Persistent Hypertension and Left Ventricular Hypertrophy After Repair of Native Coarctation of Aorta in Adults

Abstract

The purpose of this study was to describe procedural outcomes, hemodynamic improvement, regression of left ventricular mass (LV) hypertrophy and cardiovascular

The primary outcomes were procedural complications, re-interventions, and hemodynamic improvement after COA repair. The secondary outcomes were improvement in the severity of hypertension, regression of LV mass index (LVMI), and incidence of cardiovascular events (atrial fibrillation, ventricular tachycardia, heart failure hospitalization, and cardiovascular death) after COA repair. Secondary outcomes were assessed only in patients with isolated COA who had clinical and imaging follow-up at 1-year (Y1) and 3-years (Y3) post-intervention.

Of 172 patients that underwent COA repair (surgical 161; transcatheter 11), there were no procedural deaths, and all patients had residual COA gradient <20 mmHg.

Of 128 patients that met criteria for secondary outcomes assessment, 39 (36%) had a reduction in the intensity of antihypertension therapy, and cardiovascular events occurred in 16 (13%) patients. There was no significant reduction in the overall prevalence of hypertension (stage 1 and stage 2) over time (78% vs 70% vs 73%, p=0.4 at baseline, Y1 and Y3). Post-intervention hypertension (both stage 1 and 2) were independent risk factors for suboptimal LVMI regression and cardiovascular events.

Persistent hypertension was common after repair of native COA in adults, and was associated with suboptimal LVMI regression and cardiovascular events. These results suggest that optimal blood pressure control with medical therapy after COA repair may result in improved clinical outcomes.

INTRODUCTION

Several outcome studies have shown that coarctation of aorta (COA) repair in early childhood is associated with lower prevalence of hypertension and improved long-term survival.^1–3 As a result, COA repair is now routinely performed in infancy and early childhood, and the choice of surgical vs transcatheter stent therapy is based on factors such as aortic arch anatomy, associated left-sided valve lesions, body surface area, patient’s preference and institutional expertise.^{1, 4–7} However, some patients with COA go undetected for several decades, and a diagnosis may not be made until adulthood. As a result, such patients undergo primary repair of native COA in adulthood.^{8, 9} Although previous studies have estimated that COA account for less than 1% of adults of presenting with hypertension, the disease prevalence in likely underestimated because most providers may not routinely screen for COA.^{10, 11}

There have been several studies reporting long-term outcomes after COA repair, but in most of these studies, adults undergoing primary repair of native COA are often grouped together with patients of different age groups undergoing repair of native and recurrent COA.^1–3 Since adults presenting with native COA have a longer exposure to left ventricle (LV) pressure overload, and potentially have more advanced LV remodeling, we expect that the hemodynamic response and LV adaptation after COA repair will be different from that of patients that underwent COA repair in childhood. However, such data are sparse, and hence the current study. The purpose of this study was to describe the procedural outcomes, hemodynamic improvement, regression of LV hypertrophy and cardiovascular events in adults undergoing repair of native COA.

METHODS

The authors declare that all supporting data are available within the article [and its online supplementary files

Study population

This is a retrospective cohort study of adults (age ≥18 years) with native COA that received care at the adult congenital heart disease (ACHD) clinic at Mayo Clinic Rochester from January 1 2000 to December 31, 2018. The Mayo Clinic Institutional Review Board approved the study.

Outcomes

The procedural details for performing surgical and transcatheter COA repair at this institution have been extensively described.^{1, 12–14} The primary outcomes were procedural complications, re-interventions, and hemodynamic improvement after COA repair. Procedural complications were defined as aortic wall injury, femoral/access site hematoma, neurologic complications, and death prior to hospital discharge.^{1, 12–14} Hemodynamic improvement after COA repair was assessed as a change (Δ) in Doppler mean COA gradient, upper-to-lower extremity systolic blood pressure (ULE-SBP) gradient, and aortic isthmus ratio.¹⁵ The pre-intervention and post-intervention indices were based on clinical assessment, echocardiogram and cross-sectional imaging performed within 6 months prior COA repair, and within 6 months after COA repair respectively. Hemodynamically significant residual COA was defined as post-intervention Doppler mean COA gradient >20 mmHg, ULE-SBP gradient >20 mmHg, or aortic isthmus ratio <0.70.^{7, 15}

