Lack of Correlations between Electrophysiological and Anatomical-Mechanical Atrial Remodeling in Patients with Atrial Fibrillation PHILIPPE MAURY, M.D.,* EMILIE THOMSON, M.D.,* ANNE ROLLIN, M.D.,* MATHIEU BERRY, M.D.,* THOMAS COGNET, M.D.,* ALEXANDRE DUPARC, M.D.,* PIERRE MONDOLY, M.D.,* MATHIEU GAUTIER, M.D.,* OLIVIER LAIREZ, M.D.,* ´ SIMON MEJEAN, M.D.,* PIERRE MASSABUAU, M.D.,* CHRISTELLE CARDIN, M.D.,* ´ STEPHANE COMBES, M.D.,† JEAN-PAUL ALBENQUE, M.D.,† and NICOLAS COMBES, M.D.† *From the Department of Cardiology, University Hospital Rangueil, Toulouse, France; and †Department of Cardiology, Clinique Pasteur, Toulouse, France
Background: Atrial fibrillation (AF) progressively leads to electrical remodeling (ER) and anatomicalmechanical remodeling (AR), whose relationships in humans remain poorly known. Methods: ER and AR were compared in patients undergoing percutaneous radiofrequency (RF) ablation for AF. ER was defined by right and left appendage activation rates as a surrogate for atrial refractory periods. AR was approached by left atrial (LA) diameters and area and left atrial appendage (LAA) area and contractile function (mean emptying flow velocity) (LAAFV) before RF ablation. Mean duration between successive LAA contractions was considered as LAA mechanical rate. Results: Forty-one patients (31 men, age: 64 ± 9 years) with paroxysmal (27%), persistent (61%), or longpersistent AF (12%) were prospectively included (ejection fraction: 44 ± 16%). Parameters exploring AR were highly correlated to each other: LA area (28 ± 7 cm2 ), LAA area (5.7 ± 2.25 cm2 ), LA transverse (49 ± 7 mm), and anteroposterior diameter (59 ± 13 mm) or LAAFV (29 ± 13 cm/s; P < 0.05 for each comparison). Parameters exploring ER were also highly correlated: right atrial appendage (RAA; 181 ± 39 ms) and LAA (176 ± 33 ms) activation rates (P < 0.0001). There was no significant correlation between any ER and AR parameter. Only LAA mechanical rate (174 ± 36 ms) was correlated to LAA or RAA activations rates (P ࣘ 0.01). Conclusion: ER and AR are not mutually related, atrial activation rate being not correlated to LA or LAA size or function. Thus, the mechanisms leading to AF-induced atrial remodeling may differ for anatomical and electrophysiological aspects. (PACE 2015; 38:617–624) atrial fibrillation, atrial remodeling, refractory periods
Introduction Following the pioneer work by Allessie and colleagues, it has been largely demonstrated that “atrial fibrillation begets atrial fibrillation.”1 Atrial fibrillation (AF) is known to lead to a shortening of atrial action potential durations and refractoriness and to a loss of rate adaptation of atrial refractory periods.1 This happens mainly through a reduction of ICaL current, with a steady state reached after only a few days.2 This has been defined under the accepted terminology of “electrical remodeling” (ER).2 It was also early Address for reprints: Philippe Maury, M.D., Department of Cardiology, University Hospital Rangueil, 31059 Toulouse Cedex 09, France. Fax: 33 (0) 5 61 32 22 46; e-mail: [email protected] Received November 3, 2014; revised December 21, 2014; accepted January 14, 2015. doi: 10.1111/pace.12598
recognized that atrial electrical activation rate increases while atrial refractory periods decrease during AF.1 It is also well accepted that AF induces further modifications known under the term of “contractile and structural remodeling.”2 These changes associate a progressive atrial contractile dysfunction—also mainly through depression of ICaL current—together with some dedifferentiation of atrial cells and degenerative changes such as fibrosis—which follow a more progressive evolution.2–7 Experimentally, electrical and contractile/ structural remodeling follow the same time course, at least in the short term, with a parallel decrease in both atrial refractory periods and contractility over the first minutes of AF, followed by simultaneous further decrease in refractory periods and increase in atrial dimensions over the following days.2 Thus, while both electrical and contractile remodeling occurs within a few
©2015 Wiley Periodicals, Inc. PACE, Vol. 38
May 2015
617
MAURY, ET AL.
