Comprehensive First‐Line Magnetic Resonance Imaging in Hypertension: Experience From a Single‐Center Tertiary Referral Clinic

Abstract

European guidelines recommend that patients with hypertension be assessed for asymptomatic organ damage and secondary causes. The authors propose that a single magnetic resonance imaging (MRI) scan can provide comprehensive first‐line imaging of patients assessed via a specialist hypertension clinic. A total of 200 patients (56% male, aged 51±15 years, office BP 168±30/96±16 mm Hg) underwent MRI of the heart, kidneys, renal arteries, adrenals and aorta. Comparisons were made with other imaging modalities where available. A total of 61% had left ventricular hypertrophy (LVH), 14% had reduced ejection fraction, and 15 patients had myocardial infarcts. Echocardiography overdiagnosed LVH in 15% of patients and missed LVH in 14%. Secondary causes were identified in 14.5% of patients: 12 adrenal masses, 10 renal artery stenoses, seven thyroid abnormalities, one aortic coarctation, one enlarged pituitary gland, one polycystic kidney disease, and one renal coloboma syndrome. This comprehensive MRI protocol is an effective method of screening for asymptomatic organ damage and secondary causes of hypertension.

Hypertension is a global health problem. At least a quarter of the adult population has high blood pressure (BP) and the world‐wide prevalence is predicted to rise to 1.56 billion by 2025.1 The 2013 European Society of Hypertension/European Society of Cardiology guidelines for the management of arterial hypertension recommend that all patients with high BP are investigated for asymptomatic target organ damage (TOD) to enable evaluation of their cardiovascular risk and thus guide the initiation of appropriate treatment.2 Initial assessment with a 12‐lead electrocardiogram (ECG) and basic biochemical tests is advised as standard for all patients, with guidance that additional testing should be performed if indicated by history or examination findings.2 Investigation for secondary causes of hypertension is recommended where there is clinical suspicion.2 In the United Kingdom, the 2011 National Institute for Health and Care Excellence guideline on the diagnosis and management of hypertension in adults also recommends additional investigation in patients at higher risk for secondary hypertension, specifically patients with young‐onset hypertension (yHTN, <40 years), treatment‐resistant hypertension, rapidly worsening or accelerated hypertension, or in those with a history or examination findings that might indicate a secondary cause.3

In this study, we aimed to establish the utility of magnetic resonance imaging (MRI) as a screening tool in hypertension in patients assessed via a specialist hypertension clinic. The European guidelines acknowledge the value of cardiac MRI (CMR) but recommend echocardiography and peripheral arterial/abdominal ultrasound as first‐line imaging modalities.2 However, as MRI becomes increasingly available, it is likely to prove a useful tool in the assessment of patients. For example, CMR is widely regarded as the gold‐standard noninvasive imaging modality for the clinical assessment of left ventricular (LV) mass (LVM) and LV hypertrophy (LVH).4 In addition, MRI techniques can be used to image the vasculature (including the renal arteries), renal parenchyma and adrenal glands with high accuracy.5, 6 MRI could improve diagnostic efficiency in these patients as it is a highly sensitive imaging modality and would avoid the need for multiple appointments for echocardiography, renal ultrasound, and, potentially, vascular computed tomography (CT) angiography.

We propose that a single MRI scan could provide all of the routine imaging required for the evaluation of both TOD and the identification of potential secondary causes in patients in the context of a tertiary hypertension service.

Methods

Study Population

This was a retrospective analysis of 200 consecutive patients assessed with a hypertension protocol MRI scan from a prospectively gathered clinical database of patients investigated at a tertiary hypertension clinic (Bristol Heart Institute, University Hospitals Bristol NHS Foundation Trust). Eligible study participants were those referred for MRI as part of their standard clinical hypertension workup between November 2010 and July 2015, and included patients with yHTN (onset <40 years), drug‐resistant hypertension (rHTN, BP >140/90 mm Hg despite three or more antihypertensive medications), uncontrolled hypertension (ucHTN, BP >140/90 mm Hg on <3 antihypertensive medications), and other difficult‐to‐treat hypertension (eg accelerated hypertension, highly labile hypertension, or hypertension with disproportionate TOD).

