Introduction

Hypertension is typically associated with diabetes, obesity, and metabolic syndrome, serving as a primary risk factor for cardiovascular diseases1. It is estimated that there are approximately 1.39 billion adults globally are afflicted with hypertension, with three-quarters residing in developing nations and one-quarter in developed countries2. The global burden of hypertension is increasing and it is estimated that by 2025, one-third of the global population will be affected due to accelerated population aging and increasing rates of obesity3. The prevalence of hypertension among Chinese populations was 25.2% in 2012 and increased to 27.5% in 2018. Overall prevalence of hypertension among Chinese populations is on the rise5and this trend will continue for next decades4. In contrast to the high prevalence rate, the hypertension control rate for individuals aged 60 and above in China was low, e.g., 16.1% in 2012 and 18.2% in 2015, indicating that there is still a significant gap between the control rate of hypertension in Chinese elderly people and the requirement for healthy aging6.

The prevalence of hypertension among senior Chinese (aged 60 and above) is 59.2%, and 56.8% of whom have been discharged from hospitals for hypertension5, indicating that the elderly population is more severely affected. In 2017, high systolic blood pressure (SBP) led to the deaths of 2.54 million people in China, with 95.7% of them dying from cardiovascular diseases, highlighting the substantial burden posed by hypertension among the Chinese population7. Treating hypertensive patients can reduce 803,000 cardiovascular events and gain 1.2 million quality-adjusted life years (QALY) annually compared to maintaining the current situation7. Therefore, hypertension is a crucial public health issue currently faced in China.

Short-chain fatty acids (SCFAs) are known to activate various signaling pathways within the host, primarily produced by the metabolism of gut microbiota, with a small amount coming from dietary fats and liver metabolism, directly or indirectly contributing to human health8. Growing experimental evidence from rodents indicates that the gut microbiota is associated with blood pressure and SCFAs can regulate blood pressure9,10,11,12. There are some population studies showing that the level of SCFAs in the feces of hypertensive patients is higher than those of normotensive subjects13,14,15,16, and that the level of SCFAs in the serum of patients with hypertension is lower than healthy individuals13. Besides, increasing the level of SCFAs has the effect of lowering blood pressure17,18. These studies mainly focus on the relationship between SCFAs and blood pressure in the elderly. However, as far as we know, only one study from the Guangxi region has been reported in China. Due to the vast area of China, in order to eliminate the influence brought by geographical differences, we therefore conducted the study again in the Lu’an region in order to obtain comprehensive results and reveal the potential association between SCFAs and blood pressure. Though substantial amounts of studies have been conducted, our comprehension of the significance of SCFAs in human health remains restricted. This report is a cross-sectional study, we used a sample of Chinese older adults to investigate the potential association between serum SCFAs and blood pressure, separately and as the mixture. Our study is not conclusive but provides some new perspectives to understanding the role of SCFAs in human health which may lead to new means for the prevention and treatment of hypertension.

Materials and methods

Study population

We conducted a baseline study from June to September 2016 in Lu’an, China, in a cohort program called Health and Environmentally Controllable Factors in Older Adults. As the largest city in Anhui Province, Lu’an has an area of 15,451 square kilometers, with 5.8361 million people. Elderly individuals in the group were recruited by the local Center for Disease Control and Prevention together with us. Our studies have identified several risk factors related to the health of elderly people, such as hypertension drugs19, antibiotics20,21, and phthalates22.

Our previous article describes the details of the study design in detail20. Briefly, we first randomly selected two counties from Lu’an and then randomly selected one community in each county, all participants’ age ≥ 60 years, resided in the local area for at least six months, provided an informed consent, while those with mental illness or limited mobility were excluded. We made phone appointments with 1,217 older adults, 1,080 of whom completed face-to-face interviews. Participants without demographic information (N = 67) were excluded. Finally, 1,013 individuals were included in our study (Fig. 1). The process of conducting the questionnaire survey was thoroughly explained in our previous article23. The study was conducted in accordance with the Declaration of Helsinki, and approved by the Ethics Committee of Anhui Medical University (No. 20170284).

