Introduction

Heart failure (HF) syndrome is a major clinical and public health problem associated with notable morbidity and mortality [1], which affects more than 64 million people worldwide [2]. Adverse changes in cardiac structure and function, known as cardiac remodeling, play a crucial role in the development of heart failure [3]. A particularly significant aspect of this process is LVH, defined as LVMI greater than 95 g/m²in women and greater than 115 g/m²in men [4, 5]. In LVH, the thickening of the heart wall leads to increased muscle stiffness, known as diastolic dysfunction, which impairs the heart’s ability to relax and appropriately fill with blood during the resting phase of the cardiac cycle [6, 7]. Consequently, the heart’s efficiency in pumping blood is diminished, resulting in decreased cardiac output and an increased workload on the heart [8]. Over time, this can lead to the heart being unable to pump with the necessary force [9, 10]. Approximately 8.7–13% of patients with LVH develop dysfunction of the left ventricular ejection fraction, which is a significant risk factor for cardiovascular mortality [11]. Early identification of individuals at higher risk for developing LVH is crucial for reducing related disability and mortality [12, 13]. According to some cross-sectional studies, diabetic individuals with high levels of glycated hemoglobin (HbA1c) may have increased left ventricular mass, which may be related to blood glucose uncontroled [14]; the prevalence of LVH in diabetes was 17.1-20.7% in Italy and Iran [15, 16]. Most studies have concentrated on examining risk factors for heart failure (HF) among individuals with diabetes, leaving a significant void in understanding the same in populations without diabetes [17]. Triglycerides and high-density lipoprotein have been observed to be associated with LVH in the general population in China [18]. In comparison, this study aimed to investigate whether lipid profile, including TC, TG, HDL-C, LDL-C, and Non-HDL-C, are independent risk factors that will predict LVH in the non-diabetic population and to assess their predictive value.

Methods

Study population

This cross-sectional clinical data was collected from pedestrians living in Western Area Urban, Freetown, Sierra Leone, between October 2019 and October 2021 to evaluate the prevalence of cardiometabolic risk factors among adults in Sierra Leone. This study examined the prevalence of cardiometabolic risk factors, target organ damage, and their associated factors among adults [19]. This project’s data were segregated from the “DATADRYAD” database (https://datadryad.org/stash). As the study was a secondary study examining public data from a database, no ethical approval was required. This study obtained data from a database comprising information on 2,394 participants. After exclusions, which involved removing 199 participants with a history of diabetes mellitus and 103 participants with pre-diabetes, the final cohort included 2,092 individuals. Among them, 287 were identified as LVH, while the remaining 1,805 were classified as non-LVH (Fig. 1).

Fig. 1
figure 1

Study protocol. Baseline with 2398 participants, 302 were excluded. Total of 2092 subjects were finally included in the analysis

Full size image

Variables of interest

Physical examinations were conducted on all participants as part of the study, encompassing various aspects such as demographic data (age, gender, and income), lifestyle factors (fruit and vegetable consumption, smoking, and alcohol consumption), medical history (family history of hypertension and diabetes mellitus), anthropometric measurements (Body weight, height, and waist circumference), echocardiography, and blood tests. Information about the participants was collected using standard survey questionnaires. Blood pressure was measured using an Omron electronic sphygmomanometer with an appropriately sized cuff. Participants were seated during the recording, preceded by 3–5 min of rest. The average of two recordings was taken as the blood pressure reading. Weight, height, and waist circumference measurements were taken while participants were lightly clothed and barefoot. These measurements provide insight into body composition and health status. Blood samples were collected from the median cubital vein after an overnight fast (8–10 h) on the morning of the second day. Hypertension was defined as an average systolic blood pressure of 140 mm Hg or higher, diastolic blood pressure of 90 mm Hg or greater, or a participant-reported current use of antihypertensive medication. A participant who smoked more than 100 sticks of cigarettes in their lifetime and was still smoking at the interview was referred to as a smoker, while an ex-smoker was someone who had stopped smoking at least 28 days before the interview. Data on alcohol consumption were based on the WHO step survey tool. The body mass index (BMI) was calculated as a ratio of the weight in kilograms and the square of the height in meters. BMI-based body habitus was classified as obese (BMI ≥ 30 kg/m2), and abdominal obesity was defined as WC > 88 cm for women and 102 cm for men. Diabetes mellitus was defined as a fasting blood glucose (FPG) level of 7.0 mmol/L or greater, HbA1c ≥ 6.5%, or using insulin or an oral hypoglycaemic agent. Pre-diabetes was defined as FPG between 6.1 mmol/L and 6.9 mmol/L. The blood tests included analysis for total TC, TG, HDL-C, and LDL-C levels. TC calculated Non-HDL-C minus HDL-C. All lipid profile were grouped according to the interquartile method. LVMI was calculated by dividing Left ventricular mass by body surface area.

