Predictive power of late gadolinium enhancement for myocardial recovery in chronic ischaemic heart failure: a HEART sub‐study

Abstract

Background

The amount of myocardial scar measured by late gadolinium enhancement (LGE) cardiovascular magnetic resonance (CMR) imaging predicts regional recovery in wall motion following revascularization. Previous studies have been conducted in patients with a relatively recent myocardial insult and relatively preserved left ventricular (LV) function. In this sub‐study of a clinical trial, the predictive value of LGE, and other CMR‐derived data, for myocardial recovery in patients with chronic severe ischaemic cardiomyopathy was assessed.

Methods

Twenty‐two patients with severe LV impairment of ischaemic origin were enrolled as a sub‐study of a trial that randomly assigned patients to revascularization or not in addition to guideline‐indicated pharmacological therapy. Patients underwent a CMR study at baseline and 6 months. Scans were qualitatively and quantitatively assessed for wall motion, rest/stress myocardial perfusion, and LGE.

Results

The median duration of heart failure was 13 (inter‐quartile range 5–21) months. Patients had severe LV dilatation [end‐diastolic volume (EDV) 280 ± 77 mL] and reduction in LV ejection fraction (LVEF) (29 ± 10%). The percentage scar burden by LGE was 17 ± 9%. Patient characteristics of those undergoing revascularization (n = 7) or not (n = 14) were similar. Myocardial perfusion reserve index (MPRI) improved following revascularization (MPRI 1.17 vs. 1.57, P < 0.0001) but not following medical therapy (1.39 vs. 1.32, P = 0.54). However, LVEF improved in patients whether or not they had revascularization. In the revascularization group, 14% of dysfunctional segments with LGE <25% and 22% of dysfunctional segments with LGE <50% had improved contractile function. However, the transmural extent of LGE did not predict contractile recovery following revascularization or pharmacological therapy (P = 0.19, P = 0.42). LVEDV improved overall (280 ± 77 to 269 ± 83 mL, P = 0.05); improvement was associated with heart failure duration (P = 0.04).

Conclusions

In patients with chronic severe LV impairment of ischaemic origin, duration of heart failure is a better predictor of recovery than transmural extent of LGE, following medical therapy or successful revascularization. This suggests that the extent of myocardial remodelling is more important for LV recovery than the presence and extent of prior infarction alone and that LGE should not be the sole determinant of treatment method in severe LV systolic dysfunction of ischaemic origin.

Introduction

A series of trials have demonstrated the benefits of pharmacological and device therapy in appropriately selected patients. The benefit of revascularization in patients with left ventricular systolic dysfunction (LVSD) associated with an acute coronary syndrome has also been demonstrated, but the benefit of revascularization for chronic stable heart failure has not.

Early trials of revascularization for heart failure selected patients solely on the basis of coronary anatomy and systolic function, whereas more recent trials have included assessment of myocardial viability and scar. The reference standard for measurement of myocardial infarction and scar is late gadolinium enhancement (LGE) cardiovascular magnetic resonance (CMR) imaging. In observational studies, the transmural extent of LGE has predicted the recovery of contractile function and regional wall improvement following revascularization but some of these studies have enrolled patients shortly after an ischaemic insult. The ability of LGE to predict recovery in patients with chronic, severe, ischaemic cardiomyopathy and whether revascularization adds to the benefits of pharmacological therapy is unknown. CMR can also be used to assess the effect of revascularization on myocardial blood flow that may explain the success or failure of the procedure.

This study investigated the relationship between the extent of LGE and myocardial recovery in patients with long‐standing severe ischaemic cardiomyopathy treated either pharmacologically alone or with additional revascularization.

Methods

Twenty‐two patients were recruited from a multi‐centre randomized controlled study Heart Failure Revascularisation Trial (HEART‐UK ) comparing best medical treatment [Angiotensin‐converting enzyme (ACE) inhibitor, beta‐blocker, and aldosterone antagonist] with best medical therapy plus revascularization for patients with chronic heart failure and LVSD secondary to ischaemic heart disease but with little or no angina. Revascularization was by whichever conventional means (i.e. percutaneous coronary intervention or coronary artery bypass grafting) on which the attending cardiologist and cardiac surgeon reached consensus. Patients were entered into the clinical trial on an intention to treat basis, and this CMR study was a separate sub‐study.