Baseline demographics and clinical characteristics were recorded. Patient height and weight were measured with calculation of body mass index (BMI). A BMI ≥30 kg/m² was defined as obesity.7 Office systolic BP (SBP) and diastolic BP (DBP) were measured in the seated position using a standard oscillometric device and appropriately sized cuff, taking the mean of at least two repeat readings, where available.

To assess the reliability of our MRI findings, results were compared with the clinical reports from previous echocardiographic, renal ultrasound, or renal CT angiographic studies performed as part of each participant's hypertension assessment. Likewise, results from CT scans performed as part of any additional evaluation for patients with positive findings on MRI were also reviewed.

The local research ethics committee confirmed that the study conformed to the governance arrangements for research ethics committees. The study was conducted with the patients' written consent and was performed in accordance with the Declaration of Helsinki.

Protocol

Images were acquired from the level of the Circle of Willis to the level of the femoral heads.

CMR was performed at 1.5 Tesla (Avanto, Siemens, Erlangen, Germany). Steady‐state free‐precession short‐axis whole LV cines (8‐mm slice thickness, no slice gap, temporal resolution 38.1 ms, echo time 1.07 ms, representative field‐of‐view in‐plane pixel size 1.5×0.8 mm) were used for the estimation of LVM and LV volumes. Myocardial replacement fibrosis was assessed by late gadolinium enhancement (LGE).8 An inversion‐recovery fast gradient‐echo sequence was performed 10 to 15 minutes after intravenous administration of 0.1 mmol/kg gadobutrol (Gadovist, Bayer Pharma AG, Germany), in two phase‐encoding directions where there was potential artefact. Tailored inversion times were used in each patient to null normal reference myocardium.

Renovascular assessment consisted of Time‐Resolved Angiography With Interleaved Stochastic Trajectories (TWIST) contrast‐enhanced magnetic resonance angiography (MRA), which creates multiphase, multiplanar images of the thoracic and abdominal vasculature; angiography was analyzed using multiplanar reformatting post‐processing software (cmr42; Circle Cardiovascular Imaging Inc., Calgary, AB, Canada). Axial T1‐weighted images through the abdomen and pelvis with 5‐mm slice thickness were also performed.

CMR Analysis

The assessment of LV volumes and LVM were performed as previously described.9 Briefly, endocardial contours were defined at end diastole and end systole on the LV short‐axis stack using blood pool/endocardial border threshold detection software (cmr42; Circle Cardiovascular Imaging Inc.), which has been previously validated.10 Epicardial contours were defined manually at end diastole. LVM was estimated by multiplying the total myocardial volume, including papillary muscles and LV trabeculations (equivalent to LV dry weight), by 1.05 g/mL, which is the specific gravity of myocardium, as previously described.9 LVM was indexed to body surface area, calculated using the Mosteller formula. Ejection fraction was calculated from the end‐diastolic and end‐systolic endocardial volumes, and long‐axis function was assessed using mitral and tricuspid annular plane systolic excursion.

LVH was defined as indexed LVM >95th percentile of established CMR reference ranges indexed to body surface area (men: <35 years, >87 g/m²; ≥35 years, >78 g/m² and women: <35 years, >71 g/m²; ≥35 years, >70 g/m²).11 LV remodeling was defined as a ventricle with normal indexed LVM but elevated LVM/volume ratio (M/V).12 An increased M/V was defined as >95th sex‐specific percentile (men: >1.12 g/mL and women:>1.14 g/mL) from healthy volunteers, as previously described.12 The presence of LGE was quantified by visual analysis.

Statistical Analysis

Statistical analysis was performed using GraphPad Prism (La Jolla, CA). All data are presented as mean±standard deviation. Chi‐square test was used to compare categorical data. Kruskal‐Wallis test was used to compare nonparametric data, with Dunn's multiple comparison test between groups. Significance was taken as P<.05.