days, structural remodeling follows a much slower time course. In clinical practice, both AF cycle lengths8–10 and anatomical changes11,12 seem correlated to the duration of AF. Although AF leads progressively to both ER and anatomical-mechanical remodeling (AR) in experiments and in humans, the relationships and correlations between both phenomena remain poorly known in clinical practice. The aims of this study were to collect parameters exploring both ER and AR and to assess their relationship in patients undergoing AF ablation. Methods We prospectively studied 41 patients undergoing percutaneous radiofrequency (RF; n = 31) or cryoablation (n = 10) for AF at both of our centers (University Hospital Rangueuil, Toulouse, France and Clinique Pasteur, Toulouse, France). Each patient presented with symptomatic recurrent AF despite antiarrhythmic drugs. Briefly, access to the left atrium was achieved using transseptal catheterization. Pulmonary vein isolation was performed by circumferential or ostial RF ablation in case of paroxysmal AF, followed by ablation of complex fragmented electrograms and linear ablation in patients with persistent AF with the aim of AF termination according to the usual stepwise approach.13 Cryoablation was performed by achieving pulmonary vein occlusion and electrical isolation using a balloon-mounted cryoablation system and a pressure curve at the tip of the catheter.14 A standard 4-mm-tip irrigated catheter (Navistar TM or Thermocool TM Biosense Webster, Diamond Bar, CA, USA) was used for mapping and ablation in the RF group (2-mm spacing between the two distal electrodes) while a 23-mm or 28-mm cryoballoon (ArticFront, CryoCath Technology, Medtronic Inc., Minneapolis, MN, USA) was used for ablation together with a quadripolar catheter (Xtrem, Sorin Group, Clamart, France; 5-mm interelectrode spacing) for mapping in the cryoablation group. A standard quadripolar catheter (5–10-mm interelectrode spacing) was furthermore placed in the coronary sinus (CS). ER was approached by measuring the averaged cycle length during AF at the right atrial appendage (RAA) and left atrial appendage (LAA) before any RF or cryoapplication. Since atrial refractory periods and atrial cycle length during AF are highly correlated, the local fibrillation interval (average of 20 cycle lengths between successive discrete atrial activations) might be used as a surrogate for local refractory periods.1,9,15 The averaged value of local AF cycle lengths was used since minimum, mean, and median AF
618
cycle length are all well correlated with refractory periods.16 AF cycle lengths were also similarly measured at the LA-pulmonary veins junctions and at the CS in the absence of local fragmented activation. AR was approached by measuring some anatomical and contractile parameters using echocardiography. Echocardiography was performed during AF either the day before or during the procedure before any ablation. For each patient, we collected the left atrial (LA) diameters (longitudinal and transversal) and LA area in apical four-chamber view using transthoracic echocardiography, together with the LAA area and mean emptying flow velocity (LAAFV) by transesophageal echocardiography. Diameters and areas were indexed to the body surface. All measurements were performed according to the standard guidelines for echocardiography.17 We also calculated the mean duration between successive LAA contractions during transesophageal echocardiography (average of 10 successive contractions), which was considered as the LAA mechanical rate. AF was present in each case at the time of echocardiographic and electrophysiological measurements, either spontaneous or induced by atrial bursts at the beginning of the procedure for some patients with paroxysmal AF (n = 8). Statistics Continuous data are expressed as mean ± standard deviation (or median and range in case of non-Gaussian distribution of values). Continuous variables were compared using unpaired t-test or nonparametric Mann-Whitney test as suitable, while categorical variables were compared using χ 2 test or Fisher’s exact test. Correlations between numerical values were tested using Pearson’s correlation or Spearman rank correlation test as appropriate. Patients were classified according to the duration of the ongoing AF episode—hours, weeks, months, and years—and comparisons between subgroups were performed with analysis of variance. Analysis and calculations were performed using StatView TM program (Abacus Concepts, Inc., Berkeley, CA, USA, 1992–1996, version 5.0). A P value 0.8), LA diameters and LAA area (P < 0.0001, rho = 0.7), LA area and LAA area (P = 0.0003, rho 0.6), LAA velocity and LAA area (P = 0.02, rho = – 0.37), LA area and LAA velocity (P = 0.03, rho = – 0.36), and LA diameters and LAA velocity (P = 0.04, rho = – 0.34 and P = 0.008, rho = – 0.43 for transversal and longitudinal diameters, respectively). Similar correlations were still present when echocardiographic measurements were indexed to the body surface (data not shown). Both parameters exploring ER (LAA and RAA activation rates) were also highly correlated (P < 0.0001, rho = 0.8; Fig. 1). Mean AF cycle
PACE, Vol. 38
Figure 1. High correlation between LAA and RAA activation rates. LAA = left atrial appendage; RAA = right atrial appendage.
