Author + information
- Received January 8, 1997
- Revision received March 21, 1997
- Accepted April 16, 1997
- Published online August 1, 1997.
- Stephen Heupler, MDA,
- Rajendra Mehta, MDA,
- Arlene Lobo, MDA,
- Dominic Leung, MDA and
- Thomas H Marwick, MD, PhD, FACCA,* ()
- ↵*Dr. Thomas H. Marwick, Department of Cardiology, F-15, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio 44195.
Objectives. The purpose of this study was to define the value of exercise echocardiography as an independent predictor of cardiac events in women with known or suspected coronary artery disease (CAD), incremental to the data provided by clinical evaluation and exercise testing.
Background. Exercise echocardiography is more accurate than exercise electrocardiography for the identification of CAD in women. However, the prognostic implications of exercise echocardiography, especially relative to exercise electrocardiography, are undefined.
Methods. Symptom-limited exercise echocardiography was performed in 549 consecutive women between 1989 and 1993. Echocardiography and electrocardiography were performed before and after treadmill exercise; an abnormal result on exercise electrocardiography was defined by ST segment depression >0.1 mV, ischemia by exercise echocardiography as a new or worse wall motion abnormality after exercise and scar by akinesia or dyskinesia at rest. After exclusion of six patients with uninterpretable studies (1%) and 35 (6%) lost to follow-up, 508 women (mean [±SD] age 55 ± 11 years) were followed up for 41 ± 10 months for cardiac-related death, infarction or late revascularization.
Results. The group attained 92 ± 10% of age-predicted maximal heart rate, with an exercise capacity of 7 ± 2 metabolic equivalents. Of 420 women with an interpretable electrocardiogram, significant ST segment changes were present in 68 (16%). Results of exercise echocardiography were normal in 413 (81%) women, positive for ischemia in 66 (13%) and scar only in 29 (6%). No events occurred in 444 patients (89%), and 19 underwent primary revascularization (within 3 months of exercise test). Cardiac events occurred in 36 women (7%), including 17 who died of cardiac causes and 19 who had a myocardial infarction or required late revascularization for progressive symptoms. On univariate analysis, the variables associated with cardiac mortality and total cardiac events were a history of CAD, diabetes, left ventricular hypertrophy, exercise capacity and echocardiographic evidence of myocardial ischemia and infarction. A Cox proportional hazards model showed the independent predictors of outcome to be known CAD (odds ratio [OR] 6.6, 95% confidence interval [CI] 3.2 to 13.7, p < 0.00001) and echocardiographic ischemia (OR 4.3, 95% CI 2.1 to 8.7, p < 0.0001). The prognostic value of exercise echocardiography incremental to clinical and exercise variables was demonstrated using sequential Cox models.
Conclusions. In this large cohort of women, exercise echocardiography provided key prognostic information incremental to clinical and exercise testing data.
Coronary artery disease (CAD) is the leading cause of death in women in the United States . However, although the prognosis of patients with known CAD is worse in women than men, most women presenting with chest pain syndromes follow a benign course . The challenge for the physician is to identify the minority of women at high risk for cardiac events in whom outcome might be improved by further intervention.
Historically, exercise electrocardiography has been the best studied and most commonly used noninvasive test to establish prognosis [3–6]. However, the accuracy of this test for the diagnosis of CAD in women has been questioned, with studies showing a spectrum of accuracy [7–14]. Limited data suggest that the use of exercise testing is a better predictor of cardiac events in men than in women . Exercise echocardiography appears to be more accurate than exercise electrocardiography for the diagnosis of CAD in women [16–18]. The prognostic utility of exercise echocardiography has been evaluated in mixed (largely male) groups, but the ability of this test to predict outcomes in women has not been evaluated. The purpose of the present study was to determine the independent ability of exercise echocardiography to predict cardiac events in a large heterogeneous group of women with known or suspected CAD and to identify whether these data were incremental to clinical predictors of outcome.
1.1 Patient selection.
Five hundred forty-nine women with known or suspected CAD underwent exercise echocardiography between August 1989 and October 1993 for the first time at our institution. The only studies excluded were those that were uninterpretable because of inadequate image quality (n = 6).
