Association of cardiovascular events with central systolic blood pressure: A systemic review and meta‐analysis

Abstract Central blood pressure confers cardiovascular risk prediction ability, but whether the association between central systolic blood pressure (cSBP) and cardiovascular endpoints is independent of peripheral systolic blood pressure (pSBP) remains controversial. This systematic review and meta‐analysis aim to investigate the associations between cSBP and cardiovascular endpoints in models including and excluding pSBP, respectively. Observational studies assessing the risk of composite cardiovascular endpoints with baseline cSBP were searched in PubMed, Embase, Scopus, Web of Science, and Cochrane Library to May 31, 2022. Risk of bias was assessed by the Newcastle‐Ottawa Quality Assessment Scale, and random‐effects models were used to pool estimates. Finally, 48 200 participants from 19 studies with a mean age of 59.0 ± 6.9 years were included. Per 10 mmHg increase of cSBP was associated with higher risk of composite cardiovascular outcomes (risk ratio [RR]: 1.14 [95%CI 1.08–1.19]) and cardiovascular death (RR: 1.18 [95%CI 1.08–1.30]), and the associations still existed after adjusting for pSBP (RR: 1.13 [95%CI 1.05–1.21] for composite cardiovascular endpoints; RR: 1.25 [95%CI 1.09–1.43] for cardiovascular death). In pSBP‐unadjusted studies, increased cSBP was also associated with higher risk of all‐cause mortality and stroke, but not in the pSBP‐adjusted studies. Both cSBP and pSBP were similarly significantly associated with composite cardiovascular endpoints in models containing them separately and simultaneously. cSBP was significantly associated with cardiovascular events, independently of pSBP. Central or peripheral SBP could supplement cardiovascular risk assessment besides each other.


INTRODUCTION
][10][11] Nevertheless, recent studies found that cSBP was no longer a predictor for cardiovascular outcomes in multivariable-adjusted models containing peripheral systolic blood pressure (pSBP), [12][13][14] making it harder to conclude the ability of cSBP to predict future cardiovascular events independently of pSBP.Thus, the present study was conducted to provide an overview of relevant studies, and to provide an overall quantitative estimate of associations between cSBP and cardiovascular events and all-cause mortality, as well as different values between subgroups of populations and cSBP measurement methods.

METHODS
The authors declare that all supporting data are available within the article and its Supplement Materials.This study was registered with the International Prospective Register of Systematic Reviews (URL: https://www.crd.york.ac.uk/prospero/.Unique Identifier: CRD42022337668), and was reported with the PRISMA guideline. 15

Search strategy and eligibility criteria
Electronic database searches were performed in PubMed (Medline), Embase, Scopus, Web of Science, and Cochrane Library without language limitation for studies reporting associations between cSBP and cardiovascular events or all-cause mortality from database inception to May 31, 2022.Reference lists of relevant articles and reviews were also searched for additional reports.More detailed search terms are available in Table S1.K.L. and F.F. screened and evaluated the identified articles, and the independently selected studies by the two reviewers were compared to obtain the final inclusion of studies.Any conflicts were resolved by consulting a third reviewer (Y.Z.).

Outcomes
The primary outcome was composite cardiovascular endpoints, composed by at least two of the following endpoints: cardiovascular mortality, myocardial infarction, stroke, heart failure, and revascularizations (coronary artery bypass graft surgery or percutaneous coronary interventions).The detailed definitions of composite cardiovascular endpoints were summarized in Table S2.The secondary outcomes were all-cause death, cardiovascular mortality, stroke, myocardial infarction, and heart failure.Multivariable-adjusted hazard ratios (HR) or coefficients from linear regression models with their confidence intervals (CI) for the outcomes of interest were also extracted.When a study reported multiple calibration methods of central blood pressure, the results from calibration based on measured brachial mean arterial pressure and diastolic blood pressure were prioritized, because of its more accurate estimation and stronger associations with mortality independently of peripheral blood pressure. 16,17sk of bias within each study was assessed using the Newcastle-Ottawa Quality Assessment Scale for Cohort Studies (NOS), which judged a cohort study on selection of the study group (4 items, up to 4 stars), comparability (1 items, up to 2 stars), and outcome (3 items, up to 3 stars). 18Two reviewers (K.L. and F.F.) independently evaluated the included studies, and inconsistencies were resolved by consensus.

