Keywords
primary aldosteronism - cardiac magnetic resonance imaging - NT-pro-BNP - left ventricular remodeling
Introduction
Primary aldosteronism (PA) is a common form of secondary hypertension, which is
characterized by the production of excessive aldosterone and low levels of plasma
renin [1]. Such inappropriate production
of aldosterone can cause hypertension, cardiovascular damage, plasma renin
suppression, and hypokalemia, which may lead to left ventricular remodeling [2]
[3]. Left ventricular remodeling was described as compensation of the left
ventricle in response to increasing cardiac preload and afterload [4]. Clinical studies have demonstrated
that patients with PA had more obvious left ventricular remodeling, including
increased left ventricular mass and cardiac fibrosis level, than those with
essential hypertension (EH) [3]
[5]
[6]
[7]. With the development of
cardiac magnetic resonance imaging (MRI), left ventricular remodeling could be
evaluated by multiple parameters, including volume, mass, function, strain, and even
tissue characteristics, measured by the T1 mapping technique [8]
[9]
[10].
Pro B-type natriuretic peptide (pro-BNP) is synthesized in cardiomyocytes as a
pre-hormone in response to increasing ventricular wall strain. Then, pro-BNP is
cleaved into the biologically active brain natriuretic peptide (BNP) hormone and the
biologically inactive N-terminal pro-brain natriuretic peptide (NT-pro-BNP) hormone.
NT-pro-BNP is a well-established plasma biomarker of heart failure, with a longer
half-life, better in vitro stability, and relatively higher blood
concentration than BNP, another commonly used plasma biomarker of heart failure
[11]
[12]. Liu et al. found that elevated
NT-pro-BNP level is associated with myocardial fibrosis assessed by cardiac MRI
T1-mapping in a community-based population [13]. However, the relationship between NT-pro-BNP and multi-parameters on
left ventricular remodeling derived from cardiac MRI in patients with PA remains
unclear. This study aimed to assess the relationship between left ventricular
remodeling parameters of cardiac MRI and NT-pro-BNP in patients with PA. Our results
will be helpful in better comprehending the potential role of neurohormone
activation in left ventricular remodeling of patients with PA.
Materials and Methods
Study population
Patients diagnosed with PA [14] were
prospectively recruited between May 2020 and May 2022. This study was approved
by the Ethics Committee of Sichuan University West China Hospital in accordance
with the Declaration of Helsinki, as revised in 2013 (IRB No. 2016 355), and is
registered in the Chinese Clinical Trial Registry (Registration number:
ChiCTR2000031792). Each participant provided written informed consent before
enrollment. In this study, mineralocorticoid receptor antagonists and
potassium-sparing diuretics were withdrawn 4 weeks before aldosterone-to-renin
ratio (ARR) testing. Participants with a plasma ARR≥30, participants with an
ARR≥20 while plasma renin activity (PRA)<1 ng/mL/h and participants with a
plasma aldosterone concentration≥15 ng/dL were further evaluated using
confirmatory tests. Confirmatory tests included a saline infusion and/or a
captopril challenge test. A post-infusion plasma aldosterone concentration
of>10 ng/dL was the cut-off value for PA with the saline infusion test, while
a 30% captopril-induced suppression of plasma aldosterone after the captopril
challenge test indicated PA. As a control group, patients with essential
hypertension (EH) were also included in this study. EH was defined as systolic
blood pressure (SBP)≥140 mmHg or diastolic blood pressure (DBP)≥90 mmHg, while
eliminating the possibility of secondary hypertension. The exclusion criteria
were: 1) patients aged<18 years, 2) patients with known cardiovascular
disease except for hypertension, such as myocardial infarction, unstable angina,
atrial fibrillation, severe arrhythmia, systolic heart failure, cardiomyopathy,
and valvular disease, 3) patients with cardioverter defibrillator or pacemaker
implantation, 4) patients with any other known chronic disease (including
neurological disease, chronic lung disease, diabetes mellitus, cancer,
autoimmune disease, etc.), 5) patients with systemic infection, severe trauma,
or history of surgery within the past 3 months, 6) patients with claustrophobia
or other conditions that could lead to premature scan termination, and 7)
patients with artifacts on cardiac MRI. The baseline characteristics, including
demographic data, laboratory examination results, and cardiac MRI-derived
parameters, were collected. Serum levels of potassium, plasma N-terminal
pro-brain natriuretic peptide (NT-pro-BNP), troponin T(TnT), and creatine kinase
MB (CK-MB) were measured using standardized equipment by the clinical
laboratory. Normal plasma NT-pro-BNP level was defined according to the American
College of Cardiology Foundation/American College of Cardiology guidelines for
the management of heart failure [15].