1.2 Clinical evaluation.
Of the remaining 543 women (mean [±SD] age 55 ± 11 years), 95 (17%) had known CAD, defined by a history of myocardial infarction, revascularization or angiographically defined stenosis (>50% diameter) of a major vessel. In patients without a previous history of CAD, a coronary risk profile derived by the Framingham Heart Study was calculated from the lipid profile, blood pressure, smoking status and diabetes history.
1.3 Exercise testing.
All patients underwent a symptom-limited treadmill exercise test, using a protocol appropriate for each individual patient (either Bruce, modified Bruce or Naughton). Heart rate and blood pressure were measured and standard 12-lead electrocardiograms (ECGs) were recorded at rest, during each stage of exercise and during recovery. ST segment deviation was measured 0.08 s after the J point and was considered significant if there was ≥1 mm of horizontal or downgoing ST segment depression in two or more contiguous leads. ST segment changes in the presence of left bundle branch block, digoxin therapy or left ventricular (LV) hypertrophy were considered nondiagnostic. Maximal exercise capacity was recorded in estimated metabolic equivalents (METS). The impact of exercise capacity on outcome was analyzed as a continuous variable as well as on the basis of failure to attain >6 METS.
A treadmill score was calculated according to the formula described by Mark et al. : Treadmillscore= Durationofexercise(Bruceprotocol) − (5 × MaximalnetSTsegmentdeviation[mm]) − (4 × Treadmillanginaindex). The treadmill angina index was 0 if there was no angina during exercise, 1 if there was nonlimiting angina and 2 if there was angina that caused termination of exercise. A maximal heart rate response was considered to be ≥85% of the age-predicted maximal heart rate; however, patients who did not achieve this rate were also included in the analysis.
1.4 Exercise echocardiography.
Two-dimensional echocardiography was performed at rest, in the left lateral decubitus position, immediately before exercise, using standard ultrasound equipment. Images were obtained in the parasternal long- and short-axis and the apical two- and four-chamber views. All images were recorded on 0.5-in. VHS tape and digitized on-line using PreVue or ImageVue systems (Nova MicroSonics). Postexercise images were obtained, usually within 1 min of completion of exercise, using the same views. All studies were reviewed by an experienced observer. Ischemiawas defined by new or worsening regional wall motion abnormalities on exercise. Infarctionwas defined by akinesia or dyskinesia of a segment at rest.
Follow-up was performed by review of the patient’s medical record or by telephone contact 41 ± 10 months after exercise echocardiography in 508 women (94%). The measured primary end points were cardiac-related death, myocardial infarction and late myocardial revascularization (defined as coronary artery bypass graft surgery [CABG] or percutaneous transluminal coronary angioplasty [PTCA] >3 months after the test), and the date of contact or the occurrence of an end point was recorded. Non–cardiac-related deathwas defined as death not due to myocardial infarction, cardiogenic shock or cardiac arrhythmia. Primary revascularizationwas defined as any PTCA or CABG procedure that occurred <3 months after the exercise echocardiogram. Patients were censored from follow-up at the time of non–cardiac-related death or primary revascularization—the latter step was aimed at reduction of posttest referral bias.
1.6 Statistical analysis.
Descriptive statistics were generated using percentages and 95% confidence intervals (CIs) for discrete variables and mean value ± SD for continuous variables. Discrete variables were compared using chi-square analysis, and continuous variables were compared using analysis of variance; differences in outcomes were reported as odds ratios (ORs). Cardiac survival was expressed using Kaplan-Meier survival curves, which were compared by log-rank chi-square analysis. Multivariable analysis was performed using a Cox proportional hazards model.
The prognostic value of exercise echocardiography incremental to clinical and exercise testing data was assessed using sequential Cox models, whereby additional information was added to the models in the same sequence as in clinical practice. Thus, after defining the global chi-square of the clinical variables, the independent predictors from this model were added to exercise testing variables to create a second model. These predictors were then added to the results of echocardiography to create a third model. All patients participated in each model. The incremental benefit was defined by comparison of the global chi-square for each model. All statistics were performed using the SPSS for Windows statistical package.
2.1 Clinical characteristics.
Table 1contrasts the characteristics of the 508 women who were followed-up over 41 ± 10 months with 35 women (6%) who were lost to follow-up. The latter group were at somewhat lower risk than the study group because of younger age and greater exercise capacity. The study group was predominantly middle-aged (average 56 ± 11 years), with a significant prevalence of coronary disease risk factors. The diagnosis of CAD was known in 92 women (18%). Of the 416 women without a history of CAD, the 10 year cardiac risk, based on the Framingham coronary risk profile, averaged 8 ± 5% (range 0% to 38%).
