Causal Relationship Between Branched‐Chain Amino Acids and Hypertension: A Mendelian Randomization Study

Abstract

Background

This study aimed to investigate the causal relationships between branched‐chain amino acids (BCAAs) and the risks of hypertension via meta‐analysis and Mendelian randomization analysis.

Methods and Results

A meta‐analysis of 32 845 subjects was conducted to evaluate the relationships between BCAAs and hypertension. In Mendelian randomization analysis, independent single‐nucleotide polymorphisms associated with BCAAs at the genome‐wide significance level were selected as the instrumental variables. Meanwhile, the summary‐level data for essential hypertension and secondary hypertension end points were obtained from the FinnGen study. As suggested by the meta‐analysis results, elevated BCAA levels were associated with a higher risk of hypertension (isoleucine: summary odds ratio, 1.26 [95% CI, 1.08–1.47]; leucine: summary odds ratio, 1.28 [95% CI, 1.07–1.52]; valine: summary odds ratio, 1.32 [95% CI, 1.12–1.57]). Moreover, the inverse variance‐weighted method demonstrated that an elevated circulating isoleucine level might be the causal risk factor for essential hypertension but not secondary hypertension (essential hypertension: odds ratio, 1.22 [95% CI, 1.12–1.34]; secondary hypertension: odds ratio, 0.96 [95% CI, 0.54–1.68]).

Conclusions

The increased levels of 3 BCAAs positively correlated with an increased risk of hypertension. Particularly, elevated isoleucine level is a causal risk factor for essential hypertension. Increased levels of leucine and valine also tend to increase the risk of essential hypertension, but further verification is still warranted.

Nonstandard Abbreviations and Acronyms

BCAA

branched‐chain amino acid

EH

essential hypertension

IV

instrumental variable

IVW

inverse variance‐weighted

MR

Mendelian randomization

Clinical Perspective.

What Is New?

• The relationship between branched‐chain amino acids with hypertension in humans remains controversial.

• In this study, meta‐analysis was used in combination with Mendelian randomization methods, and consistent evidence favoring the causal associations of branched‐chain amino acids and the risk of hypertension was found. The present meta‐analysis provides conclusive data on a controversial issue, namely, whether branched‐chain amino acids can increase the risk of essential hypertension.

What Are the Clinical Implications?

• This study lays a theoretical foundation for clarifying the pathological process of essential hypertension and provides recommendations for dietary intervention of future hypertension treatment.

• This study sheds light on the pathogenesis of hypertension and related clinical practice, suggesting that an elevated isoleucine level could play an important role in evaluating the development of essential hypertension.

High blood pressure is one of the most significant risk factors for cardiovascular disease and premature death worldwide. It is also a significant contributor to the overall burden of disease globally. ¹ Approximately 95% of all the adult hypertension cases are identified as essential hypertension (EH), ² which affects over 100 million people worldwide and results in ≈20 000 deaths annually. ³ Although many antihypertensive medications are available, 71% of patients' blood pressure fails to reach below 130/80 mm Hg, ⁴ which highlights the limitations of antihypertensive medications. Experimental and observational clinical evidence suggests a prominent role of branched‐chain amino acids (BCAAs) in the development of hypertension. ⁵

BCAAs are composed of 3 amino acids, namely, leucine, isoleucine, and valine. These amino acids belong to a class of compounds with similar chemical structures; specifically, the α‐carbon contains branched side chains. ⁶ As indicated by evidence from in vitro and in vivo experiments, BCAAs affect the underlying processes of hypertension through regulating blood vessel constriction via oxidation and enhancing insulin release. ⁷ , ⁸ , ⁹ , ¹⁰ Although various observational studies have been conducted, the connection between BCAA intake and the risk of hypertension remains unclear since some studies yield contradictory results, ¹⁰ , ¹¹ and the exact cause‐and‐effect relationship remains to be further established. Therefore, it is necessary to conduct a meta‐analysis including all the existing epidemiological studies to evaluate the association between BCAAs and the risk of hypertension, so as to make objective preliminary conclusions. Furthermore, Mendelian randomization (MR) analysis can also be adopted to confirm the causal relationship between them. In MR analysis, the lead single‐nucleotide polymorphisms (SNPs) identified in genome‐wide association studies (GWASs) are used as genetic instruments, which can overcome the unmeasured confounders that limit observational studies and validate causality. ¹² The use of genetic variations as instrumental variables (IVs) helps to identify the underlying risk factors for hypertension, thus providing a theoretical basis for understanding the pathogenesis and prevention of hypertension. However, there is still a lack of research specifically pertaining to EH.