Cardiac magnetic resonance imaging acquisition
All subjects underwent cardiac MRI using a 3 T scanner (MAGNETOM Trio A Tim
System; Siemens Healthcare, Erlangen, Germany). All the cardiac MRI images were
acquired per the standard protocol [16]. The balanced steady-state-free-precession sequence was used to
obtain the cine images in short-axis planes from the base of the heart to the
apex, and the following scan parameters were used: repetition time (TR)/echo
time (TE), 3.4 ms/1.3 ms; field of view (FOV), 320–360 mm2; flip
angle (FA), 50º; voxel size, 1.4×1.3×8 mm3; matrix size,
256×144; and thickness, 8 mm with no gap. T1 mapping was obtained using a
motion-corrected Modified Look-Locker Inversion (MOLLI) recovery sequence with a
scanning scheme of 5b(3b)3b (where b stands for heartbeat) on the
mid-ventricular short-axis slice. The parameters for MOLLI were as follows: TR,
2.9 ms; TE, 1.12 ms; total acquisition, 11 heartbeats; in-plane spatial
resolution, 2.4×1.8 mm; FA, 35°; bandwidth, 930 Hz/pixel; inversion time (TI) of
the first experiment, 100 ms; TI increment, 80 ms; and matrix, 192×144.
Postcontrast T1 mapping was repeated approximately 15 min after intravenous
injection of gadolinium using the same MOLLI sequence (scan scheme:
4b(1b)3b(1b)2b) in the same slice. Hematocrit (HCT) was acquired to calculate
the extracellular volume (ECV) within 24 h of cardiac MRI acquisition.
Cardiac magnetic resonance imaging analysis
All subjects were analyzed using dedicated software (Argus; Siemens Healthcare,
Erlangen, Germany). Two radiologists (each with more than 5 years of experience
and 1000 cases) delineated the endocardial and epicardial contours in diastole
and systole in a stack of short-axis slices that covered the whole left
ventricle. The left ventricular function parameters, including left ventricular
end-diastolic volume, left ventricular end-systolic volume, left ventricular
ejection fraction, and left ventricular mass, were obtained. Body-surface area
indexed values, except left ventricular ejection fraction, were calculated [17].
T1 mapping measurements
Native T1 and ECV were analyzed using the QMass7.6 software (Medis, Leiden, The
Netherlands) based on the MOLLI images of mid-ventricular short-axis slice, and
the endocardial and epicardial contours were traced manually on the pre- and
post-contrast images ([Fig. 1]). For
calculating ECV, a region of interest in the center of the blood pool in the
pre- and post-contrast T1 map should be drawn, excluding papillary muscles and
trabeculae. ECV was calculated as follows [18]:
Fig. 1 Post-processing of T1 mapping in a patient with primary
aldosteronism (PA). Endocardial (red circle by solid line) and
epicardial contours (green circle by solid line) were traced manually on
the pre-contrast (a) and post-contrast (b) images of a PA
patient to calculate native T1 and post T1 values of left ventricular
myocardium. For the calculation of extracellular volume (ECV)
(c), region of interest (ROI) was drawn in the center of the
blood pool (red circle by dotted line).
Left ventricular myocardial strain
The myocardial strain was quantified using prototype analytic software
(TrufiStrain; Siemens Healthcare, Erlangen, Germany) on cine images. The endo-
and epicardial contours of the end-diastolic left ventricle were drawn manually,
and the contours on the additional cardiac phases were detected automatically
([Fig. 2]). Left ventricular
four-chamber global longitudinal strain (GLS-4ch) and four-chamber global radial
strain (GRS-4ch) were obtained in the four-chamber slices. Short-axis global
circumferential strain (GCS-sax) and short-axis global radial strain (GRS-sax)
were measured in the mid-short-axis slices.
Fig. 2 Post-processing of feature-tracking in a patient with PA.