Ethical approval was obtained for the CMR study from participating regional sites and all patients gave written informed consent prior to inclusion. Patients who fulfilled the following criteria were included: (i) chronic heart failure for at least 6 weeks; (ii) treated with diuretics; (iii) left ventricular ejection fraction (LVEF) of 35% or less on echocardiography, nuclear scintigraphy, or left ventriculogram and coronary artery disease as the cause of LV dysfunction; and (iv) at least 5 of 17 segments that showed contractile dysfunction but were viable as assessed by either stress echocardiography or nuclear myocardial perfusion scanning. Contraindications included patients who were not candidates for bypass surgery because of frailty or serious co‐morbidity; unstable angina, myocardial infarction, or stroke within the preceding 2 months or patients being considered for revascularization for the relief of angina or valve surgery. In addition to these criteria, other contra‐indications for the sub‐study were a history of airways disease or conduction abnormalities precluding pharmacological stressing with adenosine and/or any contra‐indication to CMR scanning.

Cardiovascular magnetic resonance scanning protocol

Recruited patients underwent two CMR examinations on a 1.5 T scanner (Philips Intera CV, Philips Medical Systems, Best, The Netherlands), equipped with a 5‐element cardiac synergy surface coil and vectorcardiogram gating. The first scan was performed at baseline and the second 6 months after receiving assigned therapy. Left ventricular function was assessed via contiguous multiple slice short‐axis cines using a steady state free precession sequence. Rest and stress myocardial perfusion imaging was performed using a saturation recovery fast gradient echo sequence in conjunction with sensitivity encoding parallel imaging [repetition time (TR) 2.8, echo time (TE) 1.4, flip angle 55^°; 10–14 slices covering the entire left ventricle: slice thickness 6 mm, 4 mm inter‐slice gap, acquired resolution of 1.88 × 2.21 × 10 mm³ reconstructed to 1.88 × 14.88 × 10 mm³]. A bolus of dimeglumine gadopentetate (Magnevist, Schering AG, Berlin, Germany) contrast medium was rapidly injected by a power injector into an antecubital vein via an 18 gauge peripheral cannula at a dose of 0.05 mmol/kg, followed by a flush of 20 mL normal saline for rest imaging; a second injection of 0.05 mmol/kg gadolinium was repeated after a 15 min interval at point of maximum vasodilator stress during continuous intravenous adenosine infusion (4 min into a 6 min infusion). An inversion recovery segmented k‐space T1‐weighted spoiled gradient echo sequence with a non‐selective inversion recovery pre‐pulse and trigger delay set for acquisition in mid‐diastole was used for LGE imaging (6‒8 slices to cover the entire LV; TR 7.5 ms, TE 3.8 ms, flip angle 15^°, Ti adjusted to achieve optimal suppression of normal myocardium).

Cardiovascular magnetic resonance analysis

All analyses were performed by a single experienced observer blinded to all clinical data (PS). Left ventricular volumes, mass, and ejection fraction were calculated from the short‐axis data sets, using a disc‐summation method and commercially available post‐processing software (Mass 5.0, Medis, Leiden, The Netherlands). All subsequent analysis was on a segmental basis using a 16 segment modified version (apical segment not included) of the American Heart Association model for tomographic imaging, with representative matching basal, mid, and apical slices being selected from the short axis cine and LGE data sets.

Combined wall thickening and wall motion for each segment was graded visually as follows: 0 = normal; 1 = hypokinesia; 2 = severe hypokinesia; 3 = akinesia or 4 = dyskinesia, and an LV wall motion index was subsequently calculated. An improvement of segmental function was defined as an increase in systolic wall motion of at least one grade on follow‐up, with deterioration defined as the converse. On LGE images, the extent of hyper‐enhanced tissue within each segment was measured and the percentage of hyper‐enhancing myocardium calculated for the left ventricle as a whole. In addition, the transmural extent of hyper‐enhancement was graded on a five‐point scale: 0 = none; 1 = 1–25% thickness of the myocardial segment, 2 = 26–50%, 3 = 51–75%, and 4 = 76–100 %. Semi‐quantitative assessments of myocardial perfusion for each segment on initial and follow up scan were performed on separate dedicated software (View Forum, Philips Medical Systems, Best, The Netherlands). The upslope of first pass perfusion curves created from plots of changes in myocardial signal intensity over time at rest and hyperaemic stress were calculated using a five‐point linear fit. These were corrected for baseline myocardial signal intensity pre‐contrast and arterial input as derived by blood pool signal intensity within the equivalent slice. The corrected upslope for stress was divided by the value for rest and the result expressed as a myocardial perfusion reserve index (MPRI).

