Improvements in Left Ventricular Hypertrophy and Diastolic Function Following Renal Denervation

Effects Beyond Blood Pressure and Heart Rate Reduction

Patient selection

This study was approved by the institutional medical ethics committee (No. 67/11) in accordance with the declaration of Helsinki. Sixty-six consecutive patients scheduled for RDN for treatment of resistant arterial hypertension (defined as office systolic BP [SBP] ≥140 mm Hg) were prospectively included using a prespecified analysis protocol (Symplicity extension NCT00664638, NCT00888433, and NCT01888315) if they were in sinus rhythm and did not suffer from systolic heart failure (ejection fraction <50%) or severe renal dysfunction (estimated glomerular filtration rate <45 ml/min/1.73 m² as measured by cystatin C). Patients suffering from secondary causes of hypertension were excluded.

BP and HR

Following 10 min of supine rest, BP was measured in triplicate with an automatic oscillometric monitor and HR was obtained from a 12-lead electrocardiogram. Ambulatory BP monitoring (ABPM) was performed to exclude pseudoresistant hypertension (mean 24-h SBP <130 mm Hg) (13).

Echocardiography

Transthoracic echocardiography was performed by the same operator (M.M.Y.A.S.), who was blinded to BP data and patient status, on a Vivid E9 (General Electric, Frankfurt, Germany) before and 6 months after RDN. The investigation included regular 2-dimensional echocardiography, Doppler, and tissue Doppler imaging according to the guidelines of the American Society of Echocardiography (14). All studies were read offline by an investigator blinded to patient status (S.H.S.). LV size was measured as left ventricular end-diastolic diameter (LVEDD) using M-mode. LV mass was calculated from the end-diastolic diameters of the interventricular septum (IVSd), left ventricular posterior wall (LVPWd), and LV cavity using the Devereux formula (15). Body surface area was calculated according to the Mosteller formula (16), and LV mass was indexed to height to the 2.7th power. Relative wall thickness was calculated as the sum of IVSd and LVPWd divided by LVEDD. LV diastolic function was assessed according to standardized criteria (17), using pulsed-wave (PW) Doppler of the mitral inflow and measuring deceleration time (DT) of the E-wave, E- to A-wave ratio (E/A), and isovolumetric relaxation time (IVRT) (time between closure of the aortic and opening of the mitral valve). Tissue Doppler was used for measurements of E′ septal and E′ lateral wave (peak early diastolic septal and lateral mitral annular velocities) and calculation of the ratio of the maximum E-wave velocity and the mean E′-wave. Other hemodynamic parameters were calculated as follows: stroke volume = π × VTI_LVOT × (diameter_LVOT/2)², for which VTI is velocity time integral and LVOT is LV outflow tract. Peripheral arterial compliance = stroke volume/(SBP − DBP), for which DBP is diastolic blood pressure. HR-corrected LV circumferential fiber shortening (VCFc) = FS/ET × (60/HR)^1/2, for which FS is fractional shortening and ET is ejection time. End-systolic meridional wall stress (EsMWS) was calculated according to Grossman et al. (18) and Reicheck et al. (19) as follows: (0.334 × SBP × LVESD)/(LVPWs × [1 + LVPWs/LVESD]), for which LVESD is LV end-systolic diameter and LVPWs is LV posterior wall during systole.

RDN procedure

Following renal artery angiography, RDN was performed using the Flex catheter system (Medtronic, Santa Rosa, California), as previously reported (7). In both renal arteries, radiofrequency ablations at a maximum power of 8 W lasting 2 min each were performed.

Statistical analysis

Data are presented as mean ± SEM. Data were analyzed for normal distribution using the Kolmogorov-Smirnov test. Normally distributed parameters were compared between baseline and 6-month follow-up using a paired Student t test or a Wilcoxon matched-pairs test if not normally distributed. Linear regression analyses were used to calculate the correlation between the change in BP and the change in echocardiographic parameters. Analysis of variance was performed with post-hoc testing for trend when the change in the 3 groups was compared. The Fisher exact test was used for testing associations in 2 × 2 contingency tables. Differences were considered significant if p was <0.05. SPSS version 18.0 (SPSS Inc., Chicago, Illinois) was used for statistical calculations.