Left ventricular short-axis global circumferential strain (GCS-sax) and
short-axis global radial strain (GRS-sax) were measured in the mid
short-axis slice (a). Four-chamber global longitudinal strain
(GLS-4ch) and four-chamber global radial strain (GRS-4ch) were obtained
in the four-chamber slice (b). Short-axis global radial strain
curve (c), short-axis global circumferential strain curve
(e), four-chamber global radial strain curve (d), and
four-chamber global radial strain curve (f) were also
acquired.
Statistical analysis
Statistical analysis was performed using statistical software SPSS 23 (IBM
Corporation, Chicago, USA) and GraphPad Prism 6 (GraphPad Software, San Diego,
USA). Normally distributed continuous variables were expressed as the
mean±standard deviation, and continuous variables with non-normal distribution
were expressed as the median and interquartile range. The categorical variables
were expressed as percentages. Differences in continuous variables between two
groups were analyzed using Student’s t-test or Mann–Whitney U test.
Differences in categorical variables between two groups were analyzed by
chi-squared test. Spearman’s correlation and linear regression were used to
analyze the relationships between left ventricular remodeling parameters and
plasma NT-pro-BNP level. Variables with P<0.05 on univariable regressions
were included in the multivariable regression analysis. Variables independently
related to left ventricular remodeling parameters were selected in a step-wise
method. P<0.05 was considered as statistically significant.
Results
Clinical Characteristics
A total of 74 patients with PA and 39 patients with EH were included in this
study. The clinical characteristics of patients with PA are presented in [Table 1], while the clinical
characteristics of patients with EH are presented in Supplementary Table
1. With essentially the same baseline level of gender, age, BMI, blood
pressure, and high blood pressure history, patients with PA demonstrated higher
NT-pro-BNP level than patients with EH. Patients with abnormal NT-pro-BNP level
had lower BMI, HR, HCT, and PRA and higher hsTnT and ARR than those with normal
NT-pro-BNP.
Table 1 Clinical Characteristics of Patients with Normal
and with Abnormal Plasma NT-pro BNP level.
|
Overall (N=74)
|
Normal (N=59)
|
Abnormal (N=15)
|
P
|
|
Male (%)
|
24 (32.4)
|
17 (28.8)
|
7 (46.7)
|
0.312
|
|
Age (years)
|
47.0±11.7
|
48.1±11.9
|
42.8±10.5
|
0.118
|
|
BMI (kg/m2)
|
24.8±3.6
|
25.2±3.6
|
23.1±3.4
|
0.045
|
|
SBP (mmHg)
|
146±17
|
146±16
|
147±23
|
0.833
|
|
DBP (mmHg)
|
94±12
|
94±12
|
96±11
|
0.583
|
|
HR (min-1)
|
76±11
|
78±10
|
71±11
|
0.014
|
|
HCT (%)
|
41.0 (38.0~43.0)
|
41.0 (38.5~43.0)
|
38.0
(33.5~42.0)
|
0.039
|
|
Serum potassium (mmol/L)
|
3.5 (3.1~3.9)
|
3.6 (3.1~3.9)
|
3.2 (2.9~3.7)
|
0.170
|
|
Ald (ng/dL)
|
29.1 (21.9~41.0)
|
28.5 (22.0~38.9)
|
38.9
(23.0~45.6)
|
0.259
|
|
PRA (ng/mL/h)
|
0.2 (0.0~0.5)
|
0.3 (0.1~0.6)
|
0.0 (0.0~0.3)
|
0.010
|
|
ARR
|
113.0 (49.9~534.9)
|
101.1 (45.2~381.3)
|
498.6 (127.0~955.8)
|
0.011
|
|
NT-proBNP (pg/mL)
|
86.0 (49.5~145.5)
|
70.0(42.0~96.5)
|
232.0(159.5~372.0)
|
<0.001
|
|
hsTnT (ng/L)
|
6.5 (3.8~9.7)
|
5.3 (3.5~8.9)
|
8.0 (6.6~10.8)
|
0.015
|
|
CKMB (ng/mL)
|
1.1 (0.8~1.4)
|
1.1 (0.8~1.3)
|
1.2 (0.7~1.7)
|
0.481
|
|
HBP history (months)
|
36 (12–114)
|
36 (11–96)
|
48 (12–120)
|
0.666
|
BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood
pressure; HR: heart rate; HCT: Hematocrit; Ald, aldosterone; PRA, plasma
renin activity; ARR, aldosterone-to-renin ratio; NT-pro BNP, N-terminal
pro-brain natriuretic peptide; hsTnT: high sensitivity troponin T; CKMB,
creatine kinase MB type; HBP: high blood pressure.