Statistical analysis

All data are presented as mean +/− standard deviation (continuous) or median (inter‐quartile range). Normality was determined by the Shapiro–Wilks test. The Student's t‐test was used for continuous variables and the Χ² test for categorical comparisons. Changes over time were assessed for differences between, and within, groups by paired t‐test. Data were processed by patient or segment as stated. Two‐sided significance of P < 0.05 was considered statistically significant. All statistical analysis was performed using SPSS v20 (IBM, Chicago, Illinois, USA).

Results

Baseline characteristics

One of the 22 recruited patients received an implantable cardio‐defibrillator following intervention and was excluded from analysis. Characteristics of the 21 patients in the final analysis are summarized in Tables and . Patients had severe heart failure with a mean LV end‐diastolic volume (LVEDV) of 280 ± 77 mL, and mean LVEF of 29 ± 10%. Mean time from diagnosis of heart failure was 16 ± 17 months, time between baseline and follow‐up scans was 216 ± 62 days.

Subject characteristics

	All patients (n = 21)	Revascularization (n = 7)	Conservative strategy (n = 14)	P value (revascularization vs. conservative)
Age (years)	64 ± 8	60 ± 5	66 ± 9	0.14
Male gender	19	7	12	0.47
BMI (kg/m2)	28.7 ± 3.0	29.4 ± 2.2	28.3 ± 3.3	0.45
Ex‐smoker or current smoker	13	6	7	0.11
Diabetes	7	3	4	0.51
Hypercholesterolaemia	15	6	9	0.31
Family history	17	5	12	0.43
Atrial fibrillation	8	0	8	0.01
NYHA classification
1	3	1	2
2	14	4	10
3	4	2	2
4	0	0	0
Severe symptoms (NYHA III)	4	2	2	0.43
Time since HF diagnosis (months)	13 (5,21)	22.5 (16.5,40.5)	9 (2,14)	0.01
Pre‐LVEDV (mL)	280 ± 77	320 ± 75	261 ± 73	0.1
LVEF (%)	29 ± 10	28 ± 13	29 ± 9	0.73
LVWMI	2.0 ± 0.5	2.0 ± 0.7	2.0 ± 0.4	0.83
%LV mass displaying LGE	17.6 ± 9.3	19.5 ± 11.4	16.7 ± 8.2	0.53

BMI, body mass index; HF, heart failure; LGE; late gadolinium enhancement; LV, left ventricular; LVEDV, LV end‐diastolic volume; LVEF, LV ejection fraction; NYHA, New York Heart Association; LVWMI, Left ventricular wall motion index. Study subject medication at study enrolment and completion

	All (n = 21)	Revascularization (n = 7)	Conservative (n = 14)
	Enrolment	Completion	Enrolment	Completion	Enrolment	Completion
Beta‐blocker	18	20	6	6	12	14
ACE inhibitor/ARB	17	20	6	6	11	14
Aldosterone antagonist	2	5	0	1	2	4
Digoxin	4	4	1	1	3	3
Amiodarone	0	1	0	0	0	1
Statin	14	14	5	5	9	9

ACE inhibitor, Angiotensin‐converting enzyme inhibitor; ARB, Angiotensin receptor blocker.

Of the 21 patients, 14 were randomized to the conservative strategy and 7 to revascularization. Five patients underwent coronary artery bypass graft, with a mean of three grafts, and two underwent percutaneous coronary intervention, both for two‐vessel disease. There were no significant differences in patient characteristics between the two treatment groups at baseline. Time between baseline and follow‐up scans was shorter in the medical arm at 186 days (174–196) than in the revascularization arm, at 274 days (175–373).

Segmental analysis

For the final analysis, 336 myocardial segments were available. Contractile function was abnormal in 93% of all segments at enrolment. The severity of contractile abnormalities did not differ between groups (Table ). Pre‐intervention regional wall motion correlated strongly with extent of LGE (beta coefficient 0.892, P < 0.01).