The secondary outcomes were improvement in the severity of hypertension, regression of LV mass index (LVMI), and incidence of cardiovascular events (atrial fibrillation, sustained and non-sustained ventricular tachycardia, heart failure hospitalization, and cardiovascular death) after COA repair. The assessment of secondary outcomes was restricted to patients with isolated COA (absence of hemodynamically significant LV inflow and outflow disease) that had serial clinical and imaging assessments at baseline (pre-intervention), and at 1 year (Y1) and 3 years (Y3) post-intervention. Hemodynamically significant LV inflow and outflow disease were defined as sub-valvular/valvular/supra-valvular aortic stenosis with mean gradient >20 mmHg, ≥moderate aortic regurgitation, or ≥moderate mitral regurgitation. The rationale for excluding these patients was to eliminate the confounding effect of these associated left-sided valve lesions on LVMI regression.

Hypertension was defined as having a diagnosis of hypertension requiring use of anti-hypertensive therapy prior to presentation at the Mayo ACHD clinic or having blood pressure (BP) ≥140/90 mmHg at two different settings at the time of presentation to the Mayo ACHD clinic. BP measurement was obtained from the right arm using automated BP cuff, and the recorded BP was an average of 5 measurements. Based on contemporary guidelines, we divided our cohort into 4 groups: normal BP (systolic BP <120 and diastolic BP <80 mmHg), elevated BP (systolic BP 120–129 and diastolic BP <80 mmHg), stage 1 hypertension (systolic BP 130–139 or diastolic BP 80–89 mmHg), and stage 2 hypertension (systolic BP ≥140 or diastolic BP ≥90 mmHg).¹⁶ Improvement in the severity of hypertension was assessed using 2 different metrics: (1) de-escalation of anti-hypertensive therapy defined as discontinuation or reduction in number/dose of antihypertensive medications at Y1; (2) decrease in the number of patients with systolic BP consistent with stage 1 or stage 2 hypertension at Y1.

LVMI was calculated using end-diastolic linear measurements of the ventricular septum, LV posterior wall thickness, and LV internal diameter derived from 2D echocardiography measured at the tissue-blood interphase.¹⁷ Regression of LVMI was calculated as %-change in LVMI ([LVMI pre-intervention – LVMI post-intervention]/LVMI pre-intervention x 100) at Y1 and Y3.¹⁸ The reproducibility of LVMI was assessed in 20 randomly selected patients. Intra- and interobserver agreement was evaluated after the same observer (J.W) and another observer (K.T) repeated the analysis using intraclass correlation coefficient (ICC) and 95% confidence interval. Test–retest reproducibility was assessed in 20 patients by measuring two different image sets.

We defined coronary artery disease as acute coronary syndrome (ST elevation myocardial infarction, non-ST elevation myocardial infarction or unstable angina), history of coronary revascularization (coronary artery bypass grafting [CABG] or percutaneous coronary intervention) or >50% stenosis in any vessel on invasive coronary or CT angiogram.¹⁹

Statistical analysis

Between-group differences were assessed with unpaired t-test, Wilcoxon ranked sum test, and Fisher’s exact test as appropriate. Multivariate linear regression and Cox regression analyses were used for assessment of risk factors associated with LVMI regression and cardiovascular events respectively. The regression models were constructed using stepwise forward selection of variables with a p<0.1 required for a variable to remain in the model. The variables were selected a priori based on known association with clinical outcomes, and were incorporated in the model in the following order: age, COA repair technique (surgery vs stent), pre-intervention LV global longitudinal strain, post-intervention Doppler mean COA gradient, and post-intervention Doppler mean aortic valve gradient. A p<0.05 was considered statistically significant. All statistical analyses were performed with JMP software (version 14.1.0; SAS Institute Inc, Cary NC) and GraphPad software (version 9.0.1; San Diego, CA).

