- Email: [email protected]
Synthetic Myocardial Extracellular Volume Fraction
JACC: CARDIOVASCULAR IMAGING
VOL.
ª 2017 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION
-, NO. -, 2017
ISSN 1936-878X/$36.00
PUBLISHED BY ELSEVIER
F I G U R E 1 Linear Correlation and Regression Equations at
1.5 and 3.0 Field Strengths
LETTER TO THE EDITOR
A Synthetic Myocardial Extracellular
volume
fraction
(ECV)
Hematocrit
extracellular
Hct = 922.6 (1/T1 BLOOD) - 0.1668
0.55
R 2 = 0.50 p = < 0.001
0.50
Volume Fraction
Myocardial
0.60
0.45 0.40 0.35 0.30
assessed by cardiac magnetic resonance (CMR) esti-
0.25
mates the proportion of myocardial extracellular
0.20
space relative to its cellular component. Conven-
0.50
0.55
0.60
tionally, ECV calculation requires knowledge of the patient’s hematocrit (Hct). Treibel et al. (1) recently
B
0.60
demonstrated that a synthetic ECV can be calculated
thetic ECV has only been described for data acquired with a single vendor (Siemens, Berlin, Germany) and
Hematocrit
cost of obtaining a venous Hct sample. So far, syn-
0.70
0.75
0.65
0.70
Hct = 869.7 (1/T1 BLOOD) - 0.071 R 2 = 0.46 p = < 0.001
0.55
by estimating Hct from the longitudinal relaxation rate (R1 ¼ 1/T1) of blood. This eliminates the time and
0.65
Blood R1 (s-1)
0.50 0.45 0.40
field strength (1.5-T). We hypothesized that synthetic
0.35
ECV can also be derived from other platforms and
0.30
acquisition methods, potentially broadening appli-
0.45
0.50
0.55
0.60
Blood R1 (s-1)
cability of this method. We analyzed data of 421 patients undergoing dications, 203 of whom underwent CMR scans at 1.5-T (Ingenia, Philips, Best, the Netherlands) and 218 at 3.0T (Achieva-dStream, Philips). Patients on both scanners were randomly split into equally sized derivation and validation subgroups. Derivation groups served to enable derivation of linear regression equations for the relationship of Hct and R1 of blood. This equation was
C Conventional ECV (%)
T1 mapping for research purposes and clinical in-
80 70 60 50 40 30
10
used to calculate synthetic ECV and assess its correla-
10
tion with conventionally calculated ECV in the valiwith valvular heart disease and 44 with ST-segment elevation myocardial infarction and 112 patients undergoing CMR for acute clinical reasons. The 3.0-T cohort comprised 26 healthy control subjects and 159 patients with rheumatoid arthritis and 33 patients with hypertrophic cardiomyopathy. MOdified
Look-Locker
Inversion
Recovery
(MOLLI) acquisition schemes were used to acquire T1 maps produced using vendor software before and 15 min after administration of either 0.2 mmol/kg
20
30
40
50
60
70
80
Synthetic ECV (%)
D Conventional ECV (%)
dation groups. The 1.5-T cohort comprised 47 patients
R 2 = 0.95 p = < 0.001
20
60 50 40 30 20
R 2 = 0.92 p = < 0.001
10 10
20
30
40
50
60
Synthetic ECV (%)
gadopentate dimeglumine (Magnevist, Bayer Schering, Berlin, Germany) or 0.15 mmol/kg gadobutrol (Gadovist, Bayer Schering). For all 1.5-T patients, a
(A) Linear correlation and regression equation between blood Hct and blood R1 at 1.5-T. (B) Linear correlation and regression equation between blood Hct and blood R1 at 3.0-T. (C) Linear
pre-contrast 5s(3s)3s and post-contrast 4s(1s)3s(1s)2s
correlation between conventional and synthetic ECV at 1.5-T.
scheme was used. For all 3.0-T patients the same
(D) Linear correlation between conventional and synthetic ECV
3(3s)3(3s)5 scheme was used pre- and post-contrast.
at 3.0-T. ECV ¼ extracellular volume fraction; Hct ¼ hematocrit.
