Following the first successful surgical repair of coarctation of the aorta by Gross et al. (1) in 1945, it soon became the treatment of choice for patients with this abnormality. With the advent of interventional cardiology, balloon angioplasty (2- 3) and stent (4) have become alternative methods of treatment. Long-term complications occur with all forms of treatment and include persistent or recurrent systemic hypertension, recoarctation, aneurysm formation and, uncommonly, endocarditis.

Recoarctation may lead to aggravation of systemic hypertension, coronary artery disease and congestive heart failure. Similarly, undiagnosed aneurysms may rupture with devastating consequences (5- 7). Early recognition of recoarctation and aneurysm formation is important to avoid consequent morbidity and mortality. Various noninvasive investigations are available to follow such patients. However, little is known of the sensitivity and specificity of such tests in the diagnosis of restenosis and/or aneurysm formation.

This study examined the sensitivity and specificity of symptomatology, physical examination, electrocardiogram (ECG), chest radiograph, exercise stress test and transthoracic echocardiography in an adult population with previous surgical repair or angioplasty of coarctation of the aorta, using magnetic resonance imaging (MRI) as the gold standard procedure (8- 11). The cost of the various screening tests is also analyzed with the aim of determining the most “cost-effective” way to follow such patients.

At our institution, an MRI with cine imaging and velocity mapping in selected oblique planes (12) is performed routinely every two to five years on every adult patient following surgical repair or balloon angioplasty of coarctation of the aorta together with yearly clinical follow-up, ECG, exercise testing, chest radiograph and transthoracic echocardiography. The data base of the Grown Up Congenital Heart unit was searched for patients with previous surgical repair or angioplasty of coarctation of the aorta, who had an MRI scan between October 1990 and February 1998 and were 16 years old or older at the time of their MRI.

The findings at the clinical visit in the year preceding the MRI were reviewed. Reported symptoms, medications, the presence of systemic hypertension (defined as a blood pressure greater than 140/90 on the right arm) and the presence of radiofemoral delay on physical examination were noted. Differential blood pressures in arms and legs were not used as they had not been recorded systematically. Recoarctation was said to be present on physical examination if either systolic hypertension or radiofemoral delay was present. Left ventricular hypertrophy on ECG (defined using Estes’ criteria 13) and hypertensive response on exercise stress test (defined as a systolic blood pressure >215 mm Hg 14) were also noted. One adult congenital cardiologist reviewed blindly and reported on three consecutive routine posteroanterior chest radiographs taken before the MRI scan for the presence or absence of restenosis and/or aneurysms using the classical “3 sign” and rib notching as a marker of recoarctation and abnormal discrete bulging as a marker of aneurysm formation.

Echocardiographically, a maximal instantaneous gradient equal or greater than 25 mm Hg was taken as a positive finding for recoarctation (15), and the mention of a bulge was indicative of possible aneurysm formation. The MRI scan was used as the gold standard test (8- 11) against which the individual screening tests were compared. On velocity MRI (12), recoarctation was defined as a peak velocity >2.5 m/s or, on the 2-D imaging, defined by the following equation: D Coa/D Dao < 40%, where D Coa is the diameter of the coarcted segment and D Dao is the diameter of the descending aorta at the level of the diaphragm (10). Aneurysm was defined on 2-D MRI scan as a discrete bulging of the aorta at the repair site >150% of the diameter of the descending aorta at the level of the diaphragm (16).

For each individual screening test, both the sensitivity and the specificity were calculated. Combined tests were considered positive if any one of the tests in the combination was positive. Costs assigned to each modality were based on United Kingdom National Health Service’s fees. The most “cost-effective” approach (i.e., the best combination of screening tests providing the highest sensitivity at the lowest possible cost) was calculated, assuming that if the screening test was not used, every patient would have a clinical visit and an MRI scan, and that if the screening test was used, then only patients with a positive screening test would require an MRI scan. A confirmatory MRI scan was deemed necessary because very few surgeons would operate without MRI confirmation of recoarctation or aneurysm formation. Clinical visit was automatically included in every “cost-effectiveness” calculation, assuming that a physician would want to see and examine a patient before ordering any screening test. Only combinations with high sensitivities were chosen for the cost analysis because we believe that missing recoarctation and/or, particularly, aneurysm was unacceptable for the patient. Low specificities were tolerated, as the findings of a false-positive test could be easily dealt with by performing an MRI scan.

Eighty-four adult patients (40 women, 44 men) aged 18 to 64 (mean 30.2 years) had 134 MRI scans between October 16, 1990, and February 24, 1998. Results were available on 132 MRI scans; 2 MRI scans were not completed owing to patient claustrophobia. Details of their baseline diagnosis, previous surgery and/or balloon angioplasty are provided in (Table le1). Four patients had balloon dilation of the aorta for recoarctation following surgical repair. No stent was used in these patients. Mean age of the patients at the time of their last intervention was 14 years (SD 12.2), with a mean time between their last intervention and their MRI scan of 16.2 years (SD 10.1). Mean time between the last clinic visit and related screening tests and the MRI scan was 4.5 months (SD 6.1).