Left ventricular remodeling parameters in patients with primary aldosteronism
and essential hypertension
Comparisons of left ventricular remodeling parameters between patients with PA
and EH are presented in Supplementary Figure 1. Patients with PA
demonstrated higher native T1 level than patients with EH (1227±41 vs. 1206±43
ms, P=0.015). Comparisons of left ventricular remodeling parameters between
patients with normal and abnormal NT-pro-BNP levels are presented in [Fig. 3]. Compared with normal
NT-pro-BNP patients, patients with PA having abnormal NT-pro-BNP had higher left
ventricular end-diastolic volume index (LVEDVi) (79.7±12.6 vs.
93.8±20.2 mL/m2, P=0.001); left ventricular end-systolic volume
index (LVESVi) (32.4±9.0 vs. 44.3±15.1 mL/m2, P<0.001); left
ventricular mass index (LVmassi) (56.1±13.2 vs. 76.3±21.2 g/m2,
P<0.001); ECV (26.4±2.8 vs. 28.2±4.2%, P=0.045); short-axis global
circumferential strain (GCS-sax) [(−15.5±2.6) vs. (−12.6±2.8)%, P<0.001]; and
four-chamber global longitudinal strain (GLS-4ch) [(−14.0±2.7) vs. (−12.1±1.5)%,
P=0.011] and lower left ventricular ejection fraction (LVEF) [(59.7±7.6) vs.
(53.6±8.8)%, P=0.008]; short-axis global radial strain (GRS-sax) (46.1±10.4 vs.
37.0±10.4%, P=0.003); and four-chamber global radial strain (GRS-4ch) (35.1±9.0
vs. 28.9±5.4%, P=0.013).
Fig. 3 Comparison of CMR left ventricular remodeling parameters
between patients with different plasma NT-pro BNP level. Box represents
the quartile of data. Line inside the box represents the median of data,
and the whisker represents the 95% confidence interval. Points outside
the whisker represents the data outside the 95% confidence interval.
CMR: cardiac magnetic resonance; LVEDVi, left ventricular end-diastolic
volume index; LVESVi, left ventricular end-systolic volume index;
LVmassi, left ventricular mass index; LVEF, left ventricular ejection
fraction; ECV, extracellular volume; GCS, global circumferential strain;
GRS, global radial strain; GLS, global longitudinal strain; sax: short
axis; 4ch: 4 chambers.
Relationship between NT-pro-BNP and left ventricle remodeling
parameters
Spearman’s correlation between NT-pro-BNP and left ventricle remodeling
parameters are shown in [Table 2].
LVEDVi, LVESVi, LVmassi, native T1, ECV, GCS-sax, GLS-4ch, and GRS-4ch were
significantly related to the plasma NT-pro-BNP level.
Table 2 Spearman’s Correlations Between NT-pro BNP and
Left Ventricular Remodeling Parameters.
|
r
|
P
|
|
LVEDVi(mL/m2)
|
0.315
|
0.006
|
|
LVESVi(mL/m2)
|
0.313
|
0.007
|
|
LVmassi(g/m2)
|
0.313
|
0.007
|
|
LVEF(%)
|
−0.171
|
0.145
|
|
Native T1(ms)
|
0.266
|
0.022
|
|
ECV(%)
|
0.375
|
0.001
|
|
GCS-sax(%)
|
0.241
|
0.039
|
|
GRS-sax(%)
|
−0.151
|
0.199
|
|
GLS-4ch(%)
|
0.250
|
0.032
|
|
GRS-4ch(%)
|
−0.323
|
0.005
|
LVEDVi, left ventricular end-diastolic volume index; LVESVi, left
ventricular end-systolic volume index; LVmassi, left ventricular mass
index; LVEF, left ventricular ejection fraction; ECV, extracellular
volume; GCS, global circumferential strain; GRS, global radial strain;
GLS, global longitudinal strain; sax: short axis; 4ch: 4 chambers.
Univariable and multivariable linear regression analysis was performed to
determine the factors influencing the left ventricle remodeling parameters
significantly related to plasma NT-pro-BNP level ([Table 3] , Supplementary Table
2, and Supplementary Table 3). After adjusting the statistically
significant variables in univariable linear regression analysis (P<0.05),
plasma NT-pro-BNP level was independently related to LVEDVi, LVESVi, LVmassi,
native T1, ECV, GCS-sax, GLS-4ch, and GRS-4ch.