Baseline characteristics

Patients in the treatment group had a mean age of 63.5 ± 1.2 years. Thirty-six of the 66 study patients (54.5%) were male. Average body mass index was 29.4 ± 0.6 kg/m². Patients were receiving 4.3 ± 0.1 antihypertensive drugs (Online Table 1). The RDN procedure was performed without complications in all patients.

RDN decreased BP, HR, and LV mass and improved diastolic LV function

Six months after RDN, SBP/DBP decreased from 172.9 ± 3.0/92.5 ± 2.3 mm Hg to 151.3 ± 3.2/85.5 ± 1.6 mm Hg (p < 0.001 for both), corresponding to a decrease of 21.6 ± 3.0/10.1 ± 2.0 mm Hg. Pulse pressure decreased from 81.8 ± 2.9 mm Hg to 69.0 ± 2.7 mm Hg (p < 0.001). Reduction of BP was confirmed by 24-h ABPM when completely available at both time points (n = 50). HR decreased from 67.7 ± 1.3 beats/min before to 60.5 ± 1.2 beats/min at 6 months (p < 0.001). LV mass index (LVMI) was reduced from 61.5 ± 2.0 g/m^2.7 before to 53.4 ± 1.5 g/m^2.7 at 6 months (p < 0.001). E/A in mitral PW Doppler as a measure of diastolic function increased from 0.84 ± 0.04 to 1.01 ± 0.06 (p = 0.001) (Fig. 1). DT of PW Doppler detected that E wave of the mitral inflow decreased from 252.0 ± 8.5 at baseline to 227.6 ± 5.7 ms at 6 months (p = 0.010). In tissue Doppler, an improvement of diastolic mitral annular movement was noted by an increase in E′ velocity (mean septal and lateral E′) from 6.73 ± 0.25 cm/s before RDN to 7.47 ± 0.26 cm/s at follow-up (p = 0.003). Maximum E-wave velocity increased from 66.9 ± 2.6 cm/s to 72.9 ± 2.1 cm/s (p = 0.013). Because of the increase of both E and E′, the ratio E/E′_mean as an indicator of filling pressure was unchanged (10.6 ± 0.5 before, 10.4 ± 0.4 at follow-up).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Renal Denervation Reduces Blood Pressure and Heart Rate

In the whole population of 66 patients, (A) systolic blood pressure (SBP) and (B) heart rate (HR) were reduced following renal denervation. Also, (C) left ventricular mass index (LVMI) and (D) E- to A-wave ratio (E/A) velocity as a measure of diastolic function regressed and improved, respectively.

SBP at baseline or SBP reduction did not influence LV mass regression or improvement of diastolic function

Tertiles were defined according to SBP at baseline (n = 22 patients each). In patients with a baseline pressure of 148.6 ± 1.6 mm Hg, SBP was reduced to 136.3 ± 2.9 mm Hg in those with a pressure of 169.5 ± 1.7 mm Hg to 153.2 ± 3.6 mm Hg and in those with a pressure of 200.5 ± 3.6 mm Hg to 164.4 ± 7.5 mm Hg (all p < 0.001; use-dependent effect, p for trend of SBP decrease = 0.002) (Fig. 2A). Patients were also divided into tertiles according to SBP reduction (n = 22 patients each). In the first tertile, SBP remained unchanged from baseline (166.5 ± 5 mm Hg) to 6 months' follow-up (170.8 ± 6.3 mm Hg; p = 0.081), constituting BP nonresponders (≤10 mm Hg reduction). In the second tertile, SBP was reduced by 18.6 ± 1.1 mm Hg from 161.7 ± 3.1 to 143.2 ± 3.2 mm Hg (p < 0.001). In the last tertile, SBP was reduced by 50.4 ± 3.9 mm Hg from 190.4 ± 5.1 to 140.0 ± 4.4 mm Hg (p < 0.001).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Improvements in LV Mass and Diastolic Function Are Unrelated to BP

(A) SBP decreased significantly following renal denervation in all tertiles of SBP at baseline but decreased more the higher the baseline SBP was (use-dependent effect) (B) LVMI decreased similarly in all tertiles of SBP. (C) E/A as a measure of diastolic function increased similarly in all tertiles. Abbreviations as in Figure 1.