Consequently, this study aimed to determine the cause‐and‐effect relationship between BCAAs and the risk of hypertension development by MR analysis combined with meta‐analysis.

Methods

Meta‐Analysis

Electronic databases including PubMed, Embase, Web of Science, and SCOPUS were independently searched by 2 investigators to identify relevant publications up to April 2023. The search keywords included BCAAs, isoleucine, leucine, valine, and hypertension. Both EH and secondary hypertension have been considered as forms of hypertension. Only original English literature from population studies was included. A meta‐analysis was conducted on the relationship between BCAAs and hypertension in this work according to the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses guideline. ¹³ The search strategy, study selection, and data extraction are shown in Data S1 and Figure 1.

Figure 1. Assumptions of 2‐sample Mendelian randomization analysis.

 

Quality Assessment

The Newcastle–Ottawa Scale was used to assess the quality of cohort studies. ¹⁴ , ¹⁵ It consists of 3 domains: selection, comparability, and outcome. For cross‐sectional studies, quality was assessed using a scale of 11 items recommended by the Agency for Healthcare Research and Quality. ¹⁶ , ¹⁷

Data Analysis

Data extracted for meta‐analyses assessed the summary odds ratio and its 95% CI as the effect size using a random‐effects model. ¹⁸ In the meantime, the Cochran Q tests ¹⁹ and I ² index were used to evaluate the presence of heterogeneity. ²⁰ Moreover, detailed subgroup analyses stratified by factors such as study design, sources of exposure, and assessment methods were performed to evaluate the heterogeneity. Publication bias was evaluated using the Egger test. ²¹ In cases in which publication bias was found to exist (P<0.05), a trim‐and‐fill analysis was conducted. This analysis involved trimming the asymmetric outer part of the funnel plot and then replacing the pruned study and its corresponding missing studies around the center. This approach helped to adjust for publication bias and assess the stability of the pooled results. ²² In addition, the leave‐one‐out method was applied in sensitivity analysis.

Two‐Sample MR Study

In the present work, a 2‐sample MR analysis was performed to explore the causal relationships between circulating isoleucine, leucine, and valine levels and the risks of EH and secondary hypertension. All the data supporting this MR analysis are publicly available. They can be accessed through the provided GWAS articles and database links in Table S1.

Data Source

The genetic variants associated with plasma BCAA levels were obtained for exposure data from a meta‐analysis of GWASs of 16 596 European ancestors. ²³ For the reliability of the results, a parallel verification was conducted in this study. The GWAS statistics for BCAAs (valine, leucine, and isoleucine) were collected from the Medical Research Council Integrative Epidemiology Unit OpenGWAS project (https://gwas.mrcieu.ac.uk/). It involved a total of 115 048 participants for valine (GWAS ID: met‐d‐Val), 115 074 participants for leucine (GWAS ID: met‐d‐Leu), and 115 075 participants for isoleucine (GWAS ID: met‐d‐Ile) from the UK Biobank. The data on EH and secondary hypertension outcome were obtained from the FinnGen consortium R7. All participants were of European ancestry. More detailed information can be found in Table S1.

Study Participants

Six independent SNPs (P<5×10⁻⁸) related to circulating BCAAs levels were selected from the GWAS meta‐analysis of exposure data. ²³ The studies included in the GWAS were approved by the relevant institutional review committees, and participants provided informed consent. The selection of IVs is exhibited in Data S1.

