JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY
VOL. 64, NO. 2, 2014
ª 2014 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION PUBLISHED BY ELSEVIER INC.
ISSN 0735-1097/$36.00 http://dx.doi.org/10.1016/j.jacc.2014.02.612
Prognostic Significance of LGE by CMR in Aortic Stenosis Patients Undergoing Valve Replacement Gilles Barone-Rochette, MD, Sophie Piérard, MD, Christophe De Meester de Ravenstein, MS, Stéphanie Seldrum, MD, Julie Melchior, MD, Frédéric Maes, MD, Anne-Catherine Pouleur, MD, PHD, David Vancraeynest, MD, PHD, Agnes Pasquet, MD, PHD, Jean-Louis Vanoverschelde, MD, PHD, Bernhard L. Gerber, MD, PHD
ABSTRACT BACKGROUND Prior studies have shown that late gadolinium enhancement (LGE) by cardiac magnetic resonance (CMR) can detect focal fibrosis in aortic stenosis (AS), suggesting that it might predict higher mortality risk. OBJECTIVES This study was conducted to evaluate whether LGE-CMR can predict post-operative survival in patients with severe AS undergoing aortic valve replacement (AVR). METHODS We prospectively evaluated survival (all-cause and cardiovascular disease related) according to LGE-CMR status in 154 consecutive AS patients (96 men; mean age: 74 6 years) without a history of myocardial infarction undergoing surgical AVR and in 40 AS patients undergoing transcatheter aortic valve replacement (TAVR). RESULTS LGE was present in 29% of patients undergoing surgical AVR and in 50% undergoing TAVR. During a median follow-up of 2.9 years, 21 patients undergoing surgical AVR and 20 undergoing TAVR died. In surgical AVR, the presence of LGE predicted higher post-operative mortality (odds ratio: 10.9; 95% confidence interval [CI]: 1.2 to 100.0; p ¼ 0.02) and worse all-cause survival (73% vs. 88%; p ¼ 0.02 by log-rank test) and cardiovascular disease related survival (85% vs. 95%; p ¼ 0.03 by log-rank test) on 5-year Kaplan-Meier estimates of survival after surgical AVR. Multivariate Cox analysis identified the presence of LGE (hazard ratio: 2.8; 95% CI: 1.3 to 6.9; p ¼ 0.025) and New York Heart Association functional class III/IV (hazard ratio: 3.2; 95% CI: 1.1 to 8.1; p < 0.01) as the sole independent predictors of all-cause mortality after surgical AVR. The presence of LGE also predicted higher all-cause mortality (p ¼ 0.05) and cardiovascular disease related mortality (p ¼ 0.03) in the subgroup of patients without angiographic coronary artery disease (n ¼ 110) and higher cardiovascular disease related mortality in 25 patients undergoing transfemoral TAVR. CONCLUSIONS The presence of LGE indicating focal fibrosis or unrecognized infarct by CMR is an independent predictor of mortality in patients with AS undergoing AVR and could provide additional information in the pre-operative evaluation of risk in these patients. (J Am Coll Cardiol 2014;64:144–54) © 2014 by the American College of Cardiology Foundation.
S
evere degenerative aortic stenosis (AS) is the
afterload and ventricular wall stress of this condition
most frequent valvular heart disease in indus-
stimulate left ventricular hypertrophic remodeling.
trialized countries and its prevalence steadily
Such remodeling is frequently associated with devel-
increases with age (1). The increased pressure
opment of adverse intramyocardial fibrosis (2–6),
From the Valvular Heart Disease Clinic, Division of Cardiology, Department of Cardiovascular Diseases, Cliniques Universitaires St. Luc, Pôle de Recherche Cardiovasculaire, Institut de Recherche Expérimentale et Clinique, Université Catholique de Louvain, Brussels, Belgium. Grant support was received from the Fondation Nationale de la Recherche Scientifique of the Belgian government (FRSM 3.4598.08, FRSM 3.4508.12.f, FRSM 3.4.589.06.f, and FRSM 1.A461.12). Dr. Barone-Rochette was supported by a grant from the French Federation of Cardiology. Drs. Piérard, Seldrum, and Melchior were supported by a fellowship (Aspirant Candidat Specialiste Doctorant) of the Fondation Nationale de la Recherche Scientifique of the Belgian government. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Barone-Rochette and Piérard contributed equally to this work and are joint first authors. Manuscript received August 28, 2013; revised manuscript received January 17, 2014, accepted February 26, 2014.
Barone-Rochette et al.
JACC VOL. 64, NO. 2, 2014 JULY 15, 2014:144–54
which may lead to alterations of both systolic and dia-
undergoing AVR reoperation or having had
ABBREVIATIONS
stolic left ventricular function (2,4,7,8) and functional
prior coronary surgery, and patients with
AND ACRONYMS
status (4) in patients with AS.
active malignancy or other conditions with a life expectancy of 50 years,
3 Non severe AS, 2 severe AR, 2 severe MR, 1 subvalvular stenosis
hospitalized for the pre-operative evaluation of severe degenerative AS, defined as an aortic valve area of 1% myocardial mass (upper 95% con40 healthy volunteers >50 years of age without
significant stenosis) was considered significant. The
Prior PCI GFR, ml/min/m2 Surgical risk
pattern of LGE was assessed by 2 independent observers who were blinded to the clinical data, coronary anatomy, and outcomes. Discordant findings were resolved by consensus. TRANSTHORACIC
Logistic EuroSCORE I, %
7.1 5.5
7.0 5.7
7.6 4.9
0.53 (NS)
STS score, %
2.3 1.5
2.2 1.5
2.5 1.4
0.27 (NS)
ECG
dardized
complete
echocardiography
ECHOCARDIOGRAPHY. A
comprehensive examination
was
stan-
transthoracic performed
AF
14 (9)
10 (9)
5 (9)
0.99 (NS)
using an IE33 echocardiograph (Philips Healthcare)
LBBB
14 (8)
8 (7)
6 (14)
0.35 (NS)
equipped with a 3.5/1.75-MHz phased-array trans-
SBP, mm Hg
134 20
134 21
135 20
0.77 (NS)
ages were re-analyzed off-line on Xcelera 1.2 LF 4
HR, beats/min
70 14
70 15
69 13
0.68 (NS)
(Philips Healthcare) workstations by 2 observers, who
0.71 0.17
0.71 0.16
0.70 0.18
0.73 (NS)
0.38 0.09
0.38 0.08
0.38 0.09
0.96 (NS)
PG, mm Hg
79 25
80 25
77 28
0.47 (NS)
MG, mm Hg
49 17
49 16
47 18
0.56 (NS)
44 (29)
31 (28)
13 (29)
0.87 (NS)
equation. LV volumes and EF were computed using
1.7 0.8
1.6 0.7
1.9 1.2
0.48 (NS)
the biplane Simpson method. LV mass was calculated
81 25
79 24
83 27
0.40 (NS)
ducer, according to established guidelines (15). Im-
Hemodynamics
Echocardiography AVA, cm2 AVAi, cm2/m2
were blinded to the pre-operative and follow-up clinical and CMR data. Multiple transducer positions were used to record maximal instantaneous and mean pressure gradients across the aortic valve. Aortic valve area was computed using the continuity
Angiography CAD Vessels affected CMR
using the Devereux formula.