Statistical analyses

SPSS (Version 25.0), GraphPad Prism (Version 9.5.0), and R (Version 4.3.0) were used for data analysis. Univariate analysis was performed for all variables using SPSS. Categorical variables were presented as percentages and compared using a chi-square test. Continuous variables perform a Kolmogorov-Smirnov test, The T-test was used for normal distribution, and the Mann-Whitney U was used for non-normal distribution. The correlation between LVMI and other variables was analyzed using Spearman correlation analysis. Depending on whether LVH or not, PCA was performed on a dataset containing continuous variables using R to reduce variable number. Lipid quartiles were treated as a categorical variable, with the lowest quartile as the reference category. The logistic regression test was performed to investigate the correlation between lipids and LVH using Graphpad. Three adjusted models were constructed to reduced confounders. Model I adjusted for age and income, and model II further adjusted for hypertension, alcohol consumption history, and smoking history. Model III was further adjusted for waist circumference and estimated glomerular filtration rate (eGFR). ROC curves were used to analyze the sensitivity and heterogeneity of the model and calculate the cut-off values. P < 0.05 were considered statistically significant.

Sensitivity analysis

A series of stratified analyses were conducted to evaluate the potential modification of classical risk factors, namely gender and HbA1c. Additionally, BMI adjustments will be made to mitigate the confounding effects of metabolic syndrome. Considering hypertension is an independent risk factor for LVH, its confounding effect will be further examined in the subgroup analysis stratified by hypertension (HTN).

Results

Comparison of characteristics of the LVH and Non-LVH group

Univariate analysis was performed for all variables based on the presence or absence of LVH, and the results are presented in Table 1. A significant increase in TC, TG, Non-HDL-C, and TC/HDL-C levels was observed in the LVH group when compared to the non-LVH group (P < 0.001) (Fig. 2). However, compared to the non-LVH group, patients with LVH group exhibited a significant decrease in HDL-C levels (P < 0.01). The two groups had no significant differences regarding LDL-C, TG/HDL-C, and LDL-C/HDL-C.

Table 1 Comparison of characteristics of the LVH and Non-LVH group
Full size table
Fig. 2
figure 2

Lipid levels in LVH (n=287) and Non-LVH (n=1805) groups. Data are mean ± SD. **p<0.0001

Full size image

Correlation between LVH and potential risk factors

The Spearman-correlation analysis revealed positive correlations between LVMI and age (r = 0.18, P = 0.001), alcohol consumption history(r = 0.07, P = 0.001), hypertension history(r = 0.52, P = 0.001), eGFR(r = 0.35, P = 0.001), TC(r = 0.57, P = 0.001), Non-HDL-C (r = 0.59, P = 0.001), TG(r = 0.72, P = 0.001), TC/HDL-C ratio(r = 0.44, P = 0.001), and Non-HDL-C/HDL-C ratio(r = 0.31, P = 0.001). Conversely, negative correlations were observed with income(r=-0.6, P = 0.001), smoking history(r=-0.09, P = 0.001), waist circumstance(r=-2.2, P = 0.03), LDL-C(r=-0.35, P = 0.001), HDL-C(r=-0.44, P = 0.001), TG/HDL-C (r=-0.5, P = 0.001), and LDL-C/HDL-C(r=-0.6, P = 0.001) (Table 2).