Table 3 Linear Regression Analysis between Left
Ventricular Volume and Mass Parameters and Plasma NT-pro BNP
Level.
|
LVEDVi (mL/m2) (R2=0.454)
|
LVESVi (mL/m2) (R2=0.448)
|
LVmassi (g/m2) (R2=0.423)
|
|
Univariable
|
Multivariable
|
Univariable
|
Multivariable
|
Univariable
|
Multivariable
|
|
β
|
P
|
β
|
P
|
β
|
P
|
β
|
P
|
β
|
P
|
β
|
P
|
|
Male
|
0.262
|
0.024
|
0.266
|
0.004
|
0.164
|
0.164
|
|
|
0.415
|
<0.001
|
0.421
|
<0.001
|
|
Age (years)
|
−0.201
|
0.086
|
|
|
−0.171
|
0.144
|
|
|
−0.231
|
0.048
|
−0.190
|
0.052
|
|
BMI (kg/m2)
|
0.104
|
0.379
|
|
|
0.119
|
0.311
|
|
|
0.107
|
0.364
|
|
|
|
SBP (mmHg)
|
0.233
|
0.047
|
0.031
|
0.750
|
0.162
|
0.171
|
|
|
0.274
|
0.046
|
|
|
|
DBP (mmHg)
|
0.155
|
0.190
|
|
|
0.118
|
0.318
|
|
|
0.173
|
0.043
|
|
|
|
HR (min-1)
|
−0.175
|
0.135
|
|
|
−0.062
|
0.600
|
|
|
−0.184
|
0.117
|
|
|
|
Serum potassium (mmol/L)
|
−0.280
|
0.016
|
−0.081
|
0.458
|
−0.187
|
0.111
|
|
|
−0.299
|
0.010
|
|
|
|
Ald (ng/dL)
|
0.292
|
0.012
|
0.202
|
0.030
|
0.251
|
0.031
|
0.209
|
0.024
|
0.230
|
0.049
|
|
|
|
PRA (ng/mL/h)
|
−0.219
|
0.060
|
|
|
−0.177
|
0.132
|
|
|
−0.177
|
0.130
|
|
|
|
NT-pro BNP (pg/mL)
|
0.441
|
<0.001
|
0.306
|
0.003
|
0.570
|
<0.001
|
0.445
|
<0.001
|
0.388
|
0.001
|
0.322
|
0.003
|
|
hsTnT (ng/L)
|
0.345
|
0.003
|
−0.117
|
0.324
|
0.341
|
0.004
|
0.053
|
0.604
|
0.459
|
<0.001
|
|
|
|
CKMB (ng/mL)
|
0.566
|
<0.001
|
0.403
|
<0.001
|
0.529
|
<0.001
|
0.309
|
0.003
|
0.358
|
0.001
|
0.289
|
0.007
|
|
HBP history (months)
|
−0.067
|
0.568
|
|
|
−0.041
|
0.732
|
|
|
−0.104
|
0.378
|
|
|
BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood
pressure; HR: heart rate; Ald, aldosterone; PRA, plasma renin activity;
NT-pro BNP, N-terminal pro-brain natriuretic peptide; hsTnT: high
sensitivity troponin T; CKMB, creatine kinase MB type; HBP: high blood
pressure; LVEDVi, left ventricular end-diastolic volume index; LVESVi,
left ventricular end-systolic volume index; LVmassi, left ventricular
mass index Variables with P<0.05 on univariable regressions were
included in the multivariable regression analysis R2,
adjusted R2 value of the multivariable linear regression
model.
Discussion
In this study, we found that 1) patients with PA demonstrated higher NT-pro-BNP and
native T1 level than patients with EH, 2) patients with PA with abnormal plasma
NT-pro-BNP level showed different left ventricular remodeling parameters, including
left ventricular function, strain, and T1 mapping parameters when compared with PA
patients with normal plasma NT-pro-BNP level, and 3) plasma NT-pro-BNP level was
also independently related to left ventricular remodeling parameters, including
LVEDVi, LVESVi, LVmassi, native T1, ECV, GCS-sax, GLS-4ch, and GRS-4ch in patients
with PA.