Results

Demographic Data

Data from 200 consecutive patients from our specialist hypertension clinic who underwent MRI as the imaging component of their assessment for TOD and potential secondary causes of hypertension are presented. A total of 38% of patients had rHTN, 32% had yHTN, 18% had ucHTN, and 12% had other difficult‐to‐treat hypertension. Patient demographic data are shown in Table 1. There were no adverse events as a result of the MRI scans; however, one scan was abandoned because of claustrophobia. All patients except one (renal coloboma syndrome) had an estimated glomerular filtration rate of >30 mL/min/1.73 m².

Table 1.

Patient Demographics by Hypertension Subgroup

Parameter	All Patients (N=200)	Patient Hypertension Subgroups
		Drug‐Resistant (n=76)	Young‐Onset (n=64)	Uncontrolled (n=35)	Othera (n=25)
Age, y	51±15	58±10	35±9	65±9	53±10
Male sex, No. (%)	115 (56)	46 (61)	37 (58)	17 (49)	15 (60)
BMI, kg/m2	30±6	31±5	29±7	29±5	31±6
No. of antihypertensives	3.1±2.0	4.8±1.5	1.9±1.5	2.0±1.3	2.2±1.3
SBP, mm Hg	168±30	178±29	152±22	187±27	154±23
DBP, mm Hg	96±16	99±17	94±11	100±16	91±16
LVM, g/m2	87±28	96±31	78±24	83±21	87±29

Abbreviations: BMI, body mass index; DBP, diastolic blood pressure; LVM, left ventricular mass indexed to body surface area; SBP, systolic blood pressure. ^aIncludes patients with accelerated hypertension, highly labile hypertension, or hypertension with disproportionate target organ damage warranting further investigation.

TOD Detected by MRI

Overall, 79% (157 of 200) of patients had evidence of TOD. The proportion of patients with TOD differed between the four hypertension subgroups (see Table 2, P<.0001); 92% (70 of 76) of patients with rHTN, 59% (38 of 64) with yHTN, 89% (31 of 35) with ucHTN, and 72% (18 of 25) with other HTN.

Table 2.

Prevelence of All Types of Target Organ Damage, Left Ventricular Hypertrophy, and Potential Secondary Causes of Hypertension Identified on MRI, With Comparison by Hypertensive Subgroups

Parameter	All Patients (N=200)	Patient Hypertension Subgroups
		Drug‐Resistant (n=76)	Young‐Onset (n=64)	Uncontrolled (n=35)	Othera (n=25)	P Value
Target organ damage, No. (%)	157 (79)	70 (92)	38 (59)	31 (89)	18 (72)	<.0001
Left ventricular hypertrophy, No. (%)	122 (61)	56 (74)	24 (38)	26 (74)	16 (64)	<.0001
Secondary causes, No. (%)	29 (15)	13 (17)	4 (6)	8 (23)	7 (28)	.04

^aIncludes patients with accelerated hypertension, highly labile hypertension, or hypertension with disproportionate target organ damage warranting further investigation.

In this study cohort, 61% (122 of 200) of patients had LVH and 7% (13 of 200) had LV remodeling without hypertrophy. An example of LVH is shown in Figure 1A. LGE demonstrated LV replacement fibrosis in 15% (30 of 200) of patients; the 15 patients with a noninfarction pattern of LGE all had mild patchy intramyocardial enhancement. Based on the nature of their LVH, characteristic basal mid‐wall fibrosis, and other markers, four patients were reclassified with probable hypertrophic cardiomyopathy (see example in Figure 1B).13, 14 From a functional perspective, 14% (27 of 200) of patients had reduced ejection fraction and 42% (84 of 200) had impaired long‐axis function. There was a difference in LVM between the hypertension groups (see Table 1, P<.01) with a significant difference between those with rHTN and those with yHTN on subgroup analysis (P<.05). Viewed as the proportion of patients in each hypertension subgroup with LVH, there was also a significant difference between groups (see Table 2, P<.0001); LVH in 74% (56 of 76) of those with rHTN, 38% (24 of 64) of those with yHTN, 74% (26 of 35) with ucHTN, and 64% (16 of 25) with other HTN.

Figure 1.

