Cardiovascular outcomes improve in children with renovascular hypertension following endovascular and surgical interventions

This longitudinal observational retrospective cohort study was conducted at Great Ormond Street Hospital (GOSH) and included patients diagnosed with RenoVH over a 22-year period (2000–2022).

Ethics approval for this study was waived by our local centre as all patient information was pseudoanonymised during data collection and fully anonymised prior to analysis and was only accessed by members of the study team.

Our inclusion criteria were the following:

Patients were identified using our RenoVH database; each child diagnosed with RenoVH and treated at our centre is added to this database. At our centre, the indication for interventions was confirmed diagnosis of RenoVH and uncontrolled hypertension despite maximum dose of 2 or more antihypertensives. Endovascular interventions included percutaneous transluminal angioplasty, cutting balloon angioplasty, stenting or ethanol ablation. Further 27 participants in whom endovascular procedures could not achieve sustained BP control also underwent surgical interventions.

Data for each patient were sourced from the earliest available record until their last follow-up or end of data collection, up to June 2022. Variables that were lacking in electronic health record systems were noted as missing.

We collected five subsets of data for each patient at two time points: first recorded value (before intervention) and last recorded value (after intervention) in the following categories: patient demographics (age, sex, country of origin, underlying syndrome), antihypertensive drug(s), cardiovascular (systolic BP centile, daytime systolic BP load (percentage of readings greater than a threshold value) [2]), renal (surgical interventions, kidney function) and endovascular intervention (type, location).

The blood pressure data were collected through two methods at various time points. Systolic BP was always recorded from an office BP, as not all patients, particularly those who presented earlier during the study period, underwent ambulatory blood pressure measurement (ABPM), so this allowed the same variable to be collected for all participants. Patients who did undergo ABPM had their daytime systolic BP load recorded from their ABPM report pre- and post-intervention. At the time of all recorded pre-intervention blood pressure values, all patients were on antihypertensive therapy as all patients referred to our tertiary centre were already on pharmacological management for their hypertension. Raw systolic blood pressure values were then classified as age-, sex- and height-specific centiles [2].

We documented data from every first echocardiogram which was done at the first presentation pre-procedure and patients’ last echocardiograms which were always > 6 months post-endovascular intervention. This allowed enough time for a potential echocardiographic change to be evident and detectable post-intervention. At the time of the first echocardiogram, all children were receiving maximum tolerated antihypertensive therapy but had not undergone any endovascular or surgical procedures yet. All echocardiograms were re-reviewed offline by four cardiologists solely for the purpose of the study. The reviewers were blinded to the nature of the study and goals of the study as well as the identity and clinical status of the patients at the time of echocardiogram. They recorded measurements of IVSD, PWD, LVIDD and LVIDS for each echocardiogram. Aside from re-measuring all the parameters, they also re-calculated the z-scores. The absolute values were inputted into Boston Children’s z-score calculator with the patient’s age, height, weight and sex at the time of the correlating echocardiogram to produce the corresponding z-scores. The Boston Children’s z-score calculator uses data based on over 12 years of data collection from healthy children and is used to calculate z-scores for echocardiographic measures based on a child’s age, height, weight and sex. LVM, LVMI, RWT, LVEF and FS were also then calculated (as described above). Measurements were taken under the following confines: 2-dimensional view (if this was not available, then M-mode was used) and along the parasternal short axis (if this was not available, then the parasternal long axis was used). Z-scores (Boston Children’s) were calculated based on sex, weight and height. For consistency (pairwise comparison), it was ensured that the same view was used for all measurements of each individual patient. Unfortunately, diastolic functional assessment was not available and so was not reported in this study.

The primary outcome variables included the changes in cardiac dimensions and function on echocardiograms before and after endovascular + / − surgical interventions. The secondary outcomes included BP control, preservation of kidney function and number of prescribed antihypertensive medications.

All statistical analysis was carried out on IBM Statistical Package for Social Sciences (SPSS) software version 28.0.1.0 (Chicago, IL, USA) ⁸. Before statistical testing, normality assessments with Shapiro–Wilk test were performed in which p-values < 0.05 indicated normally distributed data.