RESULTS

Baseline data

COA diagnosis and referral

We identified 176 patients with native COA at the time of their first evaluation at the Mayo ACHD clinic. Of these 176 patients, COA diagnosis was made within 1 year in 116 (66%) patients, within 1–3 years in 52 (30%) patients, and within 3–5 years in 8 (4%) prior to presentation in the Mayo ACHD clinic. The initial COA diagnosis was made by echocardiogram in 146 (83%) patients, and the indications for echocardiogram were for hypertension that was unresponsive to medical therapy (n=108), abnormal physical exam findings (n=7), atypical chest pain/dyspnea (n=22), new-onset atrial fibrillation (n=4), family history of congenital heart disease (n=15). COA diagnosis was made by cardiac CT in 24 (14%), and indications for cardiac CT were for evaluation of chest pain in the emergency department (n=9), and preoperative evaluation for aortic valve replacement (n=15). In 6 (3%) patients, COA diagnosis was made because of difficulty advancing the catheter across the aortic isthmus during preoperative coronary angiogram for aortic valve replacement. COA diagnosis was known in all patients prior to presentation to the Mayo ACHD clinic, and of the 176 patients, 142 (81%) were referred by their local physician for COA repair while 34 (19%) were self-referred for a second opinion. Of the 176 patients, 172 (98%) underwent COA repair (Supplementary Figure S1), and the baseline (pre-intervention) clinical and echocardiographic characteristics of the 172 patients that underwent COA repair are shown in Table 1. There were 13 (8%) patients with coronary artery disease, and the diagnosis of coronary artery disease was made by cardiac CT (n=10) and preoperative coronary angiogram (n=3).

Table 1:

Baseline Characteristics

Variables	All	Male (n=101)	Female (n=71)	p
Age, years	38 (27–48)	36 (25–46)	39 (27–48)	0.4
Bicuspid aortic valve	98 (57%)	62 (61%)	36 (51%)	0.02
Body mass index, kg/m2	28±5	29±5	27±4	0.3
Creatinine, mg/dl	1.22±0.38	1.28±0.31	1.17±0.29	0.1
NT-proBNP	268 (66–421)	246 (51–316)	271 (68–421)	0.4
Gynecologic data
Females	71	---	71	---
Premenopausal	57	---	57	---
Postmenopausal	3	---	3	---
Menopausal data not available	11	---	11	---
Patients on contraceptive	31	---	31	---
Patients on estrogen contraceptive	2	---	2	---
BP and HR
Systolic blood pressure, mmHg	149±26	154±24	142±21	0.03
Diastolic blood pressure, mmHg	89±15	92±12	87±13	0.06
Heart rate, bpm	76±9	74±8	77±9	0.5
CV risk factors and disease
Hypertension	127 (74%)	78 (77%)	49 (69%)	0.1
Coronary artery disease	13 (8%)	9 (9%)	4 (6%)	0.6
Diabetes	7 (4%)	4 (4%)	3 (4%)	0.9
Hyperlipidemia	29 (17%)	18 (17%)	11 (16%)	0.6
Atrial fibrillation	4 (2%)	2 (2%)	2 (3%)	0.8
Lipid profile
Total cholesterol	176±33	179±29	171±26	0.7
HDL cholesterol	46±11	44±10	49±12	0.6
LDL cholesterol	89±8	91±7	87±12	0.7
Family history of coronary artery disease	11 (6%)	5 (5%)	6 (9%)	0.4
Patient on lipid lowering therapy	26 (15%)	15 (15%)	11 (16%)	0.5
Anti-hypertensive therapy
Thiazide diuretics	25 (15%)	14 (14%)	11 (16%)	0.7
Beta blockers	69 (40%)	42 (42%)	27 (38%)	0.009
ACEI/ARB	65 (38%)	36 (36%)	29 (41%)	0.3
Calcium channel blockers	47 (27%)	29 (29%)	18 (25%)	0.5
Hydralazine	8 (5%)	4 (4%)	4 (6%)	0.3
Echocardiography
LV end-diastolic volume index, ml/m2	65±28	67±24	63±21	0.04
LV end-systolic volume index, ml/m2	25±12	27±10	24±8	0.01
LV ejection fraction, %	62±9	60±7	65±5	0.008
LV mass index, g/m2	136±32	141±32	129±32	<0.001
Concentric LV hypertrophy	96 (55%)	58 (57%)	38 (54%)	0.1
Eccentric LV hypertrophy	43 (24%)	28 (28%)	15 (21%)	0.3
Aortic valve mean gradient, mmHg	9 (7–20)	11 (8–22)	7 (7–6)	0.06
COA Doppler mean gradient, mmHg	31±12	32±10	30±8	0.7
Cross-sectional imaging
Sinus of Valsalva, mm	35±6	36±5	35±4	0.4
Mid ascending aorta, mm	12±5	13±5	12±4	0.1
Proximal arch, mm	20±3	21±3	20±4	0.2
Distal arch, mm	18±3	18±4	18±3	0.5
Isthmus, mm	9±3	9±4	9±3	0.5
Descending aorta at diaphragm, mm	17±2	18±2	16±2	0.06
Aortic isthmus ratio	0.66±0.15	0.65±0.13	0.67±0.11	0.03

