10.50	72.40 ± 10.95	.541
Laboratory measurements
TC (mmol/L)	5.59 ± 1.04	5.42 ± 0.81	.090	5.47 ± 0.59	5.31 ± 0.77	.202
TG (mmol/L)	1.45 ± 0.19	0.70 ± 0.07	<.001	1.39 ± 0.17	0.69 ± 0.08	<.001
HDL‐C (mmol/L)	1.60 ± 0.19	1.57 ± 0.15	.104	1.59 ± 0.16	1.56 ± 0.15	.255
LDL‐C (mmol/L)	2.68 ± 0.65	2.58 ± 0.58	.145	2.71 ± 0.58	2.53 ± 0.59	.076
Fasting blood glucose (mmol/L)	5.31 ± 0.48	4.98 ± 0.31	<.001	5.26 ± 0.42	4.96 ± 0.30	<.001
TyG index	9.25 ± 1.15	7.66 ± 1.20	<.001	9.05 ± 0.86	7.70 ± 1.19	<.001
CRP (mg/L)	3.14 ± 0.84	3.00 ± 0.76	.118	3.07 ± 0.93	2.92 ± 0.72	.265
RDW (%)	12.94 ± 0.43	12.89 ± 0.41	.293	12.98 ± 0.38	12.87 ± 0.40	.107
Hcy (µmol/L)	17.21 ± 3.36	16.69 ± 3.14	.156	17.35 ± 3.34	16.61 ± 3.06	.172
CysC, mg/L	0.96 ± 0.18	0.93 ± 0.19	.163	0.98 ± 0.25	0.93 ± 0.20	.176
Uric acid (µmol/L)	466.89 ± 108.80	386.61 ± 54.09	<.001	463.52 ± 141.20	372.22 ± 54.40	<.001
Estimated 24‐h USE (mmol/d)	201.80 ± 24.23	196.11 ± 21.33	.025	204.08 ± 22.69	195.41 ± 23.01	.029
eGFR (mL/min/1.73 m2)	84.87 ± 8.87	101.44 ± 8.63	<.001	86.22 ± 8.65	102.82 ± 8.49	<.001
Echocardiography
LVMI (g/m2)	124.21 ± 11.18	89.18 ± 9.20	<.001	122.33 ± 10.43	91.36 ± 8.98	<.001
LVM (g)	213.59 ± 24.33	162.99 ± 19.88	<.001	208.11 ± 24.69	165.05 ± 18.95	<.001
LVEF (%)	63.50 ± 2.32	63.83 ± 2.25	.204	64.13 ± 2.48	63.97 ± 2.19	.682

Note: The data are expressed as the mean ± SD and proportion (%).

A p value of <.05 was considered to indicate statistical significance.

Abbreviations: ABPM, ambulatory blood pressure monitoring; BMI, body mass index; CRP, C‐reactive protein; CysC, cystatin C; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; Hcy, homocysteine; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol; LVEF, left ventricular ejection fraction; LVH, left ventricular hypertrophy; LVM, left ventricular mass; LVMI, left ventricular mass index; OSAHS, obstructive sleep apnea hypopnea syndrome; RDW, red blood cell distribution width; SBP, systolic blood pressure; TC, total cholesterol; TG, triglyceride; TyG, triglyceride‐glucose; USE, urinary sodium excretion.

In addition, the concordance between the CPT and the modified acute saline loading test for the individual detection of salt sensitivity was assessed in this study. There was no statistically significant difference in salt sensitivity between the modified acute saline loading test and the CPT (χ² = 0.013, p = .908), and 732 of the 735 patients with salt‐sensitive hypertension were CPT positive according to the modified acute saline loading test (Table S3). The prevalence of LVH in patients with salt‐sensitive hypertension determined by the results of the modified acute saline loading test and CPT is shown in Figure S4.