Magnetic resonance imaging of target organ damage in hypertension. (A) Left ventricular (LV) midcavity steady‐state free precession (SSFP) short‐axis cine image at end‐diastole (Ai) LV four‐chamber/horizontal long‐axis SSFP cine image at end‐diastole (Aii) images from the same patient showing elevated indexed LV mass consistent with LV hypertrophy. (B) LV midcavity SSFP cine image at end‐diastole from another patient demonstrating LV hypertrophy (Bi) LV short‐axis midcavity magnitude inversion recovery myocardial late gadolinium enhancement image showing evidence of patchy midwall replacement myocardial fibrosis (Bii, indicated by arrow). These findings raise the possibility of previously undiagnosed hypertrophic cardiomyopathy in this case. (C and D) Phase‐sensitive inversion recovery (PSIR) images showing late gadolinium enhancement of (C) a lateral, subendocardial infarction, and (D) an inferolateral (circumflex territory), subendocardial infarction (indicated by arrows).

We identified 15 (7.5%) individuals with evidence of a previous myocardial infarction, seen as subendocardial scarring in a coronary artery distribution on LGE (Figure 1C and 1D); this was a novel diagnosis in five of these patients. A total of 44 patients (22%) had an ectatic aorta (surgical intervention not indicated) and six patients had evidence of significant preexisting cerebral or peripheral vascular disease and were taking secondary prevention medication.

Secondary Causes of Hypertension Identified by MRI Scanning

A total of 29 patients (14.5%) had potential secondary causes of hypertension identified on MRI. There was a difference in the proportion of patients with a potential secondary cause on imaging between the hypertension subgroups (Table 2, P=.04); secondary causes were identified in 17% (13 of 76) of those with rHTN, 6% (4/64) of those with yHTN, 23% (8 of 35) of those with ucHTN, and 28% (7 of 25) with other hypertension. In the full cohort, we identified 12 patients with adrenal masses/hyperplasia (including two reported as pheochromocytomas), 10 with renal artery stenoses (RAS), six with renal abnormalities potentially causing secondary hypertension (four renal atrophy, one polycystic kidney disease, one renal coloboma syndrome), seven with thyroid abnormalities, one with aortic coarctation, and one with an enlarged pituitary gland (some patients had more than one pathology identified). A total of 27% of patients had one or more accessory renal artery. Table 3 shows details, and examples are shown in Figure 2. It should also be noted that 47% of the study patients were obese; however, when comparing those with and those without obesity, a similar proportion of patients had another potential secondary cause of hypertension identified on MRI in this cohort (13 of 93 [14%] vs 26 of 107 [15%], P=not significant).

Table 3.

Secondary Causes of Hypertension, and Other Incidental Findings, as Demonstrated by MRI

Pathology	No. Cases	Details
Adrenal mass	12	Seven lesions were not hormonally active One was reported as a likely pheochromocytoma on MRI, but nonfunctional adenoma following endocrine testing and dedicated adrenal CT One patient with bilateral pheochromocytomas and a thyroid nodule diagnosed with multiple endocrine neoplasia type 2a is now under oncological treatment One patient had resolution of their hypertension following treatment with spironolactone Three patients have not had endocrine pathology excluded: One patient died of a stroke prior to investigation One patient emigrated prior to full investigation One patient is still under investigation
Renal artery stenosis	10	Cases reviewed at renal multidisciplinary team meeting: Two referred for stenting Eight for medical management
Renal abnormality	21	Six of 21 findings may reflect secondary hypertension: Two atrophy secondary to RAS Two atrophy not related to RAS One polycystic kidney disease; consistent with family history of the condition. One single hypoplastic, malrotated kidney in keeping with previous diagnosis of renal coloboma syndrome (autosomal dominant condition characterized by renal hypodysplasia, optic nerve dysplasia, and hypertension)47 Fifteen of 21 findings likely incidental: Nine unilateral or bilateral simple cyst(s) Four anatomical variants not felt to cause hypertension (eg, horseshoe kidney and duplex malrotated kidney) One patient with pelvicalyceal dilatation and possible renal parenchymal abnormality, further defined on CT and reported to be benign One previous nephrectomy for loin pain hematuria syndrome Not known to cause hypertension48; referral for renal denervation to treat concurrent resistant hypertension and, potentially, loin pain49, 50
Thyroid abnormality	7	Goiter and nodules, assessed biochemically and referred further investigation if indicated One case MEN2a, see above
Pituitary enlargement	1	Investigated with pituitary function testing and a pituitary MRI; nonfunctional
Aortic coarctation	1	Associated with bicuspid aortic valve and aortopathy Novel diagnosis, hypertension resolved following endovascular stenting
Incidental findings		Three bicuspid aortic valves (now on surveillance), one pulmonary nodule, two splenomegaly, one liver lesion (hemangioma), gallstones, uterine fibroids, and breast, liver, pancreatic, and renal cysts

