Impact of Antihypertensive Agents on Central Systolic Blood Pressure and Augmentation Index: A Meta-Analysis

Abstract

BACKGROUND

New evidence suggests that central systolic blood pressure (cSBP) and augmentation index (AI) are superior predictors of adverse cardiovascular outcomes compared to peripheral systolic BP (pSBP). We performed a meta-analysis assessing the impact of antihypertensives on cSBP and AI.

METHODS

PubMed, Cochrane Library, and CINAHL were searched until September 2014 to identify eligible articles. A DerSimonian and Laird random-effects model was used to calculate the weighted mean difference (WMD) and its 95% confidence interval (CI). Fifty-two and 58 studies incorporating 4,381 and 3,716 unique subjects were included for cSBP and AI analysis, respectively.

RESULTS

Overall, antihypertensives reduced pSBP more than cSBP (WMD 2.52mm Hg, 95% CI 1.35 to 3.69; I ² = 21.9%). β-Blockers (BBs) posed a significantly greater reduction in pSBP as compared to cSBP (WMD 5.19mm Hg, 95% CI 3.21 to 7.18). α-Blockers, angiotensin converting enzyme inhibitors, angiotensin II receptor blockers, calcium channel blockers, diuretics, renin-angiotensin aldosterone system inhibitors and nicorandil reduced cSBP and pSBP in a similar manner. The overall reduction in AI from baseline was 3.09% (95% CI 2.28 to 3.90; I ² = 84.5%). A significant reduction in AI was seen with angiotensin converting enzyme inhibitors, angiotensin II receptor blockers, calcium channel blockers, diuretics, renin-angiotensin aldosterone system inhibitors, BBs, α-blockers (ABs), nicorandil, and moxonidine reduced AI nonsignificantly.

CONCLUSIONS

BBs are not as beneficial as the other antihypertensives in reducing cSBP and AI.

Elevated blood pressure (BP) is a leading modifiable risk factor for cardiovascular disease and continues to affect approximately 1 in 3 adults, or 66.9 million people, in the United States.¹ Traditionally, hypertension is diagnosed and treated by assessing the pressure at the brachial artery (peripheral BP),² but recent evidence suggests that central hemodynamics are better predictors of cardiovascular outcomes and mortality.^3,4

Central BP is indicative of the pressure directly exerted on target organs and often varies from peripheral BP.² Aortic and carotid arteries are more elastic than fibrous peripheral vasculature and the difference in peripheral and central pressures is thought to be a result of amplification due to wave reflections caused by the variance in arterial stiffness.^2,5 As a result of arterial stiffness increasing with distance from the heart, peripheral systolic BP (pSBP) tends to be greater than central systolic BP (cSBP).⁶ Additionally, augmentation index (AI), which measures the degree of enhancement in the central pressure waveform due to reflected waves, has been shown to be an independent predictor of cardiovascular events.⁴

Recent technology has increased the availability of several noninvasive techniques to estimate central BP allowing for incorporation of these parameters in a multitude of patient populations and disease states.^7–10 Differences between the various classes of antihypertensive agents regarding their effects on central hemodynamics have been identified.^11,12 The Conduit Artery Function Evaluation (CAFE) study¹³ was one of the first trials to show differing clinical outcomes despite similar reductions in peripheral BP. In a previous meta-analysis, differing responses of β-blockers (BBs) and diuretics on central hemodynamics were implied but extrapolation of their finding was limited due to a modest number of included studies.¹² As a result of a greater number of new publications in the last few years assessing the effects of antihypertensives on central BP, we performed a meta-analysis analyzing the differential effects of antihypertensive agents on cSBP and AI. An assessment as such will help better determine the incorporation and place in therapy of the various antihypertensives in clinical practice.