NT-pro-BNP is generated along with BNP in response to the increased atrial and
ventricular wall stress [19]
[20]. Increasing wall stress, as a result
of increasing pre- or after-load in the heart, is related to ventricular remodeling.
Breetveld et al. found that increased pressure load was related to concentric left
ventricular remodeling in patients with preeclampsia [21]. Chen et al. found that left
ventricular after-load, demonstrated by blood pressure, was related to wall
thickness and LVmassi in an Asian asymptomatic cohort [22]. This could explain the phenomenon
that plasma NT-pro-BNP level was related to left ventricular remodeling
parameters.
Some clinical studies support our results and reveal the relationship between
NT-pro-BNP and left ventricle remodeling. In the PROVE-HF study, reduction of
NT-pro-BNP was correlated with left ventricle remodeling parameters, represented by
a decrease in left ventricle volume and an increase in ejection fraction measured by
echocardiography during the one-year follow-up in patients with heart failure with
reduced ejection fraction (HFrEF) treated by sacubitril-valsartan [23]. The EVALUATE-HF study reached a
similar conclusion that NT-pro-BNP was negatively correlated with the decrease of
left ventricle volume derived from echocardiography after 12-week treatment of
sacubitril-valsartan in patients with HFrEF [24]. Furthermore, in a study from the Multi-Ethnic Study of
Atherosclerosis (MESA) cohort, NT-pro-BNP was related to native T1 and ECV derived
from cardiac MRI, showing that elevation of NT-pro-BNP was an indicator of
subclinical left ventricle fibrosis, which was a sign of early left ventricle
remodeling [13].
However, these studies were not tailored to patients with PA. Our study is specific
to patients with PA and has comprehensively measured the left ventricle remodeling
parameters, including indices of left ventricle volume, function, tissue
characteristics, and strain. Little was known about the role that NT-pro-BNP plays
in patients with PA. Jakubik et al. found that BNP in patients with PA or EH did not
show significant differences, but BNP in both patients with PA and EH was higher
than in healthy controls [25]. Kato et
al. found that BNP is related to cardiac load or volume retention in patients with
PA due to adrenal adenoma [26]. In this
study, the BNP level was independently related to T1 mapping and feature tracking
parameters, which could be measured by native T1, ECV, and ventricular strain. These
parameters were proven to be imaging markers of early left ventricular remodeling
parameters [27]
[28]
[29]
[30]
[31]
[32]. To our knowledge, this is the first study to analyze the correlation
between NT-pro-BNP and early left ventricular remodeling parameters on cardiac MRI.
The results suggested that doctors should pay attention to the NT-pro-BNP level in
patients with PA to be alert to early left ventricular remodeling and perform timely
intervention. Regular follow-up of NT-pro-BNP levels in patients with PA might be
necessary. However, this study does not describe the relationship between NT-pro-BNP
and follow-up outcomes of cardiac MRI-derived left ventricular remodeling
indicators.
Our study has some limitations. First, the sample size of our study was relatively
small, which limited the application value of this study. Thus, future studies with
larger sample sizes are needed. Second, this was a cross-sectional study. The
results of this study only reflected the relationship between baseline plasma
NT-pro-BNP level and baseline left ventricular remodeling parameters, and we did not
conduct a follow-up cardiac MRI in these patients with PA. The follow-up of these
patients is needed to evaluate the relationship between baseline plasma NT-pro-BNP
level and change of the cardiac MRI-derived left ventricular remodeling parameters,
which could more intuitively reflect the phenomenon of left ventricular remodeling
than baseline left ventricular remodeling parameters.
Conclusions
In this study, we found the relationship between plasma NT-pro-BNP level and left
ventricular remodeling parameters derived from cardiac MRI in patients with PA. This
result implied that in routine clinical assessment, clinicians should pay attention
to NT-pro-BNP assessment in patients with PA.
Author contribution statement
Author contribution statement
Tao Wu and Chenxiao Xu contributed equally to the study design, data analysis and
interpretation, statistical analysis, and manuscript drafting. Jiayu Sun and Yan Ren
are the supervisors of this study and contributed to the study design, preparation,
editing, and review of the final manuscript. Lu Tang collected clinical data.
Yucheng Chen contributed to the study design and helped revise the manuscript. Xi
Wu, Xun Yue, and Pengfei Peng analyzed the imaging data. Wei Cheng, Shuai He, and
Lei Li carried out subject scanning and performed data analysis and interpretation.
All the authors read and approved the final manuscript.