EDVi, ml/m2 ESVi, ml/m2
35 25
33 23
41 28
0.07 (NS)
CORONARY ANGIOGRAPHY. Coronary angiography
EF, %
60 15
61 14
55 18
0.05
was performed in all patients using the standard
Indexed LV mass, g/m2
95 25
93 22
99 31
0.18 (NS)
Seldinger technique, and significant coronary artery disease (CAD) was considered >50% stenosis in ves-
Values are mean SD or n (%). AF ¼ atrial fibrillation; AVA ¼ aortic valve area; AVAi ¼ indexed aortic valve area; AVR ¼ aortic valve replacement; CAD ¼ coronary artery disease; CMR ¼ cardiac magnetic resonance; COPD ¼ chronic obstructive pulmonary disease; ECG ¼ electrocardiography; EDVi ¼ indexed end-diastolic volume; EF ¼ ejection fraction; ESVi ¼ indexed end-systolic volume; GFR ¼ glomerular filtration rate; HR ¼ heart rate; LBBB ¼ left bundle branch block; LGE ¼ late gadolinium enhancement; LV ¼ left ventricular; MG ¼ mean transvalvular gradient; NS ¼ nonsignificant; NYHA ¼ New York Heart Association; PAD ¼ peripheral arterial disease; PCI ¼ percutaneous coronary intervention; PG ¼ peak transvalvular aortic gradient; SBP ¼ systolic blood pressure; STS ¼ Society of Thoracic Surgery.
sels >1.5 mm in diameter. FOLLOW-UP. Survival status was obtained by phone
contact with patients, their relatives, or their physician. Patients’ history and treatment were retrieved from medical files and from review of visit or hospital records. Cause of death was categorized as cardiac or noncardiac. Cardiovascular mortality was defined as death attributable to congestive heart failure (i.e.,
with a fully automated method (14), and the results
death preceded by acute worsening or exacerbation
were expressed as a percentage of myocardial mass.
of heart failure), MI, sudden death (i.e., unexpected,
The method automatically computes mean SD of
unwitnessed or witnessed death in the absence of
Barone-Rochette et al.
JACC VOL. 64, NO. 2, 2014 JULY 15, 2014:144–54
Prognosis of LGE-CMR in Aortic Stenosis
other apparent causes) or as that occurring after the AVR procedure.
T A B L E 2 Patterns of LGE
Noninfarct LGE
STATISTICAL METHODS. Statistical analyses were
performed using SPSS version 21 (IBM SPSS Statistics, IBM Corporation, Armonk, New York) software. All tests were 2-sided, and p value 1%). The mean percent of
univariate Cox proportional hazards model, and all
myocardium affected by LGE in these patients was
significant (p < 0.10) univariate correlates of survival
3.5 2.3%. Table 1 compares the characteristics of
were entered into a forward stepwise multivariate
patients with and without LGE. LGE was more preva-
Cox model.
lent in patients with diabetes mellitus. EF was less in
F I G U R E 2 Examples of Different Patterns of LGE
(A) Absence of late gadolinium enhancement (LGE). (B to D) Noninfarct patterns: Focal nodular noninfarct LGE (B), linear midwall LGE predominantly affecting the left ventricular septum (C), and diffuse LGE pattern (D). (E) Typical transmural infarct-type pattern.
147
148
Barone-Rochette et al.
JACC VOL. 64, NO. 2, 2014 JULY 15, 2014:144–54
Prognosis of LGE-CMR in Aortic Stenosis
areas of LGE in midventricular or epicardial location.
T A B L E 3 Characteristics of Survivors Versus Nonsurvivors
Survivors (n ¼ 137)
Nonsurvivors (n ¼ 21)
Prevalence and patterns of LGE were not signifip Value
Demographics
cantly different among patients with and without CAD.