Fig. 1
figure 1

Flowchart of elders analyzed.

Full size image

Blood sample collection and physical measurement

Before collecting blood, all included participants were asked to provide their blood samples in the morning after at least an 8-h overnight fast. Upon arriving at the community health service center in the morning, a 10 ml venous blood sample is placed by a medical professional into serum separation gel tube, and gently mixed by inverting 5–6 times, allowed to stand for 30 min. The collection samples were centrifuged at 4 °C and 1000 g for 5 min. Afterwards, the samples were transported in dry ice to the local Center for Disease Control and Prevention for temporary storage at -20 °C. After all the samples were collected, they were transported to the Experimental Centre of Anhui Medical University by cold-chain truck and stored in a refrigerator at -80 °C until further analysis. When measuring height and weight, participants were required to remove their shoes, hats, and outerwear, wear lightweight clothing, had both heels together, and used a mechanical stadiometer for measuring the height of each participant. Furthermore, a lever scale was utilized to measure the weight of each participant.

A tabletop mercury column sphygmomanometer was used to measure morning blood pressure (BP). Participants were instructed to maintain in a resting condition for more than 5 min before BP was measured. We measured BP in the right arm, and BP values (mmHg) were taken as the average of three times. The following conditions are considered as hypertension: (1) either systolic blood pressure (SBP) ≥ 140 mm Hg or diastolic blood pressure (DBP) ≥ 90 mm Hg; (2) a self-reported history of hypertension; or (3) currently taking antihypertensive medications24.

Detection of short-chain fatty acids

There is still lack of a precise method available for detecting serum SCFAs in human , we developed an Ultra High Performance Liquid Chromatography-Quadrupole-Exactive-Orbitrap-Mass Spectrometry (UHPLC-QE-Orbitrap MS) method based on 3-nitrophenylhydrazine derivatization in negative electrospray ionization through parallel reaction monitoring mode for the simultaneous detection of SCFAs in the serum, and the analysis was performed on an Agilent Infinity Lab Proshell 120 EC-C18 column (2.1 mm × 100 mm, 2.7 μm, Agilent, Santa Clara, CA,USA). It demonstrated stability, sensitivity, and accuracy. The specific operation and instruments required for this method to measure SCFAs had been described in detail in our previous study25.

Statistical analysis

When conducting statistical analysis, values below the limit of detection (LOD) of SCFAs are treated as half of the LOD26. We used chi-square tests and t-tests to analyze the differences between the normotensive and hypertensive populations in terms of demographic variables such as gender and age, and analysis of the association between seven SCFAs and hypertension in the elderly using binary logistic regression models. When linear regression models were used to assess the association between SCFAs and blood pressure values, we excluded 278 older adults taking antihypertensive medication because it affects blood pressure values, which may introduce confounding variables that obscure the association between SCFAs and blood pressure values. In addition, these drugs were diverse and might have different effects. We calculated the Spearman correlation coefficients among SCFAs to assess potential confounding factors or collinearity in the regression analysis. To determine whether there is a non-linear and non-additive association between SCFAs and blood pressure, we used a Bayesian kernel machine regression (BKMR) approach, which estimates the multivariate exposure–response function utilizing a kernel machine regression model, allowing specific exposures to be set at particular quantile values and examining the association between residual exposure and results, it also allows a visual representation of the estimated exposure–response function27. We used 100 00 iterations to fit the BKMR models and presented the exposure response function in 4 different visualizations by the BKMR approach: (1) overall effect of SCFAs was obtained through comparing specific percentiles (25th, 50th and 75th percentile) of all the SCFAs with their 50th percentile; (2) univariate exposure–response curves showed the association of one SCFA with the result when all other SCFAs were fixed at their medians of exposure levels; (3) single-exposure effects (95% CI), defined as the change in the odds of hypertension associated with single SCFA as concentrations of other SCFAs set at particular percentiles (i.e., 25th, 50th, and 75th); and (4) bivariate exposure response functions were used to indicate potential interaction effects in SCFAs mixture. In the BKMR models, posterior inclusion probabilities (PIPs) represented the weight of each SCFAs in the health effect.