Statistical Analysis

For exposures with IVs ≥3 (isoleucine level), the inverse variance‐weighted (IVW) and weighted median estimator (WME) models were used to evaluate their causal relationships with the risk of hypertension‐related outcomes. Of these 2 models, the WME is used to combine data from multiple genetic variants into a single causal estimate. ²⁴ IVW was the primary causal relationship estimation model in this study, which was able to estimate the causal relationship in an unbiased and most effective manner. ²⁵ In the meta‐analysis of GWASs of European ancestors, a P‐value <0.0125 (0.05/4 IVs) for the isoleucine analysis was considered as statistically significant after Bonferroni correction. ²⁶ In the UK Biobank database, which was used for validation, the Bonferroni corrected thresholds for significance were as follows: 0.008 (0.05/6 IVs) for isoleucine, 0.005 (0.05/10 IVs) for leucine, and 0.004 (0.05/14 IVs) for valine. P values lower than these specified thresholds were considered statistically significant, respectively, and any P values <0.05 and exceeding the thresholds following Bonferroni correction were seen as requiring further confirmation for potential causal associations. ²⁷ For exposure with IVs=1 (leucine and valine levels), the Wald ratio model was adopted to evaluate their causal relationships with the risk of outcomes. Furthermore, sensitivity analysis was performed for exposures with IVs ≥3. The Cochran Q test was carried out using the IVW model to assess the heterogeneity between IVs, and a random‐effects model was applied in 2‐sample MR analysis if significant heterogeneity was detected. The consideration of weak instrument bias holds practical significance when it comes to the design and analysis of MR studies. To investigate the instrumental strength of the selected SNPs and avoid weak instrument bias, the F statistic of each SNP was calculated. ²⁸ As commonly used, F value <10 indicated that the selected SNPs were weak instrumental variables, which may lead to biased results. ²⁹ Moreover, 2 methods were used to evaluate pleiotropy in this study. The MR‐Egger intercept test was used to detect the existence of directional pleiotropy in the SNPs. ³⁰ To determine if horizontal pleiotropy is present, the global test of MR Pleiotropy Residual Sum and Outlier was conducted. ³¹ Horizontal pleiotropy, if found to be present, was then addressed by eliminating outliers and examining whether there was a significant alteration in the causal effects before and after outlier removal. To mitigate the winner's curse bias, we employed the MRlap function to adjust the IVW results. This approach considers the influence of both weak instrument bias and the winner's curse, while also accounting for potential sample overlap. This method assists in reducing these biases in our analysis. ³² In the MRlap analysis, IVs were chosen on the basis of specific criteria. These criteria include a P value threshold of 5e‐8, which is used to determine the statistical significance of the IVs. A threshold of 0.1 for linkage imbalance is used to evaluate the equilibrium between the IVs and the genetic variants with which they are associated. In addition, a distance threshold of 500 kb is used to define the maximum physical separation between the IVs and the genetic variants.

All statistical analyses were performed using Stata 15.0 (StataCorp, College Station, TX) and R 4.0.5 with TwoSampleMR package (R Foundation for Statistical Computing, Vienna, Austria). Apart from the specific cases mentioned above, P values <0.05 were considered statistically significant.

Results

Meta‐Analysis of the Relationship Between BCAAs and Hypertension

A total of 972 articles were included by searching 4 databases. After removing the duplicates, 618 records were screened, among which 610 were further excluded due to being out of scope. The selection flowchart for the 8 included studies ⁸ , ⁹ , ¹¹ , ³³ , ³⁴ , ³⁵ , ³⁶ is illustrated in Figure S1, and characteristics of these included studies are summarized in Table 1. Among the 8 studies, there were 4 cohort studies with the follow‐up period of 3 to 8.6 years, and 4 cross‐sectional studies. The total sample size of the included studies was 32 845, including 9449 patients with hypertension and 21 095 without hypertension. Hypertension is characterized by having a systolic blood pressure ≥140 mm Hg or a diastolic blood pressure ≥90 mm Hg or the use of antihypertensive medications. The quality assessment of the enrolled studies is shown in Tables S2 and S3. As evidenced by the Newcastle–Ottawa Scale scale and the Agency for Healthcare Research and Quality scale, among the 8 included studies, 6 were of high quality and two were of moderate quality.

Table 1.

General Characteristics of Included Studies for Evaluating the Associations Between BCAAs and Hypertension