Recoarctation was diagnosed in 22 patients (26.2%). Seventy-three percent of patients with recoarctation had had end-to-end repair.

The sensitivity, specificity and cost of individual screening tests for recoarctation are summarized in (Table le2). Although each screening test had a reasonable level of specificity, ranging from 74% to 89%, only echocardiography had a high sensitivity, namely 87%. Antihypertensive medication did not change significantly the sensitivity and specificity of systolic hypertension at rest or on exertion and left ventricular hypertrophy on ECG.

The sensitivity, specificity and cost of the combined tests are also shown in (Table le2). A sensitivity of 100% was reached when each screening test was performed or, alternatively, when a clinical visit and echocardiography with or without chest radiograph were performed.

Aneurysm was diagnosed in 12 patients (14.3%). There were 10 true aneurysms and 2 false aneurysms. One false aneurysm occurred following end-to-end surgical repair, and another false aneurysm occurred following interposition graft.

The sensitivity, specificity and cost of individual screening tests for aneurysm are shown in (Table le2). Each test had a specificity greater than 65%, but only chest radiograph had a sensitivity greater than 50%. Of note, hemoptysis had a sensitivity and specificity of 100% for false aneurysm in this study.

The sensitivity, specificity and cost of the combined tests are also shown in (Table le2). A sensitivity of 93% was achieved when every screening test was performed or, alternatively, when a clinical visit, chest radiograph and echocardiography were performed.

Patients had either isolated recoarctation (22 patients) or aneurysm formation (12 patients). The combination of recoarctation and aneurysm formation was not present simultaneously in any patients.

The sensitivity, specificity and cost of individual screening tests to diagnose recoarctation or aneurysm are shown in (Table le2). Even though the specificity of each test remained greater than 60%, only echocardiography maintained a high sensitivity of 85%.

The sensitivity, specificity and cost of the combined tests to detect recoarctation or aneurysm are shown in (Table le2). Sensitivities greater than 90% were reached when performing echocardiography with exercise testing or chest radiograph or clinical visit. Sensitivity of 100% was achieved when performing every screening test or a clinical visit, chest radiograph and echocardiography. Cost analysis of the most sensitive combinations (sensitivity >97%) is shown in (Table le3). The cost of performing a clinical visit and an MRI on every patient emerges as the most “cost-effective” approach. Alternatively, performing a clinical visit and an echocardiography as a screening test and an MRI on patients with positive results is also a reasonable approach, accepting that 3% of recoarctation or aneurysms will be missed (sensitivity of 97%) using this alternative.

This study was the first to evaluate the sensitivity, specificity and cost-effectiveness of various screening tests in the diagnosis of recoarctation and/or aneurysm formation in an adult population following surgical repair or angioplasty of recoarctation of the aorta. Our data show that the combination of clinical visit and MRI is the most cost-effective approach to follow such patients.

The prevalence of recoarctation reported in the literature varies widely from 7% to 60% depending on the definition used, the length of follow-up (17) and the age at surgery (17- 20). In our study, the prevalence of recoarctation was 26.2%, with 65% of recoarctation occurring following end-to-end repair. This could reflect the fact that patients with end-to-end repair were operated on at an earlier age (8 of 14 patients were less than a year old) and/or had a more complex aortic arch anatomy.

Each individual screening test had a reasonable specificity for recoarctation but had a low sensitivity.

The low sensitivity of leg weakness and radiofemoral delay on examination may relate to the high incidence in our study of “mild” recoarctation (68.7% of recoarctation had a peak gradient of 25 to 35 mm Hg on MRI scanning). Patients with mild recoarctation either may be truly asymptomatic or adjust their level of physical activity so that they do not experience symptoms. Similarly, the presence of normal pulses eliminates all but mild coarctation (21).

The low sensitivity of systolic hypertension at rest or on exertion and left ventricular hypertrophy on ECG to detect recoarctation reinforce the hypothesis that systolic hypertension after coarctation repair is probably related to abnormal aortic wall compliance, altered baroreceptor function and enhanced sympathetic and renin activity (22- 23) rather than residual mechanical obstruction at the aortic level (24). Our study reinforces the finding that there is a poor correlation between systolic hypertension at rest or on exercise and the presence of recoarctation; it thus argues against its use as a screening test to detect recoarctation.

The classical signs of coarctation on chest radiograph were not sensitive, 17%, for diagnosing recoarctation. Mild recoarctation may not cause the development of aortic bulging and collateral blood flow or, alternatively, the intrathoracic scarring after surgical repair may disguise them. The chest radiograph is thus a poor screening test for recoarctation.

