Quantification of Diffuse Myocardial Fibrosis and its Association with Myocardial Dysfunction in Congenital Heart Disease

Abstract

Background

The etiology of ventricular dysfunction in adult congenital heart disease (ACHD) is not well understood. Diffuse fibrosis is a likely common final pathway and is quantifiable using magnetic resonance imaging (MRI).

Methods and Results

ACHD patients (N=50) were studied with cardiac MRI to quantify systemic ventricular volume and function, and diffuse fibrosis. The fibrosis index for a single mid-ventricular plane of the systemic ventricle was quantified by measuring T₁ values for blood pool and myocardium before and after administration of gadolinium (0.15 mmol/kg), then adjusted for hematocrit. Results were compared to healthy volunteers (normal controls, N=14) and patients with acquired heart failure (positive controls, N=4). Patients studied (age 37±12 years, 40% female) included 11 with a systemic right ventricle (RV), 17 with tetralogy of Fallot, 10 with cyanosis and 12 with other lesions. The fibrosis index was significantly elevated in ACHD patients compared to normal controls (31.9±4.9% vs. 24.8±2.0%, p=0.001). Values were highest in systemic RV patients (35.0±5.8%, p<0.001) and cyanotic patients (33.7±5.6%, p<0.001). The fibrosis index correlated with end-diastolic volume index (r=0.60, p<0.001) and ventricular ejection fraction (r=−0.53, p<0.001), but not with age, nor oxygen saturation in cyanotic patients. Late gadolinium enhancement did not account for the differences seen.

Conclusions

ACHD patients have evidence of diffuse, extracellular matrix remodeling, similar to patients with acquired heart failure. The fibrosis index may facilitate studies on the mechanisms and treatment of myocardial fibrosis and heart failure in these patients.

Heart failure in patients with adult congenital heart disease (ACHD) is an increasingly common and life-threatening complication for a substantial portion of patients,^{1,2, 3,4} Yet etiology is not well understood, and treatment options are largely empiric. Myocardial fibrosis, or abnormal accumulation of extracellular material in the myocardium, is a common final pathway of a number of cardiovascular stresses^{5, 6} leading to diastolic dysfunction, systolic dysfunction, arrhythmia, and increased mortality; developments which all respond favorably to pharmacotherapy.^7–12 Limited studies suggest a similar pathway in ACHD, although evidence for proven response to medical therapy is lacking. Therefore, fibrosis is an ideal pathologic marker on which to focus attention.

In the 1960s, clinicians observed that exercise performance after surgical repair with congenital defects was still below that of normal individuals, and postulated the presence of a “myocardial factor” responsible for the persistent limitation.¹³ This important observation was likely the first to acknowledge that repair of the anatomic defect could not guarantee full restoration of cardiovascular function. Several years later, pathologists identified changes within the myocardium of ACHD patients including interstitial fibrosis, confirming the notion of the myocardial factor.^{14,15, 16} Few studies have continued this line of investigation. Surgical samples in young tetralogy of Fallot patients showed fibrosis in 65%.¹⁷ Recent interest has been revived by the use of late gadolinium enhancement (LGE) by magnetic resonance imaging (MRI), which has demonstrated macroscopic fibrosis in both the right ventricle (RV) and left ventricle (LV) in ACHD.^18–22 However, the method is not conducive to quantify diffuse microscopic fibrosis.

The capacity to detect, quantify, and follow myocardial fibrosis offers tremendous advantage in efforts to study etiology and response to therapy. Quantification of T₁ changes before and after gadolinium (Gd) administration during MRI has been successfully employed to differentiate normal and abnormal hearts.^23–26 We have further developed this method and used it to quantify myocardial extracellular volume fraction as a marker of extracellular matrix remodeling,^{27, 28} which we term the “fibrosis index.” The method is objective and independent of cardiac output, gadolinium dose, or imaging timing. It can be widely applied and serially obtained. We sought to apply this method for detection and quantification of fibrosis in ACHD patients.