	Study design	Age, y	Sex		Exposures	Exposure evaluation methods	Exposure comparison groups	Outcome definition	Statistical method	Covariates	Quality
First author (publication year, country)				Sample size (cases/ noncases)
Flores‐Guerrero	Cohort	Range	Both	4169	Plasma isoleucine, leucine, valine	NMR	Per SD increment	SBP/DBP ≥140/90 mm Hg	Cox regression	Age, sex, BMI, T2D, smoking, alcohol consumption, family history of hypertension, eGFR, UAE, TC/HDL‐C ratio, triglycerides, and insulin	High
(2019, Netherlands) 8	PREVEND study, follow‐up 8.6 y	28–75	(54.4% female)	(924/3245)
Mirmiran	Cohort	Range	Both	4228	Dietary isoleucine, leucine, valine	FFQ	Quartile 4 vs quartile 1	SBP/DBP ≥140/90 mm Hg	Logistic regression	Age, sex, diabetes, BMI, physical activity, smoking, and intakes of energy, carbohydrates, fat, SFAs, PUFAs, fiber, calcium, magnesium, sodium, and potassium	High
(2019, Iran) 9	TLGS, follow‐up 3.1 y	20–70	(57.9% female)	(429/3799)
Mahbub	Cross‐sectional	Range	Both	5541	Plasma isoleucine, leucine, valine	LC–MS	Quartile 4 vs quartile 1	SBP/DBP ≥140/90 mm Hg	Logistic regression	Age, sex, BMI, SBP, DBP, FPG, UA, and medication for hypertension and dyslipidemia	High
(2020, Japan) 34	Population‐based	43–64	(51.9% female)	(2736/2805)
Liu‐1	Cohort	Mean±SD	Male	3995	Dietary isoleucine, leucine, valine	FFQ	Quartile 4 vs quartile 1	SBP/DBP ≥140/90 mm Hg	COX regression	Age, ethnicity, education, urban residents, T2D, physical activity, smoking, alcohol consumption, and intakes of energy, carbohydrate, fat, protein, salt	High
(2022, China) 11	CHNS, follow‐up 4.68 y	44.4±14.2		(1115/3392)
Liu‐2	Cohort	Mean±SD	Female	4496	Dietary isoleucine, leucine, valine	FFQ	Quartile 4 vs quartile 1	SBP/DBP ≥140/90 mm Hg	COX regression	Age, ethnicity, education, urban residents, T2D, physical activity, smoking, alcohol consumption, and intakes of energy, carbohydrate, fat, protein, salt	High
(2022, China) 11	CHNS, follow‐up 4.81 y	44.8±13.9		(1104/2880)
Kubacka	Cross‐sectional	Mean±SD	Female	349	Serum Total BCAAs	Colorimetric enzyme test	Per 10 unit increment	SBP/DBP ≥140/90 mm Hg	Logistic regression	Age and BMI	Moderate
(2021, Poland) 33	Population‐based	Normoglycemia, 48.4±6.0; dysglycemia 51.2±5.51		Not report
Jennings	Cross‐sectional	Range	Female	1952	Dietary Total BCAAs	FFQ	Quartile 5 vs quartile 1	SBP/DBP ≥140/90 mm Hg	Poisson regression	Age (y), current smoking (yes or no), physical activity (inactive, moderately active, active), BMI (kg/m2), use of hormone replacement therapy (yes or no), use of diabetes or cholesterol‐lowering drugs (yes or no), use of vitamin supplements (yes or no), menopausal status (pre‐ or postmenopausal), underreporting (yes or no) and intakes of energy (kcal), carbohydrate (g), saturated fat (g), wholegrains (g), alcohol (g), and protein (g)	High
(2016, UK) 36	Population‐based	18–76		Not report
Yamaguchi	Cross‐sectional	Mean±SD	Both	8115	Plasma isoleucine, leucine, valine	LC–MS	Per SD increment	SBP/DBP ≥140/90 mm Hg	Logistic regression	Age, sex, BMI, FPG, HbA1c, LDL‐C, HDL‐C, and triglycerides	Moderate
(2017, Japan) 35	Population‐based	Hypertension, 65.2±13.0; non‐hypertension, 51.0±16.4		(3141/4974)

BCAAs indicates branched chain amino acids; BMI, body mass index; CHNS, China Health and Nutrition Survey; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; FFQ, food frequency questionnaire; FPG, fasting plasma glucose; HbA_1c, glycated hemoglobin; HDL‐C, high density lipoprotein cholesterol; LDL‐C, low‐ density lipoprotein cholesterol; LC‐MS, liquid chromatography–mass spectrometry; NMR, nuclear magnetic resonance; PREVEND, Prevention of Renal and Vascular End‐Stage Disease; PUFAs, polyunsaturated fatty acids; SBP, systolic blood pressure; SFAs, saturated fatty acids; T2D, type 2 diabetes; TC, total cholesterol; TLGS, Tehran Lipid and Glucose Study; UA, urine acid; and UAE, urine albumin excretion.