T1 values were derived by drawing a region of
2
JACC: CARDIOVASCULAR IMAGING, VOL.
Letter to the Editor
-, NO. -, 2017 - 2017:-–-
interest within the interventricular septum and
We have demonstrated that synthetic ECV, derived
blood pool at mid-ventricular level using post-
by estimating Hct from pre-contrast blood T1 values
processing software (CVI 42, Circle Cardiovascular
acquired with a MOLLI method on 1.5- and 3.0-T Philips
Imaging, Calgary, Canada). Scar was included within
systems, strongly correlates with conventionally
the interventricular septum region of interest when
calculated ECV. The correlation values demonstrated
present. Analysis was blinded. ECV was calculated
between Hct and R1 blood in the derivation cohort and
as has previously been reported (1). Statistical ana-
between conventional and synthetic ECV in the vali-
lyses were performed using SPSS version 22 (IBM
dation cohort are similar to those reported in the
Corporation, Armonk, New York). All results are
Treibel et al. (1) on a Siemens platform using both
presented as mean SD.
MOLLI and Shortened MOdified Look-Locker Inver-
There was a broad range of Hct in both 1.5-T and
sion recovery (ShMOLLI) pulse sequences. This un-
3.0-T derivation groups (0.41 0.05; range 0.27 to
derscores the accuracy of synthetic ECV and its
0.53 at 1.5-T; and 0.42 0.04; range 0.31 to 0.54 at
applicability across platforms and field strength. It
3.0-T). There was also a broad range of blood pre-
offers the potential for use on a routine clinical CMR
contrast T1 in 1.5-T and 3.0-T derivation groups
list, eliminating the need for a venous Hct sample,
(1,608 105 ms; range 1,402 to 1,912 ms at 1.5-T; and
enabling rapid clinical decision-making.
1,780 99 ms; range 1,457 to 1,993 ms at 3.0-T). The conventionally calculated ECV in the validation groups was 32 9% (range 19% to 77%) for 1.5-T and 29 5% (range 20% to 53%) for 3.0-T. The regression lines between Hct and R1 blood were linear at both field strengths (1.5-T: R 2 ¼ 0.50, p < 0.001; 3.0-T: R 2 ¼ 0.46, p < 0.001), with the following regression equations (Figure 1): 1:5-T: Synthetic Hct MOLLI ¼ ð922:6 $ ½1=T1bloodÞ 0:1668 3:0-T: Synthetic Hct MOLLI ¼ ð869:7 $ ½1=T1bloodÞ 0:071;
Graham J. Fent, MBChB Pankaj Garg, MD James R.J. Foley, MBChB Peter P. Swoboda, PhD Laura E. Dobson, MD Bara Erhayiem, BMBS John P. Greenwood, PhD Sven Plein, PhD* Thomas A. Treibel, MBBS James C. Moon, MD *Division of Biomedical Imaging Leeds Institute of Cardiovascular and Metabolic Medicine LIGHT Laboratories
where Hct is hematocrit (between 0 and 1) and
University of Leeds
R1blood is 1/T1blood in milliseconds.
LS2 9JT, United Kingdom
Using these equations to calculate synthetic ECV in
E-mail: [email protected]
both validation cohorts, conventional and synthetic
http://dx.doi.org/10.1016/j.jcmg.2016.12.007
ECV were highly correlated (Figure 1) (R 2 ¼ 0.95,
Please note: Dr. Fent is funded by a National Institute of Healthcare Research grant (1/117/27). Professor Plein is funded by a British Heart Foundation fellowship (FS 10/62/2840). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
2
p < 0.001 at 1.5-T; and R ¼ 0.92, p < 0.01 at 3.0-T). On Bland-Altman analysis, synthetic ECV demonstrated minimal bias at both 1.5-T (bias ¼ 0.81%, 95% confidence interval: 4.97% to 3.35%) and 3.0-T (bias ¼ 0.30%, 95% confidence interval: 3.92% to 3.33%).
REFERENCE 1. Treibel TA, Fontana M, Maestrini V, et al. Automatic measurement of the myocardial interstitium synthetic extracellular volume quantification without hematocrit sampling. J Am Coll Cardiol Img 2016;9:54–63.