Statistical analysis
HR was treated as risk ratio (RR) in the analysis, which was used in previous meta-analyses. 19,20RR for per 10 mmHg difference of cSBP was used to combine the results across the studies.For the studies reporting the quantiles of cSBP, the RR was calculated according to the HR of the top group compared with the bottom group. 21
The detailed risk assessment of the included studies with NOS tool is presented in Table S5.

cSBP and secondary outcomes
Secondary outcome analyses were also respectively conducted in models including or excluding pSBP (Figure S5), for associations between cSBP and endpoints of all-cause death, cardiovascular mortality, stroke, myocardial infarction, and heart failure, if these data were sufficiently available.For cSBP in models excluding pSBP, per 10 mmHg increases of baseline cSBP was significantly associated with 7% higher risk of all-cause mortality (analytical n = 14 762; RR: 1.07 [95%CI 1.02-1.12];I 2 = 44.20%),18% higher risk of cardiovascular death (analytical

Associations between cSBP and pSBP with primary outcome
Further we conducted analyses on the associations of cSBP and pSBP with cardiovascular events by pooling data from studies where effect sizes of both SBPs were reported.24 Both cSBP and pSBP were significantly associated with composite cardiovascular endpoints, when models included cSBP and pSBP separately (analytical n = 33 541; RR: 1.17

DISCUSSION
This systematic review and meta-analysis pooling data of 48 200 subjects from published studies showed that increased cSBP was significantly associated with higher risk of composite cardiovascular outcomes and cardiovascular death, independently of pSBP.Besides, a significantly positive association between cSBP and all-cause mortality was also observed in models excluding pSBP, and a stronger association between cSBP and composite cardiovascular events was observed from Asian studies in subgroup analysis, though the significances of differences disappeared in the pSBP-adjusted studies.Further sensitivity analysis of studies where effect sizes of both SBPs with composite cardiovascular events were reported showed that an increase of either SBP was associated with higher cardiovascular risks, independently of each other, though with similar effect sizes.