The Association Between Isoleucine, Leucine, and Valine Level and Hypertension

In total, 8 population studies ⁸ , ⁹ , ¹¹ , ³³ , ³⁴ , ³⁵ , ³⁶ investigated the association between isoleucine, leucine, and valine level and hypertension. Overall, 3 BCAA levels were positively correlated with the risk of hypertension, but total BCAAs was not (Figure 2). Sensitivity analysis demonstrated the robust results, but there was a high heterogeneity. In addition, upon sensitivity analysis, the heterogeneity decreased from high to moderate level after excluding 2 studies. For isoleucine, no significant publication bias was detected. But for leucine and valine, there was publication bias. Afterwards, the publication bias was corrected by using the trim‐and‐fill method with the random‐effects model, 2 hypothetical unpublished studies were added, and the effect estimate remained significant. All the above results are summarized in Table S4.

Figure 2. Forest plot of the link between BCAAs and risk of hypertension.

The overall data on isoleucine level (A), leucine level (B), and valine level (C) were positively correlated with the risk of hypertension, but total BCAA level (D) was not. BCAA indicates branched chain amino acid; and OR, odds ratio.

In subgroup analyses, there were significantly positive relationships between 3 BCAAs levels and hypertension risk in all subgroups including cohort studies, with dietary BCAAs and BCAA level being set as the categorical variables for odds ratio assessment (all P<0.05; Table S5).

MR Analysis of the Relationship Between BCAAs and Hypertension

As for isoleucine, after removing SNPs with gene polymorphism, linkage imbalance, and palindrome sequences, the remaining 4 SNPs were rs7678928, rs75950518, rs58101275, and rs1420601. For both leucine and valine, the selected SNP was rs1440581. Table 2 provides a comprehensive overview of the specific SNPs associated with levels of BCAAs.

Table 2.

Characteristics of the SNPs Associated With BCAA Levels

Exposure	Lead SNP	Chromosome	Position	Nearby gene	β*	SE	EA	OA	EAF	P value	F statistic
Isoleucine	rs7678928	4	89 222 827	PPM1K	0.09	0.0125	T	C	0.46	5.50E‐19	51.84
	rs75950518	16	70 378 917	DDX19A	0.107	0.019	C	T	0.89	2.07E‐08	31.71
	rs58101275	14	104 008 420	TRMT61A	0.085	0.0153	G	A	0.79	2.78E‐08	30.86
	rs1420601	16	49 085 649	CBLN1	0.069	0.0125	C	T	0.4	3.71E‐08	30.47
	rs1260326†	2	27 730 940	GCKR	0.06	0.012	T	C	0.41	1.14E‐09	25
Leucine	rs1440581	4	89 226 422	PPM1K	0.081	0.0132	C	T	0.53	3.86E‐25	37.65
Valine	rs1440581	4	89 226 422	PPM1K	0.098	0.0133	C	T	0.53	4.40E‐24	54.29

BCAA indicates branched chain amino acid; EA, effect allele; EAF, effect allele frequency; OA, other allele; and SNP, single‐nucleotide polymorphism.

^*β coefficients represent the change in metabolite level per effect allele in standardized units.

^† The SNP in the GCKR gene was excluded from all analyses due to multiple pleiotropic associations with potential confounders.

Evaluation of Causal Relationships Between Circulating BCAA Levels and the Risk of EH

A genetic predisposition to an elevated circulating isoleucine level was significantly associated with a higher risk of EH in the IVW model. Similarly, the same result was obtained in the WME model (IVW: odds ratio, 1.22 [95% CI, 1.12–1.34]; WME: odds ratio, 1.23 [95% CI, 1.09–1.38]; all P<0.0125; Figure 3A). In the analysis of EH, The Cochran Q test found no significant heterogeneity. The MR‐Egger intercept test did not find any pleiotropic effects, and the MR Pleiotropy Residual Sum and Outlier global test results showed that no pleiotropic SNPs were used in the MR analysis (all P>0.05). Furthermore, SNPs that greatly affected the causal estimates were not observed in the leave‐one‐out sensitivity analysis (Figure S2A; Table S6).

Figure 3. Causal estimates between BCAA levels and risk of hypertension.