VOL.
ª 2017 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION
-, NO. -, 2017
ISSN 1936-878X/$36.00
PUBLISHED BY ELSEVIER
F I G U R E 1 Linear Correlation and Regression Equations at
1.5 and 3.0 Field Strengths
LETTER TO THE EDITOR
A Synthetic Myocardial Extracellular
volume
fraction
(ECV)
Hematocrit
extracellular
Hct = 922.6 (1/T1 BLOOD) - 0.1668
0.55
R 2 = 0.50 p = < 0.001
0.50
Volume Fraction
Myocardial
0.60
0.45 0.40 0.35 0.30
assessed by cardiac magnetic resonance (CMR) esti-
0.25
mates the proportion of myocardial extracellular
0.20
space relative to its cellular component. Conven-
0.50
0.55
0.60
tionally, ECV calculation requires knowledge of the patient’s hematocrit (Hct). Treibel et al. (1) recently
B
0.60
demonstrated that a synthetic ECV can be calculated
thetic ECV has only been described for data acquired with a single vendor (Siemens, Berlin, Germany) and
Hematocrit
cost of obtaining a venous Hct sample. So far, syn-
0.70
0.75
0.65
0.70
Hct = 869.7 (1/T1 BLOOD) - 0.071 R 2 = 0.46 p = < 0.001
0.55
by estimating Hct from the longitudinal relaxation rate (R1 ¼ 1/T1) of blood. This eliminates the time and
0.65
Blood R1 (s-1)
0.50 0.45 0.40
field strength (1.5-T). We hypothesized that synthetic
0.35
ECV can also be derived from other platforms and
0.30
acquisition methods, potentially broadening appli-
0.45
0.50
0.55
0.60
Blood R1 (s-1)
cability of this method. We analyzed data of 421 patients undergoing dications, 203 of whom underwent CMR scans at 1.5-T (Ingenia, Philips, Best, the Netherlands) and 218 at 3.0T (Achieva-dStream, Philips). Patients on both scanners were randomly split into equally sized derivation and validation subgroups. Derivation groups served to enable derivation of linear regression equations for the relationship of Hct and R1 of blood. This equation was
C Conventional ECV (%)
T1 mapping for research purposes and clinical in-
80 70 60 50 40 30
10
used to calculate synthetic ECV and assess its correla-
10
tion with conventionally calculated ECV in the valiwith valvular heart disease and 44 with ST-segment elevation myocardial infarction and 112 patients undergoing CMR for acute clinical reasons. The 3.0-T cohort comprised 26 healthy control subjects and 159 patients with rheumatoid arthritis and 33 patients with hypertrophic cardiomyopathy. MOdified
Look-Locker
Inversion
Recovery
(MOLLI) acquisition schemes were used to acquire T1 maps produced using vendor software before and 15 min after administration of either 0.2 mmol/kg
20
30
40
50
60
70
80
Synthetic ECV (%)
D Conventional ECV (%)
dation groups. The 1.5-T cohort comprised 47 patients
R 2 = 0.95 p = < 0.001
20
60 50 40 30 20
R 2 = 0.92 p = < 0.001
10 10
20
30
40
50
60
Synthetic ECV (%)
gadopentate dimeglumine (Magnevist, Bayer Schering, Berlin, Germany) or 0.15 mmol/kg gadobutrol (Gadovist, Bayer Schering). For all 1.5-T patients, a
(A) Linear correlation and regression equation between blood Hct and blood R1 at 1.5-T. (B) Linear correlation and regression equation between blood Hct and blood R1 at 3.0-T. (C) Linear
pre-contrast 5s(3s)3s and post-contrast 4s(1s)3s(1s)2s
correlation between conventional and synthetic ECV at 1.5-T.
scheme was used. For all 3.0-T patients the same
(D) Linear correlation between conventional and synthetic ECV
3(3s)3(3s)5 scheme was used pre- and post-contrast.
at 3.0-T. ECV ¼ extracellular volume fraction; Hct ¼ hematocrit.
T1 values were derived by drawing a region of
2
JACC: CARDIOVASCULAR IMAGING, VOL.