lengths at the LA-pulmonary veins junction and CS were depicted in Table II and both significantly correlated to LAA and RAA cycle lengths (P < 0.05, rho > 0.5). There was no significant correlation between any ER and AR parameters: neither LA diameters, LA area, LAA area, nor LAA contractile function did significantly correlate to LAA or RAA activation rates, whether the diameters or areas were indexed to the body surface or not (see examples in Fig. 2). Only the LAA mechanical rate was correlated to both LAA (P = 0.01, rho 0.48) and RAA (P = 0.001, rho 0.61) activation rates (Fig. 3). A similar lack of correlation between AR and ER parameters was also present in the subgroups of patients without any previous atrial RF ablation, in patients with persistent AF, or in patients with paroxysmal AF (data not shown). When patients were classified according to the duration of the ongoing AF episode—hours,
May 2015
619
MAURY, ET AL.
Figure 2. Examples of lack of correlation between parameters exploring electrical remodeling and anatomical-mechanical remodeling. LA = left atrium; LAA = left atrial appendage; RAA = right atrial appendage.
Figure 3. Significant correlation between LAA mechanical rate and both LAA and RAA activation rates. LAA = left atrial appendage; RAA = right atrial appendage.
weeks, months, and years—it appeared that LA and LAA dimensions and contractile function were more altered in longer versus shorter episodes, while LAA and RAA activation rates and LAA mechanical rate were not significantly different (Table III and Fig. 4). RF ablation procedure acutely led to direct conversion in sinus rhythm in nine patients (22%) and to atrial tachycardia in four (10%), while a direct current (DC) shock was needed in 23 (56%) (performed in five other patients to allow better control of cryoablation according to the pressure curve technique14 ). Parameters exploring AR were correlated to the acute success of RF procedure: larger LAA area (6.6 ± 2.3 cm2 vs 5.1 ± 1.8 cm2 , P = 0.06), larger LA area (31 ± 7 cm2 vs 26 ± 6 cm2 , P = 0.06), and larger transversal (57 ± 10 mm vs 42 ± 14 mm, P = 0.001) and longitudinal (64 ± 9 mm vs 56 ± 14 mm, P = 0.05) diameters were present in the case of DC shock versus achievement of sinus rhythm/atrial tachycardia by ablation. There was no significant correlation between DC shock/AF termination by ablation and LAA velocity or with
620
LAA or RAA activation rates or LAA mechanical rate. AF recurred in 18 of 41 patients (44%) over a mean follow-up of 15 ± 7 months. Recurrences were correlated to some AR parameters with larger LAA area (6.7 ± 2.6 cm2 vs 4.9 ± 1.6 cm2 , P = 0.01), larger LA area (31 ± 7 cm2 vs 26 ± 6 cm2 , P = 0.03), and larger transversal (55 ± 14 mm vs 43 ± 15 mm, P = 0.02) and longitudinal (63 ± 11 mm vs 55 ± 13 mm, P = 0.04) diameters in case of recurrences, but there was no significant correlation between recurrences and LAA velocity, LAA or RAA activation rates, or LAA mechanical rate. Discussion In this study, we were not able to find any significant correlation between the parameters exploring both ER and AR in a population of 41 patients undergoing catheter ablation for AF of various durations. On the other hand, mean LAA and RAA cycle lengths—a surrogate for atrial
May 2015
PACE, Vol. 38
ELECTROPHYSIOLOGICAL AND ANATOMICAL-MECHANICAL REMODELING IN AF
Table III. ER and AR Parameters According to the Duration of the Ongoing AF Episode Hours n = 10 LA area (cm2 ) LA transversal diameter (mm) LA longitudinal diameter (mm) LAA area (cm2 ) LAA flow velocity (cm/s) RAA activation rate (ms) LAA activation rate (ms) LAA mechanical rate (ms)
21 27 40 3.5 41 172 169 172
± ± ± ± ± ± ± ±
3 3 3 0.8 17 16 13 13