Pre‐treatment cardiovascular magnetic resonance findings by segment

	All (n = 336)	Revascularization (n = 112)	Medical (n = 224)	P value (revascularization vs medical)
% Transmural enhancement
0	135 (40%)	46 (41%)	89 (40)	P = 0.82
1–25	51 (15%)	17 (15)	34 (15)
26–50	60 (18%)	19 (17)	41 (18)
51–75	52 (16%)	20 (18)	32 (14)
76–100	38 (11%)	10 (9)	28 (13)
Segmental function
Normal	24 (1%)	11 (10)	13 (6)	P = 0.41
Hypokinetic	90 (27%)	24 (21)	66 (30)
Akinetic	96 (29%)	35 (31)	61 (27)
Dyskinetic or aneurysmal	126 (38%)	42 (38)	84 (38)

The presence and degree of functional improvement from baseline to follow up were poorly predicted by the extent of LGE. Improvement in segmental contractile function relative to transmurality of LGE is shown in Figure . Overall lower extent of transmurality of LGE did not predict recovery. Pre and post‐intervention LGE did not differ significantly and was similar between groups (P = 0.59).

Overall, 20% of dysfunctional segments with LGE <25% improved, and 22% with LGE <50% improved systolic thickening by at least one grade (Table ). Improvement was similar in the revascularized and medical therapy groups (Tables and ).

Change in segmental function overall (all segments), post‐intervention, P = 0.21. Negative change in wall motion score indicates improvement

	% Transmural extent of LGE
Change in wall motion score	0	1–25	26–50	51–75	76–100
−3	1	0	0	0	0
−2	1	3	4	0	0
−1	21	12	11	11	7
0	82	30	38	31	29
1	26	6	7	9	2
2	4	0	0	1	0

LGE, late gadolinium enhancement. Revascularization: change in segmental function post intervention, P = 0.20

	% Transmural extent of LGE
Change in wall motion score	0	1–25	26–50	51–75	76–100
−3	1	0	0	0	0
−2	0	1	2	0	0
−1	4	6	2	4	4
0	33	9	13	13	6
1	8	1	2	3	0
2	0	0	0	0	0

LGE, late gadolinium enhancement. Conservative strategy: change in segmental function post intervention, P = 0.41

	% Transmural extent of LGE
Change in wall motion score	0	1–25	26–50	51–75	76–100
−3	0	0	0	0	0
−2	1	2	2	0	0
−1	17	6	9	7	3
0	49	21	25	18	23
1	18	5	5	6	2
2	4	0	0	1	0

LGE, late gadolinium enhancement.

It was not possible to calculate MPRIs in 22 segments that had either full thickness or apical infarcts with marked wall thinning. MPRI improved following revascularization (1.17 vs. 1.57, P <0.0001) but not following medical treatment (1.39 vs. 1.32, P = 0.54), see Figure . There was no correlation between improvement in segmental function and MPRI in either arm.

Per patient analysis

Following intervention, LVEDV decreased by 12 ± 25 mL overall. Improvement was seen in 13 patients, 2 of 7 in the revascularization and 11 of 14 in the medical therapy arm (P = 0.03). EDV decreased significantly following medical therapy (261 ± 73 mL to 244 ± 74 mL, P = 0.03), but not following revascularization (320 ± 28 mL to 318 ± 32 mL, P = 0.87). LVEF did not change following treatment, and no difference was seen between groups (P = 0.87), as shown in Table .

Left ventricular function and remodelling

	All patients (n = 21)	Revascularization (n = 7)	Medical (n = 14)	P value (revascularization vs medical)
LVEF post‐treatment	29 ± 10%	28 ± 14%	29 ± 8%	0.82
LVEF improved	9	3	6	1.0
Change in LVEF %	0 ± 7	0 ± 10%	0.3 ± 10.4	0.87
LVEDV after treatment (mL)	269 ± 84	318 ± 84	244 ± 74	0.05
LVEDV improved	13	2	11	0.03
Change in LVEDV (mL)	−12 ± 25	−2 ± 24	−25 ± 44	0.2

LVEDV, left ventricular end‐diastolic volume; LVEF, left ventricular ejection fraction.

Reduction in LVEDV was associated with the duration of heart failure in the entire study group (P = 0.04).

