Philippe Ritter

University of Bordeaux, Burdeos, Aquitaine, France

Are you Philippe Ritter?

Claim your profile

Publications (135)382.99 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: Background Left Bundle Branch Block (LBBB) leads to prolonged left ventricular (LV) total activation time (TAT) and ventricular electrical uncoupling (VEU: mean LV activation time minus mean right ventricular (RV) activation time); both have been shown to be preferential targets for cardiac resynchronization therapy (CRT). Whether RV apical pacing (RVAP) produces similar ventricular activation patterns has not been well studied. Objective To compare electrical ventricular activation patterns during RVAP and LBBB. Methods We performed electrocardiographic mapping (ECM) during sinus rhythm, RVAP and CRT in 24 patients with LBBB. Results We observed differences in the electrical activation pattern with RVAP compared to LBBB. During LBBB, RV activation occurred rapidly; in contrast RV activation was prolonged during RVAP (46±21 vs. 69±17 ms, p<0.001). There was no significant difference in LVTAT, however, differences in conduction pattern were observed. During LBBB LV activation was circumferential whereas with RVAP it proceeded from apex-to-base. Differences in the number, size and orientation of lines of slow conduction were also observed. With LBBB, VEU was nearly twice as long as during RVAP (73±12 vs. 38±21ms, p<0.001). CRT resulted in a greater reduction in VEU relative to LBBB activation (p<0.001). Conclusion RVAP produces significant differences in ventricular activation characteristics compared to LBBB. Significantly less VEU occurs with RVAP and as a result CRT produces a smaller relative reduction in VEU. This may explain the finding that CRT appears to be more effective in patients with LBBB than in patients upgraded because of high percentages of RV pacing.
    Heart Rhythm. 10/2014;
  • [Show abstract] [Hide abstract]
    ABSTRACT: We describe a family with suspicion of genetic arrhythmia that has benefited from a wide genetic exploration. The eldest of the siblings presented syncope at age 5.5 years and cardiac explorations were normal. A few months later, her elder sister presented a sudden death at age 4.5 years, while she was playing in the garden. The cardiac explorations showed a heart of normal structure but presence of polymorphic premature ventricular complexes. Isoprenaline test was positive. Treatment with beta-blockers (nadolol 50 mg/m2) was introduced. There was no family history of sudden death or other cardiac defects. Because of these two serious rhythmic events occurring in two young children, a genetic study was initiated by next generation sequencing of 42 genes involved in cardiac arrhythmias (long QT, Brugada, cathecholaminergic ventricular tachycardia). Two heterozygous mutations (c.613C > T/p.Gln205* and c.22 + 29 A > G) were identified in the Triadin gene, coding for a protein of the calcium release complex, recently involved in cathecholaminergic ventricular tachycardia in two families (Roux-Buisson et al., 2012). The parents of our two cases were each carriers of a heterozygous mutation and had no cardiac symptoms. Their cardiac assessment did not show any abnormality (ECG Holter, exercise test, Isoprenaline test). The nonsense p.Gln205* mutation was present in one of the published families; however the splicing mutation in intron 1 had never been identified. Minigene experiments helped to confirm its pathogenicity. Presymptomatic testing was then proposed to the third child of the family (age 3), finding the two pathogenic mutations. She was therefore put under the same treatment as her sisters. This is the second report of an autosomal recessive cathecholaminergic ventricular tachycardia due to the Triadin gene. This case illustrates the interest of Next Generation Sequencing exploring simultaneously several candidate genes, in cases of sudden death of unknown origin.
    Archives of Cardiovascular Diseases. 09/2014; 107(s 8–9):496.