METHODS

Literature search

The online databases PubMed (MEDLINE), Cochrane Library (Central), and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) were searched until 26 September 2014 for relevant studies. The following terms were used: central BP, aortic BP, carotid BP, pulse wave velocity, AI, and antihypertensives. Studies were limited to clinical trials conducted in humans and published in the English language. In searching the literature and presenting results, the guidelines provided in the Preferred Reporting Items for Systematic Reviews and Meta-Analysis: The PRISMA Statement were followed.¹⁴

Study eligibility and selection

Studies were included if they met the following criteria: (i) randomized, controlled trials in humans, (ii) reported data on pSBP and cSBP, (iii) reported data on AI, (iv) had a prescription antihypertensive medication intervention, and (v) had a duration of treatment ≥28 days. Studies were excluded if reporting incomplete or inappropriate data, used a nonprescription intervention or prescription intervention that was not an antihypertensive, were duplicate publications, or published in a language other than English.

Two researchers independently conducted an initial screening of the studies based on the abstracts. The next phase involved an examination of the full text in reference to eligibility criteria. The final eligibility of the articles was determined through agreement between the 2 reviewers, with any disagreement resolved in consultation with a third reviewer. A summary of the review is presented in the PRISMA flow chart (Figure 1). Included articles were reviewed and relevant data were extracted.

Figure 1.

Study flow diagram.

Data extraction and quality assessment

All data were extracted through the use of a standardized data abstraction tool. The following information were retrieved from each article: author, year of publication, study design, study population, study intervention, dose and duration of treatment, sample size, type of device used to measure central hemodynamics, and relevant primary and secondary endpoint data. Baseline and post-intervention pSBP and cSBP and/or AI were extracted with SD, SEM, or 95% confidence intervals (CIs) (as available) and the change from baseline was calculated. If the change from baseline was reported, this information was recorded along with SD, SEM, or 95% CIs (as available). When multiple usable arms were available within an individual study, the data were counted as another study in the meta-analysis model. Methodological quality was assessed using Jadad scores.¹⁵

Data synthesis and analysis

The weighted mean difference (WMD) for BPs was calculated as the difference between the mean change from baseline in pSBP and the mean change from baseline in cSBP (ΔpSBP – ΔcSBP). The WMD for AI was calculated as the difference in the mean change from baseline for each applicable intervention arm. A DerSimonian and Laird random-effects model was used to calculate the WMD and its 95% CIs with sample sizes halved to avoid artificial truncation of the CIs. All statistical analyses were completed using StatsDirect Version 2.6.6 (StatsDirect, Cheshire, UK). SEM was converted to SD (SE = SD/√(sample size)) and CIs were calculated (CI = (mean difference) ± [SEM × 1.96]) when necessary.

Publication bias was assessed using the Egger’s statistic and visual inspection of funnel plots. Statistical heterogeneity was assessed using the I ² statistic. To assess robustness, both the fixed and random effect models were utilized and reported. Subgroup analyses were performed to assess sources of clinical heterogeneity. Impact on cSBP and AI were assessed by analyzing data by individual antihypertensive class (α-blockers (ABs), angiotensin converting enzyme inhibitors (ACE-Is), angiotensin II receptor blockers (ARBs), BBs, calcium channel blockers (CCBs), diuretics, renin-angiotensin-aldosterone-system inhibitor (RAAS-I), nicorandil and moxonidine as well as combination interventions (ACE-I + ARB, ACE-I + diuretic, ARB + CCB, ARB + diuretic, BB + CCB, BB + diuretic, CCB + diuretic, neutral endopeptidase/ACE-I, RAAS-I + diuretic). In addition, subgroup analyses were performed by limiting to studies with a primary diagnosis of hypertension and excluding studies with Jadad scores <3.

RESULTS

A total of 52 studies incorporating 4,381 participants were including for the cSBP analysis and 58 studies incorporating 3,716 unique subjects were included for the AI analysis. Table 1 describes the study characteristics of the included studies. Fifteen of the included studies used a crossover study design while 46 used a parallel study design. Data for the following antihypertensives were available for analysis: ABs, ACE-Is, ARBs, BBs, CCBs, diuretics, RAAS-Is, nicorandil, moxonidine, and omapatrilat.

Table 1.