Table 2 Correlation analysis of LVMI and other clinical variables in non-diabetes individuals
Full size table

PCA was conducted with all continuous variables to reduce the dimensionality potentially influencing LVH; the results indicated that LVH and non-LVH populations show significant separation in the principal component space. The first principal component explained 36.8% of the variance in whether LVH occurs, suggesting that this component was the most important in distinguishing between the two groups (Fig. 3A). In the first principal component, the variables making significant contributions were the TC/HDL-C ratio, Non-HDL-C/HDL-C ratio, TG/HDL-C ratio, LDL-C/HDL-C ratio, and Non-HDL-C, each contributing around 15%. The contribution of HDL-C was slightly lower but still close to 15%, indicating it also significantly contributed to the first principal component. The contributions of LDL-C were relatively small. These results suggested lipid profile were essential in distinguishing individuals with and without LVH (Fig. 3B).

Fig. 3
figure 3

PCA in LVH and Non LVH groups

Full size image

Relationship between the levels of lipids and LVH

Univariate analysis, correlation analysis, and PCA results showed a moderate correlation between lipid levels and LVMI. Participants were categorized into quartiles based on their lipid levels. In the unadjusted model, participants in the highest quartile, as compared to those in the lowest quartile, exhibited a higher risk of LVH in the TC, Non-HDL-C, TG, TC/HDL-C ratio, Non-HDL-C/HDL-C ratio, and TG/HDL-C ratio groups (TC, OR = 2.9646, 95%CI: 2.0868–4.2115; Non-HDL-C, OR = 2.833, 95%CI: 1.982–4.049; TG, OR: 1.908, 95%CI: 1.367–2.661; TC/HDL-C, OR = 2.2182, 95%CI: 1.5445–3.1856; Non-HDL-C/HDL-C, OR = 2.1952, 95%CI: 1.5532–3.1025; TG/HDL-C, OR = 1.9166, 95%CI: 1.3831–2.6558) (Fig. 4). Adjusting for confounding factors: Model I was adjusted for age and income (Fig. 5); Model II further adjusted for hypertension, alcohol history, and smoking history (Fig. 6); Model III additionally adjusted for waist circumference and eGFR (Fig. 7). The results of Model III indicated that TC (OR = 2.773, 95%CI: 1.805–4.26), Non-HDL-C (OR = 2.74, 95%CI: 1.7723–4.236), TC/HDL-C ratio (OR = 2.237, 95%CI: 1.445–3.463), Non-HDL-C/HDL-C ratio (OR = 2.357, 95%CI: 1.548–3.588), TG/HDL-C ratio (OR = 1.513, 95%CI: 1.02–2.245) adjusted for confounding factors, the risk of LVH in participants in the highest quartile is increased compared to those in the lowest quartile.

Fig. 4
figure 4

Relationship between the levels of lipids and the prevalence of LVH (Unadjusted model)

Full size image
Fig. 5
figure 5

Relationship between the levels of lipids and the prevalence of LVH (Model 1)

Full size image
Fig. 6
figure 6

Relationship between the levels of lipids and the prevalence of LVH (Model 2)

Full size image
Fig. 7
figure 7

Relationship between the levels of lipids and the prevalence of LVH (Model 3)

Full size image

Discrimination of lipids on LVH

The ROC curve analysis revealed the predictive ability of blood lipid profile for LVH (Fig. 8), with Non-HDL-C (AUC = 0.6109) exhibiting the highest area under the curve, followed by TC (AUC = 0.6084), and the weakest predictor being TG/HDL-C ratio (AUC = 0.5787). Youden’s index was utilized to identify the best threshold value of 4.9 for TC and 3.575 for Non-HDL-C in predicting LVH, with a specificity of 47.74% and a sensitivity of 74.4%, a specificity of 50.52% and a sensitivity of 70.86%, respectively (Table 3).

Fig. 8
figure 8

Receiver operating characters (ROC) Curve and area under eclipse

Full size image
Table 3 Areas under the receiver operating characteristic curves (ROC) for lipid profiles in LVH
Full size table

Sensibility analysis

After further adjustment for sex and HbA1c, the results remained consistent with those before adjustment (Supplement 1). Similarly, after adjusting for BMI, the results remained consistent (Supplement 2). Upon stratification analysis for hypertension, the results were consistent among hypertensive patients (Supplement 3).

Discussion

This study is the first to delve into the relationship between lipid components such as TC, Non-HDL-C, TG, TC/HDL-C ratio, Non-HDL-C/HDL-C ratio, TG/HDL-C ratio, and LVH in non-diabetic populations. Our result showed that Non-HDL-C, TC, TC/HDL-C ratio, Non-HDL-C/HDL-C ratio, and TG/HDL-C ratio can serve as independent risk factors for predicting the preservation of LVH in non-diabetic individuals. Additionally, Non-HDL-C and TC can be utilized as predictive markers due to their higher area beneath the curve. Furthermore, this correlation exhibited notable strength and consistency across correction groups for gender and HbA1c, BMI correction group, and stratified analysis of hypertension.