F I G U R E 2
Associations between cSBP and primary outcomes in models excluding (A) and including pSBP (B).Relative risk and 95% confidence interval were presented for a 10 mmHg increase of cSBP.Results in male and female were separately reported in Janner (2012).
According to multivariable models, previous studies on the associations of cSBP and cardiovascular endpoints could be divided into those with pSBP adjusted or unadjusted.The pooled results of unadjusted models in our study were not surprising, because the conclusions obtained from most unadjusted models were consistent, [10][11][12][13]26,27,30 which confirmed the positive associations between cSBP and cardiovascular outcomes. However,ere are still controversies if the associations are independent of pSBP.14,24,35,36 An individual-level meta-analysis showed that cSBP measured by SphygmoCor CvMS device was not associated with composite cardiovascular outcome independently of pSBP, explained by the high correlations between cSBP and pSBP.13 In our analysis, cSBP is still associated with composite cardiovascular outcomes in pooled results from studies in which pSBP is adjusted, possibly due to inclusion of various devices to measure cSBP with different central pulse waveform derivation and calibration methods.Though subgroup analyses show that the association between cSBP and cardiovascular outcomes do not differ according to central pulse waveform derivation and calibration methods, there is a trend towards stronger association between higher cSBP from carotid artery and cardiovascular outcomes, indicating that carotid artery-derived cSBP may be more valuable than peripheral artery-derived cSBP, but few studies compared the associations of cSBP measured from different pulse waveform recording sites and cardiovascular outcomes in a specific population, and we therefore expect more studies to confirm this hypothesis.
In this study, subgroup analyses are carried out by pooling analyses of studies containing cSBP and pSBP separately or simultaneously, to identify the characteristics that may cause the differences and provide new thoughts for future researches.In subgroup analysis of studies using models excluding pSBP, a stronger association between cSBP and cardiovascular events is observed from pooling results in Asian studies than non-Asian studies, though comparable between the groups in models including pSBP.However, the included Asian studies in subgroup analysis of studies using models excluding pSBP have different cSBP obtaining methods and baseline characteristics.Therefore, it is hard to make a conclusion of greater benefit of cSBP measurement among Asian population, and future studies are warranted to find out whether it is the additional value of central blood pressure measurement for Asian population, or the heterogeneities that leaded to the difference.
In the current study, we conducted a sensitivity analysis on the studies where coefficients of both SBPs with cardiovascular outcomes were reported to quantify the associations.Considering the controversy over the collinearity between cSBP and pSBP, the analyses were respectively performed in studies where multivariable-adjusted models contained cSBP and pSBP separately or simultaneously, and similar results were observed.In simultaneously-contained analysis, the pooled results of both SBPs are slightly reduced compared with that in separately-contained analysis, but both SBPs are still associated with cardiovascular events, suggesting that either SBP may deliver supplementary information of cardiovascular risk assessment besides each other.Cardiovascular risk assessment is vital to both primary and secondary prevention of cardiovascular diseases, [37][38][39] and a recent TA B L E 3 Subgroup analysis of associations between central systolic blood pressure primary outcome in studies reporting models including peripheral systolic blood pressure.study showed that central hypertension was associated with higher cardiovascular risk regardless of peripheral blood pressure. 36Therefore, future studies should focus on whether the integration of both SBPs would yield better ability of cardiovascular risk assessment.In general, the quantity of studies reporting effect sizes of both SBPs in separate/simultaneous models was relatively small, and more cohort studies are expected to report the results of both SBPs contained in the multivariable-adjusted models separately and simultaneously when discussing the associations of the two SBPs with cardiovascular outcomes.

Subgroups
Some limitations need to be addressed.Firstly, as previously stated, we included studies using various devices to measure cSBP with different central pulse waveform derivation and calibration methods, but we also performed subgroup analysis to compare the methods and techniques for cSBP estimation.Secondly, we included studies whose participants had different characteristics, representing different F I G U R E 3 Associations of cSBP and pSBP with primary outcomes in models including cSBP and pSBP, separately (A) and simultaneously (B).
Relative risk and 95% confidence interval were presented for a 10 mmHg increase of cSBP.cardiovascular risks at baseline.Thirdly, composite cardiovascular endpoints were defined differently among the included studies.
Fourthly, the available data were insufficient to fully discuss the heterogeneity sources and factors of associations.Fifthly, meta-analyses of observational studies are prone to study-specific biases and unique problems arising from differences in study design.Lastly, cSBP was only analyzed as continuous variable but not categorical variable in current analyses, because there were insufficient categorical data for pooling.

CONCLUSIONS
cSBP was significantly associated with cardiovascular events, independently of pSBP.Despite comparable effect sizes of the 2 SBPs, both SBPs were significantly associated with cardiovascular outcomes in models containing each other, suggesting that central or peripheral SBP could supplement cardiovascular risk assessment.Future studies should provide data on the influences of acquisition methods of cSBP and its values among various regions, as well as the significance of integrating both SBPs on cardiovascular disease prevention.
Sensitivity analyses were conducted by recalculating the pooled RR after omitting one study at a time.All results were reported with 95% CI, and two-sided p < .05 was regarded as statistically significant.All analyses were performed by Review Manager 5.4 (Cochrane Collaboration, Copenhagen, Denmark) and STATA version 16.0 (Stata Corp LP, College Station, Texas, United States).
Subgroup analysis of associations between central systolic blood pressure primary outcome in studies reporting models excluding peripheral systolic blood pressure.
a Data in male and female were separately reported in Janner (2013).