Characteristics of included studies

Author, year	Patient population	Drug class(s)	Sample size	Study design	SBP/AI	Jadad
Baschiera et al. 201416	Essential hypertension	RAAS-I vs. ACE-I	154	Parallel	SBP only	5
Bhagatwala et al. 201417	Prehypertension	Diuretic	17	Non-controlled	SBP & AI	1
Kim et al. 201418	Hypertension	ARB vs. BB	182	Parallel	SBP & AI	3
Peters et al. 201419	Hemodialysis patients	ARB vs. placebo	41	Parallel	SBP & AI	4
Agnoletti et al. 201320	Essential hypertension	CCB vs. ARB vs. Diuretic vs. placebo	110	Parallel	SBP & AI	3
Briasoulis et al. 201321	Type II diabetes	BB vs. nebivolol	61	Parallel	SBP & AI	2
Davis et al. 201322	Prehypertension	Nebivolol vs. placebo	25	Parallel	AI only	4
Dorresteijn et al. 201323	Obesity-related hypertension	RAAS-I vs. moxonidine vs. diuretic vs. placebo	30	Crossover	AI only	5
Eeftinck et al. 201324	Hypertension	Nebivolol + diuretic vs. BB + diuretic	22	Crossover	SBP & AI	3
Ihm et al. 201325	Essential hypertension	CCB vs. ARB	161	Parallel	SBP & AI	2
Kubota et al. 201326	Essential hypertension	RAAS-I vs. RAAS-I + diuretic	30	Parallel	SBP & AI	3
Kwon et al. 201327	Hypertension	ARB + diuretic vs. ARB + diuretic	28	Crossover	SBP & AI	3
Park et al. 201328	Hypertension	BB vs. BB	191	Parallel	SBP & AI	2
Radchenko et al. 201329	Essential hypertension	ARB + diuretic vs. BB + diuretic	59	Parallel	SBP & AI	2
Shahin et al. 201330	Intermittent claudication	ACE-I vs. placebo	14	Parallel	SBP & AI	5
Spanos et al. 201331	Hypertension	RAAS-I vs. ARB	29	Parallel	SBP & AI	2
Studinger et al. 201332	Hypertension	BB vs. BB vs. nebivolol	60	Parallel	SBP & AI	5
Takami et al. 201333	Hypertension	ARB + CCB vs. ARB + CCB	52	Parallel	SBP & AI	3
Zhou et al. 201334	Essential hypertension	BB vs. BB	109	Parallel	SBP & AI	3
Frimodt-Moller 2012 et al. 35	CKD + hypertension	ACE-I + ARB	57	Parallel	SBP & AI	3
Hayoz et al. 201236	Postmenopausal women	ARB + diuretic vs. CCB + diuretic	125	Parallel	SBP only	3
Kanaoka et al. 201237	Essential hypertension	RAAS-I	17	Non-controlled	SBP & AI	1
Takenaka et al. 201238	CKD + hypertension	CCB vs. CCB	59	Parallel	SBP & AI	3
Virdis et al. 201239	Essential hypertension	RAAS-I vs. ACE-I	50	Parallel	SBP & AI	2
Vitale et al. 201240	Essential hypertension	ARB + diuretic vs. nebivolol + diuretic	65	Parallel	SBP & AI	4
Williams et al. 201241	Marfan syndrome	BB vs. ACE-I vs. CCB	14	Crossover	SBP & AI	5
Ferdinand et al. 201142	Hypertension	RAAS-I + diuretic vs. CCB	324	Parallel	SBP only	5
Hefferman et al. 201143	Hypertension	BB vs. BB	24	Crossover	AI only	2
Kampus et al. 201144	Hypertension	BB vs. nebivolol	63	Parallel	SBP & AI	3
Kandavar et al. 201145	Hypertension	BB vs. nebivolol	41	Parallel	AI only	1
Kimura et al. 201146	CKD	Nicorandil	9	Non-controlled	SBP & AI	3
Matsui et al. 201147	Hypertension	ARB + CCB vs. ARB + diuretic	207	Parallel	AI only	3