3.2. Screening for predictive factors

LASSO regression analysis was performed to determine potential predictors with nonzero coefficients, as shown in Figure 2. These variables included age, sex, alcohol intake, office systolic blood pressure, duration of hypertension, abdominal obesity, TyG index, uric acid, estimated 24‐h urinary sodium excretion, and eGFR. Subsequently, a multivariate logistic regression analysis identified seven factors as independent predictors of LVH, yielding the following results: age (OR: 1.830; 95% CI: 1.202–2.460; p = .002), female sex (OR: 1.571; 95% CI: 1.122–2.038; p = .003), abdominal obesity (OR: 2.175; 95% CI: 1.516–2.854; p < .001), duration of hypertension (OR: 2.357; 95% CI: 1.460–3.309; p < .001), office systolic blood pressure (OR: 4.035; 95% CI: 2.654–5.459; p < .001), TyG index (OR: 3.665; 95% CI: 2.380–4.986; p < .001), and eGFR (OR: 3.067; 95% CI: 2.202–3.961; p < .001) (Figure 3). The importance of each variable in the multivariate logistic regression model is illustrated in Figure S5.

FIGURE 2.

The risk factors for LVH were screened using the LASSO binary logistic regression model. (A) The optimal parameters (lambda) of the LASSO model were determined by 10‐fold cross‐validation. The black and red dotted vertical lines indicate the optimal lambda values that were achieved using the minimum criterion and one standard error of the minimum criterion (1‐SE criterion), respectively. Ten variables with nonzero coefficients were selected based on the 1‐SE criterion. (B) LASSO coefficient profiles of all the candidate variables. The coefficient profile plot was produced against the log (lambda) sequence. LASSO, least absolute shrinkage and selection operator.

FIGURE 3.

Multivariate analysis of risk factors for LVH. Estimated odds ratios determined in a logistic regression model (stepwise selection: forward). CI, confidence interval; eGFR, estimated glomerular filtration rate; LVH, left ventricular hypertrophy; OR, odds ratio; SBP, systolic blood pressure; TyG, triglyceride‐glucose; USE, urinary sodium excretion.

3.3. Risk prediction nomogram development

The final nomogram model was constructed using multivariate analysis and included seven independent risk factors: age, sex, office systolic blood pressure, duration of hypertension, abdominal obesity, TyG index, and eGFR (Figure 4A). The nomogram converts the risk associated with each indicator into a numerical value. The probability corresponding to the sum of these values represents the likelihood of developing LVH.

FIGURE 4.

Development of the risk nomogram. (A) A nomogram for predicting the risk probability of LVH in young adult patients with salt‐sensitive hypertension. (B) The dynamic nomogram used as an example on the shinyapps.io platform (https://cmulvh.shinyapps.io/dynnomapp/). Users can calculate the risk probability of LVH for young adult patients with salt‐sensitive hypertension by submitting information on seven related traits online via mobile phones or computers. eGFR, estimated glomerular filtration rate; LVH, left ventricular hypertrophy; SBP, systolic blood pressure; TyG, triglyceride‐glucose.

To simplify its use in clinical settings, we developed an online predictive nomogram (Figure 4B). Clinicians can input the seven metrics from the full model into the designated text box on the website for calculation (available at https://cmulvh.shinyapps.io/dynnomapp/).

3.4. Evaluations of the nomogram performance

The nomogram was evaluated internally using the area under the curve (AUC) and calibrated using 1000 bootstrap samples. In the training set (Figure 5A), the AUC was 0.863 (95% CI: 0.831–0.894), while in the validation set (Figure 5B), the AUC was 0.803 (95% CI: 0.766–0.840). Furthermore, we evaluated each variable in the nomogram model using ROC curve analysis, as shown in Figure S6. These findings indicated that the nomogram has good discriminatory power and predictive value, enabling accurate differentiation between LVH and non‐LVH patients.

FIGURE 5.