Weeks n = 11 30 55 65 6.8 28 192 183 191
± ± ± ± ± ± ± ±
7 12 5 2.5 8 38 35 60
Months n = 15 31 55 66 6.1 21 181 182 166
± ± ± ± ± ± ± ±
7 7 8 1.9 9 44 44 30
Years n=5
P
± ± ± ± ± ± ± ±
0.001
Background: Atrial fibrillation (AF) progressively leads to electrical remodeling (ER) and anatomicalmechanical remodeling (AR), whose relationships in humans remain poorly known. Methods: ER and AR were compared in patients undergoing percutaneous radiofrequency (RF) ablation for AF. ER was defined by right and left appendage activation rates as a surrogate for atrial refractory periods. AR was approached by left atrial (LA) diameters and area and left atrial appendage (LAA) area and contractile function (mean emptying flow velocity) (LAAFV) before RF ablation. Mean duration between successive LAA contractions was considered as LAA mechanical rate. Results: Forty-one patients (31 men, age: 64 ± 9 years) with paroxysmal (27%), persistent (61%), or longpersistent AF (12%) were prospectively included (ejection fraction: 44 ± 16%). Parameters exploring AR were highly correlated to each other: LA area (28 ± 7 cm2 ), LAA area (5.7 ± 2.25 cm2 ), LA transverse (49 ± 7 mm), and anteroposterior diameter (59 ± 13 mm) or LAAFV (29 ± 13 cm/s; P < 0.05 for each comparison). Parameters exploring ER were also highly correlated: right atrial appendage (RAA; 181 ± 39 ms) and LAA (176 ± 33 ms) activation rates (P < 0.0001). There was no significant correlation between any ER and AR parameter. Only LAA mechanical rate (174 ± 36 ms) was correlated to LAA or RAA activations rates (P ࣘ 0.01). Conclusion: ER and AR are not mutually related, atrial activation rate being not correlated to LA or LAA size or function. Thus, the mechanisms leading to AF-induced atrial remodeling may differ for anatomical and electrophysiological aspects. (PACE 2015; 38:617–624) atrial fibrillation, atrial remodeling, refractory periods
Introduction Following the pioneer work by Allessie and colleagues, it has been largely demonstrated that “atrial fibrillation begets atrial fibrillation.”1 Atrial fibrillation (AF) is known to lead to a shortening of atrial action potential durations and refractoriness and to a loss of rate adaptation of atrial refractory periods.1 This happens mainly through a reduction of ICaL current, with a steady state reached after only a few days.2 This has been defined under the accepted terminology of “electrical remodeling” (ER).2 It was also early Address for reprints: Philippe Maury, M.D., Department of Cardiology, University Hospital Rangueil, 31059 Toulouse Cedex 09, France. Fax: 33 (0) 5 61 32 22 46; e-mail: [email protected] Received November 3, 2014; revised December 21, 2014; accepted January 14, 2015. doi: 10.1111/pace.12598
recognized that atrial electrical activation rate increases while atrial refractory periods decrease during AF.1 It is also well accepted that AF induces further modifications known under the term of “contractile and structural remodeling.”2 These changes associate a progressive atrial contractile dysfunction—also mainly through depression of ICaL current—together with some dedifferentiation of atrial cells and degenerative changes such as fibrosis—which follow a more progressive evolution.2–7 Experimentally, electrical and contractile/ structural remodeling follow the same time course, at least in the short term, with a parallel decrease in both atrial refractory periods and contractility over the first minutes of AF, followed by simultaneous further decrease in refractory periods and increase in atrial dimensions over the following days.2 Thus, while both electrical and contractile remodeling occurs within a few
©2015 Wiley Periodicals, Inc. PACE, Vol. 38
May 2015
617
MAURY, ET AL.