In addition, we used restricted cubic splines (RCS) to further investigate the potential non-linear relationship between SCFAs and blood pressure. Based on the concentration distribution of SCFAs, we applied 3 knots at specific locations.

In the sensitivity analysis, we included older adults taking antihypertensive drugs to verify the robustness of our main analysis.

To reduce the interference of confounding factors, we included variables such as age (60 ≤ age < 70 years old; age ≥ 70 years old), gender(male; female), education(illiteracy; primary school; middle school; high school and above), household income (< 30,000yuan/year; ≥ 30,000yuan/year; decline to answer), body mass index (BMI = weight (kg)/height^ 2 (m)), smoking status(no and yes), alcohol drinking status(no and yes), diabetes(no and yes) and taking antihypertensive medication(no and yes) as covariates in the models for the analysis. Of these covariates, BMI was divided into four groups (BMI < 18.5 kg/m2;18.5 ≤ BMI < 24 kg/m2; 24 ≤ BMI < 28 kg/m2, BMI ≥ 28 kg/m2). Smoking was defined as “yes” if the participant had smoked more than one cigarette per day and for above six months. Drinking was described as “yes” if the participant had drunk once or more per week and for over six months. Diabetes was determined as a history of self-declared diabetes, insulin or glucose-lowering drug use, or fasting glucose ≥ 7.0 mmol/L. Taking antihypertensive medication was defined as “yes” if the participant took antihypertensive medication at least once a day in last month.

We used the bkmr package and rms package in R software (version 4.3.0) to build the BKMR models and RCS models. We performed all remaining analyses using Statis-tical Package for the Social Sciences (SPSS, version 27.0), and statistical significance was defined as having a p-value less than 0.05.

Results

Characteristics of the study population

The demographic profile of the study participants was presented in Table 1. The study subjects comprised 1,013 elderly individuals with an average age of 71.31 years. The mean SBP was 139.53 ± 20.83 mm Hg and the mean DBP was 80.50 ± 10.70 mm Hg. Among all participants, 55.68% were female, 46.10% were illiterate, 59.52% household income lower than RMB 30,000, and 43.63% BMI in the normal range of values. Most of those subjects were also non-smokers or non-drinkers. Besides, 30.70% had no hypertension and 69.30% had hypertension. Among the hypertensive population, 39.60% were taking antihypertensive medication. Table 1 also showed a comparison of the levels of 7 SCFAs between the normal blood pressure population and the one with hypertension. There were no significant differences in age, gender, education, annual household income, smoking status, drinking status and level of SCFAs. The detection probability (DP) and LODs of the 7 SCFAs were shown in Table 2. The LODs for acetic acid (AA), propanoic acid (PA), butyric acid (BA), isobutyric acid(iso-BA), valeric acid.

Table 1 Comparison of basic characteristics between normotensive and hypertensive participants (n = 1013).
Full size table
Table 2 Detection probability and serum concentration of SCFAs in the study population.
Full size table

(VA), isovaleric acid (iso-VA), and caproic acid (CA) were 2,000 pg/ml, 20 pg/ml, 0.5 pg/ml, 0.2 pg/ml, 10 pg/ml, 1 pg/ml, and 50 pg/ml, respectively, and except the DP of VA was 99.61%, the remaining SCFAs were 100%. The Spearman’s rank correlation coefficients between SCFAs were displayed in Figure S1, and SCFAs were significantly correlated with each other (all P-values were less than 0.05), with Spearman’s rank correlation (rs) values ranging from 0.12 to 0.67.

Associations between single SCFA and blood pressure

The outcome of between single SCFA and hypertension analyses were shown in Table 3. Significant correlations are not found in the unadjusted model nor in the adjusted single-SCFA model.

Table 3 The associations between SCFAs and hypertension using logistic regression.
Full size table

Following adjustment for covariates and remaining SCFAs, the significant negative association of AA (OR = 0.696, 95%CI: 0.501–0.966) and VA (OR = 0.713, 95%CI: 0.516–0.985) with hypertension are observed (Table 3).