ACEI: Angiotensin converting enzyme inhibitor; ARB: Angiotensin-II receptor blockers; LV: Left ventricle; COA: Coarctation of aorta.

^*Thoracic aorta dimensions were based on cross-section imaging data.

Surgical and transcatheter COA Repair [n=172]

Procedural details

Of the 172 patients that underwent COA repair, 11 (6%) patients had transcatheter stent therapy, of which 10 patients received covered Cheatham Platinum stents (29, 34, 38 mm), and 1 patient received an eV3 Intrastent LD Max open-cell stent (36 mm). The other 161 (94%) patients underwent surgical COA repair using the following techniques: interposition graft (n=76), patch aortoplasty +/− reverse hemi-arch repair (n=56), ascending-to-descending aorta bypass graft (n=31), and extended end-to-end anastomosis (n=9). Other concomitant surgical procedures performed at the time of COA repair were aortic valve replacement (n=31), aortic root/ascending aorta replacement (n=18) resection of subaortic stenosis (n=8), and coronary bypass grafting (n=8). The decision to perform surgical vs transcatheter stent therapy was based on shared decision making between the patient, cardiologist, and surgeon. The patients that underwent surgery had smaller proximal arch dimension (19±4 vs 23±3 mm, p=0.02) and distal arch dimension (17±3 vs 20±3 mm, p=0.04), and were more likely to have concomitant mitral valve and LV outflow tract disease. There were no procedural complications in either group.

Hemodynamic response and reinterventions

Compared to pre-intervention COA indices, the Doppler mean gradient decrease by 24±8 mmHg, ULE-SBP decreased by 28 (17–31) mmHg, and aortic isthmus ratio increased by 0.21±0.06 (Table 2). None of the patients had hemodynamically significant residual coarctation. There was one reintervention in the transcatheter stent therapy group, and this involved re-dilation of a 36 mm EV3 Max LD stent 12 months post-implantation in a 47-year-old male. There were no reinterventions in the surgical group.

Table 2:

Postoperative Hemodynamic Changes After COA Repair (n=172)

Variables	All (n=172)	Surgery (n=161)	Stent (n=11)	p
COA mean gradient (pre-intervention), mmHg	31±12	32±11	31±8	0.7
COA mean gradient (post-intervention), mmHg	8±4	7±4	10±3	0.02
Δ mean gradient, mmHg	24±8	25±7	22±5	0.04
*ULE-SBP gradient (pre-intervention), mmHg	33 (18–42)	33 (17–42)	32 (20–39)	0.7
*ULE-SBP gradient (post-intervention), mmHg	5 (0–8)	5 (0–9)	5 (0–8)	0.9
Δ ULE-SBP gradient, mmHg	28 (17–31)	28 (17–33)	27 (18–32)	0.6
Aortic isthmus ratio (pre-intervention)	0.66±0.15	0.66±0.13	0.65±0.09	0.5
Aortic isthmus ratio (post-intervention)	0.87±0.11	0.87±0.10	0.84±0.12	0.1
Δ Aortic isthmus ratio	0.21±0.06	0.21±0.06	0.20±0.07	0.3

COA: Coarctation of aorta; ULE-SBP: Upper-to-lower extremity systolic blood pressure; Δ denotes difference between pre- and post-intervention indices;

^*ULE-SBP gradient data (pre- and post-intervention) were available in 155 patients.