  • [Show abstract] [Hide abstract]
    ABSTRACT: Background Operators who extract cardiac devices are exposed to considerable irradiation and excess risk of radiation-induced disorders. A dedicated radioprotection cabin was developed to offer complete protection against radiation. This randomized study was designed to ascertain the protection against radiation conferred by a radioprotection cabin and the safety during extraction of cardiac devices. Methods 37 consecutive patients presenting with indication for extraction of cardiac device were randomly assigned to standard extraction technique (n=19), versus extraction with the use of a radiation protection cabin (n=18). Fluoroscopic exposure was compared using electronic dosimeters placed on the thorax, back, foot and head of the operator. Results The procedural times and total fluoroscopic exposure times and the complication rates were not significantly different between the 2 groups. The mean dose of radiation delivered to the thorax and back was similar in both groups (p=0.3 and p=0.8, respectively). In contrast, the mean doses of radiation delivered to the head and to the feet were respectively 68 and 390 times lower in the cabin group than in the control group (p<0.001). Conclusion The cabin offers a nearly full body radioprotection and eliminated the need to wear a lead apron, without increasing procedural time or complication rate during cardiac device extraction.
    The Canadian journal of cardiology 08/2014; · 3.12 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: -A specific non-invasive signal processing was applied to identify drivers in distinct categories of persistent atrial fibrillation (PsAF).
    Circulation 07/2014; · 15.20 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: -Although the Task Force Criteria (TFC) for ARVC have recently been updated the diagnosis remains challenging in the early stages. The aim of this study was to evaluate the diagnostic value of β-adrenergic stimulation in arrhythmogenic right ventricular cardiomyopathy (ARVC).
    Circulation Arrhythmia and Electrophysiology 06/2014; · 5.95 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Recent studies have demonstrated that LV pacing site is a critical parameter in optimizing cardiac resynchronization therapy (CRT). The present study evaluates the effect of pacing from different LV locations on QRS duration (QRSd) and their relationship to acute hemodynamic response in congestive heart failure (CHF) patients.
    Journal of Cardiovascular Electrophysiology 06/2014; · 3.48 Impact Factor
  • Cardiac electrophysiology clinics 06/2014; 6(2):207–216.
  • [Show abstract] [Hide abstract]
    ABSTRACT: Acute haemodynamic measurements can be used for AV delay (AVD) optimisation of cardiac resynchronisation therapy (CRT). It is uncertain whether non-invasive measurements are as reliable as those measured invasively.
    Heart (British Cardiac Society) 06/2014; 100 Suppl 3:A22-3. · 5.01 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Left pre-ejection period (LPEP), the interval from onset of ventricular depolarization to the beginning of aortic ejection, is shortened with CRT. We studied the effect of altering AV delay on LPEP in patients with CRT.
    Heart (British Cardiac Society) 06/2014; 100(Suppl 3):A22. · 5.01 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Insufficient correction of mechanical dyssynchrony is a cause of non-response to cardiac resynchronization therapy (CRT). To determine if CRT delivery could be optimized during the implantation procedure by choosing the number and location of pacing sites using echocardiography guidance. In patients with a QRS≥150ms or a QRS<150ms and criteria for mechanical dyssynchrony, the objective of the implantation procedure was to shorten the left pre-ejection interval (LPEI), measured online, by at least 10ms compared with standard biventricular configuration, by moving the right ventricular (RV) lead at different locations and, if necessary, by adding a second RV lead. Ninety-one patients (70 men; mean age 73±10 years; left ventricular [LV] ejection fraction 29±10%) were included. The final pacing configuration was standard biventricular in 15 (17%) patients, optimized biventricular in 22 (24%) and triple-site ventricular in 54 (59%). LPEI was shortened by ≥ 10ms compared with standard biventricular stimulation in 73 (80%) patients. Compared with standard biventricular pacing, the final optimized pacing configuration improved global intraventricular synchrony (decreasing LPEI from 158±36ms to 134±29ms; P<0.001), LV systolic efficiency (decreasing LPEI/LV ejection time from 0.58±0.18 to 0.46±0.13; P<0.001) and LV filling (increasing LV filling time/RR from 44±8% to 47±7%; P<0.001) and decreased mitral valve regurgitation. Intraoperative echocardiography-guided placement of RV lead(s) during CRT implantation is feasible and acutely improves LV synchrony compared with standard biventricular stimulation.