Methods

Patient Selection

Consecutive adults with ACHD referred for MRI at our institution were studied according to accepted clinical standards including the use of gadolinium. Specifically, though not exclusively, we focused on three diagnostic subgroups; A) patients with a systemic RV (L-transposition of the great arteries or D-transposition with prior atrial redirection surgery, N=11), B) repaired tetralogy of Fallot (N=17), and C) cyanotic heart disease including Eisenmenger syndrome (N=10). These were targeted because systemic ventricular dysfunction is a known complication in each, they are commonly referred for MRI, and because fibrosis has been shown by LGE in all three groups.^18–22 Additional patients with other congenital abnormalities with an assumed increased likelihood of diffuse fibrosis were also included (N=12). In addition, we studied patients with dilated or ischemic cardiomyopathy with visible replacement fibrosis (positive controls, N=4) and healthy volunteers (negative controls, N=14). Groups are summarized in Table 1. The protocol was approved by the Institutional Review Board, and healthy volunteers signed informed consent prior to participation.

Table 1.

Subgroup Considerations

	N	M/F	Age (yrs)	EF (%)	EDVi (ml/m2)
			Mean ± SD	Mean ± SD	Mean ± SD
Healthy (neg. control)	14	8/6	37.7 ± 11.9	66 ± 7	79 ± 19
Systemic RV	11	9/2	36.1 ± 13.6	46 ± 11	111 ± 32
Tetralogy	17	8/9	34.1 ± 11.7	57 ± 12	82 ± 16
Cyanosis	10	4/6	42.3 ± 10.9	54 ± 11	142 ± 69
Other Congenital	12	9/3	32.9 ± 12.4	63 ± 10	95 ± 33
Acquired (pos. control)	4	2/2	58.3 ± 4.54*	50 ± 21	105 ± 37

Study population by subgroup. All values are for the systemic ventricle in each group.

^*Acquired heart disease patients were older by ANOVA. EF = Ejection Fraction, EDVi = End-diastolic Volume Index. RV = right ventricle.

MRI Protocol

Patients were studied using either a 1.5 Tesla (Philips Achieva) or a 3 Tesla (Philips Intera) scanner, depending solely on scanner availability. Short axis images through the heart were obtained to quantify RV and LV volume and function using established methods.^{18, 19} Phase velocity flow mapping through the aorta and pulmonary artery were used as confirmation of ventricular stroke volumes. Measurements of T₁ were made using a Look-Locker technique^29–31 (a gradient echo sequence with a non-slice selective inversion pulse initially after the R wave, followed by segmented gradient echo acquisition for 21 cardiac phases over 2–3 cardiac cycles; temporal resolution 40 ms, slice thickness 8 mm; repetition time > 3 R-to-R intervals) for a single mid-ventricular plane. This type of sequence is often employed for determination of optimal nulling for delayed enhancement (“inversion time scout”). This same Look-Locker acquisition was repeated, using the same mid-ventricular plane, and at various intervals 3–25 minutes after injection of Gd-DTPA (Omniscan, 0.15 mmol/kg), interwoven during the LGE portion of the study. LGE was studied in the usual manner 10–20 minutes after gadolinium administration.^{18, 19} Total scan times were typically 50–90 minutes depending on the clinical questions addressed. Scans were done according to clinical need, not purely for study purposes.

Fibrosis Index Quantification

For each T₁ sequence the endo- and epicardial borders of the systemic ventricle were traced (Figure 1A) and divided into 6 standard segments (Mass CMR software, Medis, The Netherlands). We did not study the sub-pulmonic ventricle in any patient. The signal intensity vs. time curve for each segment and blood pool was used to determine a segmental T₁ through exponential fitting (Figure 1B), and its reciprocal, R₁. This was done for each Look-Locker acquisition. The slope of the linear relationship between R₁ for myocardium vs. blood from all R₁ measurements before and after gadolinium administration defined the partition coefficient for gadolinium, or lambda³⁰ (Figure 1C), and values for all 6 myocardial segments were averaged. To obtain the myocardial volume of distribution of gadolinium, or extracellular volume fraction, the partition coefficient was multiplied by (1-hematocrit/100)^{27, 31} which we refer to as the “fibrosis index.”