Echocardiography was the most sensitive screening test, 87%. The accuracy of 2-D and Doppler echocardiography in diagnosing native or recurrent coarctation has been documented previously, with angiographic correlation varying between 0.89 to 0.91 (10- 11,15). The major shortcoming of echocardiography lies in the poor image resolution obtained in adults and in the failure to obtain a Doppler gradient in the presence of significant collateral blood flow (15). In our study, most of the patients with recoarctation were diagnosed using Doppler signal. The absence of significant collateral blood flow in our patients with recoarctation probably contributed to the high sensitivity of Doppler echocardiography.

The prevalence of aneurysm in this study was 14.3%, with the majority of cases (50%) occurring following end-to-end repair. Surgical repair by Dacron patch aortoplasty is the most widely recognized technique associated with aneurysm formation, with a reported incidence of between 2% to 27% (5- 7,16). The “natural history” of aneurysm formation is unknown. Aneurysms have been reported to occur as early as three years, and as late as 18 years, following surgery (5,7,16). Aneurysm formation following end-to-end repair is unusual. Few other studies, however, have systematically examined patients after surgical repair of coarctation of the aorta, and most studies have targeted patients following patch aortoplasty repair. Given our results, more scrupulous follow-up of patients following end-to-end repair is advisable.

The sensitivity of general symptoms to detect aneurysm formation was low, 31%, because aneurysms are usually clinically silent. Hemoptysis emerged, however, as a very sensitive and specific symptom of false aneurysm. The poor sensitivity of systolic hypertension to detect aneurysm formation makes it unlikely to be causal in the development of aneurysms.

The sensitivity of echocardiography to detect aneurysm was surprisingly low, 29%. Even though echocardiography has been shown to be good for the diagnosis of recoarctation, its sensitivity in discovering aneurysms in an adult population has not been evaluated before. Bromberg et al. (16) evaluated prospectively the accuracy of echocardiography and chest radiograph in detecting aortic aneurysm in 29 children following patch aortoplasty repair of coarctation of the aorta. The reported sensitivity and specificity of echocardiography was 71% and 76%, respectively, differing significantly from our result. Our lower sensitivity may relate in part to the poorer image resolution of the descending aorta in the adult population compared to children, resulting in higher false-negative studies in adults. Second, angiography, the standard used in the study (16), was performed only in patients with a positive screening test, making the interpretation of the reported sensitivity difficult.

Similarly, the reported sensitivity of chest radiograph to diagnose aneurysm in that study was 100% compared to ours, 67%. Again, by not performing angiographic studies on every patient, it is probable that the numbers of false-negative chest radiographs were underestimated, falsely raising its diagnostic value.

The sensitivity of individual screening tests for diagnosing recoarctation or aneurysm formation was greater than the sensitivity of individual tests in diagnosing aneurysm alone because the prevalence of the disease (recoarctation or aneurysm) was greater than the prevalence of aneurysm alone.

Various combinations of screening tests achieved a sensitivity of greater than 90%, but few combinations had a specificity greater than 50%. The cost of performing a clinical visit and an MRI on every patient without any prior screening test emerged as the most “cost-effective” approach to diagnose recoarctation or aneurysm formation because of the high false-positive rate (low specificity) of the combined screening tests requiring confirmatory MRI scanning. Alternatively, performing a clinical visit and an echocardiography as a screening test and an MRI on patients with positive results was also a reasonable approach if MRI resources are limited, with the knowledge, however, that 3% of recoarctation or aneurysms will be missed (sensitivity of 97%) using this alternative.

The frequency at which these screening tests should be done remains to be determined and depends on the yearly prevalence or “rate of progression” of recoarctation and/or aneurysm formation. This could not be assessed by our study given that it did not look at the MRI scans of every patient operated on in a prospective way. Second, noninvasive testing such as echocardiography may offer additional information not readily available from MRI testing as of yet, such as aortic valve mobility and systolic and diastolic ventricular function. Such information may be required for a given patient, and follow-up in such cases should therefore be tailored to the patient’s specific needs.

Recoarctation and aneurysm formation following surgical repair or angioplasty of coarctation of the aorta are common and important to diagnose. Prior to this study, little was known of the sensitivity and specificity of various screening modalities to detect recoarctation or aneurysm, and consequently the best approach to follow these patients was undefined. The combination of clinical visit and an MRI on every patient without any prior screening test emerged as the most “cost-effective” approach to diagnose either recoarctation or aneurysm formation in an adult population following surgical repair or angioplasty of recoarctation of the aorta. Alternatively, performing a clinical visit with an echocardiogram and an MRI on patients with positive results is also a reasonable approach if MRI scanning resources are limited. We would suggest that such patients be referred to a specialty unit with experience and appropriate facilities for their follow-up.