Abbreviations: CT, computed tomography; MRI, magnetic resonance imaging; RAS, renal artery stenosis.

Figure 2.

Secondary causes of hypertension demonstrated on magnetic resonance imaging. (A) Maximal intensity, arterial phase, coronal image (time‐resolved imaging with stochastic trajectories‐magnetic resonance [TWIST‐MR] angiography) showing left ostial renal artery stenosis (indicated by arrow) with left accessory renal artery inferior to main renal artery. (B) TWIST‐MRA showing right accessory renal artery (incidental finding, indicated by arrow). (C) Single left malrotated and inferiorly positioned kidney in a patient with renal coloboma syndrome (delayed‐phase coronal image from TWIST‐MRA showing arterial and venous phase imaging). (D) Multiple low signal, well‐defined entities in both renal cortices on nephrographic phase imaging from coronal TWIST‐MRA, which represent renal cysts in a patient with polycystic kidney disease (see arrows). (E) Maximum‐intensity projection sagittal image showing coarctation of the aorta just distal to the left subclavian artery (marked by arrow) and numerous collateral vessels. (F) Right benign adrenal nodule (see arrow). (G) Bilateral adrenal pheochromocytomas in a patient with multiple endocrine neoplasia type IIa (see arrows). (H) Large left thyroid nodule (see arrow). (F, G, and H) Axial half‐Fourier acquisition single‐shot turbo spin‐echo (HASTE) images.

MRI vs Other Imaging Modalities

MRI results were compared with conventional echocardiographic, renal ultrasound, or CT scans performed as part of a participant's hypertension assessment, where available. In this cohort, 84 patients had previous echocardiograms, 81 had renal ultrasound, and 11 had CT imaging (adrenal protocol CT or renal CT angiography) performed either prior to their MRI scan or as part of a more focused investigation of their MRI findings.

When compared with CMR, echocardiography overdiagnosed LVH in 15% of cases (false‐positive) and missed LVH in 14% (false‐negative). Echocardiography had a sensitivity of 79% and specificity of 64% for the detection of LVH in this dataset (see Supplementary Table 1 for predictive values). Simple renal ultrasound assessment failed to detect four cases of RAS. One patient who had “normal” adrenal glands reported on MRI had a 6‐mm adenoma identified on subsequent focused adrenal CT imaging, which was performed because of positive biochemistry indicating Conn syndrome; this was retrospectively not visible on the initial MRI. In another case, an adrenal mass was reported as a likely pheochromocytomas on MRI but was felt to be benign following endocrine testing and a dedicated adrenal CT (see Table 3). Following initial investigation with MRI, further imaging was recommended in 24 of 200 patients; this included focused CT to further characterize RAS and adrenal masses, ultrasound of thyroid abnormalities (all benign except an MEN2a case), or imaging of other incidental findings.

Discussion

We have shown that MRI is a safe and effective imaging strategy for evaluating TOD and screening for secondary causes in patients assessed within the context of a specialist hypertension clinic. MRI has the significant benefit of being able to image multiple systems in one session and is a low‐risk investigation that does not involve ionizing radiation. This is particularly relevant when investigating patients with young‐onset disease who may go on to have repeated imaging during their lifetimes. The high prevalence of pathology seen in this study justifies the use of more advanced imaging techniques in higher‐risk/more complex patients with hypertension as recommended by European guidelines.2

Asymptomatic TOD was identified in over three quarters of our study population, and, most notably, LVH was demonstrated in 61% of the patients. The prevalence of TOD and LVH differed between the subgroups of hypertensive patients in this study. Fewer patients with yHTN had LVH, and while this may reflect the fact that these patients have been exposed to high BP for a shorter period of time, the development of different hypertensive heart disease phenotypes is likely to be multifactorial.15 The level of LVH across the full study cohort is higher than the 36% to 41% prevalence seen in the general hypertensive population16; however, we present data for patients who required assessment in a tertiary clinic. TOD is highly prevalent in established HTN (echocardiographic LVH seen in 55% to 75% of patients with rHTN17), and healthcare professionals should have a low threshold to request further imaging in these individuals, particularly if this will alter management as a result of increased cardiovascular risk.

