Background: Cardiac magnetic resonance (CMR) is routinely used to risk stratify patients with heart failure through the evaluation of cardiac function and myocardial tissue. CMR derived lung water is an emerging technique in this risk assessment but has not been evaluated in patients with acute decompensated heart failure. Furthermore, changes in CMR-derived lung water during active diuretic treatment has not been characterized.

Methods: Patients admitted with decompensated heart failure requiring intravenous diuretic therapy were prospectively recruited to participate in this single site observational study. They underwent a baseline non-contrast CMR at 1.5 T within 48 hours of admission and at discharge. The imaging protocol included conventional cine imaging for cardiac volumes and function, native myocardial and hepatic T1 and lung water using a custom Cartesian ultrashort TE pulse sequence (TE = 140 ms) with free-breathing acquisitions (1). CMR images were analyzed using commercial software (CVI42, Circle Inc) by an experienced reader blinded to clinical outcomes. Global lung water density (LWD) at functional residual capacity (minimum lung volume) was quantified with a user-independent machine learning lung segmentation approach (Figure 1) (2). Heart failure readmission events within 6 months from the admission CMR were collected and univariate analyses evaluated potential associations of clinical and imaging metrics with outcome.

Results: The study cohort included 43 patients (29 male, mean age 67±15 years) after excluding 3 unable to complete the initial scan (Table 1). Medical history was notable for prior heart failure in 28, atrial fibrillation in 23, coronary artery disease in 16 and chronic obstructive lung disease in 6. Admission mean NT-proBNP was 6929±8635 ng/L and the median daily dose of furosemide was 80±55 mg. On the admission CMR, mean LVEF was 28±16% and 31 patients were identified as HFrEF, 6 as HFmrEF and 9 as HFpEF. Mean LWD was 32.5±7.3% and was moderately correlated to NT-proBNP (Figure 2). Normative LWD values using similar methods are 27.2±5.0% (3-5). A discharge CMR was obtained in 29 patients at a median of 5 days from admission and mean LWD decreased by 10.4±13% relative to baseline, p< 0.001 (Figure 3). During post-discharge follow-up, 7 patients were readmitted with heart failure within 180 days. NT-proBNP and indexed LV end-diastolic volume were each associated with heart failure readmission, p < 0.05, and there was a similar trend for LWD at admission, p=0.057 on univariate testing.

Conclusion: Lung water density imaging is feasible in patients admitted with acute decompensated heart failure and diuretic therapy was associated with objectively measured pulmonary decongestion. Ongoing work will establish its utility for characterizing cardiovascular risk in this setting.

Comparison of BNP and Imaging Data in Patients With and Without Heart Failure Readmissions

BNP and Imaging Comparison: CHF Readmission within 180 Days (Medians with IQRs)

Variable	Readmit < =180d	No readmit < =180d	P value
BNP Value	9.39e+03 [1.26e+04]	3.78e+03 [3.68e+03]	0.004847
Follow-up BNP	3.52e+03 [1.69e+03]	851 [1.12e+03]	0.008582
Indexed LVEDV (ml/m2)	149 [73.7]	92.8 [68.9]	0.02541
Total Lung Water Fraction (%)	35.2 [6.21]	30.9 [7.53]	0.06715
Myocardial T1 (ms)	1.06e+03 [31.5]	1.03e+03 [94]	0.09937
Indexed LV mass (g/m2)	76.9 [23.1]	63.3 [30.3]	0.2377
T1 Liver	653 [35.5]	636 [156]	0.2887
LVEF (%)	18.6 [12.5]	23.7 [28]	0.3485
Abs diff in lung water %	2.29 [10.2]	2.05 [3.4]	0.3551
Total Lung Volume (mL)	1.95e+03 [695]	2.3e+03 [633]	0.4098
BNP reduction	17.3 [45.8]	63 [37.7]	0.495
Total Lung Water (mL)	655 [130]	647 [231]	0.5294
Total Vascular Water Volume (mL)	146 [38.6]	129 [49.4]	0.5294
Follow-up Total Lung Volume (mL)	2.4e+03 [853]	2.33e+03 [867]	0.5327
Follow-up Total Lung Water (mL)	638 [245]	673 [279]	0.7897
Follow-up Total Vascular Water Volume (mL)	130 [43.8]	129 [68.5]	0.8358
Follow-up Total Lung Water Fraction (%)	30.3 [6.99]	29.7 [6.89]	0.9292

Baseline Demographic, Clinical and Imaging Data

Age	66.5±14
Male Sex	29 (67.4%)
Body Mass Index, kg/m2	30.2±5.9
Systolic Blood Pressure, mmHg	127±24
Heart Rate, /min	92±29
Respiratory Rate	21±4
Prior CHF	29
Atrial Fibrillation	23
Coronary Artery Disease	16
Chronic Obstructive Lung Disease	6
NT-proBNP, ng/L	6929±8635
Creatinine, um/L	123±59
Hemoglobin, g/L	132±22
LVEF, %	28±16
Indexed LV mass, g/m2	70±21
Indexed LVEDV, ml/m2	114±56
Myocardial T1, ms	1005±139
Hepatic T1, ms	656±121
Lung water density, %	32.5±7.4

Figure 1. Example of Lung Water Density Image in a Patient with Heart Failure
Figure 2. Correlation of Admission NT-proBNP and Lung Water Density
Figure 3. Comparison Between Admission and Discharge Lung Water Density