Evaluation of the nomogram. (A) ROC curve for the training set. (B) ROC curve for the validation set. (C) Calibration curve for the training set. (D) Calibration curve for the validation set. (E) Decision curve analysis for the training set. (F) Decision curve analysis for the validation set. AUC, area under the curve.

The Hosmer−Lemeshow test was performed to assess the fit of the nomogram, and the results demonstrated that the nomogram was a good fit for both the training set (p = .424) and the validation set (p = .372). The calibration curves in the training (Figure 5C) and validation (Figure 5D) sets closely resembled the ideal diagonal line. Moreover, the clinical validity of the model was evaluated using the DCA method (Figure 5E, F). The DCA demonstrated a significantly favorable net benefit and predictive accuracy of the nomogram model in the training and validation sets. These findings indicate that our nomogram has significant potential for informing clinical decision‐making.

When comparing the predictive performance of our nomogram to that of a previously established tool ¹⁴ for predicting LVH, we observed that the AUC of our nomogram was greater in the training set (0.863 vs. 0.637; p < .01) (Figure S7, A). In the validation set, our nomogram had an AUC of 0.803, which was greater than that of the existing tool (0.624 [95% CI, 0.575–0.699]; p < .01) (Figure S7, B).

4. DISCUSSION

The main finding of the present study was the development and validation of a prediction model for developing LVH in young patients with salt‐sensitive hypertension. The final nomogram yielded an estimated probability of LVH for each patient based on information from seven objective clinical items, which included age, sex, office systolic blood pressure, duration of hypertension, abdominal obesity, the TyG index, and the eGFR. In addition, the nomogram also showed good discrimination and calibration.

The pathogenesis of LVH is a complex process influenced by various factors. First, it was observed that age plays a significant role in predicting the development of LVH in young patients with salt‐sensitive hypertension. The findings indicated that individuals aged ≥40 years with salt‐sensitive hypertension have a greater likelihood of developing LVH than those under 40 years of age, which aligns with the findings of previous researches. ¹² , ¹³ , ³⁴ It is well known that left ventricular mass increases with age. ³⁴ Early‐onset hypertension is associated with greater odds of cardiac damage and cardiovascular mortality than late‐onset hypertension. ³⁴ , ³⁵ Second, our predictive model showed that female sex was associated with LVH. Previous studies have also established a correlation between female sex and LVH and demonstrated that being female is a strong predictor of LVH. ¹² , ¹³ Recent evidence suggests that women exhibit increased salt sensitivity compared to men, irrespective of ethnicity or menopausal status. ³⁶ Female patients with salt‐sensitive hypertension demonstrated a greater likelihood of developing LVH, potentially due to aldosterone‐mediated endothelial dysfunction in women. ³⁶ Third, these findings support previous observations showing the association of LVH with office systolic blood pressure and the duration of hypertension. ¹² , ¹³ , ³⁷ Patients with high systolic blood pressure often exhibit LVH and accelerated to concentric cardiac remodeling. ³⁷ Emerging evidence indicates that lowering intensive systolic blood pressure protects against LVH development in older hypertensive patients. ³⁸ Similarly, we also found that a shorter duration of hypertension was associated with a lower risk of developing LVH. As a reflector of cumulative exposure to high blood pressure, the duration of hypertension may be considered an essential factor for LVH. ³⁹

Moreover, the current study revealed a correlation between abdominal obesity and LVH in patients with salt‐sensitive hypertension, which is consistent with the findings of previous studies. ⁴⁰ , ⁴¹ Waist circumference serves as a crucial indicator for diagnosing abdominal obesity and has been associated with cardiovascular and metabolic disorders. ⁴² The relationship between abdominal obesity and LVH can be attributed to the strong correlation between waist circumference and excessive visceral adipose tissue accumulation in the body. ¹⁹ , ⁴² Visceral adipose tissue is metabolically active and requires more energy than subcutaneous adipose tissue. ¹⁹ This increased energy demand further burdens the heart in hypertensive patients, ultimately leading to the development of LVH. ¹⁹ , ⁴¹ While BMI is commonly used to assess general obesity, it fails to account for body fat distribution. ¹⁹ , ⁴² Numerous studies have shown that waist circumference, which measures abdominal obesity, is a better predictor of cardiovascular disease risk factors and mortality than BMI. ⁴² , ⁴³