Hypertension, regression of LV hypertrophy and cardiovascular events [n=128]

Of the 172 patients that underwent COA repair, 128 (74%) met the inclusion criteria for the analyses of secondary outcomes (Supplementary Figure S1).

Hypertension

Of the 108 patients that were on anti-hypertensive therapy prior to COA repair, 39 (36%) patients had de-escalation of antihypertensive therapy at Y1 (reduction in the number/dose of anti-hypertensive medications [n=21] and discontinuation of anti-hypertensive therapy [n=18]), resulting in a reduction in the number of patients on anti-hypertensive therapy from 108 (84%) pre-intervention to 90 (70%) at Y1 (Table 3). Of the 39 patients that had de-escalation of anti-hypertensive therapy, there were 8 (20%) with normal BP, 16 (41%) with elevated BP, and 15 (39%) with stage 1 hypertension.

Table 3:

Postoperative Changes in Hypertension Severity (n=128)

	Baseline		Y1		Y3
Variables	Male (n=77)	Female (n=51)	Male (n=77)	Female (n=51)	Male (n=77)	Female (n=51)
Blood pressure data
Systolic blood pressure, mmHg	148±21	139±22	138±16	133±14	134±21	132±23
Diastolic blood pressure, mmHg	91±14	83±13	80±13	76±11	78±14	75±15
Hypertension severity group
Normal blood pressure	0	1 (2%)	5 (7%)	3 (6%)	5 (7%)	3 (6%)
Elevated blood pressure	15 (20%)	8 (16%)	14 (18%)	9 (18%)	14 (18%)	12 (24%)
Stage 1 hypertension	19 (25%)	14 (28%)	46 (60%)	26 (51%)	46 (60%)	29 (57%)
Stage 2 hypertension	46 (60%)	25 (49%)*	16 (21%)	9 (18%)	19 (15%)	19 (15%)
Intensity of therapy
Any anti-hypertensive medication	68 (88%)	40 (78%)*	55 (71%)	35 (69%)	54 (70%)	40 (78%)
1 anti-hypertensive medication	36 (47%)	22 (43%)	29 (38%)	20 (39%)	31 (40%)	23 (45%)
2 anti-hypertensive medication	27 (35%)	15 (29%)*	26 (34%)	15 (29%)	25 (33%)	15 (29%)
≥3 anti-hypertensive medications	5 (7%)	3 (6%)	---	---	---	---
Anti-hypertensive therapy
Thiazide diuretics	12 (16%)	10 (20%)	14 (14%)	8 (16%)	14 (14%)	8 (16%)
Beta blockers	37 (48%)	21 (41%)	16 (21%)	12 (24%)	16 (21%)	11 (20%)
ACEI/ARB	36 (47%)	17 (33%)*	26 (33%)	16 (31%)	26 (34%)	15 (30%)
Calcium channel blockers	25 (33%)	13 (26%)	12 (16%)	7 (14%)	13 (19%)	10 (20%)
Hydralazine	2 (3%)	2 (4%)	---	---	---	---

: Angiotensin converting enzyme inhibitor; ARB: Angiotensin-II receptor blockers;

^*signifies p value <0.05

Although there was a temporal reduction in the prevalence of stage 2 hypertension and corresponding increase in the prevalence of stage 1 hypertension (Table 3), the overall prevalence of hypertension (stage 1 and stage 2) did not change over time (78% vs 70% vs 73%, p=0.4 at baseline, Y1 and Y3 respectively). The post-intervention residual Doppler mean COA gradient was similar across the 4 hypertension severity groups (7±4 vs 8±4 vs 6±2 vs 7±3 mmHg for normal BP, elevated BP, stage 1 hypertension and stage 2 hypertension respectively, p=0.7), demonstrating that the difference in systolic BP was not due to residual COA.

Regression of LVMI

At baseline, the mean LVMI was 134±33 g/m² for all patients; 141±27 g/m² for males and 128±22 g/m² for females. There was excellent intraobserver (ICC 0.92 [0.88–0.97]), interobserver (ICC 0.87 [0.82–0.92]), and test-retest (ICC 0.90 [0.86–0.94]), reproducibility for the assessment of LVMI.