LVMI decreased by the same extent in all tertiles of baseline SBP (from 57.6 ± 3.2 g/m^2.7 to 50.8 ± 2.3 g/m^2.7 [p = 0.004] in the first, from 64.8 ± 3.3 g/m^2.7 to 56.6 ± 2.4 g/m^2.7 [p = 0.001] in the second, and from 62.2 g/m^2.7 ± 3.7 to 52.6 ± 3.2 g/m^2.7 [p < 0.001] in the third tertile [p for trend of LVMI decrease = 0.529]) (Fig. 2B). LVMI also regressed comparably in all tertiles of SBP reduction: from 64.0 ± 4.1 g/m^2.7 to 53.2 ± 2.7 g/m^2.7 (p < 0.001) in the first, from 59.6 ± 2.9 g/m^2.7 to 51.6 ± 2.3 g/m^2.7 (p < 0.001) in the second, and from 61.7 ± 3.3 g/m^2.7 to 55.3 ± 3.0 g/m^2.7 (p = 0.002) in the third tertile. Comparable LVMI regression was observed in all tertiles of 24-h BP at baseline and 24-h BP reduction (measured by ABPM) (Online Fig. 1). End-diastolic diameter of the interventricular septum and the LV posterior wall decreased accordingly in all tertiles. Linear regression analysis of the relative change in BP and the relative change in LVMI proved that there was no significant interaction of the change in SBP and LVMI regression (r² = 0.024; p = 0.280). The number of patients with LVH (>44 g/m^2.7 for women, >48 g/m^2.7 for men) decreased from 55 patients at baseline to 39 patients at follow-up (p = 0.004). Similar to wall thickness, LVEDD decreased in all tertiles. LVESD and FS did not change significantly (data not shown).

Next, the relation of diastolic function with SBP was investigated. In the baseline SBP tertiles, E/A increased by 0.19 ± 0.09 (p = 0.018), 0.09 ± 0.05 (p = 0.064), and 0.09 ± 0.07 (p = 0.220), respectively (p for trend of E/A increase = 0.260) (Fig. 2C). DT and IVRT were reduced in all tertiles of baseline SBP. E/A, DT, and IVRT also improved in all tertiles of SBP reduction, and linear regression analyses confirmed lack of relation of improvement of mitral inflow parameters with SBP reduction (r² = 0.001; p = 0.770) for DT, (r² = 0.000; p = 0.907) for E/A, and (r² = 0.014; p = 0.351) for IVRT. Also, E′ velocity increased comparably in all tertiles of SBP reduction.

HR at baseline or HR reduction does not predict LV mass regression or improvement of diastolic function

Although HR reduction correlated with baseline HR (Fig. 3A), it was similar in all tertiles of baseline SBP (data not shown). In the tertiles of SBP reduction, HR also decreased similarly from 69.1 ± 2.4 beats/min to 61.2 ± 2.3 beats/min (p < 0.001), from 64.6 ± 2.2 beats/min to 58.1 ± 2.3 beats/min (p = 0.009), and from 69.7 ± 2.3 beats/min to 62.3 ± 1.7 beats/min (p = 0.001) (i.e., by 9.1 ± 3.2, 10.8 ± 3.7, and 10.9 ± 3.1 beats/min, respectively). Linear regression analysis corroborated lack of association of HR reduction with BP reduction (r² = 0.017; p = 0.314). We next analyzed the relation of HR reduction and change in LV mass. Decrease in LVMI was present to a similar extent in all tertiles of baseline HR and HR reduction (Fig. 3B). Increase of E/A (Fig. 3C) and decrease of IVRT also occurred to a similar extent in all tertiles of baseline HR and HR reduction.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Improvements in LV Mass and Diastolic Function Are Unrelated to HR

(A) HR decreased significantly following renal denervation in all tertiles of HR at baseline but decreased more the higher the baseline HR was (use-dependent effect). (B) LVMI decreased similarly in all tertiles of HR. (C) E/A as a measure of diastolic function increased similarly in all tertiles. Abbreviations as in Figure 1.