Age, yrs
74 9
76 6
0.15 (NS)
SURGERY. AVR was performed using a bioprosthesis
Male
80 (60)
16 (76)
0.16 (NS)
in 148 patients (96%) (126 stented and 22 stentless
Hyperlipidemia
83 (62)
12 (57)
0.65 (NS)
Hypertension
80 (60)
16 (76)
0.16 (NS)
Former smoker
32 (24)
8 (38)
0.17 (NS)
Current smoker
18 (14)
2 (10)
0.87 (NS)
coronary artery bypass grafts was 1.8 0.8, including
Diabetes
31 (23)
4 (19)
0.88 (NS)
80% left internal mammary grafts. Postoperative
Family history of CAD
27 (20)
4 (19)
0.89 (NS)
echocardiography demonstrated correct function of
NYHA functional class III/IV
31 (24)
11 (54)
0.005
Chest pain
29 (22)
2 (10)
0.32 (NS)
9 (7)
2 (9)
0.65 (NS)
Risk factors
valves) and using a mechanical valve in 6 patients
Smoking history
Symptoms
(4%). One hundred ten patients underwent isolated AVR, whereas 44 patients received associated coronary artery bypass grafting. The mean number of
the prosthesis, without patient–prosthesis mismatch,
Syncope
in all patients. EVENTS AFTER SURGICAL AVR. Over a median
follow-up of 2.9 years (100% complete), 21 patients
History/comorbidity COPD
14 (10)
4 (19)
0.27 (NS)
died, 11 of cardiovascular disease related causes. Five
PAD
14 (10)
2 (10)
0.99 (NS)
of those 11 were post-operative deaths that occurred
Stroke
14 (10)
1 (5)
0.69 (NS)
within 30 days after AVR or during hospitalization,
3 (2)
2 (9)
0.14 (NS)
76 29
65 17
0.03
Logistic EuroSCORE I, %
6.9 5.6
8.6 4.9
0.18 (NS)
the 11 cardiovascular disease related deaths occurred
STS score, %
2.3 1.5
2.6 1.3
0.30 (NS)
after 30 days, including 3 sudden deaths, 1 due to
AF
13 (10)
1 (5)
0.69 (NS)
LBBB
10 (7)
4 (19)
0.11 (NS)
Prior PCI 2
GFR, ml/min/m Surgical risk
including 3 sudden deaths, 1 related to post-operative heart failure, and 1 due to perioperative stroke. Six of
heart failure, 1 due to infective endocarditis, and 1
ECG
Echocardiography 2
2
due to aneurysm rupture. The 10 noncardiac deaths were caused by cancer (n ¼ 7), sepsis (n ¼ 1), cerebral hemorrhage after an accidental fall (n ¼ 1), and sui-
AVAi, cm /m
0.38 .09
0.38 .07
0.99 (NS)
PG, mm Hg
80 25
74 23
0.31 (NS)
MG, mm Hg
49 17
45 15
0.29 (NS)
Stented bioprosthesis
109 (82)
17 (81)
0.88 (NS)
Stentless bioprosthesis
18 (14)
4 (19)
0.68 (NS)
6 (5)
0
0.60 (NS)
Valve size, mm
24 2
24 0.2
0.78 (NS)
Associated CABG
35 (26)
9 (43)
0.12 (NS)
EDVi, ml/m2
80 25
82 24
0.74 (NS)
ence
ESVi, ml/m2
35 25
37 26
0.76 (NS)
sole predictors of all-cause survival. Multivariate Cox
EF, %
60 15
59 17
0.80 (NS)
analysis demonstrated that the presence of LGE
Ind. LV mass, g/m2
93 23
101 33
0.22 (NS)
and NYHA functional class were the only indepen-
Type of surgery
Mechanical valve
cide (n ¼ 1). LGE AND PREDICTION OF SURVIVAL IN PATIENTS UNDERGOING SURGICAL AVR. The clinical charac-
teristics of the patients according to survival are shown in Table 3. Univariate and multivariate predictors of all-cause survival after AVR are shown in Table 4. The presence of LGE, New York Heart Association (NYHA) functional class III/IV, and the pres-
CMR
Values are mean SD or n (%). CABG ¼ coronary artery bypass grafting; other abbreviations as in Table 1.
of
left
bundle
branch
block
were
the
dent determinants of all-cause survival. Also, for cardiovascular disease related survival, the presence of LGE was the sole predictor on multivariate Cox analysis. Kaplan-Meier curves for all-cause and cardiovascular disease related survival, according to pre-
patients with LGE than in those without LGE. All
operative LGE status, are shown in Figure 3. Patients
other parameters, and in particular the severity of
with pre-operative LGE had significantly worse all-
AS, and the prevalence of CAD were not significantly
cause and cardiovascular disease related survival
different among the 2 groups. LGE was considered
than did patients without LGE. This resulted mainly
infarct type if its location was subendocardial or
from higher post-operative mortality (OR: 10.9; 95%
transmural; otherwise, it was considered noninfarct
CI: 1.18 to 100; p ¼ 0.02), whereas the risks for late all-
type (Table 2, Fig. 2). Noninfarct patterns were
cause mortality (p ¼ 0.25 by log-rank test) and car-
focal lines (septal stripe), focal spots, or diffuse
diovascular disease related mortality (p ¼ 0.60 by
Barone-Rochette et al.
JACC VOL. 64, NO. 2, 2014 JULY 15, 2014:144–54
149
Prognosis of LGE-CMR in Aortic Stenosis
log-rank test) were not significantly increased. The presence of LGE also remained a significant predictor
T A B L E 4 Predictors of Overall Survival After Surgical AVR on Univariate and
Multivariate Cox Analysis
of increased all-cause and cardiovascular disease Univariate
related mortality (p ¼ 0.05 and p ¼ 0.03, respectively,
HR (95% CI)
by log-rank test) if we considered only patients without CAD (Fig. 4). CHARACTERISTICS
AND
SURVIVAL
OF
PATIENTS
Multivariate p Value
HR (95% CI)
p Value
Presence of LGE
2.8 (1.1–6.7)
0.02
2.8 (1.1–6.9)
0.025
NYHA functional class III/IV
3.2 (1.3–8.0)
0.007
3.2 (1.1–8.1)
0.01
LBBB
2.8 (0.92–8.3)
0.06
–
–
UNDERGOING TAVR. TAVR patients were significantly
older, had higher comorbidities, a higher NYHA
CI ¼ confidence interval; HR ¼ hazard ratio; other abbreviations as in Table 1.
functional class, a lower glomerular filtration rate, a lower EF, and more severe CAD than did patients undergoing conventional AVR (Table 5). Also, they often had a history of MI or cardiac surgery, and LGE was present more frequently in patients undergoing undergoing surgical AVR. TAVR was performed using the Sapien valve (Edwards Lifesciences Corporation, Irvine, California), by transfemoral access in 25 patients and by transapical access in 15 patients. Median follow-up after TAVR was 3.9 years. During this period, 20 patients died, 15 of cardiovascular causes. In the subgroup of 25 patients undergoing transfemoral TAVR (Fig. 5), the presence of LGE was associated with significantly worse (p ¼ 0.045) car-
A
100
All-Cause Survival (%)
TAVR (20 of 40 [50%]; p ¼ 0.005) than in patients
80 60
No LGE LGE
40 20
diovascular disease related survival and a trend for a
0
lower but statistically nonsignificant difference (p ¼ 0.09) all-cause survival. By contrast, survival in the 15
Log Rank p=0.02
0
1
2
3
4
5
Time (Years)
patients undergoing transapical TAVR was not pre-
no LGE
110
92
66
48
12
5
dicted by LGE. Indeed, this probably occurred
LGE
44
33
30
21
9
8
because these patients had significantly higher perioperative mortality (25% vs. 0% in transfemoral
B
100
and an early learning curve with the transapical technique.
