Although each of the three methods used to quantify the myocardial scar using LGE in CMR has a different concept to calculate the size of the scar, they all showed comparable results.
The manual method is known to be the most accurate, however, extremely time consuming. In addition, we face some technical problems when there are two or more non-connected scar mass in the same short axis slice. The second method using the total number of segments involving any late gadolinium enhancement is theoretically less accurate as it consider one segment affected even the late enhancement is focal or minimal. On the other hand it is the least time consuming method by only giving a general impression about how many of the myocardial segments include a scarred tissue.
However, the 3rd method by summation of the amount of each segment with scarred myocardium (% of each segment with a scar), is less time consuming than the manual one with considerable accuracy representing how much of the myocardium is unhealthy.
Many other methods were tested and compared for the accuracy to quantify the myocardial scar; e.g. Flett et al.[4] studied the reproducibility of LGE quantification techniques in 3 different pathological conditions; acute myocardial infarction (AMI), chronic myocardial infarction (CMI) and hypertrophic cardiomyopathy (HCM), using 7 techniques; including manual quantification, automatic methods including; thresholding by 2, 3, 4, 5, or 6 SDs above remote myocardium, and the full width at half maximum (FWHM) technique. They concluded that regardless of the underlying disease, the FWHM technique for LGE quantification gives LGE volume mean results similar to manual quantification and is statistically the most reproducible, reducing required sample sizes by up to one-half.[4]
Another study by Gao et al.[5] using automatic thresholding; measured a 50% larger scar size going from 5-SD to 2-SD thresholds above remote myocardial signal intensity. Neilan et al.[6] found that scar size was, on average, 50% greater using the 2-SD technique versus the FWHM technique (9 ± 5% by 2-SD method vs. 6 ± 4% by FWHM method); however, there was close correlation between both the measurements (r = 0.92, p < 0.001), and more importantly, both methods of quantification showed robust prognostic association.
This means; despite the different ways used to quantify the myocardial scar, the manual assessment is considered one of the most accurate methods, however, no one specific method agreed to be the standard yet.
In our study, we are mainly concerned about the prognostic value of the scar value and if there is a linked to the clinical outcome of cardiomyopathy. It was clearly observed that there is a clear linear relationship between the size of the myocardial scar (using any of the 3 methods) and the severity of the clinical events. More serious cardiac events eg; hospitalization, serious arrhythmias and sudden cardiac arrest were more detected in patients with larger scar size in both groups (P value < 0.001), (Fig. 4,5,6).
In our study, we were not just interested in the sudden cardiac arrest as the main outcome, however, the wider spectrum of the clinical events is thought to give more data about the expected clinical pattern, hence, the treatment plan and cardiac patient expected quality of life.
It was observed that serious cardiac events were less seen in patient with mean scar mass < 5.4–8.4%. On the contrary, patients who have experienced sudden cardiac arrest have a mean scar mass of 15.9 % and patients with ventricular tachycardia with mean of 9.8% (Table 1). It was also interesting that the hospitalization with heart failure manifestations is significantly more frequent with high scar mass specially in ischaemic patients with mean of 27.9%. This may be explained as the scar amount in ischaemic cardiomyopathy is directly reflecting the severity of the baseline coronary artery disease and number of territories involved. Another finding that may support this theory is the mean EF for those patients was < 45%.
Many of the other studies were focused mainly on the link of the myocardial scar and sudden cardiac death to guide the indication for ICD implantation, giving less attention to the rest of the clinical spectrum of the outcome. For example, Neilan et al.[6] demonstrated that for every 1% of LV mass increase in scar size, the risk of cardiovascular death or ventricular arrhythmia increases by 15%. This relationship was similar whether scar size was measured using the 2-SD method (HR, 1.15; 95% confidence interval, 1.12–1.18) or the FWHM-method (HR, 1.16; 95% confidence interval, 1.12–1.20). When only arrhythmic events were considered, the extent of scar was again associated with higher event risk (HR, 1.17 for each 1% absolute increase in scar size; 95% confidence interval, 1.12–1.22). Similar results were reported by Gulati et al.[7] for each percent scar extent the risk of all-cause mortality was increased by 11% (HR, 1.11; 95% confidence interval, 1.06–1.16), and the risk for arrhythmic events was increased by 10% (HR, 1.10; 95% confidence interval, 1.05–1.16).