Among the 128 patients that met the inclusion criteria for the analyses of secondary outcomes, the mean %-change in LVMI was 9±3% and 13±4% at Y1 and Y3 respectively, leading to a temporal reduction in LVMI from 136±25 g/m² (pre-intervention) to 122±21 g/m² (Y1) to 119 ±17 g/m² (Y3), p=0.01). The patients with history of hypertension pre-intervention had less robust regression of LVMI at Y1 and Y3 (Figure 1). Similarly, the patients with stage 1 and stage 2 hypertension also had less robust regression of LVMI at Y1 and Y3 as compared to those with normal/elevated BP (Figure 1). The determinants of LVMI regression were older age at time of presentation and post-intervention hypertension (Table 4), but we did not observe any association between LVMI regression and type of COA intervention or residual COA (Table 4). We did not observe any correlation between LVMI regression and the type of antihypertensive therapy used pre- or post-therapy. Subgroup analysis performed based on sex, showed that both groups age and hypertension were independent risk factors for LVMI regression in both groups.

Figure 1.

(Top): Between-group comparison of %change in left ventricular mass index (LVMI) regression at Y1 and Y3 (A) and cardiovascular events (B) based on hypertension status pre-intervention. Patients with history of hypertension (red) had less LVMI regression and more cardiovascular events as compared to patients without history of hypertension (blue)

(Bottom): Between-group comparison of %change in LVMI regression at Y1 and Y3 (A) and cardiovascular events (B) based on hypertension status at 1 year post-intervention. Patients with stage 2 hypertension (red) and stage 1 hypertension (black) had less LVMI regression and more cardiovascular events as compared to patients with normal/elevated blood pressure (blue) Pairwise comparison of LVMI regression at Y1 (*), LVMI regression at Y3 (**), and cardiovascular events (***) showed no significant differences between patients with stage 2 vs stage 1 hypertension

Table 4:

Multivariate Linear Regression Model Showing Determinants of LVMI Regression

Variables	β coefficient ± SE	p
Age, per 5y increment	−0.28±0.14	0.02
Male sex	0.18±0.24	0.1
Stage 1 hypertension at Y1*	−0.31±0.18	0.02
Stage 2 hypertension at Y1*	−0.38±0.21	0.009
Doppler COA mean gradient	0.16±0.43	0.5
Doppler aortic valve mean gradient	−0.08±0.14	0.1
Transcatheter stent therapy	0.23±0.35	0.3

COA: Coarctation of aorta; SE: standard error; LVMI: Left ventricular mass index

Severity of hypertension at Y1 modeled as categorical variable using the normal/elevated BP group as the reference group.

A diagnosis of hypertension pre-intervention was also independently associated with regression of LVMI (β coefficient ± SE: −0.49±0.23, p0.005). However, we did not input both pre-intervention hypertension and post-intervention hypertension in the same model because of collinearity

Cardiovascular events

The median follow-up was 8±4 years, and during this period, 9 (7%) had new onset atrial fibrillation, 6 (5%) had non-sustained ventricular tachycardia and 4 (3%) required hospitalization for heart failure in the context of atrial arrhythmias, and 3 (2%) died from cardiovascular causes (sudden cardiac death n=1, stroke-related death n=2). The combined outcome of cardiovascular events occurred in 16 (13%) patients. The 10-year cumulative incidence of cardiovascular events was higher in the patients with hypertension pre-intervention as compared to those without hypertension (14% vs 0%, p<0.001). Compared to group with normal/elevated BP, those with stage 1 and 2 hypertension had a higher cumulative 10-year incidence of cardiovascular events (Figure 1). Persistent hypertension post-intervention and suboptimal regression of LVMI were independent risk factors for cardiovascular events (Table 5). We did not observe any correlation between cardiovascular events and the type of antihypertensive therapy used pre- or post-therapy. Subgroup analysis performed based on sex, showed that hypertension and suboptimal LV mass regression were independent risk factors for cardiovascular events in both groups.

Table 5.