DISCUSSION Our study revealed that in patients with severe AS without a history of MI undergoing surgical AVR, the presence and extent of myocardial focal fibrosis and necrosis detected by LGE-CMR predict increased
Cardiovascular Disease Related Survival (%)
TAVR; p ¼ 0.04), likely reflecting higher comorbidity
80 No LGE LGE
60 40 20
perioperative risk and worse all-cause and cardio-
0
vascular disease related survival. This was also true in
Log Rank p=0.03
0
1
We also observed that LGE can predict higher cardiovascular
disease
related
mortality
risk
after
transfemoral TAVR. FIBROSIS AND LGE-CMR IN AS. In response to the
2
3
4
5
Time (Years)
patients without angiographically significant CAD. no LGE
110
92
66
48
12
5
LGE
44
33
30
21
9
8
F I G U R E 3 Kaplan-Meier Survival in Patients With AS Undergoing Surgical AVR,
by LGE Status
chronic pressure overload of severe AS, the LV reacts by compensatory concentric hypertrophic remodeling. This phenomenon involves not only increased
All-cause (A) and cardiovascular disease related (B) survival in 154 patients undergoing surgical aortic valve replacement (AVR), according to the presence or absence of late gadolinium enhancement (LGE) pre-operatively detected by cardiac magnetic resonance
myocyte volume but also coordinated remodeling and
(CMR). Kaplan-Meier graphs show that patients without LGE had greater all-cause and
increased extracellular matrix (16). By histopathology,
cardiovascular survival than did patients with LGE.
hearts with severe AS present with both interstitial
Barone-Rochette et al.
JACC VOL. 64, NO. 2, 2014 JULY 15, 2014:144–54
Prognosis of LGE-CMR in Aortic Stenosis
present in a substantial proportion of patients with
A
AS, or to imbalance between reduced myocardial
All-Cause Survival (%)
100
supply resulting from decreased coronary perfusion pressure and increased myocardial metabolic demand
80
subsequent to the increased LV workload in severe AS (18,19). Indeed, a direct association between the
60
magnitude of ischemia and the development of No LGE
fibrosis in AS has been described (19).
LGE
40
LGE-MRI is the technique of choice to noninvasively detect myocardial fibrosis and necrosis
20
in the myocardium (9). Indeed, gadolinium-based contrast agents have extravascular distribution vol-
Log Rank p=0.05
0
0
1
2
3
4
5
6 6
3 4
ume, which increases when the extracellular matrix proliferates or when myocytes are replaced by focal
Time (Years) no LGE LGE
B
79 31
66 21
42 19
26 13
fibrosis. Imaging performed several minutes after a gadolinium injection, when the contrast agent has equilibrated with the blood, can therefore reveal
100
focal areas of intramyocardial fibrosis as bright areas of gadolinium accumulation relative to normal
Cardiovascular Disease Related Survival (%)
150
80
tissue. Although the detection of LGE is likely a measurement of focal replacement fibrosis and is less sensitive for diffuse interstitial fibrosis, several
60 40
No LGE
studies have indicated a good correlation between
LGE
the size of the LGE by MRI and histologically measured fibrosis (4,5,10) in AS. Our present study used an automated-analysis method for the quanti-
20
fication of LGE, with high reproducibility, which was
Log Rank p=0.03
0
0
validated in animals against histology (14). Similar to
1
2
3
4
5
previous studies (5,10,11), we found that LGE indicating focal replacement fibrosis occurs in approxi-
Time (Years) no LGE LGE
79 31
66 21
42 19
26 13
6 6
mately 30% of patients with AS and may present in
3 4
different
distribution
patterns,
either
mirroring
typical ischemic LGE or in focal and diffuse nonF I G U R E 4 Kaplan-Meier Survival in Patients With AS Without CAD Undergoing
ischemic patterns. Because we did not perform his-
Surgical AVR, by LGE Status
topathology in our patients, we can only speculate
All-cause (A) and cardiovascular disease related (B) survival in a subgroup of 110 patients
on the underlying pathophysiological mechanisms of
without angiographically detected coronary artery disease (CAD) undergoing surgical AVR,
macroscopically
according to the presence or absence of LGE pre-operatively detected by CMR. Kaplan-
endocardial infarct-type LGE might represent un-
Meier graphs show that patients without LGE had greater all-cause and cardiovascular
recognized infarcts but also could result from
survival than did patients with LGE. Abbreviations as in Figure 3.
detected
LGE
by
MRI.
Sub-
longstanding ischemia due to oxygen supply–demand mismatch in severe AS. Focal spots of LGE could
result
from
distal
microembolization
or
diffuse reactive fibrosis as well as focal replacement
represent areas of more dense replacement fibrosis,
fibrosis (2–5). Diffuse fibrosis appears to result from
similar to midwall LGE. Diffuse LGE patterns likely
alterations
metal-
result from increased gadolinium distribution vol-
loproteinases and their specific tissue inhibitors (3)
ume in areas with diffuse histological interstitial
subsequent
in to
the the
balance
of
matrix
overstimulation
of
the
angiotensin-renin system (17) (Central Illustration).
fibrosis. IMPLICATIONS OF FIBROSIS IN AS ON LV FUNCTION
Replacement fibrosis involves myocyte death by
AND OUTCOME. Our study confirms the previously
apoptosis or necrosis. This can occur in a direct
reported association of LGE with LV dysfunction
response to the mechanical forces associated with
(4,5,8,10). Indeed, patients with LGE had slightly
increased afterload or from the effects of myocardial
lower EF than patients without LGE. More impor-
ischemia (6), due either to CAD, a coexisting condition
tantly, we showed that high amounts of LGE by CMR
Barone-Rochette et al.
JACC VOL. 64, NO. 2, 2014 JULY 15, 2014:144–54
predicted worse all-cause and cardiovascular disease related survival after AVR. This is consistent with
T A B L E 5 Characteristics of Patients Undergoing TAVR
TAVR (n ¼ 40)
findings from a recent study that demonstrated that histopathologically measured fibrosis is associated with poorer outcomes in patients after AVR (20). Another study reported high post-operative mortality among 58 patients with high amounts of histopathological fibrosis; however, it was not powered to
151
Prognosis of LGE-CMR in Aortic Stenosis
p Value All TAVR vs. Surgical AVR
All TAVR (n ¼ 40)
Transfemoral (n ¼ 25)
Transapical (n ¼ 15)
Age, yrs
83 5
83 5
84 4
VOL. 64, NO. 2, 2014
ª 2014 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION PUBLISHED BY ELSEVIER INC.