days, structural remodeling follows a much slower time course. In clinical practice, both AF cycle lengths8–10 and anatomical changes11,12 seem correlated to the duration of AF. Although AF leads progressively to both ER and anatomical-mechanical remodeling (AR) in experiments and in humans, the relationships and correlations between both phenomena remain poorly known in clinical practice. The aims of this study were to collect parameters exploring both ER and AR and to assess their relationship in patients undergoing AF ablation. Methods We prospectively studied 41 patients undergoing percutaneous radiofrequency (RF; n = 31) or cryoablation (n = 10) for AF at both of our centers (University Hospital Rangueuil, Toulouse, France and Clinique Pasteur, Toulouse, France). Each patient presented with symptomatic recurrent AF despite antiarrhythmic drugs. Briefly, access to the left atrium was achieved using transseptal catheterization. Pulmonary vein isolation was performed by circumferential or ostial RF ablation in case of paroxysmal AF, followed by ablation of complex fragmented electrograms and linear ablation in patients with persistent AF with the aim of AF termination according to the usual stepwise approach.13 Cryoablation was performed by achieving pulmonary vein occlusion and electrical isolation using a balloon-mounted cryoablation system and a pressure curve at the tip of the catheter.14 A standard 4-mm-tip irrigated catheter (Navistar TM or Thermocool TM Biosense Webster, Diamond Bar, CA, USA) was used for mapping and ablation in the RF group (2-mm spacing between the two distal electrodes) while a 23-mm or 28-mm cryoballoon (ArticFront, CryoCath Technology, Medtronic Inc., Minneapolis, MN, USA) was used for ablation together with a quadripolar catheter (Xtrem, Sorin Group, Clamart, France; 5-mm interelectrode spacing) for mapping in the cryoablation group. A standard quadripolar catheter (5–10-mm interelectrode spacing) was furthermore placed in the coronary sinus (CS). ER was approached by measuring the averaged cycle length during AF at the right atrial appendage (RAA) and left atrial appendage (LAA) before any RF or cryoapplication. Since atrial refractory periods and atrial cycle length during AF are highly correlated, the local fibrillation interval (average of 20 cycle lengths between successive discrete atrial activations) might be used as a surrogate for local refractory periods.1,9,15 The averaged value of local AF cycle lengths was used since minimum, mean, and median AF
618
cycle length are all well correlated with refractory periods.16 AF cycle lengths were also similarly measured at the LA-pulmonary veins junctions and at the CS in the absence of local fragmented activation. AR was approached by measuring some anatomical and contractile parameters using echocardiography. Echocardiography was performed during AF either the day before or during the procedure before any ablation. For each patient, we collected the left atrial (LA) diameters (longitudinal and transversal) and LA area in apical four-chamber view using transthoracic echocardiography, together with the LAA area and mean emptying flow velocity (LAAFV) by transesophageal echocardiography. Diameters and areas were indexed to the body surface. All measurements were performed according to the standard guidelines for echocardiography.17 We also calculated the mean duration between successive LAA contractions during transesophageal echocardiography (average of 10 successive contractions), which was considered as the LAA mechanical rate. AF was present in each case at the time of echocardiographic and electrophysiological measurements, either spontaneous or induced by atrial bursts at the beginning of the procedure for some patients with paroxysmal AF (n = 8). Statistics Continuous data are expressed as mean ± standard deviation (or median and range in case of non-Gaussian distribution of values). Continuous variables were compared using unpaired t-test or nonparametric Mann-Whitney test as suitable, while categorical variables were compared using χ 2 test or Fisher’s exact test. Correlations between numerical values were tested using Pearson’s correlation or Spearman rank correlation test as appropriate. Patients were classified according to the duration of the ongoing AF episode—hours, weeks, months, and years—and comparisons between subgroups were performed with analysis of variance. Analysis and calculations were performed using StatView TM program (Abacus Concepts, Inc., Berkeley, CA, USA, 1992–1996, version 5.0). A P value 0.8), LA diameters and LAA area (P < 0.0001, rho = 0.7), LA area and LAA area (P = 0.0003, rho 0.6), LAA velocity and LAA area (P = 0.02, rho = – 0.37), LA area and LAA velocity (P = 0.03, rho = – 0.36), and LA diameters and LAA velocity (P = 0.04, rho = – 0.34 and P = 0.008, rho = – 0.43 for transversal and longitudinal diameters, respectively). Similar correlations were still present when echocardiographic measurements were indexed to the body surface (data not shown). Both parameters exploring ER (LAA and RAA activation rates) were also highly correlated (P < 0.0001, rho = 0.8; Fig. 1). Mean AF cycle
PACE, Vol. 38
Figure 1. High correlation between LAA and RAA activation rates. LAA = left atrial appendage; RAA = right atrial appendage.