The association between SCFAs and blood pressure were shown in Table 4. After adjusting for age, gender, education, household income, BMI, smoking status, alcohol drinking status, diabetes and other SCFAs, we observe a significant negative association of AA (β = -3.89, 95% CI: -7.12 – -0.66) with the SBP levels, and a significant negative association between iso-VA (β = -2.11, 95% CI: -3.94 – -0.29) and DBP levels. Whether in unadjusted or adjusted linear regression models, we all observe significant positive associations between CA and BP levels.

Table 4 The associations between SCFAs and blood pressure using linear regression.
Full size table

Associations between the SCFAs mixture and blood pressure

Figure 2A demonstrated the joint effect of the mixture of SCFAs on hypertension, BKMR plots reveal an inverse trend between the mixture of SCFAs and hypertension, even if not statistically significant. The relationships between single SCFA and hypertension were illustrated in Fig. 2B, with the remaining SCFAs set at their median levels, in line with findings from logistic regression models, AA, VA, and iso-VA exhibit an inverse trend with the hypertension. Figure 2C exhibited the change in the odds of hypertension associated with single SCFA as concentrations of other SCFAs set at particular percentiles (i.e., 25th, 50th, and 75th). Eventually, bivariate exposure response functions were used to indicate potential interaction effects in SCFAs mixture (Fig. 2D). The bivariate interaction plots visualized by BKMR do not show a strong interaction between SCFAs and hypertension (Fig. 2C and 2D). The PIPs for conditional inclusion into the BKMR model between SCFAs and hypertension were summarized in Table S1, and CA has the highest PIP value (PIP = 0.275).

Figure S2A showed the analysis of the joint effect of mixture of SCFAs on SBP, with BKMR plots. The result shows that there is an inverse trend between mixture of SCFAs and SBP levels, although not significant. The relationships between single SCFA and SBP were illustrated in Figure S2B, with the remaining SCFAs set at their median levels, which exhibited the single dose–response association and the 95% CI. Similar to the results of the linear regression models, we observe a significant negative correlation between AA and SBP, while CA is significantly positively correlated with SBP. Figure S2C indicated the change in the effect of single SCFA on SBP levels when the remaining SCFAs were kept at specific percentiles (i.e., 25th, 50th, and 75th). Finally, bivariate exposure–response curves for SCFAs were evaluated (Figure S2D). The bivariate interaction plots visualized by BKMR do not show a strong interaction between SCFAs and SBP (Figure S2C and S2D). Table S2 summarized the PIP values between SCFAs and SBP in BKMR model in which AA has the highest PIP value (PIP = 0.156). The joint effect analysis of SCFAs mixture on the DBP was displayed in Figure S3A. The BKMR plots indicate that there is no association between mixture of SCFAs and DBP levels. Figure S3B showed the exposure–response association and 95% CI for single SCFA. Similar to the results of the linear regression models, we observe a significant negative association between iso-VA and DBP levels, and CA is significantly positively correlated with DBP. Figure S3C indicated the variation in the effect of single SCFAs on DBP levels when the remaining SCFAs were kept at specific percentiles (i.e., 25th, 50th, and 75th). Finally, bivariate exposure–response curves for SCFAs were evaluated (Figure S3D). The bivariate interaction plots visualized by BKMR also do not show a strong interaction of SCFAs on diastolic blood pressure (Figure S3C and S3D). The PIPs for conditional inclusion into the BKMR model between SCFAs and DBP were summarized in Table S3, and PA has the largest PIP value (PIP = 0.399).

The results of the RCS were shown in the supplemental material Figure S4-6. SCFAs are not nonlinearly associated with the prevalence of hypertension, SBP and DBP levels (all P for nonlinear > 0.05).

Sensitivity analyses

After excluding 278 older adults who were taking antihypertensive medication, we further analyzed logistic regression and BKMR models with the presence of hypertension as the dependent variable. Similar results to the primary outcome were observed (Table S4 and Figure S7), but the joint effect of SCFAs (Figure S7A) no longer shows an inverse trend with hypertension, which might be due to the insufficient number of cases, making it difficult for us to detect potential association. Whereas, in the linear regression and BKMR models analyses with systolic and diastolic blood pressure as dependent variables, we included older adults who were taking antihypertensive medication, and the results shows that our findings are robust (Table S5 and Figures S8-9).