Figure 1.

A) Images from Look-Locker sequence for a patient with tetralogy of Fallot. Regions of interest for the LV myocardium and blood pool are drawn. B) The time vs. signal intensity curves for the blood pool (red) and myocardium (blue), which are fitted to determine the decay constant T₁. The inverse, R₁, for each is determined before gadolinium administration and at several time points after. C) The relationship of R₁ signals between the blood pool and the myocardium before and after gadolinium is linear. The slope of this relationship is the partition coefficient of gadolinium, and determines the fibrosis index. Examples from a patient with cyanosis (Eisenmenger syndrome, blue) and a normal control (pink) are shown. The steeper slope reflects the increase in gadolinium present due to an enlarged extracellular space.

Late Gadolinium Enhancement (LGE)

Areas of late enhancement (unrelated to sites of expected surgical intervention such as a VSD patch) were identified and categorized by location. To determine whether the presence of LGE influenced the fibrosis index, we quantified the area of LGE within the mid-ventricular plane used for the fibrosis index, expressed as a percentage of total myocardial area in the plane (% area LGE). In this quantification, we included any enhancement in the RV-LV junction that overlapped with the region of interest drawn for the fibrosis index.

Statistical Analysis

Results are presented as mean ± SD. One way ANOVA was used to compare fibrosis index values for all ACHD patient subgroups and both positive and negative controls, with Tukey’s post-hoc analysis (SPSS, version 11.0). Age, systemic ventricular ejection fraction (EF), and end-diastolic volume index (EDVi) were also compared in the same manner. Student’s t-test was used to compare patients with vs. without LGE. Pearson’s correlation coefficient was used to correlate the fibrosis index with age, oxygen saturation in cyanotics, ventricular volumes, ejection fraction, and % area of LGE. To test whether the association of EDVi, EF, and age with the fibrosis index varied between diagnostic groups, an interaction between fibrosis index and diagnostic group was included in linear regression models. Non-parametric methods (Kruskal Wallis test with follow-up Mann-Whitney comparisons and Spearmen Correlation coefficients) were also used to assess the sensitivity of the results to standard parametric assumptions. One way ANOVA for repeated measures was used to detect a possible association between mean fibrosis index and the index in specific myocardial segments.

Results

Fifty adult ACHD patients were studied (age 37±12 years, range 18–71, 20 female, 30 male). In addition to the target subgroups (Table 1), we studied 4 patients with pressure overloading (3 with coarctation of the aorta and one with bicuspid aortic stenosis), 5 with volume loading (systemic atrioventricular valvular regurgitation or shunts), 2 with single ventricle Fontan palliation, and 1 with prior Rastelli repair. In addition we studied 4 positive controls and 14 healthy volunteers as negative controls. Positive controls were significantly older. From the entire group (including controls), 25 studies were performed at 3.0 Tesla, and 38 at 1.5 Tesla, with no overall difference in the fibrosis index between them (p=0.40), including no differences between healthy controls specifically.