In this cohort there was also a relatively high (14.5%) prevalence of potential secondary causes of hypertension compared with the 5% to 10% rate of secondary hypertension reported among general hypertensive populations.18, 19, 20 Once parallel biochemical testing is taken into account, the prevalence of secondary hypertension seen in this study cohort may be more consistent with rates of secondary hypertension of 20% to 31% seen in patients with more severe or treatment‐resistant hypertension.21, 22

There was a significant difference between the prevalence of secondary causes between the subgroups of hypertensive patients in this study. The highest prevalence of secondary causes was in the “other” hypertension group. This reflects the nature of the patients in this group, some of whom were referred to the clinic for additional assessment of known secondary hypertension. However, it is interesting that fewer of those with yHTN had possible evidence of secondary hypertension compared with the other groups. It might have been expected that patients with yHTN would be more likely to have a vascular or endocrine condition as a driver for the early onset of their disease. Conversely, 23% of those with ucHTN had a potential secondary cause of hypertension on MRI, even though their persistently high BP could be attributed to insufficient pharmacotherapy and these patients would not necessarily be considered to be at higher risk for secondary hypertension.

MRI is widely regarded as the gold‐standard noninvasive technique for assessing LV volumes and LVM4 and also provides information about systolic and diastolic cardiac function.9 While ECG is widely available as a screening tool, it has a relatively low sensitivity for LVH, particularly in patients with obesity.23, 24 MRI is also superior to echocardiography for the assessment of LVM25, 26, 27 (see Supplementary Table 1 for data on the accuracy of diagnosis of LVH using ECG or echocardiography vs CMR). The European guidelines do not place much emphasis on myocardial dysfunction as a marker of TOD2; however, in this study, 14% of patients had reduced ejection fraction and 42% had impaired long‐axis function. Ejection fraction was generally preserved in our cohort, but using this parameter alone may miss significant cardiac systolic dysfunction in the presence of LVH.28

MRI can also be used to characterize any change in the geometry of the left ventricle. Patients with normal LVM but concentric LV remodeling are at increased cardiovascular risk compared with those with normal LVM and LV structure.29 The other additional advantage of MRI is the ability to evaluate replacement fibrosis using LGE.30 In this study, five of 15 myocardial infarctions seen on MRI were unexpected diagnoses and resulted in initiation of secondary prevention medication. Accurate evaluation of LVM, morphology, and fibrosis also helps to differentiate between hypertensive heart disease and hypertrophic cardiomyopathy, which makes MRI an important tool for assessing patients with significant LVH and concurrent hypertension.6, 14, 31, 32 The use of gadolinium as an MRI contrast agent carries a risk of nephrogenic systemic fibrosis (NSF), quantified as an incidence of 0.02% in one retrospective, multicenter study of 83,121 patients.33 Those with an estimated glomerular filtration rate <30 mL/min/1.73 m² are at highest risk for NSF.34 Novel MRI techniques provide even more additional information; T1 mapping can quantify diffuse interstitial fibrosis35, 36 and wave intensity analysis can be used to assess vascular stiffness.37 Table 4 summarizes the comparison between different investigation modalities for the evaluation of LVH.

Table 4.