Fig. 2
figure 2

Associations between 7 SCFAs and the hypertension among older adults were estimated using Bayesian kernel machine regression (BKMR). Adjustment variables for BKMR are consistent with logistic and linear regression models, including age, sex, education, household income, BMI, smoking status, alcohol drinking status and diabetes. (A) overall effect of SCFAs was obtained through comparing specific percentiles (25th, 50th and 75th percentile) of all the SCFAs with their 50th percentile; (B) univariate exposure–response curves showed the association of one SCFA with the hypertension when all other SCFAs were fixed at their medians of exposure levels; (C) single-exposure effects (95% CI), defined as the change in the odds of hypertension associated with single SCFA as concentrations of other SCFAs set at particular percentiles (i.e., 25th, 50th, and 75th); and (D) bivariate exposure response functions were used to indicate potential interaction effects in SCFAs mixture. Abbreviation: AA acetic acid; PA propanoic acid; BA butyric acid; iso-BA isobutyric acid; VA valeric acid; iso-VA isovaleric acid; CA caproic acid.

Full size image

Discussion

In our study, we used logistic regression and BKMR approaches for comprehensively evaluating the single and joint associations between serum SCFAs and hypertension, and comprehensively assessed single and joint associations between SCFAs and blood pressure values using linear regression and BKMR models. In our study, individual associations of AA and VA with the odds of hypertension in logistic regression models are significantly negative, while their independent associations in BKMR are also negative but not significant. Overall, we can find that SCFAs mixture are negatively linked with the odds of hypertension, although no significance is observed, this may be because single SCFA analyses underestimate the protective effect of SCFAs on blood pressure levels in the elderly and mixture analyses are warranted.

In the linear regression and BKMR models, we observe a significant negative association of AA with the SBP levels, a significant negative association between iso-VA and DBP levels, and CA is significantly positively correlated with SBP and DBP. In addition, the BKMR plots show that there is no association of SCFAs mixture with SBP and DBP levels, but we can observe that the SCFAs mixture has an opposite trend to the SBP (albeit not statistically significant). To our knowledge, this study represents the initial investigation into the correlation between serum SCFAs and BP among elderly Chinese individuals.

Analyses of single SCFA

In our study, we observe that there is a negative correlation between AA, VA and hypertension. Specifically, we observe a significant negative correlation between AA and SBP, and a significant negative association between iso-VA and DBP levels, while CA is significantly positively correlated with SBP and DBP. These results are biologically reasonable. In line with our data, previous studies have shown that supplementation with oral acetate28,29,30and valerate31reduces blood pressure in animal experiments. SCFAs can regulate blood pressure through the G protein-coupled receptors (GPCRs) which play a significant role in the development of hypertension32,33,34. SCFAs were found to interact with at least four GPCRs to affect host blood pressure35, namely GPCR41, 43, 109A, and olfactory G protein-coupled receptor 78 (olfr78). Acetate is one of the most potent agonists of GPCR41 and GPR4336, and AA promotes the release of intracellular calcium ions through the activation of GPCR41 and GPCR43, thereby lowering blood pressure37. The PA can partially regulate the renin levels through olfr78, thus increasing blood pressure and antagonizing the hypotensive effect mediated by GPCR4138. CA were shown to enhance differentiation and proliferation of T helper 1 (Th1) and Th17 cells and to suppress that of Treg lymphocytes, thus supporting inflammation39,40, this could explain why we observe a significant positive correlation between CA and SBP, DBP. In addition, animal studies have shown that SCFAs promote the level of interleukin-10 (IL-10)41, a GPCR43-mediated anti-inflammatory factor, and attenuates the inflammatory response mediated by helper T cells42and regulatory T cells42,43. Furthermore, hyperactivation of the sympathetic nervous system is one of the key hallmarks of hypertension44. It has been shown that SCFAs can directly modulate the sympathetic nervous system via GPCR4145. Therefore, we speculate that AA and VA may negatively regulate blood pressure through above mentioned pathways. In supporting this speculation, a study of long-lived people in Guangxi, China, found that AA counteracts the damage caused by hypertension to maintain a healthy body balance46. A cohort study including participants from six different ethnic groups showed that participants with lower SBP tended to have lower levels of AA in their feces, which may be due to the fact that more SCFAs are absorbed into the body and thus less being excreted out of the body47.