Fibrosis index values for all groups are shown (Table 2). Overall comparisons among groups was highly significant (ANOVA p<0.0001). The fibrosis index was significantly elevated in ACHD patients compared to negative controls (31.9±4.9% vs. 24.8±2.0%, p=0.001) and was comparable to the mean fibrosis index in positive controls (36.2±5.7%, p=0.1). Values were highest in systemic RV patients (35.0±5.8%, p<0.001 vs. normal controls) and cyanotic patients (33.7±5.6%, p<0.001 vs. normal controls). The fibrosis index correlated with EDVi (r=0.60, 95% CI 0.42–0.73, p<0.001, Figure 2) and EF (r=−0.53, 95% CI −0.68 - −0.34, p<0.001, Figure 3), but not with age (r=0.11, 95% CI −0.13–0.34, p=0.37). There was no evidence that the association of EDVi, EF, and age with the fibrosis index varied between diagnostic groups, and therefore the data for ACHD patients were pooled. Results were similar when using non-parametric procedures. As mean + 2 SD for the control group was 28.8%, we defined an abnormal fibrosis index as any value ≥29%.

Table 2.

Fibrosis index

	Mean ± SD	Min.	Max.	p vs Healthy	p vs Acquired
Healthy (neg. control)	24.8% ± 2.0%	21.6%	28.1%	--	<0.001
Systemic RV	35.0% ± 5.8%	25.3%	45.5%	<0.001	0.997
Tetralogy	29.2% ± 3.4%	23.9%	38.0%	0.051	0.039
Cyanosis	33.7% ± 5.6%	23.5%	41.9%	<0.001	0.926
Other Congenital	31.4% ± 3.2%	26.2%	36.5%	0.002	0.436
Acquired (pos. control)	36.2% ± 5.8%	32.7%	44.9%	0.003	--

Results of the fibrosis index by subgroup. Omnibus p-value <0 .0001. Tukey’s method used for follow-up comparisons. RV=right ventricle.

Figure 2.

Relationship between systemic ventricular volume index and the fibrosis index (r=0.60, 95% CI 0.42–0.73, p<0.001). Dotted vertical line is the upper limit of normal.

Figure 3.

Relationship between systemic ventricular ejection fraction and the fibrosis index is linear. (r=-0.53, 95% CI -0.68 - -0.34, p<0.001). Dotted vertical line is the upper limit of normal.

Subgroup Differences

Of the 11 systemic RV patients, 10 had abnormal fibrosis index values. There was a positive relationship between EDVi (morphologic RV) and the fibrosis index, (r=0.65, p=0.03) and a non-significant trend with EF (r=−0.40, p=0.2). There was no association with age. Two patients had severe tricuspid regurgitation, including the patient with the highest value of 46%, who had an ejection fraction of 45%. There was no association with tricuspid valve regurgitant fraction (estimated by stroke volume comparisons).

All tetralogy patients had been repaired and none were cyanotic. The mean fibrosis index for the LV was higher than for healthy controls (29.2±3.4%), though this did not reach statistical significance (p=0.051). The value was abnormal in 9 of the 17 patients. The patient with the highest fibrosis index (value of 38%), had severe LV dysfunction (EF 30%). Overall there was no correlation with LV EDVi or EF as seen in the other subgroups. There was also no correlation between the fibrosis index (for the LV) and RV size or function. We did not measure the fibrosis index for the RV. There was a non-significant correlation between the pulmonary valve regurgitant fraction and the fibrosis index (r=0.44, p=0.086). Patients with an abnormal fibrosis index (N=9) compared to those with normal values (N=8) more often had moderate to severe pulmonary regurgitation (p=0.02, Fisher’s exact test), and had fewer prior cardiac surgeries (range 1- 3 surgeries, p=0.034 by chi-square test). We found no significant differences in prior shunt placement, age at time of shunt placement, or age at time of repair.