2.2 Exercise testing.
Most patients exercised adequately, to a mean workload of 7 ± 2 METS and 92 ± 10% of maximal predicted heart rate. However, 100 women (20%) exercised submaximally (<85% of maximal heart rate), and the rate–pressure product ranged from 10.5 to 45.1 × 103beats/min × mm Hg (average 27.7 ± 5.2 × 103beats/min × mm Hg). Ninety-nine women (19%) developed angina during the exercise test, which was nonlimiting in 66 and severe in the other 33.
The exercise ECG was nondiagnostic in 88 women (17%) because of baseline abnormalities. Of the remaining 420 women, 68 (16%) had significant ST segment changes on exercise. The average treadmill score was 3 ± 4 (range −23 to 21), and 256 women (50%) had a score ≥5, which placed them in a low risk category for cardiac events. Only 15 women (3%) were characterized as high risk on the basis of their treadmill score.
2.3 Exercise echocardiography.
Results of exercise echocardiography were abnormal in 95 women (19%). A pattern consistent with infarction (a wall motion abnormality at rest without exercise-induced abnormalities) was present in 29 women. Ischemia (evidenced by a normal baseline study results and an exercise-induced wall motion abnormality) was present in 42 women. In the remaining 24 women, a wall motion abnormality was present at rest and a new one developed with exercise, consistent with ischemia in the setting of a previous infarction. Of the 95 women with abnormal results on exercise echocardiography, 56 (59%) had a history of CAD. Only 36 of the 413 patients (9%) with normal exercise echocardiographic results had a previous history of CAD.
No events occurred during follow-up in 444 women (87%). Cardiac-related death occurred in 17 women (3%), and myocardial infarction or late revascularization occurred in 19 (4%). Patients having a revascularization procedure within 3 months of the exercise test (n = 19 [4%]) or dying of noncardiac causes (n = 9 [2%]) were censored at the time of these events. Non–cardiac-related deaths were due to malignancy in six patients, chronic pulmonary disease in one, renal failure in one and noncardiac complications of surgical intervention in one.
2.5 Univariate correlation of outcomes and clinical and exercise variables.
Table 2summarizes the correlation between clinical variables and cardiac-related events. Both cardiac mortality and total cardiac events were associated with a previous history of CAD, diabetes and ECG evidence of myocardial infarction or LV hypertrophy. Ninety-five patients had a previous history of CAD, and after exclusion of 15 with primary revascularization or non–cardiac-related death, 23 of the remaining patients (30%) with known CAD had a cardiac-related event during follow-up compared with only 13 (3%) of 403 patients without a history of CAD who had an event. After exclusion of 5 patients who underwent primary revascularization, a cardiac event occurred in 10 (21%) of 48 patients with versus only 26 (6%) of 432 without diabetes.
Of the 68 women who had significant ST segment changes on exercise, 8 underwent primary revascularization or died of noncardiac causes. Cardiac events occurred in 9 (15%) of the remaining 60 women with ST segment changes during the follow-up period compared with only 18 (5%) of 337 without ECG changes (OR 3.1, p = 0.01). However, ST segment changes were not significantly correlated with cardiac-related death.
The influence of exercise capacity on outcome was analyzed by classifying the study cohort into two subgroups: women who achieved ≥6 METS (n = 343) and those who achieved <6 METS (n = 165). There were 16 cardiac events (5%) in the group that achieved ≥6 METS and 20 primary events (13%) in those who achieved <6 METS (OR 3.5, p = 0.001). Exercise capacity was also associated with cardiac-related death (Table 2). The treadmill score was associated with cardiac events but not with cardiac-related death in women.
The association of echocardiographic evidence of ischemia and infarction with cardiac events in women is summarized in Fig. 1. Cardiac events occurred in 17 (31%) of 54 women with echocardiographic evidence of ischemia compared with 19 (4%) of 426 without ischemia (OR 9.8, 95% CI 4.4 to 21.9, p < 0.00001). The event rate of women with echocardiographic evidence of infarction was 31% and only 5% in the remainder (OR 8.5, 95% CI 3.7 to 19.4, p < 0.00001).