Shah et al. 201148	Essential hypertension	BB vs. BB	41	Parallel	SBP & AI	1
Boutouyrie et al. 201049	Essential hypertension	ARB + CCB vs. BB + CCB	331	Parallel	SBP & AI	2
Doi et al. 201050	Hypertension	CCB vs. diuretic	37	Parallel	SBP & AI	2
Kinouchi et al. 201051	Diabetes + hypertension	ARB vs. ARB + diuretic	25	Crossover	AI only	2
Mackenzie et al. 200952	Isolated systolic hypertension	ACE-I vs. BB vs. CCB vs. diuretic	59	Parallel	SBP & AI	2
Matsui et al. 200953	Hypertension	ARB + CCB vs. ARB + diuretic	207	Parallel	SBP only	3
Palombo et al. 200954	Essential hypertension	CCB	21	Non-controlled	AI only	1
Ahimastos et al. 200855	Peripheral arterial disease	ACE-I	20	Parallel	SBP & AI	2
Alem et al. 200856	Stroke	Diuretic vs. diuretic	23	Parallel	AI only	4
Dhakam et al. 200857	Isolated systolic hypertension	BB vs. nebivolol	16	Crossover	SBP & AI	3
Karalliedde et al. 200858	Type II diabetes	ARB vs. CCB	131	Parallel	SBP & AI	4
Mahmud and Feely 200859	Hypertension	BB vs. nebivolol	39	Parallel	SBP & AI	3
Schneider et al. 200860	Essential hypertension	ARB vs. BB	156	Parallel	SBP & AI	2
Dart et al. 200761	Hypertension	ACE-I vs. diuretic	479	Parallel	SBP only	2
Jiang et al. 200762	Essential hypertension	ACE-I vs. diuretic	101	Parallel	SBP & AI	3
Dhakam et al. 200663	Hypertension	BB vs. ARB	21	Crossover	SBP & AI	3
Rajagopalan et al. 200664	Normotensive	ARB vs. placebo	33	Crossover	AI only	3
Tsang et al. 200665	Isolated diastolic dysfunction	ACE-I vs. placebo	9	Parallel	AI only	4
Davies et al. 200566	Hypertension	BB vs. ARB	17	Crossover	AI only	2
Mahmud and Feely 200567	Essential hypertension	Diuretic vs. diuretic	24	Crossover	SBP & AI	2
Mitchell et al. 200568	Hypertension	Omapatrilat vs. ACE-I	159	Parallel	SBP & AI	3
de Luca et al. 200469	Hypertension	ACE-I + duretic vs. BB	52	Parallel	SBP & AI	3
London et al. 200470	Essential hypertension	ACE-I + diuretic vs. BB	181	Parallel	SBP & AI	3
Neal et al. 200471	Liver transplant + hypertension	CCB vs. BB vs. ACE-I	24	Crossover	SBP & AI	2
Deary et al. 200272	Essential hypertension	CCB vs. AB vs. ACE-I vs. BB vs. diuretic vs. placebo	30	Crossover	SBP & AI	1
Klingbeil et al. 200273	Essential hypertension	ARB vs. diuretic	40	Parallel	SBP & AI	5
Mahmud and Feely 2002(a)74	Hypertension	ARB vs. diuretic	11	Crossover	SBP only	2
Mahmud and Feely 2002(b)75	Hypertension	ARB vs. ACE-I	12	Crossover	SBP only	1
Asmar et al. 200176	Essential hypertension	ACE-I + diuretic vs. BB	144	Parallel	AI only	2
Dart et al. 200177	Hypertension	ACE-I vs. “usual care”	51	Parallel	SBP only	3
Spratt et al. 200178	Hypertension	ARB + diuretic	35	Non-controlled	SBP & AI	1
Chen et al. 199579	Essential hypertension	ACE-I vs. BB	79	Parallel	AI only	2
London et al. 199480	ESRD	ACE-I vs. CCB	24	Parallel	SBP & AI	4
Guerin et al. 199281	Hypertension	BB vs. CCB	20	Parallel	AI only	2