In order to determine the influence of the RV on septal fibrosis in tetralogy of Fallot, we compared fibrosis index values of the 6 segments of the myocardial wall in all 17 patients using a random effects model. Range of fibrosis index measures for the tetralogy of Fallot patients by segment were as follows: Anterior 21.7–37.1, Anterior-septal 23.7–39.5, Inferior-septal 21.1–41.5, Inferior 24.9–40.8, Lateral 23.1–34.9, Posterior 23.5–33.2 %. Similar ranges for normal controls were: Anterior 20.6–29.3, Anterior-septal 18.7–29.9, Inferior-septal 21.4–29.5, Inferior 21.6 − 31.5, Lateral 19.6–27.1, Posterior 19.6–26.3 %. ANOVA for repeated measures indicated that in tetralogy patients there was a significant variation of the means of the fibrosis index among segments (p=0.009). In healthy volunteers the same test did not indicate any significant variation between segments (p=0.127). The largest values for the fibrosis segment in tetralogy patients were observed in the inferior segments. Post-hoc paired comparisons confirmed that among the 6 myocardial segments the inferior segment had higher fibrosis index values compared with all other segments (p<0.05 with Bonferroni correction) except the inferior-septal segment. Relative to the anterior segment this difference of fibrosis index only amounted to a trend (p<0.1). Therefore, the septum was not disproportionately affected compared to other myocardial segments.

In cyanotic patients (mean oxygen saturation 80 ± 4%), 9 out of 10 had abnormal values. There was a strong correlation between the fibrosis index and systemic EDVi (r=0.72, p=0.019) and EF (r=−0.86, p<0.001). There was no correlation of the fibrosis index with resting oxygen saturation (mean 80 ± 4%), or age. Even after exclusion of cyanotic patients from the rest of the cohort, relationships of the fibrosis index with EDVi and EF remained significant for the remaining patients (r=0.493, p<0.001 and r=−0.451, p=0.001 respectively).

Ten of the 12 other ACHD patients studied had abnormal fibrosis coefficient values; including all 3 coarctation patients, 5 patients with volume loading, and both Fontan patients. One patient with a prior surgical repair of a VSD was normal, as was a patient with bicuspid aortic valve stenosis.

Late Gadolinium Enhancement (LGE)

Of the total ACHD cohort, two patients (1 cyanotic and 1 systemic RV) had poor LGE data from difficulty nulling and were not included. Of the remaining, LGE was present in the systemic ventricle in 27 (56%), specifically 5/10 transposition patients, 12/17 tetralogy patients, 5/9 cyanotic patients, and 5/12 other ACHD patients. The most common LGE finding was enhancement at the RV-LV junction, and in 12 patients (7 with tetralogy) this was the only area of enhancement present. The majority of other positive LGE findings involved small portions of septum or papillary muscle. Only one patient had transmural enhancement consistent with a small myocardial infarction, known from a perioperative coronary embolism. There was no difference in the fibrosis index between patients with vs. without LGE (31.2±4.8% vs. 33.2±5.1%, p=0.17). All patients with acquired heart disease had visible LGE within the plane selected for the fibrosis index.

The % area LGE in ACHD patients was significantly lower than the amount seen in acquired heart disease (3.0±2.8% vs. 14.1±6.3%, p<0.001). For all ACHD patients, there was no correlation between the fibrosis index and % area LGE (r=0.22, Figure 4). In 5 patients, the LGE findings were not present in the plane used for LGE. Even excluding these patients, there was no relationship between % area LGE and the fibrosis index (r=0.27).

Figure 4.

Relationship between fibrosis index and the area LGE (area of positive late enhancement / total area of myocardium in the selected plane) for ACHD patients. There was no correlation, and many patients with significantly elevated fibrosis index had little/no detectable LGE. By comparison, patients with acquired heart disease had area LGE values of 16–20% (data not shown). Dotted vertical line is the upper limit of normal.

Discussion

ACHD patients show evidence of myocardial extracellular matrix expansion causing an enlargement of the volume of distribution of gadolinium, namely the fibrosis index. We found a relationship between this index and ventricular volume and systolic function across a subset of ACHD patients. This is congruous with histopathologic studies demonstrating diffuse fibrosis in ACHD.^{14, 16, 17} Together with other data showing neurohormonal activation,³² exercise intolerance,³³ and mortality⁴ similar to patients with acquired heart failure, our findings provide additional evidence that ACHD is a form of progressive heart failure with a pathologic process similar to other etiologies.