    Archives of cardiovascular diseases 04/2014; · 0.66 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Although cardiac resynchronization therapy (CRT) is effective in most patients with heart failure (HF) and ventricular dyssynchrony, a significant minority of patients (approximately 30%) are non-responders. Optimal atrioventricular and interventricular delays often change over time and reprogramming these intervals might increase CRT effectiveness. The SonR algorithm automatically optimizes atrioventricular and interventricular intervals each week using an accelerometer to measure change in the SonR signal, which was shown previously to correlate with hemodynamic improvement (left ventricular [LV] dP/dtmax). The RESPOND CRT trial will evaluate the effectiveness and safety of the SonR optimization system in patients with HF New York Heart Association class III or ambulatory IV eligible for a CRT-D device. Enrolled patients will be randomized in a 2:1 ratio to either SonR CRT optimization or to a control arm employing echocardiographic optimization. All patients will be followed for at least 24 months in a double-blinded fashion. The primary effectiveness end point will be evaluated for non-inferiority, with a nested test of superiority, based on the proportion of responders (defined as alive, free from HF-related events, with improvements in New York Heart Association class or improvement in Kansas City Cardiomyopathy Questionnaire quality of life score) at 12 months. The required sample size is 876 patients. The two primary safety end points are acute and chronic SonR lead-related complication rates, respectively. Secondary end points include proportion of patients free from death or HF hospitalization, proportion of patients worsened, and lead electrical performance, assessed at 12 months. The RESPOND CRT trial will also examine associated reverse remodeling at 1 year.
    American heart journal 04/2014; 167(4):429-36. · 4.65 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To enhance understanding of the working mechanism of cardiac resynchronization therapy (CRT) by comparing animal experimental, clinical, and computational data on the hemodynamic and electromechanical consequences of left ventricular and biventricular pacing (LVP and BiVP, respectively). It is unclear why LVP and BiVP have comparative positive effects on hemodynamic function of patients with dyssynchronous heart failure (HF). Hemodynamic response to LVP and BiVP (%-change LVdP/dtmax) was measured in 6 dogs and 24 patients with HF and left-bundle branch block (LBBB), followed by computer simulations of local myofiber mechanics during LVP and BiVP in the failing heart with LBBB. Pacing-induced changes of electrical activation were measured in dogs using contact mapping and in patients using a noninvasive multielectrode electrocardiographic mapping technique. LVP and BiVP similarly increased LVdP/dtmax in dogs and in patients, but only BiVP significantly decreased electrical dyssynchrony. In the simulations, LVP and BiVP increased total ventricular myofiber work to the same extent. While the LVP-induced increase was entirely due to enhanced right ventricular (RV) myofiber work, the BiVP-induced increase was due to enhanced myofiber work of both the LV and RV. Overall, LVdP/dtmax correlated better with total ventricular myofiber work than with LV or RV myofiber work alone. Experimental, human, and computational data support the similarity of hemodynamic response to LVP and BiVP, despite differences in electrical dyssynchrony. The simulations provide the novel insight that, through ventricular interaction, the RV myocardium importantly contributes to the improvement in LV pump function induced by CRT.
    Journal of the American College of Cardiology 08/2013; · 14.09 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To analyze left ventricular (LV) obstruction in hypertrophic cardiomyopathy (HCM) during exercise echocardiography. Despite the association of symptoms with LV outflow tract obstruction in HCM, there exist paradoxical situations in which significant intraventricular gradients (>50 mmHg) at rest occur in conjunction with excellent exercise tolerance. To examine this phenomenon we performed exercise echocardiography and analyzed the clinical status in 107 HCM patients with and without resting obstruction. At rest, 69 patients had no obstruction while 38 exhibited an intraventricular gradient, 9 of whom exhibited a decrease in gradient of at least 30 mmHg (99±35 to 30±14 mmHg, p<0.001) during exercise (paradoxical response to exercise or PRE). PRE patients presented a significantly lower NYHA clinical class and higher left ventricular volumes and arterial pressure both at rest and during exercise than HCM patients in whom the gradient increased or did not change during stress echocardiography. Finally, PRE patients exhibited a trend toward a reduced rate of cardiac events. Our study reports a subgroup of HCM patients, designated PRE, based on a decreased intraventricular gradient during exercise. The reduced exertional obstruction may account for the better functional class and trend to less clinical events in PRE patients.