Abbreviations: AB, α-blockers; ACE-I, angiotensin converting enzyme inhibitor; AI, augmentation index; ARB, angiotensin II receptor blocker; BB, β-blocker; CKD, chronic kidney disease; CCB, calcium channel blocker; ESRD, end stage renal disease; RAAS-I, renin-angiotensin-aldosterone-system inhibitor; SBP, systolic blood pressure.

Overall, antihypertensives reduced pSBP more than cSBP (2.52mm Hg, 95% CI 1.35 to 3.69; I ² = 21.9%) (Figure 2). This effect was maintained when limiting to patient populations with a primary diagnosis of prehypertension or hypertension (2.82mm Hg, 95% CI 1.46 to 4.17, I ² = 29%). Additionally, utilizing the fixed-effects model (2.40mm Hg, 95% CI 1.46 to 3.34) and limiting to Jadad scores ≥3 (2.57mm Hg, 95% CI 1.32 to 3.82) did not change the outcomes (Table 2).

Figure 2.

Overall effect of antihypertensives on peripheral systolic blood pressure (pSBP) and central systolic blood pressure (cSBP).

Table 2.

Antihypertensive class effect on peripheral systolic blood pressure minus central systolic blood pressure and augmentation index

Subgroup	ΔpSBP – ΔcSBP (mm Hg)			AI (%)
	N (n)	Pooled WMD	95% CI	N (n)	Pooled WMD	95% CI
Overall (random-effects)	52 (4,381)	2.52	1.35 to 3.69	58 (3716)	3.09	2.28 to 3.90
Overall (fixed-effects)	52 (4,381)	2.40	1.46 to 3.34	58 (3716)	2.45	2.19 to 2.71
Jadad ≥3	31 (2,540)	2.57	1.32 to 3.82	33 (2381)	3.68	2.82 to 4.55
HTN diagnosis	40 (3,791)	2.82	1.46 to 4.17	44 (3226)	2.75	1.79 to 3.71
AB	1 (30)	−0.12	−14.30 to 14.06	1 (30)	3.95	−2.51 to 10.42
ACE-I	14 (677)	−2.40	−4.89 to 0.08	12 (330)	5.61	4.27 to 6.95
ARB	9 (393)	1.12	−2.25 to 4.49	11 (445)	5.28	1.95 to 8.61
BB	17 (963)	5.19	3.21 to 7.18	24 (1192)	0.32	−0.84 to 1.48
CCB	11 (510)	1.01	−2.17 to 4.19	12 (379)	5.36	3.77 to 6.95
Diuretic	10 (449)	0.65	−2.47 to 3.77	10 (268)	3.24	1.03 to 5.45
RAAS-I	5 (150)	1.09	−4.80 to 6.99	5 (100)	3.39	1.83 to 4.95
Nicorandil	1 (9)	−4.60	−56.33 to 47.13	1 (9)	7.10	−5.43 to 19.63
Moxonidine	N/A	N/A	N/A	1 (30)	0.30	−2.80 to 3.40
ACE-I + ARB	1 (57)	1.00	−17.19 to 19.19	1 (57)	2.00	−4.02 to 8.02
ACE-I + diuretic	2 (118)	0.29	−3.74 to 4.32	3 (188)	4.21	1.53 to 6.88
ARB + CCB	3 (324)	2.81	−3.31 to 8.94	3 (324)	4.60	3.09 to 6.10
ARB + diuretic	7 (350)	4.28	−3.27 to 11.82	7 (340)	3.46	1.66 to 5.27
BB + CCB	1 (162)	2.10	−2.21 to 6.41	1 (162)	2.81	1.16 to 4.46
BB + diuretic	3 (80)	10.60	−3.47 to 24.67	3 (80)	0.59	−4.19 to 5.36
CCB + diuretic	1 (62)	6.00	−0.08 to 12.08	N/A	N/A	N/A
NEP/ACE-I	1 (75)	−4.90	−12.22 to 2.42	1 (75)	8.00	5.26 to 10.74
RAAS-I + diuretic	2 (177)	−1.29	−8.90 to 6.31	1 (15)	2.30	−12.05 to 16.65

Abbreviations: AB, α-blockers; ACE-I, angiotensin converting enzyme inhibitor; ARB, angiotensin II receptor blocker; BB, β-blocker; CCB, calcium channel blocker; HTN, hypertension; N, number of studies; n, sample size; NEP, neuroendopeptidase; RAAS-I, renin-angiotensin-aldosterone-system inhibitor; WMD, weighted mean difference.