A, Causal relationship between BCAA levels and risk of EH. B, Causal relationship between BCAA levels and risk of secondary hypertension. BCAA indicates branched chain amino acid; IVW, inverse variance‐weighted; OR, odds ratio; SNP, single‐nucleotide polymorphism; and WME, weighted median estimator.

Based on the Wald ratio model, when the levels of BCAAs (leucine and valine) increase by 1 unit, the risk of EH increases by 31% for leucine and 25% for valine (all P<0.01; Figure 3A). In the validation data set from the UK Biobank, evidence based on 6 SNPs was noted between isoleucine levels and increased risk of EH, with a P value of 0.027. Similarly, evidence was observed between higher levels of circulating leucine and a higher risk of EH, with a P value greater than the threshold draw from Bonferroni correction (P _leucine=0.010). Additionally, there was evidence of an association between elevated valine levels and an increased risk of EH, based on 14 SNPs, with a P value of 0.006. The results of the 3 associations above exhibit heterogeneity and did not demonstrate pleiotropy. After conducting MRlap analysis, the corrected P values for the effects of isoleucine, leucine, and valine on EH were found to be 0.0005, 0.0035, and 0.0004, respectively. These values are all lower than the thresholds determined by the Bonferroni correction (Table S7).

Evaluation of the Causal Relationships Between Circulating BCAAs Levels and the Risk of Secondary Hypertension

The findings obtained from the IVW model indicate that there is an uncertain causality between the genetic predisposition to circulating isoleucine levels and the risk of developing secondary hypertension (Data S2; Figure 3B; Figure S2B; Tables S6 and S7).

Discussion

In this study, a meta‐analysis of all the existing high‐quality studies was conducted, which showed the significant associations between elevated levels of isoleucine, leucine, and valine and the increased risk of hypertension. Further, the cause‐and‐effect relationships between elevated levels of isoleucine, leucine, and valine and the higher risks of hypertension and essential hypertension were demonstrated by the 2‐sample MR analysis. However, our results demonstrated an uncertain causality with the risk of secondary hypertension.

According to results in the preliminary meta‐analysis, the increased levels of isoleucine, leucine, and valine were significantly associated with a higher risk of hypertension, and the sensitivity analysis also revealed the stable results. According to the heterogeneity analysis, the Flores‐Guerrero study ⁸ and the Yamaguchi study ³⁵ contributed to the heterogeneity observed in the results. The Flores‐Guerrero study used BCAA levels as a continuous variable for assessing odds ratio and also adjusted for a family history of hypertension, which is a confounding factor that other studies did not take into account. On the other hand, the Yamaguchi study was considered to be of moderate quality and had more lenient inclusion and exclusion criteria compared with other studies. Additionally, while the above‐mentioned articles included blood glucose and lipid indicators as covariates, other studies did not include these variables, which could also contribute to the observed heterogeneity. After excluding these 2 studies, the association remained significant and the heterogeneity across studies reduced to <50%. In the corresponding subgroup analyses of categorical variables, the summary odds ratio values of the relationships between isoleucine, leucine, and valine levels and the risk of hypertension were 1.55 (95% CI, 1.25–1.93), 1.55 (95% CI, 1.32–1.82), and 1.63 (95% CI, 1.39–1.91), respectively, and the heterogeneity was 0. We also carried out publication bias tests on the included studies for each of the 3 BCAAs. For isoleucine, no significant publication bias was detected in its relationship with an increased risk of hypertension. However, a significant publication bias was found in the associations between leucine and valine levels and the risk of hypertension in the preliminary study results. Additionally, the trim‐and‐fill method was adopted to correct publication bias, unpublished studies with 2 additional hypotheses were added, and the effect estimate indicated that this association remained significant.