    Journal of the American College of Cardiology 06/2013; · 14.09 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: OBJECTIVES: To investigate whether noninvasive electrocardiographic activation mapping (ECM) is a useful method for predicting response to Cardiac Resynchronization Therapy (CRT). BACKGROUND: One third of the patients appear not to respond to CRT when they are selected according to QRS duration. METHODS: We performed ECM in 33 consecutive CRT-candidates (QRS≥120ms). In 18 patients the 12-lead ECG morphology was Left Bundle-Branch Block (LBBB) and in 15 Nonspecific Intraventricular Conduction Disturbance (NICD).Three indices of electrical dyssynchrony were derived from intrinsic maps: right and left ventricular total activation times and ventricular electrical uncoupling (VEU: difference between the LV and RV mean activation times). We assessed the ability of these parameters to predict response, measured using a clinical composite score, after 6 months treatment with CRT. RESULTS: Electrocardiographic maps revealed homogeneous patterns of activation and consistently greater VEU and LV total activation time in patients with LBBB compared to heterogeneous activation sequences and shorter VEU and LV total activation time in NICD patients (VEU: 75±12ms vs. 40±22ms; p<0.001/ LVTAT: 115±21ms vs 91±34ms; p=0.03). LBBB and NICD patients had similar RV total activation times (62±30ms vs 58±26ms; p=0.7). The area under the receiver operating characteristic curve indicated that VEU (AUC: 0.88) was significantly superior to QRS duration (AUC: 0.73) and LVTAT (AUC: 0.72) for predicting CRT response (p<0.05). With a 50ms cut-off value, VEU identified CRT responders with 90% sensitivity and 82% specificity whether LBBB was present or not. CONCLUSION: Ventricular electrical uncoupling measured by electrocardiographic mapping predicted clinical CRT response better than QRS duration or the presence of LBBB.
    Journal of the American College of Cardiology 04/2013; · 14.09 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: AIMS: The long-term clinical value of the optimization of atrioventricular (AVD) and interventricular (VVD) delays in cardiac resynchronization therapy (CRT) remains controversial. We studied retrospectively the association between the frequency of AVD and VVD optimization and 1-year clinical outcomes in the 199 CRT patients who completed the Clinical Evaluation on Advanced Resynchronization study.METHODS AND RESULTS: From the 199 patients assigned to CRT-pacemaker (CRT-P) (New York Heart Association, NYHA, class III/IV, left ventricular ejection fraction <35%), two groups were retrospectively composed a posteriori on the basis of the frequency of their AVD and VVD optimization: Group 1 (n = 66) was composed of patients 'systematically' optimized at implant, at 3 and 6 months; Group 2 (n = 133) was composed of all other patients optimized 'non-systematically' (less than three times) during the 1 year study. The primary endpoint was a composite of all-cause mortality, heart failure-related hospitalization, NYHA functional class, and Quality of Life score, at 1 year. Systematic CRT optimization was associated with a higher percentage of improved patients based on the composite endpoint (85% in Group 1 vs. 61% in Group 2, P < 0.001), with fewer deaths (3% in Group 1 vs. 14% in Group 2, P = 0.014) and fewer hospitalizations (8% in Group 1 vs. 23% in Group 2, P = 0.007), at 1 year.CONCLUSION: These results further suggest that AVD and VVD frequent optimization (at implant, at 3 and 6 months) is associated with improved long-term clinical response in CRT-P patients.
    Europace 03/2013; · 2.77 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: BACKGROUND: Reproducibility and hemodynamic efficacy of optimization of AV delay (AVD) of cardiac resynchronization therapy (CRT) using invasive LV dp/dtmax are unknown. METHOD AND RESULTS: 25 patients underwent AV delay (AVD) optimisation twice, using continuous left ventricular (LV) dp/dtmax, systolic blood pressure (SBP) and pulse pressure (PP). We compared 4 protocols for comparing dp/dtmax between AV delays: We assessed for dp/dtmax, LVSBP and LVPP, test-retest reproducibility of the optimum. Optimization using immediate absolute dp/dtmax had poor reproducibility (SDD of replicate optima=41ms; R2=0.45) as did delayed absolute (SDD 39ms; R2=0.50). Multiple relative had better reproducibility: SDD 23ms, R2=0.76, and (p<0.01 by F test). Compared with AAI pacing, the hemodynamic increment from CRT, with the nominal AV delay was LVSBP 2% and LVdp/dtmax 5%, while CRT with pre-determined optimal AVD gave 6% and 9% respectively. CONCLUSIONS: Because of inevitable background fluctuations, optimization by absolute dp/dtmax has poor same-day reproducibility, unsuitable for clinical or research purposes. Reproducibility is improved by comparing to a reference AVD and making multiple consecutive measurements. More than 6 measurements would be required for even more precise optimization - and might be advisable for future study designs. With optimal AVD, instead of nominal, the hemodynamic increment of CRT is approximately doubled.