Subgroup analyses of individual drug classes showed a nonsignificant difference between pSBP and cSBP with ACE-Is, ARBs, CCBs, diuretics, and RAAS-Is (−2.40mm Hg, 95% CI −4.89 to 0.08; 1.12mm Hg, 95% CI −2.25 to 4.49; 1.01mm Hg, 95% CI −2.17 to 4.19; 0.65mm Hg, 95% CI −2.47 to 3.77; 1.09mm Hg, 95% CI −4.80 to 6.99, respectively) (Table 2). BBs posed a significantly greater reduction in pSBP as compared to cSBP (5.19mm Hg, 95% CI 3.21 to 7.18) (Table 2). Only one study was available for analysis of ABs and nicorandil, which showed no difference in the reduction between pSBP and cSBP (−0.12mm Hg, 95% CI −14.30 to 14.06; −4.60mm Hg, 95% CI −56.33 to 47.13; −4.90mm Hg, respectively) (Table 2). None of the combination antihypertensive regimens revealed a significant difference in their effect on pSBP compared to cSBP (Table 2).

The overall reduction in AI from baseline was 3.09% (95% CI 2.28 to 3.90; I ² = 84.5%) (Figure 3). When only the studies containing a patient population with a primary diagnosis of prehypertension or hypertension were analyzed, the change in AI from baseline was 2.75% (95% CI 1.79 to 3.71; I ² = 84.8%). Additionally, utilizing the fixed-effects model (2.45%, 95% CI 2.19 to 2.71; I ² = 84.5%) and limiting to Jadad scores ≥3 (3.68%, 95% CI 2.82 to 4.55; I ² = 64.1%) did not change the outcomes (Table 2).

Figure 3.

Overall effect of antihypertensives on augmentation index (AI).

Subgroup analyses of individual antihypertensive classes showed a significant reduction in AI with ACE-Is, ARBs, CCBs, diuretics, and RAAS-Is (5.61%, 95% CI 4.27 to 6.95; 5.28%, 95% CI 1.95 to 8.61; 5.36%, 95% CI 3.77 to 6.95; 3.24%, 95% CI 1.03 to 5.45; 3.39%, 95% CI 1.83 to 4.95, respectively) (Table 2). BBs had a nonsignificant reduction in AI (0.32%, 95% CI −0.84 to 1.48) along with AB, nicorandil, and moxonidine (3.95%, 95% CI −2.51 to 10.42; 7.10%, 95% CI −5.43 to 19.63; 0.30%, 95% CI −2.80 to 3.40, respectively) (Table 2).

While the Egger’s statistic showed a lack of publication bias (P > 0.976 for CBP and P > 0.125 for AI), it cannot be ruled out based on visual inspection of funnel plots (Figure 4).

Figure 4.

(A) Bias assessment plot for peripheral systolic blood pressure (pSBP) minus central systolic blood pressure (cSBP) and (B) augmentation index (AI).

DISCUSSION

Overall, antihypertensives significantly reduce pSBP more than cSBP and AI from baseline. An analysis by antihypertensive class showed that the significantly greater reduction of pSBP compared to cSBP was true only for BBs but not the other mono- or combination antihypertensive therapy groups. ACE-Is, ARBs, CCBs, diuretics, and RAAS-Is all significantly reduced AI from baseline, while BBs, ABs, nicorandil, and moxonidine did not.

In the Heart Outcome Prevention Evaluation study, a modest reduction of only 3.3mm Hg in pSBP post-ACE-I therapy was associated with a 22% relative risk reduction in cardiovascular death, myocardial infarction, or stroke.⁸² However, the same magnitude of benefit should not be extrapolated post-BB therapy due to the significant difference evident in the reduction in pSBP and cSBP. As such, in a meta-analysis of 3,285 subjects, an increase in cSBP of 10mm Hg corresponded to an 8.8% increased risk of total CV events.⁴ Our results are in line with previous evidence showing that BBs may be inferior to other antihypertensives in their ability to prevent cardiovascular events, specifically stroke.^83–85 The diverse and variable mechanisms of action of BBs could attribute to this inferior ability.⁸⁶ BBs demonstrate antihypertensive action by decreasing cardiac output, inhibiting the release of renin and production of angiotensin II, and blocking presynaptic α-adrenoreceptors.⁸⁶ However, BBs also exhibit effects on glucose and lipid metabolism⁸³ and their long-term hemodynamic effect of increasing peripheral resistance may lead to their inefficacy in reducing morbidity and mortality.⁸⁷

Another important finding of this meta-analysis was the effect of antihypertensive classes on AI. A 10% increase in AI has been shown to increase risk of total cardiovascular events by 32% and total mortality by 38%.⁴ The significant reduction in AI with ACE-Is, ARBs, CCBs, diuretics, and RAAS-Is exhibits additional evidence of the beneficial effects of these drug classes.