Multivariate Cox Model Showing Risk Factors Associated with Cardiovascular Events

Variables	HR (95%CI)	p
Age, per 5y increment	1.16 (1.05–1.27)	0.003
Stage 1 hypertension at Y1*	1.19 (1.08–1.30)	0.007
Stage 2 hypertension at Y1*	1.27 (1.06–1.48)	0.01
LV mass index regression, per 5% change	0.89 (0.84–0.95)	0.002
LV global longitudinal strain, %	0.97 (0.93–1.01)	0.08

COA: Coarctation of aorta; LV: Left ventricle; HR: Hazard ratio; CI: Confidence interval

Severity of hypertension at Y1 modeled as categorical variable using the normal/elevated BP group as the reference group.

A diagnosis of hypertension pre-intervention was also independently associated with regression of LVMI (β coefficient ± SE: −0.49±0.23, p0.005). However, we did not input both pre-intervention hypertension and post-intervention hypertension in the same model because of collinearity

DISCUSSION

COA repair is routinely performed in infancy and early childhood in the modern era.^{5, 6} However, some patients with COA can go undetected for several decades resulting in the need for a primary repair of native COA in adulthood. We reviewed outcomes in 172 adult patients that underwent repair of native COA, and these are the main finding: (1) COA repair in adulthood was safe, effective, and associated with low risk of intervention; (2) Hypertension was common prior to COA repair, persisted in more than 70% of the patients after COA repair, and was associated with suboptimal regression of LVMI; (3) Hypertension and sub-optimal LVMI regression were independent risk factors for cardiovascular events during follow-up.

The excellent hemodynamic result, low risk of procedural complications and reintervention rate observed in the current study is similar to outcomes reported in a previous study of 52 adult patients that underwent transcatheter COA repair for native COA, even though the majority of our patients underwent surgical repair which more invasive.⁹ We postulate that excellent procedural outcomes in this series may be due to appropriate and individualized selection of COA repair techniques, identifying the procedure best suited for the thoracic aorta anatomy of each patients, improvement in procedural techniques over time (all interventions were performed within the last 2 decades), and institutional expertise (procedures were performed in a referral center with high case volume). These factors have been shown to be associated with good procedural outcomes.^{1, 5, 6}

Persistent hypertension after COA repair is a well-recognized entity, and occurs in 44–60% of adult patients with prior COA repair.^{3, 20, 21} The etiology of persistent hypertension after COA repair is due to a combination of endothelial dysfunction, abnormal arterial smooth muscle reactivity, changes in material properties of the aorta leading to increased arterial stiffness, and reduced baroreceptor sensitivity leading to increased sympathetic activation.^22–24 The prevalence of hypertension pre- and post-intervention observed is our cohort is significantly higher that estimates for prior studies, and we postulate that this may be related to the older age at time of COA repair in the current study.^{1, 14, 23} Consistent with our postulation, de Divitiis et al previously demonstrated an inverse relationship between age at the time of COA repair and the extent of persistent endothelial dysfunction, arterial stiffness and impaired conduit function, which in turn results in persistent hypertension even after a successful COA repair.²³

Increased LVMI is a physiologic response to LV pressure or volume overload, and to normalized wall stress.^{18, 25} Regression of LVMI typically occurs following interventions that normalize loading conditions, and the extent of LVMI regression is a reflection of the adequacy of ‘LV unloading’.^{18, 25} In the current study, although all patients had effective LV unloading based the low post-intervention residual COA gradients, we observed that patients with hypertension (even stage 1 hypertension) had less regression of LVMI suggesting a significant ongoing LV pressure overload. Chronic LV pressure overload from hypertension has been shown to cause increased LV stiffness, myocardial fibrosis, and LV systolic/diastolic dysfunction.²⁵ These pathologic changes ultimately manifest clinically as atrial fibrillation and heart failure (from diastolic dysfunction and left atrial remodeling), and ventricular tachycardia (from scars and fibrosis of LV myocardium) as observed in the current study.^26–28 These finding have important clinical implications as discussed below.