We employed a technique for assessing diffuse fibrosis based on changes of T₁ after gadolinium administration. Gd-contrast based changes in myocardial R₁ accurately reflect Gd concentration and relative distribution volume with good reproducibility,^23,24 and correlate with histological evidence of fibrosis even in the absence of LGE.²⁵ The methodology avoids the necessity of defining “normal” myocardium for nulling as is done with LGE imaging.³⁴ However, reliance on T₁ alone as an indicator of extracellular remodelling requires consistency in gadolinium dosing and timing of post-contrast image acquisition.^{34, 35} The method used here overcomes this by measuring the myocardial partition coefficient for Gd-based contrast whereby the R₁ changes in myocardium are normalized by the R₁ changes in the blood. Correcting for hematocrit to account for changes in plasma Gd concentration gives the fibrosis index, a measure of extracellular volume fraction. Increased extracellular volume implies the presence of diffuse interstitial fibrosis. The normal range in our study agrees with other estimates of the extracellular myocardial volume.²⁶ Our data cannot exclude the possibility that replacement fibrosis (or frank myocyte necrosis) contributed to the observed findings. However, elevation of the fibrosis index was not explained by visible LGE, and was found diffusely across wall segments. Although the presence of LGE will raise the fibrosis index, as we found in positive controls, the ACHD patients had fibrosis index values far greater than what could be explained by visible LGE alone (Figure 4), and 21 patients had no LGE in the selected plane despite an elevated fibrosis index. These findings are consistent with the interpretation that diffuse microscopic fibrosis is present.

The amount of LGE seen in target subgroups was generally consistent with previous reports.^18–20 Although degree of LGE in tetralogy and D-transposition patients correlates with adverse clinical variables,^{18, 19} the clinical relevance of LGE in Eisenmenger patients remains undetermined.²⁰ One possible reason for this may be the difficulty of quantifying LGE when the fibrotic change may be much more diffuse, as has been described in cardiac amyloidosis.³⁶ Indeed, difficulty nulling may reflect the diffuse nature of extracellular myocardial change, and nulling with shorter inversion times would effectively suppress the evidence of diffuse fibrosis.

Our use of positive controls specifically included patients with visible fibrosis on LGE, including patients with both ischemic and non-ischemic cardiomyopathy. We recognize that not all non-ischemic cardiomyopathy patients will necessarily have visible fibrosis by LGE.^{37, 38} Therefore, we also made comparisons on patients with idiopathic cardiomyopathy from our previously study.³⁹ We identified 7 patients (5 men, age 59±16 years) with reduced LV function (LVEF 32±14%) and quantified the fibrosis index for each. The fibrosis index (mean 31.4±4.6%) was significantly higher than our control group (p<0.001), but not different from our ACHD cohort nor positive controls (p=0.78 and p=0.16 respectively). There were significant correlations within this group between the fibrosis index and EDVi and EF (r=0.51 and −0.59 respectively), and the index did not increase with age. These confirmatory findings also support the similarities between ACHD and acquired heart failure.

Specific Subgroup Considerations

Extracellular remodeling was present in all three diagnostic subgroups (though less so in tetralogy) despite likely differences in the pathogenesis of myocyte dysfunction in each such as volume and/or pressure loading, abnormal ventricular geometry, previous cardiopulmonary bypass, or chronic myocardial ischemia. The finding that diffuse interstitial fibrosis is present to some degree in all three groups underscores this as a common final pathway and invites further study into specific inciting mechanisms in each subgroup.