Discussion

Dysfunctional myocardium with LGE <50% is generally considered to be ‘viable’ and expected to recover contractile function following revascularization. Previous series have reported rates of contractile recovery of 60–78% in patients with no or limited LGE. However, in these studies, overall LV systolic function was not severely impaired, a relatively few segments were dysfunctional, and time from to insult to revascularization was short.

In contrast, patients in this study had a longstanding diagnosis of ischaemic cardiomyopathy with widespread contractile dysfunction and severely impaired LVEF. Data for prediction of recovery of function in long‐standing ischaemia and contractile dysfunction are sparse. A recent report suggested that in the absence of extensive infarction, severely remodelled myocardium may recover systolic function following revascularization.

In this population, functional recovery was much lower than in previous reports, with only 23/162 (14%) of dysfunctional segments with LGE <25% improving at follow‐up. This was irrespective of whether patients were treated pharmacologically or with revascularization.

The likely cause for the reduced functional recovery in this study was the long duration of ischaemia and its severity. Chronic ‘hibernation’ leads to changes in myocyte metabolism and the extra‐cellular matrix. These changes become more severe with prolonged ischaemic injury and duration of hibernation. Delayed improvement in contractile function has previously been shown to be associated with greater duration of hibernation. The low rate of recovery seen in this population is therefore likely to be due to established changes in the extra‐cellular matrix and cellular metabolism and to cardiac myocyte de‐differentiation.

The duration of hibernation and time to recover systolic function correlate with recovery times following revascularization ranging for days to months. In previous studies, follow up imaging of patients was performed at approximately 3 to 6 months. In this study patients were scanned 186 ± 7 days after commencing medical therapy and 274 ± 79 days after revascularization. Although the follow‐up duration was thus longer than in previous reports, it may still have been insufficient to detect functional recovery in a cohort of patients with long‐standing and severe ischaemic heart failure.

This study is too small to derive definitive comparisons between medical treatment and revascularization. However, several observations can be made. Segmental recovery was seen in both medical and revascularization groups. Contemporary medical therapy with ACE inhibitor and beta‐blockers are known to improve LV function in heart failure, especially if the LV is not severely dilated. However, on average, LVEF did not improve significantly in either group. This could reflect the small number of patients. Most patients had received pharmacological treatment for a long time prior to the study and might have already received any expected benefit. However, LVEDV appeared to improve in the conservatively managed group compared with those who were revascularized. This could reflect myocardial damage caused by the revascularization procedure. Differences in the severity of LV dilatation and duration of disease between the two groups and a chance statistical difference due to multiple comparisons provide alternative explanations. Ultimately, in a study of this size, small differences should not be over‐interpreted. However, this is a randomized study suggesting that revascularization does not cause large, consistent benefits on LV function in patients with chronic ischaemic cardiomyopathy and is therefore an important finding.

This study also showed no correlation between the transmural extent of LGE and recovery following revascularization, despite clear improvement in MPRI demonstrating successful revascularization. Previous studies have not measured perfusion alongside LGE, and our study provides the first evidence that improved flow does not necessarily lead to improved function when assessed at 6 months. The lack of functional improvement despite improved perfusion suggests established structural and metabolic change within hibernating myocardium as described earlier that are not reversible within the follow‐up period.

Limitations

The main limitation of this study is its sample size, and it therefore has to be considered as hypothesis‐generating. However, it does show that improved LV function with revascularization is not universal, consistent, nor to be taken for granted. Larger studies should test the ability of LGE to predict more modest or erratic recovery in function. Such studies might also use T1 mapping and extra cellular volume mapping that may better predict the likelihood of recovery following revascularization in patients with ischaemic cardiomyopathy who do not have visually detectable LGE.

Conclusions

In patients with long‐standing severe LV impairment of ischaemic origin, duration of heart failure is a better predictor of global functional recovery than transmural extent of LGE, following medical therapy or revascularization. This suggests that the transmurality of LGE, commonly used as the primary basis of revascularization decision making, may fail to capture the extent of remodelling and the potential for recovery in severe long‐standing LVSD.

Further studies should be performed that are powered to detect indices that predict response following revascularization in severe LVSD, possibly using CMR techniques better at assessing diffuse myocardial change, including extracellular volume fraction (ECV) calculation with T1 mapping. More importantly, more randomized trials should be conducted to show that the benefits of revascularization justify the risk and cost, even in carefully selected patients with heart failure.

Funding

The study was supported by a British Heart Foundation research grant.