Moreover, hypertension was further classified into essential and secondary categories in the MR analysis, and the effects of BCAAs on each other were analyzed. First, SNPs with gene pleiotropy were eliminated on the basis of the PhenoScanner database. For EH, our MR analysis results showed that a higher isoleucine level led to an increased risk of EH. Based on the results of the pleiotropy analysis, the P value of the MR‐Egger intercept test is >0.05. This suggests that there is no directional pleiotropy of the IVs, and the observed association between the genetic variations and the traits being studied is not influenced by directional pleiotropy. After conducting further MR Pleiotropy Residual Sum and Outlier analysis, it was determined that there were no significant outliers found (with a global test P value >0.05). As a result, it was concluded that there was not enough evidence to support the presence of horizontal pleiotropy in the relationship between these BCAAs and EH. After correction by Bonferroni, the higher isoleucine level was found to induce an increased risk of EH in the IVW model, with no heterogeneity between IVs, and similar results were obtained in the WME model. Further, the results of leave‐one‐out sensitivity analysis suggested the robust causal association. The Wald ratio model identified the significant causal relationships between the elevated leucine and valine levels and an increased risk of EH. These findings support the results of our meta‐analysis and further confirm the significant causal relationships between isoleucine, leucine, and valine levels and the higher risk of EH. To confirm the presence of the cause‐and‐effect relationships, we also analyzed the UK Biobank database as a means of independent validation. The findings suggest that the P values for the associations between the 3 types of BCAAs and EH were <0.05 and exceed the thresholds after Bonferroni correction. ²⁶ , ²⁷ , ³⁷ , ³⁸ , ³⁹

The winner's curse bias occurs in MR when there is overlap between the data set used to select genetic variants and the data set used to estimate genetic associations. To examine the potential bias of the winner's curse, an MRlap analysis was conducted in the validation data set as part of this study. In this study, we found that the effect size in the validation database was statistically significant after applying MRlap correction. Additionally, it was observed that the IVW method underestimated the true causal effects. It is worth noting that the MRlap corrected effects significantly differed from the observed effect in IVW. As a result, priority consideration should be given to the corrected effects of MRlap on the basis of prior theories. ³² Therefore, there is no winner's curse bias due to sample overlap. The MRlap function takes into account not only the winner's curse but also addresses the issue of weak instrument bias. ³² In the MR analysis, the F statistic for each SNP included was >30, indicating the absence of bias caused by weak instrumental variables. With regard to the underlying mechanism, high levels of BCAAs can continuously activate the mammalian target of rapamycin complex 1, which then phosphorylates insulin receptor substrate 1, leading to insulin resistance. ⁴⁰ Insulin resistance plays a key role in the progression of hypertension, mainly through increasing the sodium reabsorption in renal tubular cells, activating the sympathetic nervous system and increasing the calcium concentration in vascular smooth muscle cells, ultimately enhancing vascular resistance and affecting blood pressure. ⁴¹ , ⁴² Research on the respective mechanisms of isoleucine and valine in hypertension is still lacking, but the limited research results indicate that dietary isoleucine intake is correlated with the increased body mass index, ⁴³ which is known to increase the risk of hypertension. ⁴⁴

Regarding secondary hypertension, the primary MR analysis did not identify any significant causal relationship between circulating levels of isoleucine, leucine, and valine and the risk of developing secondary hypertension. To validate these results, the UK Biobank database was used. The initial analysis results indicate that further confirmation is required for the potential causal relationships between isoleucine, leucine, and secondary hypertension. There is no causal relationship found between valine and secondary hypertension. To investigate the presence of winner's curse bias in this study, an additional MRlap analysis was performed. The MRlap analysis showed that the adjusted effect was not statistically significant. Furthermore, the MRlap‐corrected effects were shown to be statistically different from the observed effect in the IVW analysis. Therefore, the conclusive result should be based on the MRlap‐corrected effects. Consequently, there is no evidence to support the notion that the 3 BCAAs are directly responsible for causing secondary hypertension. According to the existing research results, there may not be a direct association between BCAAs and secondary hypertension. Neither animal nor in vitro studies have identified a direct association with BCAAs. In addition, the common causes of secondary hypertension include obstructive sleep apnea, renal parenchymal disease, renal artery stenosis, and primary hyperaldosteronism. ⁴⁵ However, so far, no research has been conducted regarding the alterations in levels of BCAAs in patients with hypertension associated with the aforementioned diseases. Therefore, it is necessary to exercise caution when using BCAA supplements to treat secondary hypertension in humans.