SBP or HR reduction occurs independently of LVMI reduction or change in diastolic LV function

LVMI regression was present in all tertiles of LVMI at baseline but showed a use-dependent effect with stronger regression the higher the baseline LVMI was (p for trend < 0.001; data not shown). When the whole patient group was divided into tertiles according to LVMI regression, SBP decrease was similar across all tertiles of LVMI regression (decrease of 30.0 ± 5.4, 18.6 ± 6.0, and 18.8 ± 5.3 mm Hg, respectively; all p < 0.01) (Fig. 4A). We also observed HR reduction to be similar in all tertiles of LVMI reduction (Fig. 4B). Lack of correlation of HR and LVMI reduction was underlined in linear regression analysis (r = 0.008; p = 0.478). SBP and HR reduction occurred similarly in all tertiles of E/A increase (Figs. 4C and 4D). A calculation of 4 groups according to the presence of a decrease in BP and HR showed that LV mass regression was present in all groups independent of either BP or HR reduction (Fig. 5).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

BP and HR Reduction Are Decreased Independently of LV Mass and Diastolic Function

In tertiles of LVMI reduction, (A) SBP and (B) HR decreased to a similar extent. In tertiles of E/A change as a parameter of diastolic function, (C) SBP and (D) HR decreased to a similar extent. Abbreviations as in Figure 1.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

LV Mass Decreased Evenly in Patients With or Without BP or HR Reduction

After investigation of the 4 groups (n = 16 each) according to the decrease (or not) in BP and the decrease (or not) in HR, LV mass was reduced to the same extent in all groups. Abbreviations as in Figure 1.

Functional cardiovascular parameters improve depending on BP reduction

Vascular compliance increased from 1.05 ± 0.05 before RDN to 1.39 ± 0.06 ml/mm Hg at 6 months (p < 0.001). Improvement in compliance was not evident in the first tertile of BP reduction (1.18 ± 0.10 ml/mm Hg before, 1.15 ± 0.09 ml/mm Hg 6 months after RDN; p = 0.890), more pronounced in the second tertile (increase from 1.17 ± 0.08 ml/mm Hg to 1.55 ± 0.09 ml/mm Hg; p = 0.005), and most pronounced in the tertile with strongest BP decrease (compliance increase from 0.93 ± 0.06 ml/mm Hg to 1.40 ± 0.09 ml/mm Hg; p < 0.001). In linear regression analysis, increase in compliance was dependent on BP decrease (r² = 0.288; p < 0.001). VCFc was not affected by RDN (data not shown). EsMWS decreased from 66.9 ± 2.9 to 59.4 ± 2.2 × 10³ dynes/cm² (p = 0.042). Again, because of the dependency of the calculation of the parameter on BP, its decrease was present only in the tertile with greatest BP reduction (regression coefficient r² = 0.130; p = 0.013). Interestingly, compliance and EsMWS were not associated with HR reduction. For calculation of contractile function independently of afterload, VCFc was correlated with EsMWS (20,21). The y-axis intercept, representing intrinsic contractility at a hypothesized afterload of zero, increased mildly but significantly following RDN (from 0.746 ± 0.062 before to 0.820 ± 0067 after RDN; p < 0.001) (Online Fig. 2).

Study limitations

Magnetic resonance imaging is considered the gold standard for the detection of LV mass (36). However, echocardiography as a widely accessible and inexpensive tool is known to provide valid and reproducible LVH measurements (15). Our data on LVH regression are confirmed by unpublished but presented magnetic resonance imaging data (37). Nevertheless, the present data are still preliminary because they are derived from a small single-center patient population. One cannot exclude that the differences in office BP and LVMI reductions found herein were partially influenced by other factors, including a possible regression to the mean of office BP readings over repeated visits.

Accurate assessment of LV filling pressure in a noninvasive approach has limitations. The combination of tissue Doppler and the mitral annulus and mitral inflow velocity curves has been proven to yield the most valid results when compared with invasive measurements of LV end-diastolic pressures in left heart catheterization (38). The fact that despite improvement in diastolic function (reduction in left atrial volume, DT, IVRT, increase in E/A), E/E′ remained unchanged in our study is due to a strong rise in E wave. The latter can be explained by the HR decline induced by RDN because HR reduction shifts ventricular filling toward the early phase (E wave) (39). A relevant direct effect of the decrease in BP on E/A is unlikely due to the lack of correlation.