ISSN 0735-1097/$36.00 http://dx.doi.org/10.1016/j.jacc.2014.02.612
Prognostic Significance of LGE by CMR in Aortic Stenosis Patients Undergoing Valve Replacement Gilles Barone-Rochette, MD, Sophie Piérard, MD, Christophe De Meester de Ravenstein, MS, Stéphanie Seldrum, MD, Julie Melchior, MD, Frédéric Maes, MD, Anne-Catherine Pouleur, MD, PHD, David Vancraeynest, MD, PHD, Agnes Pasquet, MD, PHD, Jean-Louis Vanoverschelde, MD, PHD, Bernhard L. Gerber, MD, PHD
ABSTRACT BACKGROUND Prior studies have shown that late gadolinium enhancement (LGE) by cardiac magnetic resonance (CMR) can detect focal fibrosis in aortic stenosis (AS), suggesting that it might predict higher mortality risk. OBJECTIVES This study was conducted to evaluate whether LGE-CMR can predict post-operative survival in patients with severe AS undergoing aortic valve replacement (AVR). METHODS We prospectively evaluated survival (all-cause and cardiovascular disease related) according to LGE-CMR status in 154 consecutive AS patients (96 men; mean age: 74 6 years) without a history of myocardial infarction undergoing surgical AVR and in 40 AS patients undergoing transcatheter aortic valve replacement (TAVR). RESULTS LGE was present in 29% of patients undergoing surgical AVR and in 50% undergoing TAVR. During a median follow-up of 2.9 years, 21 patients undergoing surgical AVR and 20 undergoing TAVR died. In surgical AVR, the presence of LGE predicted higher post-operative mortality (odds ratio: 10.9; 95% confidence interval [CI]: 1.2 to 100.0; p ¼ 0.02) and worse all-cause survival (73% vs. 88%; p ¼ 0.02 by log-rank test) and cardiovascular disease related survival (85% vs. 95%; p ¼ 0.03 by log-rank test) on 5-year Kaplan-Meier estimates of survival after surgical AVR. Multivariate Cox analysis identified the presence of LGE (hazard ratio: 2.8; 95% CI: 1.3 to 6.9; p ¼ 0.025) and New York Heart Association functional class III/IV (hazard ratio: 3.2; 95% CI: 1.1 to 8.1; p < 0.01) as the sole independent predictors of all-cause mortality after surgical AVR. The presence of LGE also predicted higher all-cause mortality (p ¼ 0.05) and cardiovascular disease related mortality (p ¼ 0.03) in the subgroup of patients without angiographic coronary artery disease (n ¼ 110) and higher cardiovascular disease related mortality in 25 patients undergoing transfemoral TAVR. CONCLUSIONS The presence of LGE indicating focal fibrosis or unrecognized infarct by CMR is an independent predictor of mortality in patients with AS undergoing AVR and could provide additional information in the pre-operative evaluation of risk in these patients. (J Am Coll Cardiol 2014;64:144–54) © 2014 by the American College of Cardiology Foundation.
S
evere degenerative aortic stenosis (AS) is the
afterload and ventricular wall stress of this condition
most frequent valvular heart disease in indus-
stimulate left ventricular hypertrophic remodeling.
trialized countries and its prevalence steadily
Such remodeling is frequently associated with devel-
increases with age (1). The increased pressure
opment of adverse intramyocardial fibrosis (2–6),
From the Valvular Heart Disease Clinic, Division of Cardiology, Department of Cardiovascular Diseases, Cliniques Universitaires St. Luc, Pôle de Recherche Cardiovasculaire, Institut de Recherche Expérimentale et Clinique, Université Catholique de Louvain, Brussels, Belgium. Grant support was received from the Fondation Nationale de la Recherche Scientifique of the Belgian government (FRSM 3.4598.08, FRSM 3.4508.12.f, FRSM 3.4.589.06.f, and FRSM 1.A461.12). Dr. Barone-Rochette was supported by a grant from the French Federation of Cardiology. Drs. Piérard, Seldrum, and Melchior were supported by a fellowship (Aspirant Candidat Specialiste Doctorant) of the Fondation Nationale de la Recherche Scientifique of the Belgian government. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Barone-Rochette and Piérard contributed equally to this work and are joint first authors. Manuscript received August 28, 2013; revised manuscript received January 17, 2014, accepted February 26, 2014.
Barone-Rochette et al.
JACC VOL. 64, NO. 2, 2014 JULY 15, 2014:144–54
which may lead to alterations of both systolic and dia-
undergoing AVR reoperation or having had
ABBREVIATIONS
stolic left ventricular function (2,4,7,8) and functional
prior coronary surgery, and patients with
AND ACRONYMS
status (4) in patients with AS.
active malignancy or other conditions with a life expectancy of 50 years,
3 Non severe AS, 2 severe AR, 2 severe MR, 1 subvalvular stenosis
hospitalized for the pre-operative evaluation of severe degenerative AS, defined as an aortic valve area of 1% myocardial mass (upper 95% con40 healthy volunteers >50 years of age without
significant stenosis) was considered significant. The
Prior PCI GFR, ml/min/m2 Surgical risk
pattern of LGE was assessed by 2 independent observers who were blinded to the clinical data, coronary anatomy, and outcomes. Discordant findings were resolved by consensus. TRANSTHORACIC
Logistic EuroSCORE I, %
7.1 5.5
7.0 5.7
7.6 4.9
0.53 (NS)
STS score, %
2.3 1.5
2.2 1.5
2.5 1.4
0.27 (NS)
ECG
dardized
complete
echocardiography
ECHOCARDIOGRAPHY. A
comprehensive examination
was
stan-
transthoracic performed
AF
14 (9)
10 (9)
5 (9)
0.99 (NS)
using an IE33 echocardiograph (Philips Healthcare)
LBBB
14 (8)
8 (7)
6 (14)
0.35 (NS)
equipped with a 3.5/1.75-MHz phased-array trans-
SBP, mm Hg
134 20
134 21
135 20
0.77 (NS)
ages were re-analyzed off-line on Xcelera 1.2 LF 4
HR, beats/min
70 14
70 15
69 13
0.68 (NS)
(Philips Healthcare) workstations by 2 observers, who
0.71 0.17
0.71 0.16
0.70 0.18
0.73 (NS)
0.38 0.09
0.38 0.08
0.38 0.09
0.96 (NS)
PG, mm Hg
79 25
80 25
77 28
0.47 (NS)
MG, mm Hg
49 17
49 16
47 18
0.56 (NS)
44 (29)
31 (28)
13 (29)
0.87 (NS)
equation. LV volumes and EF were computed using
1.7 0.8
1.6 0.7
1.9 1.2
0.48 (NS)
the biplane Simpson method. LV mass was calculated
81 25
79 24
83 27
0.40 (NS)
ducer, according to established guidelines (15). Im-
Hemodynamics
Echocardiography AVA, cm2 AVAi, cm2/m2
were blinded to the pre-operative and follow-up clinical and CMR data. Multiple transducer positions were used to record maximal instantaneous and mean pressure gradients across the aortic valve. Aortic valve area was computed using the continuity
Angiography CAD Vessels affected CMR
using the Devereux formula.