Li et al.[8] aimed to develop a risk score (LGE Based Prediction of SCD Risk in Non-ischemic Dilated Cardiomyopathy [ESTIMATED]) based on late gadolinium enhancement (LGE) in CMR to predict sudden cardiac death (SCD) in patients with non-ischemic dilated cardiomyopathy (NIDCM) and left ventricular EF ≤ 35%. They followed up 395 patients with NIDCM for 3 years for SCD events. The ESTIMATED score (constructed by the LGE extent > 14%, syncope, atrial flutter/fibrillation, non-sustained ventricular tachycardia, advanced atrioventricular block, and age ≤ 20 or > 50 years) showed good calibrations for SCD prediction.[8] By the score, 20.3% of primary prevention patients were categorized as high risk (≥ 3 points), 28.1% as intermediate risk (2 points), and 51.6% as low risk (0–1 points) for 3-year SCD events (45.9% vs 20.1% vs 5.1%, P < 0.0001). The 3-year SCD events were also well in agreement with the score stratification in patients without implantable cardioverter-defibrillator.[8]
Their study suggested LGE-based (ESTIMATED) risk score to be validated in providing refined SCD prediction. The score may help to identify candidates for primary prevention implantable cardioverter-defibrillator in patients with NIDCM.[8]
However, our study included both ischaemic and non-ischaemic spectrum, studied them independently and also compared to each other. Moreover, the follow up included a wide spectrum of clinical events ranging from mild chest pain or shortness of breath, passing through hospitalization due to decompensated heart failure and up to malignant arrhythmia and sudden cardiac arrest. We were more concerned about the clinical pattern of the patients regarding their morbidity, hospitalization and quality of life.
The mean EF was 51% +/-14 (18–77%) in group I (32% of group I with EF < 45% and 69% with EF > 45%) and 48% +/-16 (9–77%) in group II (37% of group I with EF < 45% and 63% with EF > 45%).
There was a statistically significant linear relationship between the LV systolic dysfunction represented by LVEF and event severity in group I (ischaemic) with P value of 0.013. On the opposite for the non-ischaemic cardiomyopathy, there was no clear relationship between the LVEF and event severity (P value 0.150). In this group (non-ischaemic), the main predictor of cardiac events was the scar mass. For example; the four patients in group II that experienced sudden cardiac arrest the EF was above 45% but with high scar mass average (13.99 gm ± 13.77).
It was also observed in group I (ischaemic), the lower EF was more linked with hospitalization due to decompensated heart failure (63% of the hospital admissions has EF < 45%). In contrary, most of the patients with no events or mild chest pain or dyspnoea have EF > 45%. This could be explained as in ischaemic patients the amount of the scarred myocardium is directly linked to the severity of the underlying coronary artery disease and the amount of scar mass and its distribution may also indicated the number of coronary territory affected.
In the non-ischaemic patients, usually the preserved LVEF is usually misleading because it is not indicating the degree of the underlying myocardial pathology. However, LGE in cardiac MRI is more precise in tissue characterization and spotting the unhealthy myocardium that is usually a substrate for serious arrhythmogenic events and subsequently sudden cardiac arrest even in cases of preserved EF.
In Dokainish et al.[9] they had similar outcome to ours when they evaluated the prognostic implications of left ventricle (LV) systolic and diastolic dysfunction early post-acute ST-segment elevation myocardial infarction (STEMI). Patients with LV ejection fraction (LVEF) ≤ 45% and restrictive diastolic function (RDF) were at greatly increased risk of MACE (hazard ratio [HR] = 8.85, 95% CI, 4.21–18.60) compared to patients with LVEF ≥ 45% and without RDF. RDF remained a strong predictor for MACE in patients with LVEF ≥ 45% (HR = 4.38, 95% CI, 1.52–12.60) and in multivariate models adjusted for LVEF, LV end-systolic volume, and clinical variables.
Regarding non-ischaemic cardiomyopathy, Ge et al.[10] investigated if structural abnormality on cardiac magnetic resonance imaging (CMR) represented by LGE may be a predictor of major adverse cardiac events (MACE) in patients with non-sustained ventricular tachycardia (NSVT) and ventricular tachycardia (VT)/sudden cardiac death (SCD).
They studied 651 patients (ages, 54 ± 15; 61% men) referred to CMR for ventricular arrhythmia were divided into 2 groups, according to the presence of NSVT (53%) or sustained VT/aborted SCD (47%). MACE was a composite of cardiovascular death, a need for heart transplantation or left ventricular assist device, and recurrent VT/ventricular fibrillation needing therapy. The mean left ventricular ejection fraction was 54 ± 13%, and late gadolinium enhancement (LGE) was present in 39% of patients (mean, 9.5 ± 8%).[10]
A structurally abnormal heart, defined by LGE, abnormalities in wall motion, or impaired systolic function, was observed in 52% of patients (n = 336). A change in diagnostic impression based on CMR took place for 27% of patients with NSVT vs 40% of patients with VT/SCD (P < .001). A total of 122 patients experienced a MACE during the follow-up period (median, 3.6 years). Structural abnormality detected on CMR was found to be an independent predictor of MACE (hazard ratio, 3.65; 95% CI, 2.09–6.27; P < .001).[10]
The small number of patients included in our study, especially in each type of non-ischaemic cardiomyopathy is one of the limitations for our study. In addition, longer follow up duration would have added more significant predictive value.