This study has several advantages. First, this study is the first to use MR analysis combined with meta‐analysis to assess the associations between BCAA levels and the risk of hypertension. The included studies were of high quality, and sensitivity analysis showed the result stability and statistical significance after adjusting for heterogeneity and publication bias. Notably, both expression quantitative trait localization mapping and MR are widely used genetic analysis methods. However, expression quantitative trait localization mapping is known to have certain limitations. It tends to have a detection bias, as it only identifies formal relationships without establishing causal relationships. Moreover, the results of expression quantitative trait localization mapping can be influenced by multiple genotypes impacting a single expression level, as well as other factors like age, sex, and environment. ⁴⁶ MR analysis can effectively control confounding factors and minimize the reverse causality to confirm the causal relationships. ⁴⁷ Second, the GWAS data used for MR analysis in this study were corrected for factors such as age, sex, and genotype, thereby avoiding confounding factors like age, sex, and race. ⁴⁸ Third, to enhance the conclusiveness of the findings, we conducted MRlap analysis to evaluate the winner's curse bias. Nevertheless, certain limitations should also be noted in this study. First, the MR database is derived from the European populations, which needs to be verified in other ethnic groups by epidemiological studies in the future. Second, this study performed a meta‐analysis to examine the literature on blood sources and dietary sources of BCAAs. This approach is supported by previous research that has highlighted the significance of both plasma and dietary sources of BCAAs in relation to the risk of hypertension. ⁹ , ³⁴ BCAAs are essential amino acids that cannot be synthesized by human being themselves and must be ingested from the outside world; therefore, increased levels of plasma concentration primarily occur due to consuming high amounts of food on a daily basis, ⁴⁹ and previous epidemiological research has affirmed this. ⁵⁰ , ⁵¹ In the past, various meta‐analyses have been carried out on BCAAs and phytochemicals following the mentioned principles. These analyses involved combining BCAAs obtained from both blood samples and dietary sources to create a pooled analysis. Similarly, other meta‐analyses have been conducted on phytochemicals using similar methods. These studies also involved analyzing a mixture of phytochemicals derived from both blood and diet. ⁵² , ⁵³ To demonstrate the reliability of our results, this study further conducted subgroup analyses stratified by the study type, BCAA data source, and available data types for analysis. The study's findings suggested that there were meaningful and positive associations between the levels of 3 specific amino acids known as BCAAs and the risk of hypertension within the subgroup of individuals who consumed BCAAs through their diet. However, the available studies analyzing the relationship between plasma BCAAs levels and hypertension are limited in number. Despite the inability to perform a subgroup analysis, the results of both studies consistently demonstrate findings aligned with dietary sources of BCAA. Third, in the MR analysis of leucine and valine levels, only 1 SNP associated with leucine and valine was extracted from the existing research database so far, which did not rule out the bias of results due to the small number of SNPs. However, the MR analysis results should be combined with the current meta‐analysis results and mechanism research results for comprehensive judgment. As shown in our meta‐analysis results, both leucine and valine levels were significantly correlated with the risk of hypertension. In terms of the underlying mechanism, previous studies have found that valine plays a role in establishing an inflammatory state, ⁵⁴ which stimulates the production of reactive oxygen species via 2 subunits of the reduced form of nicotinamide‐adenine dinucleotide phosphate oxidase, and increases reactive oxygen species production in mitochondria. ⁵⁴ Moreover, valine can activate nuclear factors in monocytes by targeting the mammalian target of rapamycin complex 1 κβ pathways, thereby producing some inflammatory mediators (like interleukin‐6 and tumor necrosis factor‐α). ⁵⁵ It has been proven that inflammatory states may affect blood pressure by means of endothelial dysfunction. ⁵⁶ Leucine can induce insulin release, ⁵⁷ which then affects blood pressure through insulin resistance. On the other hand, valine is an important nutritional signal that directly or indirectly affects metabolism. Based on the above research findings, this study suggests that the increased levels of leucine and valine may predict the increased risks of EH. In any case, larger genetic studies and related SNP tests are warranted in the future to further confirm the above causal relationships.

Conclusions

Increased levels of 3 BCAAs are significantly associated with the increased risk of hypertension. In particular, elevated isoleucine level serves as a causal risk factor for EH. In addition, increased levels of leucine and valine also tend to increase the risk of hypertension, but further verification is still needed.

Sources of Funding

This study was supported by grants from the scientific research project of Shanghai Municipal Health Commission (201940188) and the Three‐Year Action Program of Shanghai Municipality for Strengthening the Construction of Public Health System (GWV‐10.2‐YQ3).

Disclosures

None.

Supporting information

Data S1–S2

Tables S1–S7

Figures S1–S2

Reference ⁵⁸