LVH, characterized by thickening of the myocardial wall of the left ventricle or dilation of the left ventricular chamber, poses a significant risk, with individuals displaying consistently high LVH showing the highest mortality risk [16]. Early detection and intervention and reducing the independent risk factors for LVH can serve as early preventive and intervention measures for the disease [18, 20]. Hypertension, diabetes, and renal insufficiency are major known risk factors for LVH [21]. A study in Mauritius found a higher prevalence of LVH among individuals with hypertension [22]. In pre-dialysis patients, the ratio of HDL-C to c-reactive protein and LVMI have been linked to baseline renal function and have shown an independent correlation with the progression of chronic kidney disease [23]. Additionally, African Americans with non-glomerular chronic kidney disease showed a higher prevalence of ambulatory hypertension and increased LVMI [24].

LVH is characterized by progressive impairment in β-oxidation, resulting in dysregulated fatty acid oxidation and significant metabolic remodeling of the heart [25]. Several studies suggest that alterations in lipid profile play a role in the pathophysiology of LVH. However, a consensus regarding the relationship between dyslipidemia and LVH is yet to be reached. Elevated HDL-C levels in mice models have been shown to impede LVH progression, evidenced by reduced left ventricular weight in mice fed a high-fat diet [26]. Change in leptin-signal transducer and activator of transcription three pathway signaling is related to the occurrence of LVH in Rats [27]. While drug interventions targeting lipid profile have shown promise in alleviating LVH by reducing TC, TG, and LDL-C while increasing HDL-C, the exact relationship between lipid regulation and LVH improvement remains uncertain [28]. Evidence suggests that individuals with dyslipidemia, defined by non-traditional lipid indices such as high atherosclerosis index and high TC/HDL-C, along with hypertension, are at higher risk of developing LVH [29]. A 7-year longitudinal and mediation study on increased lipids with risk of worsening cardiac damage in 1595 adolescents recognized that increased TC, Non-HDL-C, and TG were associated with worsening diastolic function and higher LV filling pressure [30]. In non-diabetic patients, the elevated TG in males with hypertension is independently associated with LVH [31]. This may be related to high TG causing myocardial steatosis, leading to left ventricular hypertrophy [32]. Increased TG/HDL-C ratio in obese children has been associated with eccentric LVH development [33]. The uncertainty of the effect of blood lipids on LVH may be related to insulin resistance and abnormal blood pressure caused by metabolic syndrome caused by dyslipidemia [27].

Furthermore, the most significant discovery in this study is that a comprehensive lipid profile has varying degrees of impact on the increase in LVH in the non-diabetic population. While lipids are affected by factors such as glucose metabolism, insulin resistance, and obesity, their influence on LVH remains notable [34]. Meanwhile, due to the variety of lipid types, they become an undeniable independent influencing factor in LVH. This study provides new evidence for the direct correlation between blood lipids and LVH and a new direction for managing and treating this disease.

Study strengths and limitations

This study is the first to investigate the risk determinant linked to LVH in a non-diabetic Sierra Leone population. PCA was used to determine the relative importance of blood lipids in the study, an innovation. Multiple corrections were performed to remove confounding factors and demonstrate the correlation between lipids and LVH. But it has some limitations. First, data focuses on individuals in Africa, which is constricted. It is better to enlarge samples and include other regions. Second, if participants are tracked and followed up, monitoring blood lipids will provide a more comprehensive understanding of the impact of blood lipids on LVH. Finally, conducting lipidomics analysis on the population to focus on the effect of the lipid profile on LVH would be even more convincing.

Conclusion

Lipid profile, represented by TC, Non-HDL-C, TC/HDL-C, Non-HDL-C/HDL-C ratio, and TG/HDL-C ratio, are independent risk factors of LVH in non-diabetic people. In addition, Non-HDL-C, with a cutoff value of 3.575, and TC, with a cutoff value of 4.9, were found to be an essential indicator for predicting the prevalence LVH disease with specific predictive values.