Comparison Between Different Investigation Modalities for the Evaluation of LVH

Investigation	Benefits	Limitations	Supporting References
Electrocardiography	Safe, cheap and accessible Prognostic data	Low sensitivity for LVH, especially in obesity	Cuspidi et al, 201451 Bacharova et al, 201523 Rodrigues et al, 201624
Two‐dimensional echocardiography	Safe, low cost, and widely available Good sensitivity for LVH and significant myocardial infarction Prognostic data	Images can be limited by body habitus Geometrical assumptions made for the quantification of LV volumes	Levy et al, 199052 Verdecchia et al, 200153 Myerson et al, 200225
Three‐dimensional echocardiography	Safe Similar accuracy for estimation of LVM to CMR	Specialist equipment and trained operators required with limited availability	Nosir et al, 199854 Kuhl et al, 200055
Cardiac CT	Multiple‐images slices attained in a single breath hold Reasonable accuracy Prognostic data	Exposure to ionizing radiation and requires intravenous contrast agent	Mousseaux et al, 199456 Klein et al, 201657
CMR	Safe, nonionizing Superior accuracy and reproducibility for the quantification of LVM compared with two‐dimensional echocardiography Late gadolinium enhancement assesses fibrosis/scar Helpful to define etiology of LVH Prognostic data Can reduce sample size in research studies	Limited accessibility outside specialist centers, high cost Contraindicated for device implants, cerebral clips, claustrophobia, etc Possible incidental findings Risk of nephrogenic sclerosing fibrosis in patients with significant renal impairment	Bottini et al, 199527 Bellenger et al, 200058 Grothues et al, 200259 Myerson et al, 20024 Chirinos et al, 201060

Abbreviations: CMR, cardiac magnetic resonance; CT, computed tomography; LVH, left ventricular hypertrophy; LVM, left ventricular mass.

MRI is also a useful diagnostic tool for identifying potential secondary causes of hypertension.31 RAS was identified in 5% of our study population but has been reported in as many as 24% of patients with resistant hypertension.38 European guidelines recommend duplex ultrasound as the first‐line investigation for RAS.2 This technique is noninvasive, but ultrasound can be limited by body habitus, does not penetrate air from bowel gas, and is not the “gold‐standard” examination for excluding RAS. Duplex ultrasound is also highly operator dependent, with a sensitivity of 75% and specificity of 89.6% in experienced operators.39 More robust alternatives are therefore CT angiography and gadolinium‐enhanced MRA. CT angiography has excellent spatial and temporal resolution with a sensitivity of 94% and specificity of 93% for detecting RAS39; however, CT angiography exposes patients to ionizing radiation and requires the use of an iodinated contrast agent, with a risk of contrast‐induced nephropathy in patients with renal impairment.40 Gadolinium‐enhanced MRA, as used in this study, has now become much more widely available and has a similar sensitivity and specificity to CT angiography (90% and 94.1%, respectively), although it does carry a very small risk of NSF in patients with renal impairment, as described above.

Adrenal masses and hyperplasia can be detected using MRI; however, the frequent occurrence of incidental adrenal masses, commonly nonactive adenomas, means that biochemical assessment should be the primary screening tool for endocrine hypertension.31 Our MRI protocol includes a simple single breath‐hold axial T1‐weighted imaging stack (5‐mm slice thickness) through the abdomen to screen for “resolvable” adrenal lesions, and therefore small abnormalities may be missed. High‐resolution adrenal CT or MRI, beyond our study protocol, is therefore required in patients with positive biochemistry to identify microadenomas and differentiate between pheochromocytomas and adenomas presenting as Conn or Cushing syndrome.6, 31 Pituitary lesions can be seen on standard MRI; however, a dedicated contrast‐enhanced MRI scan is also required to localize pituitary microadenomas.6

In this study, only two of 12 adrenal masses were hormonally active and only two of 10 cases of RAS went on to be stented. Furthermore, following initial investigation with MRI, further imaging was recommended in 12% of patients (Table 3). Did these other findings and investigations generate unnecessary patient stress or healthcare costs?

Incidental findings may be identified via any imaging modality; however, it is more likely when performing cross‐sectional imaging of large areas of the body as seen with MRI and CT. The patients in this study were already undergoing assessment for secondary hypertension with concurrent endocrine testing, and we were therefore able to interpret any adrenal abnormalities in the context of the patient's endocrine profile. We have found these investigations to be complimentary in clinical practice. On the other hand, patients with “normal” adrenals on MRI but a positive endocrine diagnosis would in practice require additional, high‐resolution focused adrenal imaging.