The data provided by electrocardiography and echocardiography are complementary. The outcomes associated with combinations of these results are illustrated in Fig. 2. Women who had no ischemia by either modality (n = 328) had a 96% event-free survival rate at 30 months, whereas those who had ischemia by both (n = 16) had only a 55% event-free survival rate. Women with conflicting ECG and echocardiographic results had event rates intermediate between these extremes, but women with positive exercise echocardiographic and negative ECG findings (n = 27) had a lower event-free survival rate (73% at 30 months) than those without echocardiographic evidence of ischemia but with ST segment changes (n = 52 [91% event-free survival rate at 30 months]).
2.6 Multivariate correlation of outcomes and clinical and exercise variables.
The number of variables that could be incorporated into a multivariate model was limited by the occurrence of cardiac events in only 36 women. Thus, a Cox proportional hazards model was based on the relation between outcome and past cardiac history, diabetes, exercise capacity and exercise echocardiographic findings. Of these variables, only a past cardiac history (OR 6.6, 95% CI 3.2 to 13.7, p < 0.00001) and echocardiographic evidence of ischemia (OR 4.3, 95% CI 2.1 to 8.7, p = 0.0001) were independent predictors of outcome.
2.7 Incremental value of exercise echocardiography.
A sequence of Cox proportional hazard models were constructed, in which the number of patients in each model was the same, to assess the prognostic value of exercise echocardiography incremental to other clinical and exercise variables. The initial model, based on clinical variables, revealed a previous history of CAD (OR 10.2, 95% CI 5.2 to 20.1, p < 0.0001) to be the only independent predictor of future cardiac events. In the second (clinical plus exercise) model, cardiac history was combined with cardiac workload, ST segment change and treadmill score. The independent predictors were a previous history of CAD and the treadmill score (OR 0.94, 95% CI 0.90 to 0.98, p = 0.004). In the third model, the latter two variables were combined with evidence of ischemia or infarction at exercise echocardiography. The independent predictors were a previous history of CAD and the presence of echocardiographic ischemia (OR 4.3, 95% CI 2.1 to 8.7, p = 0.0001). The incremental value of each step in the prediction of cardiac events is depicted by the global chi-square of the clinical, exercise and echocardiographic models in Fig. 3.
Because a known history of CAD was such a powerful predictor for future cardiac-related events in our cohort, patients were classified on the basis of the presence or absence of known CAD. The sequential models were reapplied to these subgroups (Fig. 4). In women without previous CAD, clinical variables that were significantly associated with future cardiac outcomes were diabetes (OR 4.3, 95% CI 1.2 to 15.5, p = 0.03) and ECG evidence of LV hypertrophy (OR 6.9, 95% CI 1.5 to 31.7, p = 0.01). The only exercise variable that predicted on adverse outcome was the treadmill score (OR 0.89, 95% CI 0.83 to 0.97, p = 0.004). When these features were combined with the echocardiographic data in the third model, the presence of ischemia by exercise echocardiography (OR 5.4, 95% CI 1.5 to 19.6, p = 0.01) added incremental prognostic data.
In women with a history of CAD, there were no other clinical predictors of adverse events. In the model that included exercise variables, only maximal rate–pressure product had a borderline significant predictive value (p = 0.05), but treadmill score and ECG ischemia were not predictive of adverse cardiac events. The presence of ischemia by exercise echocardiography was the only predictor of cardiac events in the patients with known CAD (OR 3.5, 95% CI 1.5 to 8.1, p = 0.003).