    International journal of cardiology 03/2013; · 6.18 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: BACKGROUND: -Contact force (CF) is an important determinant of lesion formation for atrial endocardial radiofrequency (RF) ablation. There are minimal published data on CF and ventricular lesion formation. We studied the impact of CF on lesion formation using an ovine model both endo and epicardially. METHODS AND RESULTS: -Twenty sheep received 160 epicardial and 160 endocardial ventricular RF applications using either a 3.5 mm irrigated-tip catheter (Thermocool, Biosense-Webster, n=160) or a 3.5 irrigated-tip catheter with CF assessment (Tacticath, Endosense, n=160), via percutaneous access. Power was delivered at 30 watts for 60 seconds when either catheter/tissue contact was felt to be good or when CF>10g with Tacticath. Following completion of all lesions acute dimensions were taken at pathology. Identifiable lesion formation from RF application was improved with the aid of CF information, from 78% to 98% on the endocardium (p<0.001) and from 90% to 100% on the epicardium (p=0.02). The mean total force was greater on the endocardium (39±18g vs 21±14g for the epicardium, p<0.001) mainly due to axial force. Despite the Force-Time-Integral being greater endocardially, epicardial lesions were larger (231±182mm(3) vs 209±131mm(3); p=0.02) probably due to the absence of the heat sink effect of the circulating blood and covered a greater area (41±27 vs 29±17mm(2); p=0.03) due to catheter orientation. CONCLUSIONS: -In the absence of CF feedback, 22% of endocardial RF applications that are felt to have good contact didn't result in lesion formation. Epicardial ablation is associated with larger lesions.
    Circulation Arrhythmia and Electrophysiology 02/2013; · 5.95 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: INTRODUCTION: Response rate after cardiac resynchronization therapy (CRT) remains suboptimal. We sought to identify pre- and intraprocedural predictors of response using MRI. METHODS AND RESULTS: Sixty patients underwent MRI before CRT. Left ventricular (LV) volumes and ejection fraction were assessed on cine images. Intra-LV dyssynchrony was defined as the maximal delay between first peaks of radial wall motion over 20 segments. Myocardial scar extent was quantified using delayed-enhanced MRI. After CRT, the paced LV segment was characterized on preprocedural MRI with respect to presence of scar and mechanical delay, the latter being quantified using time to first peak of wall motion, expressed in percentage of the total LV activation. Echocardiography was performed before and 6 months after CRT to quantify reverse remodeling (RR). Mean RR at 6 months was 30 ± 29% of baseline LV end-systolic volume. At univariate analysis, RR related to baseline LV end-diastolic and end-systolic volumes (R(2) = 0.101, P = 0.01; R(2) = 0.072, P = 0.04), intra-LV mechanical dyssynchrony (R(2) = 0.351, P < 0.0001), scar extent (R(2) = 0.273, P < 0.0001), and presence of scar at pacing site (R(2) = 0.100, P = 0.01). QRS duration and mechanical delay at pacing site were not found related to RR (R(2) = 0.041, P = 0.12 and R(2) = 0.012, P = 0.4, respectively). At multivariate analysis intra-LV mechanical dyssynchrony, scar extent, and LV end-diastolic volume were independent predictors of RR (R(2) = 0.307, P = 0.001; R(2) = 0.096, P = 0.002, R(2) = 0.078, P = 0.005, respectively). CONCLUSION: Intra-LV dyssynchrony and scar extent are independent predictors of RR after CRT. Scar at pacing site is associated to a lesser response to CRT. Mechanical delay at this site has no impact on the response.