EDVi, ml/m2 ESVi, ml/m2
35 25
33 23
41 28
0.07 (NS)
CORONARY ANGIOGRAPHY. Coronary angiography
EF, %
60 15
61 14
55 18
0.05
was performed in all patients using the standard
Indexed LV mass, g/m2
95 25
93 22
99 31
0.18 (NS)
Seldinger technique, and significant coronary artery disease (CAD) was considered >50% stenosis in ves-
Values are mean SD or n (%). AF ¼ atrial fibrillation; AVA ¼ aortic valve area; AVAi ¼ indexed aortic valve area; AVR ¼ aortic valve replacement; CAD ¼ coronary artery disease; CMR ¼ cardiac magnetic resonance; COPD ¼ chronic obstructive pulmonary disease; ECG ¼ electrocardiography; EDVi ¼ indexed end-diastolic volume; EF ¼ ejection fraction; ESVi ¼ indexed end-systolic volume; GFR ¼ glomerular filtration rate; HR ¼ heart rate; LBBB ¼ left bundle branch block; LGE ¼ late gadolinium enhancement; LV ¼ left ventricular; MG ¼ mean transvalvular gradient; NS ¼ nonsignificant; NYHA ¼ New York Heart Association; PAD ¼ peripheral arterial disease; PCI ¼ percutaneous coronary intervention; PG ¼ peak transvalvular aortic gradient; SBP ¼ systolic blood pressure; STS ¼ Society of Thoracic Surgery.
sels >1.5 mm in diameter. FOLLOW-UP. Survival status was obtained by phone
contact with patients, their relatives, or their physician. Patients’ history and treatment were retrieved from medical files and from review of visit or hospital records. Cause of death was categorized as cardiac or noncardiac. Cardiovascular mortality was defined as death attributable to congestive heart failure (i.e.,
with a fully automated method (14), and the results
death preceded by acute worsening or exacerbation
were expressed as a percentage of myocardial mass.
of heart failure), MI, sudden death (i.e., unexpected,
The method automatically computes mean SD of
unwitnessed or witnessed death in the absence of
Barone-Rochette et al.
JACC VOL. 64, NO. 2, 2014 JULY 15, 2014:144–54
Prognosis of LGE-CMR in Aortic Stenosis
other apparent causes) or as that occurring after the AVR procedure.
T A B L E 2 Patterns of LGE
Noninfarct LGE
STATISTICAL METHODS. Statistical analyses were
performed using SPSS version 21 (IBM SPSS Statistics, IBM Corporation, Armonk, New York) software. All tests were 2-sided, and p value 1%). The mean percent of
univariate Cox proportional hazards model, and all
myocardium affected by LGE in these patients was
significant (p < 0.10) univariate correlates of survival
3.5 2.3%. Table 1 compares the characteristics of
were entered into a forward stepwise multivariate
patients with and without LGE. LGE was more preva-
Cox model.
lent in patients with diabetes mellitus. EF was less in
F I G U R E 2 Examples of Different Patterns of LGE
(A) Absence of late gadolinium enhancement (LGE). (B to D) Noninfarct patterns: Focal nodular noninfarct LGE (B), linear midwall LGE predominantly affecting the left ventricular septum (C), and diffuse LGE pattern (D). (E) Typical transmural infarct-type pattern.
147
148
Barone-Rochette et al.
JACC VOL. 64, NO. 2, 2014 JULY 15, 2014:144–54
Prognosis of LGE-CMR in Aortic Stenosis
areas of LGE in midventricular or epicardial location.
T A B L E 3 Characteristics of Survivors Versus Nonsurvivors
Survivors (n ¼ 137)
Nonsurvivors (n ¼ 21)
Prevalence and patterns of LGE were not signifip Value
Demographics
cantly different among patients with and without CAD.