2.8 Comparison of predictive value of exercise testing and exercise echocardiography.
Of the exercise variables, the best univariate predictor of cardiac events was exercise capacity <6 METS (Table 2). To further assess the incremental benefit of echocardiographic imaging during the exercise test, the patients were classified into low and high risk groups on the basis of their exercise capacity, and then these subgroups were further stratified according to their echocardiographic results. Of 165 patients in the high risk subgroup based on an exercise capacity <6 METS, 35 had echocardiographic ischemia. The event rate was 41% in this subgroup, with concordant high risk test results. Exercise echocardiography showed no ischemia in 130 patients who exercised to <6 METS, and the event rate was 7% in this subgroup. There were 343 patients who exercised to ≥6 METS, and 312 of these had no ischemia; the event rate was 4% in this subgroup, with concordant low risk test results. The remaining 31 patients exercised to ≥6 METS and had ischemia, with an event rate of 20%. Thus, 161 patients had discordant risk stratification based on their exercise capacity and ischemia. Sixteen people in this group had primary revascularization or non-cardiac-related death. Of the remaining 145 people, the exercise echocardiogram correctly predicted the outcome in 117, and exercise capacity predicted the outcome in 28. Therefore, the sum prognostic benefit for echocardiographic compared to exercise capacity in this study was 85 patients correctly classified into an appropriate risk group. After exclusion of the primary revascularization and patients who died of noncardiac causes the total cohort is 480 women, and of this group, 85 (18%) would be better risk stratified with exercise echocardiography than exercise capacity alone.
In this study, the detection of exercise-induced wall motion abnormalities was an independent predictor of cardiac events in 508 women followed up for 41 ± 10 months after exercise echocardiography. Evidence of ischemia by echocardiography was a better predictor of future cardiac events than exercise capacity or ST segment changes. The information provided by exercise echocardiography was incremental to that provided by clinical and exercise ECG variables, both in the overall study cohort and in subgroups with and without a previous history of CAD.
3.1 Diagnosis of CAD in women.
The reported accuracy of the ST segment response for the diagnosis of CAD in women has been variable. A large series, obtained from the Coronary Artery Surgery Study , found no gender differences in the false positive rate. However, a number of smaller studies [9–13]have identified either lower sensitivity or specificity in women. The sensitivity of exercise testing may be compromised by a number of factors, including lower exercise capacity. Premenopausal women have a lower pretest probability of CAD than men , which poses a problem for any noninvasive approaches to the diagnosis of CAD. Moreover, various non-Bayesian factors influence the diagnostic performance of the tests, including mitral valve prolapse and syndrome X .
Bayesian factors may compromise the accuracy of myocardial perfusion scintigraphy. Anterior wall image artifacts due to breast attenuation may cause reduced specificity and may be avoided by judicious interpretation of transaxial images. Newer technologies, such as attenuation correction and use of technetium isotopes , which enable the recognition of normal function in an apparent perfusion defect, may also reduce false positive results. However, the sensitivity of myocardial perfusion imaging may be lower in women than men . Other non-Bayesian factors that may influence the sensitivity of this test include the smaller size of the female heart, the greater prevalence of mild CAD and the performance of submaximal exercise in women.
Exercise echocardiography is an accurate and cost-efficient test for the diagnosis of CAD in women [16–18]. Even after the exclusion of patients influenced by referral bias, the specificity of exercise echocardiography exceeds that of ST segment evaluation alone . However, the sensitivity of this technique is less than that in mixed populations, which probably reflects the performance of submaximal exercise, as discussed earlier. The use of pharmacologic stress testing may be a means of circumventing this limitation .
3.2 Prognostic evaluation of known or suspected CAD in women.
Few data have addressed gender-based differences in the risk stratification of CAD. In the single published study of the prognostic significance of exercise testing according to gender , exercise testing was helpful in both men and women for the assessment of long-term survival. However, only in men were the results of exercise testing able to identify a high risk subset in whom operation was justified. Although reduced exercise capacity and ST segment depression correlated with reduced survival in our study, these factors were not independently predictive of adverse outcome.
Despite the diagnostic limitations of exercise stress myocardial perfusion imaging in women, patients with three-vessel or left main CAD may be accurately identified using this technique . Several recent studies have indicated that the ability of myocardial perfusion imaging to identify women with high risk coronary anatomy is paralleled by ability to predict subjects liable to have cardiac complications during follow-up. In a study of 80 women followed up over 2 years, among whom the cardiac event rate was 1.3%/year, the predictive value of negative dobutamine–atropine stress results combined with technetium 99m-sestamibi tomography was high . In 212 women with normal LV function studied by Pancholy et al. , age and large myocardial perfusion abnormalities were independent predictors of death and myocardial infarction. A recent study by Hachamovitch et al. showed that the value of perfusion imaging incremental to clinical and exercise testing data was greater in women than men.
3.3 Use of exercise echocardiography for prognostic evaluation.
The prognostic implications of exercise echocardiography in populations of mixed (predominantly male) gender have been demonstrated in previous studies. This test is effective for determining prognosis in patients with chest pain syndromes and after myocardial infarction . However, no study to date has examined the prognostic utility of exercise echocardiography in a large group of women.