This study also revealed a significant association between a high TyG index and an increased risk of LVH, which is in line with findings from a previous study. ⁴⁴ The TyG index, a simple and cost‐effective marker for insulin resistance, has gained popularity among clinicians due to its ease of use. ²¹ Multiple studies have consistently shown that patients with salt‐sensitive hypertension are more likely to have insulin resistance compared to those without. ⁷ , ¹⁶ , ⁴⁵ An increasing body of evidence suggests that insulin resistance negatively affects diastolic function and contributes to LVH, which is indicative of a poor prognosis. ⁸ , ⁴² This phenomenon might be explained by insulin resistance through three potential mechanisms: inflammatory endothelial dysfunction, ectopic synthesis of angiotensinogen, and hyperinsulinemia causing excessive stimulation of the renin–angiotensin–aldosterone system. ⁴⁴ , ⁴⁵ In addition, our findings support previous studies conducted in the general population demonstrating that kidney dysfunction is associated with a greater LVMI. ³⁴ , ⁴⁶ Furthermore, it is commonly observed that LVH and kidney dysfunction coexist. ⁴⁶ Existing evidence indicates that cardiac remodeling, characterized by LVH and myocardial fibrosis development, occurs as kidney disease progresses. ³⁴ , ⁴⁶ Additionally, 24‐h urinary sodium excretion and nighttime systolic blood pressure were excluded from the final model because their correlation with LVH was present only in the univariate analysis.

Various protocols have been employed to assess salt sensitivity, yet a universally recognized method has not been established in clinical settings. Typically, two primary approaches have been utilized in clinical research: acute salt loading and chronic low‐ and high‐sodium dietary interventions. ¹⁶ , ⁴⁷ Acute saline loading involves the administration of 2000 mL of normal saline infusion over 4 h on day 1, followed by a low‐salt diet and three doses of oral diuretics on day 2. ¹⁶ , ⁴⁷ Chronic interventions with low‐ and high‐sodium diets are implemented to observe blood pressure responses to low‐ and high‐salt consumption over a period of 1–2 weeks. ¹⁶ , ⁴⁷ However, the established methods of salt sensitivity determination are too complicated for screening at the population level. In our study, we utilized the modified acute saline loading method to assess salt sensitivity, in line with the latest Chinese expert consensus on diagnosing salt‐sensitive hypertension. ¹⁶ This method can provide a reliable response within a relatively short timeframe (within a single day), suggesting its potential suitability for large‐scale epidemiological investigations in the Chinese population. ¹⁶ However, it is important to note that the definition of salt sensitivity and the associated procedures vary significantly among studies. ¹⁶ , ⁴⁷ Therefore, there is a pressing need for more research focusing on developing simpler and more universally accepted methods for testing salt sensitivity.

One of the key strengths of this study was that it was the first to propose a nomogram model based on readily available and inexpensive clinical routine data for predicting the risk of LVH in young patients with salt‐sensitive hypertension. This tool might help clinicians stratify young patients with salt‐sensitive hypertension according to LVH. By employing this approach, clinicians could stratify the risk of LVH in these patients more comprehensively. Our simplified model was well calibrated and relied on only seven routine variables. Generally, the application of salt load and diuretics to identify salt‐sensitive hypertension is only carried out in wards, and it is difficult to conduct such experiments in outpatient clinics and communities. Thus, our investigation indicated that CPT may be a simple, accessible, and alternative measure for assessing salt sensitivity in the clinic and in larger populations. In addition, we developed a simple and convenient dynamic nomogram that allows clinicians to calculate individualized risks of LVH. This user‐friendly approach could easily be applied in clinical settings and other cohorts, thereby advancing clinical decision‐making.