Age, yrs
74 9
76 6
0.15 (NS)
SURGERY. AVR was performed using a bioprosthesis
Male
80 (60)
16 (76)
0.16 (NS)
in 148 patients (96%) (126 stented and 22 stentless
Hyperlipidemia
83 (62)
12 (57)
0.65 (NS)
Hypertension
80 (60)
16 (76)
0.16 (NS)
Former smoker
32 (24)
8 (38)
0.17 (NS)
Current smoker
18 (14)
2 (10)
0.87 (NS)
coronary artery bypass grafts was 1.8 0.8, including
Diabetes
31 (23)
4 (19)
0.88 (NS)
80% left internal mammary grafts. Postoperative
Family history of CAD
27 (20)
4 (19)
0.89 (NS)
echocardiography demonstrated correct function of
NYHA functional class III/IV
31 (24)
11 (54)
0.005
Chest pain
29 (22)
2 (10)
0.32 (NS)
9 (7)
2 (9)
0.65 (NS)
Risk factors
valves) and using a mechanical valve in 6 patients
Smoking history
Symptoms
(4%). One hundred ten patients underwent isolated AVR, whereas 44 patients received associated coronary artery bypass grafting. The mean number of
the prosthesis, without patient–prosthesis mismatch,
Syncope
in all patients. EVENTS AFTER SURGICAL AVR. Over a median
follow-up of 2.9 years (100% complete), 21 patients
History/comorbidity COPD
14 (10)
4 (19)
0.27 (NS)
died, 11 of cardiovascular disease related causes. Five
PAD
14 (10)
2 (10)
0.99 (NS)
of those 11 were post-operative deaths that occurred
Stroke
14 (10)
1 (5)
0.69 (NS)
within 30 days after AVR or during hospitalization,
3 (2)
2 (9)
0.14 (NS)
76 29
65 17
0.03
Logistic EuroSCORE I, %
6.9 5.6
8.6 4.9
0.18 (NS)
the 11 cardiovascular disease related deaths occurred
STS score, %
2.3 1.5
2.6 1.3
0.30 (NS)
after 30 days, including 3 sudden deaths, 1 due to
AF
13 (10)
1 (5)
0.69 (NS)
LBBB
10 (7)
4 (19)
0.11 (NS)
Prior PCI 2
GFR, ml/min/m Surgical risk
including 3 sudden deaths, 1 related to post-operative heart failure, and 1 due to perioperative stroke. Six of
heart failure, 1 due to infective endocarditis, and 1
ECG
Echocardiography 2
2
due to aneurysm rupture. The 10 noncardiac deaths were caused by cancer (n ¼ 7), sepsis (n ¼ 1), cerebral hemorrhage after an accidental fall (n ¼ 1), and sui-
AVAi, cm /m
0.38 .09
0.38 .07
0.99 (NS)
PG, mm Hg
80 25
74 23
0.31 (NS)
MG, mm Hg
49 17
45 15
0.29 (NS)
Stented bioprosthesis
109 (82)
17 (81)
0.88 (NS)
Stentless bioprosthesis
18 (14)
4 (19)
0.68 (NS)
6 (5)
0
0.60 (NS)
Valve size, mm
24 2
24 0.2
0.78 (NS)
Associated CABG
35 (26)
9 (43)
0.12 (NS)
EDVi, ml/m2
80 25
82 24
0.74 (NS)
ence
ESVi, ml/m2
35 25
37 26
0.76 (NS)
sole predictors of all-cause survival. Multivariate Cox
EF, %
60 15
59 17
0.80 (NS)
analysis demonstrated that the presence of LGE
Ind. LV mass, g/m2
93 23
101 33
0.22 (NS)
and NYHA functional class were the only indepen-
Type of surgery
Mechanical valve
cide (n ¼ 1). LGE AND PREDICTION OF SURVIVAL IN PATIENTS UNDERGOING SURGICAL AVR. The clinical charac-
teristics of the patients according to survival are shown in Table 3. Univariate and multivariate predictors of all-cause survival after AVR are shown in Table 4. The presence of LGE, New York Heart Association (NYHA) functional class III/IV, and the pres-
CMR
Values are mean SD or n (%). CABG ¼ coronary artery bypass grafting; other abbreviations as in Table 1.
of
left
bundle
branch
block
were
the
dent determinants of all-cause survival. Also, for cardiovascular disease related survival, the presence of LGE was the sole predictor on multivariate Cox analysis. Kaplan-Meier curves for all-cause and cardiovascular disease related survival, according to pre-
patients with LGE than in those without LGE. All
operative LGE status, are shown in Figure 3. Patients
other parameters, and in particular the severity of
with pre-operative LGE had significantly worse all-
AS, and the prevalence of CAD were not significantly
cause and cardiovascular disease related survival
different among the 2 groups. LGE was considered
than did patients without LGE. This resulted mainly
infarct type if its location was subendocardial or
from higher post-operative mortality (OR: 10.9; 95%
transmural; otherwise, it was considered noninfarct
CI: 1.18 to 100; p ¼ 0.02), whereas the risks for late all-
type (Table 2, Fig. 2). Noninfarct patterns were
cause mortality (p ¼ 0.25 by log-rank test) and car-
focal lines (septal stripe), focal spots, or diffuse
diovascular disease related mortality (p ¼ 0.60 by
Barone-Rochette et al.
JACC VOL. 64, NO. 2, 2014 JULY 15, 2014:144–54
149
Prognosis of LGE-CMR in Aortic Stenosis
log-rank test) were not significantly increased. The presence of LGE also remained a significant predictor
T A B L E 4 Predictors of Overall Survival After Surgical AVR on Univariate and
Multivariate Cox Analysis
of increased all-cause and cardiovascular disease Univariate
related mortality (p ¼ 0.05 and p ¼ 0.03, respectively,
HR (95% CI)
by log-rank test) if we considered only patients without CAD (Fig. 4). CHARACTERISTICS
AND
SURVIVAL
OF
PATIENTS
Multivariate p Value
HR (95% CI)
p Value
Presence of LGE
2.8 (1.1–6.7)
0.02
2.8 (1.1–6.9)
0.025
NYHA functional class III/IV
3.2 (1.3–8.0)
0.007
3.2 (1.1–8.1)
0.01
LBBB
2.8 (0.92–8.3)
0.06
–
–
UNDERGOING TAVR. TAVR patients were significantly
older, had higher comorbidities, a higher NYHA
CI ¼ confidence interval; HR ¼ hazard ratio; other abbreviations as in Table 1.
functional class, a lower glomerular filtration rate, a lower EF, and more severe CAD than did patients undergoing conventional AVR (Table 5). Also, they often had a history of MI or cardiac surgery, and LGE was present more frequently in patients undergoing undergoing surgical AVR. TAVR was performed using the Sapien valve (Edwards Lifesciences Corporation, Irvine, California), by transfemoral access in 25 patients and by transapical access in 15 patients. Median follow-up after TAVR was 3.9 years. During this period, 20 patients died, 15 of cardiovascular causes. In the subgroup of 25 patients undergoing transfemoral TAVR (Fig. 5), the presence of LGE was associated with significantly worse (p ¼ 0.045) car-
A
100
All-Cause Survival (%)
TAVR (20 of 40 [50%]; p ¼ 0.005) than in patients
80 60
No LGE LGE
40 20
diovascular disease related survival and a trend for a
0
lower but statistically nonsignificant difference (p ¼ 0.09) all-cause survival. By contrast, survival in the 15
Log Rank p=0.02
0
1
2
3
4
5
Time (Years)
patients undergoing transapical TAVR was not pre-
no LGE
110
92
66
48
12
5
dicted by LGE. Indeed, this probably occurred
LGE
44
33
30
21
9
8
because these patients had significantly higher perioperative mortality (25% vs. 0% in transfemoral
B
100
and an early learning curve with the transapical technique.