lengths at the LA-pulmonary veins junction and CS were depicted in Table II and both significantly correlated to LAA and RAA cycle lengths (P < 0.05, rho > 0.5). There was no significant correlation between any ER and AR parameters: neither LA diameters, LA area, LAA area, nor LAA contractile function did significantly correlate to LAA or RAA activation rates, whether the diameters or areas were indexed to the body surface or not (see examples in Fig. 2). Only the LAA mechanical rate was correlated to both LAA (P = 0.01, rho 0.48) and RAA (P = 0.001, rho 0.61) activation rates (Fig. 3). A similar lack of correlation between AR and ER parameters was also present in the subgroups of patients without any previous atrial RF ablation, in patients with persistent AF, or in patients with paroxysmal AF (data not shown). When patients were classified according to the duration of the ongoing AF episode—hours,
May 2015
619
MAURY, ET AL.
Figure 2. Examples of lack of correlation between parameters exploring electrical remodeling and anatomical-mechanical remodeling. LA = left atrium; LAA = left atrial appendage; RAA = right atrial appendage.
Figure 3. Significant correlation between LAA mechanical rate and both LAA and RAA activation rates. LAA = left atrial appendage; RAA = right atrial appendage.
weeks, months, and years—it appeared that LA and LAA dimensions and contractile function were more altered in longer versus shorter episodes, while LAA and RAA activation rates and LAA mechanical rate were not significantly different (Table III and Fig. 4). RF ablation procedure acutely led to direct conversion in sinus rhythm in nine patients (22%) and to atrial tachycardia in four (10%), while a direct current (DC) shock was needed in 23 (56%) (performed in five other patients to allow better control of cryoablation according to the pressure curve technique14 ). Parameters exploring AR were correlated to the acute success of RF procedure: larger LAA area (6.6 ± 2.3 cm2 vs 5.1 ± 1.8 cm2 , P = 0.06), larger LA area (31 ± 7 cm2 vs 26 ± 6 cm2 , P = 0.06), and larger transversal (57 ± 10 mm vs 42 ± 14 mm, P = 0.001) and longitudinal (64 ± 9 mm vs 56 ± 14 mm, P = 0.05) diameters were present in the case of DC shock versus achievement of sinus rhythm/atrial tachycardia by ablation. There was no significant correlation between DC shock/AF termination by ablation and LAA velocity or with
620
LAA or RAA activation rates or LAA mechanical rate. AF recurred in 18 of 41 patients (44%) over a mean follow-up of 15 ± 7 months. Recurrences were correlated to some AR parameters with larger LAA area (6.7 ± 2.6 cm2 vs 4.9 ± 1.6 cm2 , P = 0.01), larger LA area (31 ± 7 cm2 vs 26 ± 6 cm2 , P = 0.03), and larger transversal (55 ± 14 mm vs 43 ± 15 mm, P = 0.02) and longitudinal (63 ± 11 mm vs 55 ± 13 mm, P = 0.04) diameters in case of recurrences, but there was no significant correlation between recurrences and LAA velocity, LAA or RAA activation rates, or LAA mechanical rate. Discussion In this study, we were not able to find any significant correlation between the parameters exploring both ER and AR in a population of 41 patients undergoing catheter ablation for AF of various durations. On the other hand, mean LAA and RAA cycle lengths—a surrogate for atrial
May 2015
PACE, Vol. 38
ELECTROPHYSIOLOGICAL AND ANATOMICAL-MECHANICAL REMODELING IN AF
Table III. ER and AR Parameters According to the Duration of the Ongoing AF Episode Hours n = 10 LA area (cm2 ) LA transversal diameter (mm) LA longitudinal diameter (mm) LAA area (cm2 ) LAA flow velocity (cm/s) RAA activation rate (ms) LAA activation rate (ms) LAA mechanical rate (ms)
21 27 40 3.5 41 172 169 172
± ± ± ± ± ± ± ±
3 3 3 0.8 17 16 13 13
Weeks n = 11 30 55 65 6.8 28 192 183 191
± ± ± ± ± ± ± ±
7 12 5 2.5 8 38 35 60
Months n = 15 31 55 66 6.1 21 181 182 166
± ± ± ± ± ± ± ±
7 7 8 1.9 9 44 44 30
Years n=5
P
± ± ± ± ± ± ± ±
0.001