Descriptive statistics summarised the variables within the dataset. To compare mean differences between subgroups at different time points (before and after intervention or between first and last recorded values), continuous data were analysed with the paired sample t-test for parametric data or the Wilcoxon matched-pair signed-rank tests for non-parametric data. Proportions were analysed by means of the chi-square tests. All results are reported with 95% confidence intervals, with a p-value < 0.05 indicating statistical significance. Multivariate linear regression was also carried out to investigate the possibility of any confounding relationships between blood pressure and antihypertensive medication with cardiac changes (changes in LVMI). The covariates included SBP centiles, age, presence of MAS, number of antihypertensive medications, types of antihypertensive medications, number of interventions and if the patient underwent surgical interventions.

An initial cohort of 172 children who were referred to our centre and underwent endovascular interventions/surgery were identified over the study period. However, 20 patients did not have digital records of their endovascular interventions, so these were excluded which resulted in a final cohort of 152 patients.

Over half the patients (n = 95, 62.5%) were male, and the median age at the time of referral was 6.9 years (IQR 3–9). The majority (n = 99, 65.1%) of patients resided in the UK at the time of the referral, with others residing in 20 other countries worldwide. Less than half (n = 43, 28.3%) of patients had an underlying diagnosis, with the most common being neurofibromatosis type 1 (NF1) (n = 22, 14.5%). Overall, 59 children (38.8%) presented with middle aortic syndrome (MAS). Almost half (n = 61, 42%) of patients had unilateral lesions, 74 (51%) had bilateral lesions and 9 patients (6%) had disease in accessory renal arteries but not the main renal arteries. See Supplementary File 1 for full patient demographics.

All patients had at least one endovascular intervention, and 27 participants (17.7%) also received a surgical intervention. The surgical interventions included unilateral (n = 15, 9%) and bilateral nephrectomies (n = 2, 1%), unilateral (n = 10, 6%) and bilateral auto-transplant (n = 1, 1%) and living donor (n = 1, 1%) and deceased donor (n = 4, 2%) kidney transplants. The median number of interventions performed in the study was 2 (IQR 1–4). The median overall follow-up time from the date of referral was 3.8 years, with the median time between first (pre-procedure) and last (post-procedure) echocardiogram being 2.6 years (IQR 1.1–6.7). At last follow-up, the mean patient age was 9.5 years, with 16 patients (11%) being older than 18 years at last follow-up. The last echocardiograms were at least 6 months post-procedure to allow enough time for cardiac remodelling to occur.

As displayed in Fig. 1, 139 patients (91.4%) had SBP readings above the 95th centile, and 129 patients (84.6%) had a SBP above the 99th centile at presentation. By last follow-up (median follow-up time between first and last BP was 3.8 years), the number of patients with SBP above the 95th and 99th centile significantly decreased to 69 patients (45.6%) and 30 patients (19.7%), respectively (p < 0.001). Overall, 68 patients (45%) had pre- and post-procedure ABPM measurements which allowed assessment of daytime average SBP load. The mean daytime average SBP load also significantly reduced post-intervention from 60.4 to 44.4 (p = 0.02).

Our centre’s approach to antihypertensive therapy involves calcium channel blockers first-line followed by beta-blockers, then alpha-blockers, then diuretics, and if patients still had uncontrolled hypertension, the use of ACE inhibitors and angiotensin-receptor blockers was carefully considered. The maximum dose of each drug was tried before the addition of another antihypertensive.

At referral to our centre, 140 patients (91.9%) were receiving antihypertensive therapy with 77 patients (50.7%) requiring 3 or more antihypertensives. This significantly reduced to 120 patients (78.7%) (p < 0.01) still needing antihypertensive therapy and 58 patients (38.2%) (p = 0.02) requiring 3 or more antihypertensive post-procedure. Overall, 70 patients (45.9%) had seen a reduction in the number of prescribed antihypertensive post-intervention. The differences in cardiac outcomes based on the type of antihypertensives used were not investigated in this study.