DISCUSSION Our study revealed that in patients with severe AS without a history of MI undergoing surgical AVR, the presence and extent of myocardial focal fibrosis and necrosis detected by LGE-CMR predict increased
Cardiovascular Disease Related Survival (%)
TAVR; p ¼ 0.04), likely reflecting higher comorbidity
80 No LGE LGE
60 40 20
perioperative risk and worse all-cause and cardio-
0
vascular disease related survival. This was also true in
Log Rank p=0.03
0
1
We also observed that LGE can predict higher cardiovascular
disease
related
mortality
risk
after
transfemoral TAVR. FIBROSIS AND LGE-CMR IN AS. In response to the
2
3
4
5
Time (Years)
patients without angiographically significant CAD. no LGE
110
92
66
48
12
5
LGE
44
33
30
21
9
8
F I G U R E 3 Kaplan-Meier Survival in Patients With AS Undergoing Surgical AVR,
by LGE Status
chronic pressure overload of severe AS, the LV reacts by compensatory concentric hypertrophic remodeling. This phenomenon involves not only increased
All-cause (A) and cardiovascular disease related (B) survival in 154 patients undergoing surgical aortic valve replacement (AVR), according to the presence or absence of late gadolinium enhancement (LGE) pre-operatively detected by cardiac magnetic resonance
myocyte volume but also coordinated remodeling and
(CMR). Kaplan-Meier graphs show that patients without LGE had greater all-cause and
increased extracellular matrix (16). By histopathology,
cardiovascular survival than did patients with LGE.
hearts with severe AS present with both interstitial
Barone-Rochette et al.
JACC VOL. 64, NO. 2, 2014 JULY 15, 2014:144–54
Prognosis of LGE-CMR in Aortic Stenosis
present in a substantial proportion of patients with
A
AS, or to imbalance between reduced myocardial
All-Cause Survival (%)
100
supply resulting from decreased coronary perfusion pressure and increased myocardial metabolic demand
80
subsequent to the increased LV workload in severe AS (18,19). Indeed, a direct association between the
60
magnitude of ischemia and the development of No LGE
fibrosis in AS has been described (19).
LGE
40
LGE-MRI is the technique of choice to noninvasively detect myocardial fibrosis and necrosis
20
in the myocardium (9). Indeed, gadolinium-based contrast agents have extravascular distribution vol-
Log Rank p=0.05
0
0
1
2
3
4
5
6 6
3 4
ume, which increases when the extracellular matrix proliferates or when myocytes are replaced by focal
Time (Years) no LGE LGE
B
79 31
66 21
42 19
26 13
fibrosis. Imaging performed several minutes after a gadolinium injection, when the contrast agent has equilibrated with the blood, can therefore reveal
100
focal areas of intramyocardial fibrosis as bright areas of gadolinium accumulation relative to normal
Cardiovascular Disease Related Survival (%)
150
80
tissue. Although the detection of LGE is likely a measurement of focal replacement fibrosis and is less sensitive for diffuse interstitial fibrosis, several
60 40
No LGE
studies have indicated a good correlation between
LGE
the size of the LGE by MRI and histologically measured fibrosis (4,5,10) in AS. Our present study used an automated-analysis method for the quanti-
20
fication of LGE, with high reproducibility, which was
Log Rank p=0.03
0
0
validated in animals against histology (14). Similar to
1
2
3
4
5
previous studies (5,10,11), we found that LGE indicating focal replacement fibrosis occurs in approxi-
Time (Years) no LGE LGE
79 31
66 21
42 19
26 13
6 6
mately 30% of patients with AS and may present in
3 4
different
distribution
patterns,
either
mirroring
typical ischemic LGE or in focal and diffuse nonF I G U R E 4 Kaplan-Meier Survival in Patients With AS Without CAD Undergoing
ischemic patterns. Because we did not perform his-
Surgical AVR, by LGE Status
topathology in our patients, we can only speculate
All-cause (A) and cardiovascular disease related (B) survival in a subgroup of 110 patients
on the underlying pathophysiological mechanisms of
without angiographically detected coronary artery disease (CAD) undergoing surgical AVR,
macroscopically
according to the presence or absence of LGE pre-operatively detected by CMR. Kaplan-
endocardial infarct-type LGE might represent un-
Meier graphs show that patients without LGE had greater all-cause and cardiovascular
recognized infarcts but also could result from
survival than did patients with LGE. Abbreviations as in Figure 3.
detected
LGE
by
MRI.
Sub-
longstanding ischemia due to oxygen supply–demand mismatch in severe AS. Focal spots of LGE could
result
from
distal
microembolization
or
diffuse reactive fibrosis as well as focal replacement
represent areas of more dense replacement fibrosis,
fibrosis (2–5). Diffuse fibrosis appears to result from
similar to midwall LGE. Diffuse LGE patterns likely
alterations
metal-
result from increased gadolinium distribution vol-
loproteinases and their specific tissue inhibitors (3)
ume in areas with diffuse histological interstitial
subsequent
in to
the the
balance
of
matrix
overstimulation
of
the
angiotensin-renin system (17) (Central Illustration).
fibrosis. IMPLICATIONS OF FIBROSIS IN AS ON LV FUNCTION
Replacement fibrosis involves myocyte death by
AND OUTCOME. Our study confirms the previously
apoptosis or necrosis. This can occur in a direct
reported association of LGE with LV dysfunction
response to the mechanical forces associated with
(4,5,8,10). Indeed, patients with LGE had slightly
increased afterload or from the effects of myocardial
lower EF than patients without LGE. More impor-
ischemia (6), due either to CAD, a coexisting condition
tantly, we showed that high amounts of LGE by CMR
Barone-Rochette et al.
JACC VOL. 64, NO. 2, 2014 JULY 15, 2014:144–54
predicted worse all-cause and cardiovascular disease related survival after AVR. This is consistent with
T A B L E 5 Characteristics of Patients Undergoing TAVR
TAVR (n ¼ 40)
findings from a recent study that demonstrated that histopathologically measured fibrosis is associated with poorer outcomes in patients after AVR (20). Another study reported high post-operative mortality among 58 patients with high amounts of histopathological fibrosis; however, it was not powered to
151
Prognosis of LGE-CMR in Aortic Stenosis
p Value All TAVR vs. Surgical AVR
All TAVR (n ¼ 40)
Transfemoral (n ¼ 25)
Transapical (n ¼ 15)
Age, yrs
83 5
83 5
84 4