    Journal of Cardiovascular Electrophysiology 01/2013; · 3.48 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Although cardiac resynchronization therapy (CRT) is effective in most patients with heart failure (HF) and ventricular dyssynchrony, a significant minority of patients (approximately 30%) are non-responders. Optimal atrioventricular (AV) and interventricular (VV) delays often change over time and reprogramming these intervals might increase CRT effectiveness. The SonR™ algorithm automatically optimizes AV and VV intervals each week using an accelerometer to measure change in the SonR signal, which was shown previously to correlate with hemodynamic improvement (LV dP/dtmax). The RESPOND CRT trial will evaluate the effectiveness and safety of the SonR optimization system in patients with HF NYHA class III or ambulatory IV eligible for a CRT-D device. Enrolled patients will be randomized in a 2:1 ratio to either SonR CRT optimization or to a control arm employing echocardiographic optimization. All patients will be followed for at least 24 months in a double-blinded fashion. The primary effectiveness end point will be evaluated for non-inferiority, with a nested test of superiority, based on the proportion of responders (defined as alive, free from HF-related events, with improvements in NYHA class or improvement in Kansas City Cardiomyopathy Questionnaire quality of life score) at 12 months. The required sample size is 876 patients. The two primary safety end points are acute and chronic SonR lead-related complication rates, respectively. Secondary end points include proportion of patients free from death or HF hospitalization, proportion of patients worsened, and lead electrical performance, assessed at 12 months. The RESPOND CRT trial will also examine associated reverse remodeling at 1 year.
    American Heart Journal. 01/2013;
  • [Show abstract] [Hide abstract]
    ABSTRACT: OBJECTIVES: Management of pacemaker infection in pacing-dependent patients is often challenging. Typically, temporary pacing is used while antibiotic therapy is given for a number of days before reimplantation of a new endocardial system. This results in a prolonged hospital stay and complications associated with temporary pacing. In this study, we examine the feasibility of performing a single combined procedure of epicardial pacemaker implantation followed by system extraction. METHODS: One hundred consecutive infected pacemaker-dependent patients underwent implantation of 2 epicardial ventricular leads and were converted to a ventricular demand pacing system. The infected pacing system was then extracted during the same procedure. Patients were followed up for 12 months. RESULTS: Significant pericardial bleeding developed during the procedure in 3 patients. The presence of the pericardial drain positioned during the implantation of the epicardial pacing system meant that cardiac tamponade did not occur, allowing surgical repair with sternotomy to be carried out under stable hemodynamic conditions. Two of these 100 patients died in the 30-day postoperative period; 1 death was due to septic shock and 1 to pulmonary distress. Median 1-year epicardial pacing thresholds were stable and excellent (1.4 ± 0.9 volts). However, 1 of the 2 leads developed increased thresholds in 6 patients, which led to the exclusive use of other ventricular lead. CONCLUSIONS: A single combined procedure of surgical epicardial pacemaker implantation and pacemaker system extraction appears to be a safe and effective method for managing pacemaker-dependent patients with infected pacemakers.
    The Journal of thoracic and cardiovascular surgery 09/2012; · 3.41 Impact Factor

Publication Stats

3k Citations
382.99 Total Impact Points

Institutions

  • 2009–2014
    • University of Bordeaux
      Burdeos, Aquitaine, France
    • Université Victor Segalen Bordeaux 2
      Burdeos, Aquitaine, France
    • Université de Rennes 2
      Roazhon, Brittany, France
  • 2013
    • Imperial College London
      • International Centre for Circulatory Health
      Londinium, England, United Kingdom
    • Maastricht Universitair Medisch Centrum
      Maestricht, Limburg, Netherlands
  • 2010–2012
    • Centre Hospitalier Universitaire de Bordeaux
      Burdeos, Aquitaine, France
  • 2008
    • University Hospital of Heraklion
      Irákleio, Attica, Greece
  • 2001
    • Centre Hospitalier Universitaire de Rennes
      • Service de cardiologie et maladies vasculaires
      Roazhon, Brittany, France
  • 1994
    • St George's, University of London
      Londinium, England, United Kingdom