Our findings significantly bolster the results of a previous meta-analysis (n = 24) driven by the incorporation of a larger number of recently published studies (44 additional studies plus 22 from previous meta-analysis). While their findings suggested a significantly differing effect on pSBP and cSBP with BB and diuretics, our results only suggest this to be significantly true post-BB therapy. Our meta-analysis also supports the notion that clinical practice should not be averse to appropriate diuretics use. Their results also suggested a significant increase in AI with BB therapy; this significance was not evident in the current meta-analysis. This may be due to the fact that they were limited by the availability of only 3 studies while our analysis incorporated 24 studies. BBs also have a direct impact on the heart rate, confounding the interpretation of the unadjusted AI. Future clinical trials should use heart rate adjusted AI to assess the efficacy of existing and novel drugs. While we included a total of 66 studies overall, the results for ABs, nicorandil, moxonidine, and certain combinations should be interpreted with caution due to the low sample sizes.

There are several limitations to this meta-analysis. The dosage of antihypertensives varied between the trials included in this meta-analysis. Many trials contained titration protocols and the total dose of antihypertensive that was ultimately used in each participant is unknown. Also, some study protocols allowed for the addition of diuretics when BP goals were not met with the study intervention alone. However, these confounders should play a limited role in the interpretation of our results as the change from baseline was assessed for all endpoints. Moreover, while studies were only included in the analysis if duration of treatment was at least 28 days, there was wide variance in the duration of the studies from 4 weeks^{17,24,41,43,46,48,56,59,74,75,81} to 4 years.⁶¹ Furthermore, the tonometry site for measurement of central hemodynamics varied amongst studies and a recent review of available techniques for measuring central BP cautioned that noninvasive cSBP estimation is device/technique-dependent.¹⁰

One study, the CAFE study,¹³ which was included in the previous meta-analysis, was not included in this analysis for not reporting baseline central BP data and also for not identifying which patients were on monotherapy or specific combinations of antihypertensives. This study however included 2,073 participants and had the ability to have a sizable impact on our results.

The number of randomized, controlled trials including central hemodynamic endpoints has substantially increased in the past decade. However, many of these trials include small patient populations and are unable to individually show overall class effects of antihypertensives. This systematic review highlighted the need for further studies on some antihypertensive drug classes lacking evidence, such as ABs and RAAS-Is, and emerging agents, such as moxonidine and omapatrilat, which have an increasingly important role in the management of cardiovascular disease. Further, while we compared inter-antihypertensive class differences, the availability of more studies should allow for future meta-analyses assessing intra-class differences (e.g., dihydropyridine vs. non-dihydropyridine) and the impact of specific agents. The paucity of usable information limits this meta-analysis from performing detailed subgroup analyses.

It is well established that improving pSBP leads to better cardiovascular outcomes. The CAFE trial first demonstrated differing central hemodynamic effects with 2 different antihypertensive agents while having identical reductions in pSBP. The preponderance of current evidence suggests that central hemodynamic endpoints (cSBP and AI) are better predictors of long-term outcomes over pSBP. The results of this meta-analysis showed that BBs are less effective than other antihypertensives at reducing both cSBP and AI. Our results bolster the recommendations of the recently published guidelines for the management of high BP, which have deferred the initiation of BBs in the treatment algorithm.⁸⁸ Future trials are needed to determine the true magnitude of benefit of current and emerging pharmacological and lifestyle interventions in regards to central hemodynamic endpoints.

DISCLOSURE

The views expressed in this material are those of the authors, and do not reflect the official policy or position of the U.S. Government, the Department of Defense, or the Department of the Air Force. These data were presented at the American Heart Association’s Epidemiology and Prevention/Lifestyles 2015 Scientific Sessions.