We found comparatively less fibrosis in the tetralogy group, which generally had less evidence of heart failure. Although systolic dysfunction is more common in the RV in this group, we measured fibrosis in the LV. LV dysfunction is increasingly common and has prognostic significance,^{40, 41} though very little is known about its etiology. In practical terms, it is harder to reliably demarcate the thin-walled RV. In this limited subgroup, we found no evidence of interventricular interaction that might explain the development of fibrosis in the LV through electromechanical and/or neurohormonal coupling with the RV. However, the finding of more severe PR in patients with higher LV fibrosis does suggest a role of RV loading.

Other patients included those with pressure loading. All coarctation patients were previously repaired and did not have restenosis at the time of the study. The bicuspid valve patient had moderate stenosis, though the fibrosis value was not elevated as we would have expected. The patient was asymptomatic and it is possible that fibrosis had not yet occurred in response to the load.

Clinical Implications and Future Directions

The importance of our data lies not only in demonstration of diffuse fibrosis but also in the potential application of the fibrosis index as a tool for future studies. The quantification of diffuse fibrosis serially over time has tremendous implications for both therapeutic and mechanistic studies.

Therapeutically, studies of specific medications known to attenuate or reverse fibrosis can employ this method for detecting change non-invasively in specific patient groups. To date, studies have generally not succeeded in showing a favorable impact of these drugs using clinical endpoints.^{42, 43} An exception to this was the report that tetralogy patients with restrictive RV physiology randomized to lisinopril vs. placebo for 6 months had a favorable increase in LV ejection fraction.⁴⁴ This is interesting in light of our data, which suggest a possible rationale for the findings; namely that restriction is known to be associated with fibrosis, and fibrosis should improve with ACE inhibitor therapy. Small studies with surrogate endpoints such as the fibrosis index will be easier to fund and coordinate. These in turn can provide important pilot data to support larger multicenter studies with hard clinical endpoints including exercise capacity, arrhythmia, or even mortality, which are tremendously lacking in the field.

Mechanistically, the identification of key inciting events that trigger fibrogenesis as well as their biochemical pathways will manifest differences between subgroups, and be the key to prevention. Using the fibrosis index in either humans or animal models coupled with biomarkers of inflammation, neurohormonal activation, or collagen deposition, we may demonstrate to what extent fibrogenesis is induced by pressure loading, myocardial ischemia, cardiopulmonary bypass, or microvascular thrombosis. In addition, the method can be applied to other patient groups susceptible to ventricular dysfunction.

Limitations

Our study employs an approach that is relatively novel. Only a single mid-ventricular slice was quantified, and analysis is time-consuming. The fibrosis index measures extracellular volume, which may be increased from processes other than fibrosis (such as acute inflammation or infiltrative processes). The relation between the partition coefficient of gadolinium (on which the fibrosis index is based) and collagen volume fraction (the gold standard for quantifying fibrosis) has only been tested ex-vivo,²⁸ which by necessity employs a different method of gadolinium diffusion. Obtaining histologic confirmation was not feasible here, although correlation between histology and T₁ measurements has been demonstrated.²⁵ Only systolic function was measured, although diastolic function would be expected to correlate with fibrosis. Further study will be required for additional validation. The method requires exclusion of pacemaker/ICD patients. Since the need for pacing and/or arrhythmia may be associated with myocardial fibrosis, this selection bias may exclude patients with more extreme fibrosis. Our study has limited power, and thus the lack of association/correlation does not necessarily mean that an association would not be found with a larger sample size. This is especially relevant for tests involving the 4 positive controls.

Conclusions

In congenital heart disease, it is becoming increasingly evident that protection of the myocardium should be given as much attention as the hemodynamics early on in a patient’s life, both for preservation of ventricular function and improved survival. This study demonstrates that remodeling of the extracellular matrix, or myocardial fibrosis, which can be quantified serially and non-invasively, is associated with adverse ventricular enlargement and declining ventricular function across different forms of congenital heart disease. The data confirm that myocardial fibrosis is a common final pathway leading to myocardial dysfunction, and opens further possibilities for much needed research into both the etiology and treatment of these patients.