Letter to the Editor
-, NO. -, 2017 - 2017:-–-
interest within the interventricular septum and
We have demonstrated that synthetic ECV, derived
blood pool at mid-ventricular level using post-
by estimating Hct from pre-contrast blood T1 values
processing software (CVI 42, Circle Cardiovascular
acquired with a MOLLI method on 1.5- and 3.0-T Philips
Imaging, Calgary, Canada). Scar was included within
systems, strongly correlates with conventionally
the interventricular septum region of interest when
calculated ECV. The correlation values demonstrated
present. Analysis was blinded. ECV was calculated
between Hct and R1 blood in the derivation cohort and
as has previously been reported (1). Statistical ana-
between conventional and synthetic ECV in the vali-
lyses were performed using SPSS version 22 (IBM
dation cohort are similar to those reported in the
Corporation, Armonk, New York). All results are
Treibel et al. (1) on a Siemens platform using both
presented as mean SD.
MOLLI and Shortened MOdified Look-Locker Inver-
There was a broad range of Hct in both 1.5-T and
sion recovery (ShMOLLI) pulse sequences. This un-
3.0-T derivation groups (0.41 0.05; range 0.27 to
derscores the accuracy of synthetic ECV and its
0.53 at 1.5-T; and 0.42 0.04; range 0.31 to 0.54 at
applicability across platforms and field strength. It
3.0-T). There was also a broad range of blood pre-
offers the potential for use on a routine clinical CMR
contrast T1 in 1.5-T and 3.0-T derivation groups
list, eliminating the need for a venous Hct sample,
(1,608 105 ms; range 1,402 to 1,912 ms at 1.5-T; and
enabling rapid clinical decision-making.
1,780 99 ms; range 1,457 to 1,993 ms at 3.0-T). The conventionally calculated ECV in the validation groups was 32 9% (range 19% to 77%) for 1.5-T and 29 5% (range 20% to 53%) for 3.0-T. The regression lines between Hct and R1 blood were linear at both field strengths (1.5-T: R 2 ¼ 0.50, p < 0.001; 3.0-T: R 2 ¼ 0.46, p < 0.001), with the following regression equations (Figure 1): 1:5-T: Synthetic Hct MOLLI ¼ ð922:6 $ ½1=T1bloodÞ 0:1668 3:0-T: Synthetic Hct MOLLI ¼ ð869:7 $ ½1=T1bloodÞ 0:071;
Graham J. Fent, MBChB Pankaj Garg, MD James R.J. Foley, MBChB Peter P. Swoboda, PhD Laura E. Dobson, MD Bara Erhayiem, BMBS John P. Greenwood, PhD Sven Plein, PhD* Thomas A. Treibel, MBBS James C. Moon, MD *Division of Biomedical Imaging Leeds Institute of Cardiovascular and Metabolic Medicine LIGHT Laboratories
where Hct is hematocrit (between 0 and 1) and
University of Leeds
R1blood is 1/T1blood in milliseconds.
LS2 9JT, United Kingdom
Using these equations to calculate synthetic ECV in
E-mail: [email protected]
both validation cohorts, conventional and synthetic
http://dx.doi.org/10.1016/j.jcmg.2016.12.007
ECV were highly correlated (Figure 1) (R 2 ¼ 0.95,
Please note: Dr. Fent is funded by a National Institute of Healthcare Research grant (1/117/27). Professor Plein is funded by a British Heart Foundation fellowship (FS 10/62/2840). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
2
p < 0.001 at 1.5-T; and R ¼ 0.92, p < 0.01 at 3.0-T). On Bland-Altman analysis, synthetic ECV demonstrated minimal bias at both 1.5-T (bias ¼ 0.81%, 95% confidence interval: 4.97% to 3.35%) and 3.0-T (bias ¼ 0.30%, 95% confidence interval: 3.92% to 3.33%).
REFERENCE 1. Treibel TA, Fontana M, Maestrini V, et al. Automatic measurement of the myocardial interstitium synthetic extracellular volume quantification without hematocrit sampling. J Am Coll Cardiol Img 2016;9:54–63.