All first echocardiograms were carried out prior to any interventions, and all last echocardiograms were performed at least 6 months post-intervention with a median time interval of 2.6 years. LVMI significantly decreased by a mean of 18.3 g/m^2.7 from pre- to post-intervention (p < 0.01), and the number of patients with LVH also decreased from 99 patients (65.4%) to 80 patients (53.1%) although this was not statistically significant. RWT also significantly decreased by 0.04 post-intervention (p = 0.01). LVEF and FS improved post-intervention although not to a statistically significant degree. There was also a significant improvement in average IVSD and PWD (p = 0.01 and p < 0.01) as demonstrated by a reduction in z-scores. The mean z-score for LVIDD on last echocardiograms remained within normal limits but increased marginally by 0.08 (− 0.13 to − 0.05), as shown in Table 1.

Almost half of the patients (n = 60, 39.5%) had at least one abnormal echocardiographic parameter on either their first or last echocardiogram, and of those, 47 children (30.9%) had at least one severely abnormal parameter. As exhibited in Fig. 2, there was a trend of improvement in all cardiac parameters whereby wall thickness z-scores decreased, LVMI decreased and FS increased over time, indicating an echocardiographic improvement in those who presented with abnormal echocardiographic parameters.

A subgroup analysis focussing only on patients with at least one abnormal or severely abnormal echocardiographic parameter was performed, comparing values at baseline and at their last available follow-up. IVSD (p ≤ 0.04) and PWD (p ≤ 0.001) decreased significantly in both subgroups, as shown in Table 2. The proportion of patients with at least moderately impaired contractility (FS ≤ 20) fell from 7 patients (4.8%) to 4 patients (2.5%), and those with LVH fell from 99 patients (65.4%) to 80 patients (53.1%), but these were not significant (p = 0.36 and p = 0.08, respectively).

At presentation, only 47 patients (30.9%) had normal cardiac geometry which increased to 63 patients (41.7%) post-intervention. At presentation, 46 patients (30.1%) showed concentric hypertrophy which significantly decreased to 24 patients (15.5%) post-intervention (p = 0.03). Further details on cardiac remodelling can be seen in Table 3.

At our centre, 6 children (3.9%) presented with heart failure, as defined by a severely abnormal FS value (≤ 20%) on their first echocardiogram. All children presenting with heart failure had improved findings in all echocardiographic parameters (IVSD, PWD, LVIDD, LVMI, RWT, LVEF and FS) after intervention. However, one child remained in heart failure and had persistently abnormal echocardiographic parameters, while all other patients with resolving heart failure showed a disappearance of severely abnormal echocardiographic parameters post-intervention. The greatest reduction in number of children with abnormal echocardiographic parameters was seen in PWD in which there was a 60% reduction (5/6 pre-intervention to 2/6 post-intervention). Just one child (16.7%) remained with a severely abnormal FS compared to six pre-intervention.

At presentation, 5 of the children presenting with heart failure had SBP readings above the 99th centile. At last follow-up post-intervention, only one patient had SBP readings above the 99th centile. The number of antihypertensives also reduced post-intervention in these children; at referral, 4 children required 3 or more antihypertensives and post-procedure, only one child required 3 or more antihypertensives and the remaining children either required none or just one antihypertensive agent.

There was a reduction in the proportion of children with abnormal echocardiographic parameters after at least one surgical and endovascular intervention, and no children had severely abnormal cardiac measurements for IVSD, PWD, LVIDD, LVMI, RWT, FS or LVEF on their last echocardiogram.

At referral, the median number of medications children received was 3.0 (IQR 4.0–2.0), but at last follow-up, this decreased to a mean of 1.4 (SD ± 1.7).

Linear regression analysis was carried out to investigate the possibility of a confounding relationship between change in blood pressure centiles and cardiac dimensions. Other covariates included age, presence of MAS, number and type of antihypertensives and number of interventions. This produced a regression model with an R value of 0.73 (p = 0.001). This analysis found no significant relationship between LVMI and blood pressure; however, it did find that the children with MAS and children undergoing surgical interventions were more strongly associated with a reduction in LVMI post-intervention. Other covariates did not influence change in LVMI post-intervention. Further details can be seen in Table 4.

There was a 2.3% increase in mean eGFR across the cohort post-endovascular and/or surgical intervention. While this was not statistically significant, it does indicate the preservation of kidney function post-procedure.