This study has several limitations. First, this was a retrospective observational design, which may limit causal inference and generalizability. Our cohort may exhibit referral bias, as our center is an academic hospital, potentially resulting in an underrepresentation of young patients with mild hypertension and low‐moderate cardiovascular risk in the general population. Although the nomogram was internally validated, it is necessary to conduct studies involving diverse ethnic groups and prospective multicenter studies to validate the results obtained in our study externally. Second, the sample size was relatively small. There is no widely or globally approved gold standard for diagnosing salt‐sensitive hypertension. Notably, the salt sensitivity test via the acute saline loading method is complex and time‐consuming in actual clinical practice. Therefore, CPT, as a simple, practical, and alternative method, might be considered for screening salt‐sensitive hypertension, ²³ , ⁴⁸ and further larger‐scale studies should be conducted to provide more evidence. Third, ambulatory blood pressure measurements were performed only once at baseline. Nocturnal hypertension is highly prevalent among Chinese and Asian populations and is mainly attributed to high salt intake and high salt sensitivity. ¹⁸ In this study, nocturnal hypertension was common in young adults with salt‐sensitive hypertension, although nighttime systolic blood pressure was not included in the final prediction model. Considering the moderate reproducibility of ambulatory blood pressure monitoring, it is better to evaluate nocturnal hypertension based on repeated 24‐h ambulatory blood pressure measurements. Another potential limitation of our study is that the method used to estimate sodium intake was based on a formula‐derived estimate of 24‐h urinary excretion rather than actual 24‐h urinary collection. ²² To address variations in daily salt intake among individuals and minimize bias, collecting 24‐h and spot urine samples on multiple consecutive days is needed in future studies. Fourth, several studies have shown that treatment with various classes of antihypertensive drugs can effectively reduce left ventricular mass and influence left ventricular geometry. ³⁷ , ⁴⁹ Our study revealed that a majority of young adults with salt‐sensitive hypertension did not receive antihypertensive treatment, largely due to low awareness rates in this population. ² As a result, the number of patients taking antihypertensive drugs in our study was relatively small, potentially leading to the oversight of some true associations due to insufficient statistical power. Therefore, future research should investigate the impact of different antihypertensive therapies on the risk of LVH in this specific population. Fifth, LVH is recognized as a complex and multifactorial condition that encompasses biological, behavioral, and psychological factors. While we considered 36 variables, our findings may cover only some relevant literature. Hence, future studies incorporating a more comprehensive range of variables could improve the understanding of how risk prediction is performed in diverse patient cohorts.

5. CONCLUSIONS

Higher salt sensitivity and salt intake are characteristics of hypertension in Asian populations than in Western populations. Thus, the early management of hypertension‐mediated cardiac damage is vital in Asian populations. This study established a simplified and validated nomogram model for accurately predicting LVH in young patients with salt‐sensitive hypertension. The predictive model holds significant value in stratifying patients with salt‐sensitive hypertension and identifying those at high risk for LVH.

AUTHOR CONTRIBUTIONS

Jindong Wan: Formal analysis; resources and writing original draft. Peijian Wang: Data curation and supervision. Sen Liu: Resources and validation. Xinquan Wang: Funding acquisition and data curation. Peng Zhou: Validation and supervision. Jian Yang: Conceptualization; methodology and writing review & editing. All the authors critically reviewed and approved the final paper.

CONFLICT OF INTEREST STATEMENT

There are no conflicts of interest.

Supporting information

Supplemental material for this article is available online.

Wan J, Wang P, Liu S, Wang X, Zhou P, Yang J. Risk factors and a predictive model for left ventricular hypertrophy in young adults with salt‐sensitive hypertension. J Clin Hypertens. 2024;26:933–944. 10.1111/jch.14863

DATA AVAILABILITY STATEMENT

The data are available upon reasonable request from the corresponding author.