Animal studies have shown that PA and BA both lower blood pressure in mice30,48,49,50. Cross-sectional human studies have also shown that there is a negative correlation between serum BA and hypertension13,51. It has also been shown that butyrate levels are negatively correlated with portal hypertension52and can be potentially used for blood pressure-lowering interventions53, e.g., a randomized, double-blind, placebo-controlled trial has shown that butyrate can reduce blood pressure in patients with type 2 diabetes mellitus54. In line with these reports, our study also show that there is a negative correlation between butyric acid and hypertension in Chinese elderly population, although not statistically significant.

From the current studies conducted in young and older adults, the findings are nearly identical, i.e., AA may be a potential treatment for hypertension; our cohort study was specifically conducted in older adults and called Health and Environmentally Controllable Factors in Older Adults, and therefore there are no data from the younger populations in the same geographic area, and future studies need to focus on this issue. Most notably, our findings may only apply to older adults in Lu 'an City and cannot be generalized to the entire population.

Analyses of SCFAs mixture

Epidemiological studies on the association between SCFAs mixture in serum and blood pressure are limited. As far as we are concerned, previous studies have not addressed this issue. In this study, we applied the BKMR models to explore the correlation between the SCFAs mixture in serum and blood pressure. We observe that the mixture of SCFAs has an inverse trend with hypertension and SBP, but not significant. Yet, the relationship between SCFAs mixture in serum and hypertension, blood pressure still warrants further investigation.

Strengths and limitations

This study has the following advantages: (1) This is the first study to explore the correlation between serum SCFAs and hypertension use elderly residents in China and our results provide initial indications for further studying the influence of serum SCFAs mixture on hypertension; (2) In this report, BKMR model was used to investigate potential non-linear, holistic and interactive relationships between mixture of SCFAs and hypertension, blood pressure; and (3) The chemical properties of SCFAs are unique, and currently there is still lack of a precise method available for detecting serum SCFAs in human, we developed an UHPLC-QE-Orbitrap MS method that demonstrated stability, sensitivity, and accuracy.

Nevertheless, several limitations should be noted. Firstly, owing to the constraints of the cross-sectional design, we were unable to establish a causal relationship between serum SCFAs mixture and hypertension or blood pressure. Secondly, only one-time measurement of serum SCFAs was conducted for elder individuals, which may not adequately illustrate the long-term SCFAs levels in older adults and could introduce measurement bias. Thirdly, we can’t rule out the effects of all confounding factors, such as antibiotics, as most SCFAs in vivo are produced by the fermentation of dietary fibers by the gut microbiome, which may affect the abundance and community of the gut microbiome. Fourthly, we limited our analyses to 7 SCFAs that were most abundant in our samples, although other SCFAs may also be involved in the regulation of blood pressure. Fifthly, the collection or storage of biological samples may affect the stability of SCFAs, which may introduce bias into the measurement of SCFAs concentrations. Sixthly, several lifestyle components were self-reported and thus, recall bias is inevitable. For example, the missing use of medications such as anti-inflammatory drugs and antibiotics is a limitation of our study. Seventhly, in addition to diabetes, some individuals may also suffer from other comorbidities of hypertension, which may prompt study subjects to undergo drug treatment or lifestyle changes. This could lead to changes in their SCFAs concentrations, thereby forming a reverse causal relationship. Finally, our findings may only apply to older adults in Lu 'an City and cannot be generalized to the entire population.

Conclusions

In conclusion, while the precise mechanism underlying SCFAs’ impact on blood pressure remains elusive, our results advocate for considering SCFA as a potential intervention to lower blood pressure, and especially AA may be a possible target for research. Further investigation is warranted to confirm these preliminary findings and to gain deeper insights into the mechanisms through which SCFAs contribute to blood pressure regulation.