Clinical implications and future directions

Contemporary guidelines do no recommend anti-hypertensive therapy in patients with stage 1 hypertension in the absence of associated atherosclerotic cardiovascular disease risk factors.¹⁶ As a result, antihypertensive therapy is typically discontinued after COA repair in patients systolic BP less than the guideline-directed threshold for medical therapy (systolic BP >140 mmHg).^{1, 4, 9} The consistent relationship between stage 1 hypertension, LVMI regression and cardiovascular events in the current study, challenges this paradigm, and suggest that intensification of anti-hypertensive therapy (rather than weaning off anti-hypertensive therapy) may be the right management strategy after COA repair in this population. Consistent with this postulate, data from the SPRINT trial demonstrated a survival benefit in patients receiving intensive antihypertensive therapy with systolic BP target <130mmHg as compared to systolic BP target of 130–139 mmHg.²⁹ While we cannot directly extrapolate data from the SPRINT trial to the COA population, the results of the current study provide the scientific premise for clinical trials to determine the optimal systolic BP target that will provide the most survival benefit with the least adverse effects.

The increase in cardiovascular risk observed in our study is likely related to arterial stiffness and other unique hemodynamic factors present in patients with COA. Each time the LV ejects blood, there is an incident (forward) wave that is transmitted by the conduit arteries, and then reflected back at multiple sites of impedance mismatch throughout the arterial system.^{25, 28} The timing of arrival of the reflected waves to the proximal aorta depends on aortic stiffness of the conduit arteries which transmits both the forward and backwards traveling waves. In a healthy aorta, pulse wave velocity is slow, and the reflected wave arrives at the proximal aorta in diastole, thus producing an augmentation central diastolic BP [DBP] without significant increase in central systolic BP [SBP]. In a stiff aorta, on the other hand, the reflected waves arrive at the proximal aorta in late systole producing augmentation of central SBP (LV afterload) and loss of augmentation of central DBP (coronary perfusion pressure).²⁵ In addition to having increased aortic stiffness, COA patients develop fibrotic changes at the aortic isthmus (site of COA repair), and this area of impedance mismatch creates foci of wave reflection,³⁰ and the proximity of the site of wave reflection leads to arrival of the reflected wave much earlier in the cardiac cycle as compared to non-COA patients with similar degree of arterial stiffness.³⁰ This further exacerbates central aortic hemodynamic abnormalities in this population.

Limitations

This is a retrospective study of COA patients referred to a single tertiary center, and hence it is prone to selection and ascertainment bias. We relied on office BP measurements rather that home or ambulatory BP measurement, which are the preferred methods for longitudinal BP monitoring. We did not observe any correlation between clinical outcomes and type of antihypertensive therapy used, but we had no of verifying whether the patients were compliant with their medical therapy because of the retrospective study design. These factors are potential confounders, and hence the need for prospective studies to bridge the knowledge gaps highlighted above.

Conclusions

Primary COA repair in adults with native COA in safe and effective for elimination of aortic isthmus stenosis. However, persistent hypertension was common even after a successful repair. Hypertension (including stage 1 hypertension) was associated with suboptimal regression of LVMI and cardiovascular events. These results are concerning and highlight the importance of early COA diagnosis and repair, as well as close follow-up after COA repair. While we may not always be able to influence the age of presentation of patients with native COA, the results of this study suggest that optimal BP control with medical therapy after COA repair may be associated with improve clinical outcomes.

PERSPECTIVES.

There are limited outcome data about adults undergoing repair of native COA. The purpose of this study was to describe the procedural outcomes, hemodynamic improvement, regression of LV hypertrophy and cardiovascular events in adults undergoing repair of native COA. We observed that persistent hypertension was common after repair of native COA in adults, and was associated with suboptimal LVMI regression and cardiovascular events. These results suggest that optimal blood pressure control with medical therapy after COA repair may result in improved clinical outcomes.

NOVELTY AND SIGNIFICANCE.

What is new?

Persistent hypertension was common after repair of native COA in adults, and was associated with suboptimal LVMI regression and cardiovascular events.

What is relevant?

Diagnosis and treatment of hypertension after COA will potentially improve clinical outcomes.

Summary:

These results are concerning and highlight the importance of early COA diagnosis and repair, as well as close follow-up after COA repair. While we may not always be able to influence the age of presentation of patients with native COA, the results of this study suggest that optimal BP control with medical therapy after COA repair may be associated with improve clinical outcomes.

Abbreviations:

COA

Coarctation of aorta

LV

Left ventricle

ACHD

Adult congenital heart disease

ULE-SBP

Upper-to-lower extremity systolic blood pressure

BP

Blood pressure

LVMI

LV mass index

ICC

Intraclass correlation coefficient