The ability of exercise echocardiography to predict future cardiac events in women might be different than in men for several reasons: 1) From a clinical standpoint, women with CAD present at an older age, have a higher incidence of diabetes, have less extensive coronary disease and have a worse postinfarction outcome than men . Each of these clinical variables has obvious prognostic implications. Also, there are differences in exercise physiology and the cardiac response to exercise between women and men. Women generally have a lower exercise capacity and are more likely to have a submaximal heart rate response than men, which may adversely affect the accuracy of exercise echocardiography [18, 31]. Moreover, although the interpretation of this test is dependent on analysis of regional as well as global LV function, an abnormal ejection fraction response to exercise is a less reliable indicator of CAD in women than men . Nonetheless, the current study confirms the independent and incremental prognostic value of exercise echocardiography in a large heterogeneous group of women.
3.4 Study limitations.
Some limitations are inherent in this study. The report is potentially biased by involving patients at a tertiary referral center. However, appropriate training has been shown to optimize the accuracy of observers, thus enhancing the applicability of these data in other circumstances.
Women who had revascularization performed within 3 months of exercise echocardiography were excluded because the test may have influenced this outcome. However, although inclusion of early revascularization as an end point would have unfairly improved the positive predictive value of the test, their exclusion probably reduced the complication rate in women with positive test results. Moreover, women with chest pain syndromes generally have a more benign prognosis than men—attested to in this study by an event rate of only 7% (n = 36) in the 480 women who did not have primary revascularization or non–cardiac-related death. For this reason, a larger number of women will be necessary to show the utility of the test in the prediction of cardiac-related death.
Exercise echocardiography offers valuable prognostic information for defining high risk groups of women with chest pain syndromes. This test may offer important information to help the physician distinguish those women who require further expensive diagnostic and therapeutic therapies from those patients who can be treated more conservatively.
- coronary artery bypass graft surgery
- coronary artery disease
- confidence interval
- electrocardiogram, electrocardiographic
- left ventricular
- metabolic equivalents
- odds ratio
- percutaneous transluminal coronary angioplasty
- Received January 8, 1997.
- Revision received March 21, 1997.
- Accepted April 16, 1997.
- The American College of Cardiology
- ↵(1996) Heart and Stroke Facts: 1996 Statistical Supplement (American Heart Association, Dallas (TX)), pp 2–3.
- Orencia A,
- Bailey K,
- Yawn BP,
- Kottke TE
- Weiner DA,
- Ryan TJ,
- McCabe CH,
- et al.
- McNeer JF,
- Margolis JR,
- Lee KL,
- et al.
- Barolsky SM,
- Gilbert CA,
- Faruqui A,
- Nutter DO,
- Schlant RC
- Melin JA,
- Wijns W,
- Vanbutsele RJ,
- et al.
- Hung J,
- Chaitman BR,
- Lam J,
- Lesperance J,
- Dupras G,
- Fines P
- Guiteras P,
- Chaitman BR,
- Waters DD,
- Bourassa MG,
- Scholl JM,
- Ferguson RJ,
- Wagniart P
- Linhart JW,
- Laws JG,
- Satinsky JD
- Weiner DA,
- McCabe CH,
- Fisher L
- Sawada SG,
- Ryan T,
- Fineberg NS,
- Armstrong WF,
- Judson WE,
- McHenry PL
- Marwick TH,
- Anderson T,
- Williams MJ,
- et al.
- Anderson KM,
- Wilson PWF,
- Odell PM,
- Kannel WB
- Amanullah AM,
- Kiat H,
- Friedman JD,
- Berman DS
- Chae SC,
- Heo J,
- Iskandrian AS,
- Wasserleben V,
- Cave V
- Pancholy S,
- Fattah AA,
- Kamal AM,
- et al.
- Hachamovitch R,
- Berman DS,
- Kiat H,
- et al.
- Marwick TH,
- Nemec JJ,
- Pashkow FJ,
- Stewart WJ,
- Salcedo EE
- Hanley PC,
- Zinsmeister AR,
- Clements IP,
- Bove AA,
- Brown ML,
- Gibbons RJ
- Picano E,
- Lattanzi F,
- Orlandini A,
- Marini C,
- L’Abbate A