RAS is a recognized cause of hypertension38; however, the management of RAS has become controversial following data from studies, including the Angioplasty and Stenting for Renal Artery Lesions (ASTRAL) 41 and Cardiovascular Outcomes in Renal Atherosclerotic Lesions (CORAL)42 trials, which showed no benefit of renal artery stenting over medical management with renin‐angiotensin system blockade. The picture has been further complicated by the advent of renal denervation. The assessment of renal artery anatomy is used to screen for suitability for this interventional treatment for hypertension.43, 44

Finally, other incidental findings may have both positive and negative impacts on patients. For example, the novel finding of a bicuspid aortic valve means that patients can be put under surveillance for significant valvular dysfunction or aortopathy, and while this may create additional anxiety, early identification of these issues may prevent serious complications. In light of these concerns, we would emphasize that MRI is best targeted at patients in whom the increased likelihood of target organ or secondary causes of hypertension can justify the impact of any incidental findings.

Limitations

This study is a retrospective analysis of prospectively gathered clinical MRI data. As such, we had no control group and therefore cannot comment on the prevalence of TOD or secondary hypertension found on imaging in comparison with a normotensive population. Additionally, we are unable to present a comparison between our MRI data and baseline data acquired using more conventional investigation techniques across our full cohort, and where comparisons were made with other imaging modalities, the data were in the form of nonstandardized clinical reports. There was also a proportion of patients seen in the hypertension clinic in whom MRI was not indicated (especially if previously investigated), and patients who were not referred for MRI because of a history of claustrophobia, metal foreign body or implant, or morbid obesity (their girth was too large for the MRI scanner), this may represent a selection bias. We clearly recognize that this is a skewed population of hypertensive patients who have been referred for further assessment in a specialist clinic, and these data cannot be extrapolated to the general hypertensive population who are routinely managed in primary care.

Finally, the cost of MRI and access to this imaging modality can be extremely variable. We acknowledge that no formal cost‐benefit analysis has been performed for MRI assessment vs echocardiography and renal ultrasound/CT in this cohort (the approximate costs of these different investigations are shown in Supplementary Table 2). Existing data indicate that there may be a cost‐benefit of echocardiography over ECG in the diagnosis of LVH in select populations, but there are few formal data on any potential cost‐benefit of MRI vs standard techniques and further research is required.45, 46

Conclusions

MRI is a safe and effective imaging modality for screening patients with high BP for asymptomatic TOD and secondary causes of hypertension. It provides high diagnostic specificity and sensitivity for identifying LVH and RAS, and while CMR is not one of the primary imaging modalities recommended in clinical hypertension guidelines,2 it is the current gold standard for the noninvasive assessment of LV volumes, mass, and function.6, 31 We acknowledge that MRI remains a relatively expensive investigation and is not routinely available in the community; however, diagnostic information from multiple organ systems can be obtained from a 1‐hour scanning session, and this increased efficiency will be of benefit to patients. TOD was identified in >75% of our study population, resulting in the initiation of secondary prevention medication in those with newly identified cardiovascular disease and triggering the initiation of antihypertensive medication in many of those with yHTN. Given the high prevalence of TOD reported in this study and the clinical impact of our findings, we recommend that MRI be used as the primary imaging modality in tertiary hypertension clinics.

Disclosures

This work was supported by the NIHR Bristol Cardiovascular Biomedical Research Unit, Bristol Heart Institute. The views expressed are those of the authors and not necessarily those of the National Health Service, National Institute for Health Research, or Department of Health. AEB is funded by a University Hospitals Bristol NHS Foundation Trust Clinical Research Fellowship. JCLR is funded by Clinical Society of Bath Postgraduate Research Bursary 2014 and Royal College of Radiologists Kodak Research Scholarship 2014. ECH and JFRP are funded by the British Heart Foundation.

Supporting information

Table S1. Examples of sensitivities, specificities, and predictive values for the diagnosis of left ventricular hypertrophy by different techniques vs cardiac magnetic resonance (CMR).

Table S2. Costs for different imaging modalities.

J Clin Hypertens (Greenwich). 2017;19:13–22. DOI: 10.1111/jch.12920. © 2016 Wiley Periodicals, Inc.