Left ventricular global function index is associated with myocardial iron overload and heart failure in thalassemia major patients

Purpose: The left ventricular global function index (LVGFI) is a comprehensive marker of cardiac performance, integrating LV morphology with global function. We explored the cross-sectional association of LVGFI with myocardial iron overload (MIO), LV ejection fraction (LVEF), myocardial fibrosis, and heart failure (HF) in β-thalassemia major (TM) patients. Methods: We considered 1352 adult TM patients (708 females, 32.79 ± 7.16years) enrolled in the Myocardial Iron Overload in Thalassemia Network and 112 healthy subjects (50 females, 32.09 ± 6.08years). LVGFI and LVEF were assessed by cine images and MIO by multislice multiecho T2* technique. Replacement myocardial fibrosis was detected by late gadolinium enhancement technique. Results: LVGFI and LVEF were significantly lower in patients with significant MIO (global heart T2*<20ms) than in patients without MIO and in healthy subjects but were comparable between TM patients without MIO and healthy subjects. In TM, LVGFI was significantly associated with LVEF (R = 0.733; p < 0.0001). Global heart T2* values were significantly associated with both LVGFI and LVEF, but the correlation with LVGFI was significantly stronger (p = 0.0001). Male sex, diabetes mellitus, significant MIO, and replacement myocardial fibrosis were the strongest predictors of LVGFI. Eighty-six patients had a history of HF and showed significantly lower global heart T2* values, LVEF, and LVGFI than HF-free patients. A LVGFI ≤ 44.9% predicted the presence of HF. The LVGFI showed a diagnostic performance superior to that of LVEF (area under the curve: 0.67 vs. 0.62; p = 0.039). Conclusion: In TM patients the LVGFI correlates with MIO and provides incremental diagnostic value for HF detection compared with LVEF.


Introduction
Beta thalassemia major (β-TM) is a hereditary disorder characterized by a deficiency in the synthesis of beta-globin chains [1]. This results in chronic hemolytic anemia that is treated with multiple blood transfusions. The chronic administration of large amounts of blood combined with extravascular haemolysis and increased intestinal absorption of iron lead to iron accumulation into liver, heart, and other organs [2]. Cardiac disease caused by iron overload remains the main cause of death in patients with TM, despite the improvements in survival over the past 20 years [3][4][5][6][7]. Iron overload-related cardiomyopathy may be reversed when intensive chelation therapy is started early [8][9][10]; in clinical practice the diagnosis is often delayed because cardiac symptoms are a late manifestation.
The assessment and quantification of cardiac iron load is the key for the clinical management of TM patients. Cardiac Magnetic Resonance (CMR) using the T2* technique meets this need [11][12][13]. T2* is a magnetic relaxation property of the tissue and it is inversely related to intracellular iron stores [14]. The use of a multislice T2* technique allows a global analysis of the left ventricle (LV) [13,15].
CMR also is the gold standard for the non-invasive evaluation of cardiac function, volumes, and mass. The most frequently used index of LV function in clinical practice is the LV ejection fraction (LVEF). Reduced LVEF has emerged as a strong and reliable independent predictor of heart failure (HF) and cardiac complications in TM, overshadowing the importance of a multi-parametric CMR approach (which is typically one of the strongest aspect of CMR assessment of cardiac diseases) in this population [16]. However, LVEF has two important limitations: it does not account for the relationship between LV mass and LV dimensions [17] [18] and it fails to index diastolic dysfunction [19]. These limitations have prompted a search for novel synthetic indicators of left ventricular performance and the left ventricular global function index (LVGFI) is one of the most promising. Conversely to the LVEF, the LVGFI comprises information on both cardiac function and morphology [20], reflecting cardiac performance for different degrees of structural LV remodeling. Mewton et al. demonstrated that in a large multi-ethnic cohort of healthy subjects, without symptoms of cardiovascular disease, the LVGFI was superior to the LVEF in predicting heart failure, hard cardiovascular events, and a composite end point including all events [20]. Moreover, the LVGFI was demonstrated a useful functional parameter of the LV also in ST-segment elevation myocardial infarction (STEMI) patients, being related to infarct size and other markers of myocardial and microvascular damage and offering prognostic information beyond traditional cardiac risk factors including the LVEF [21,22].
Because of its utility, prognostic role, and simplicity, the LVGFI is an appealing index that deserves to be investigated in different population settings. To the best of our knowledge, there is no data about the distribution and the clinical correlates of LVGFI in TM.
The aim of this multicenter study was to systematically explore the cross-sectional association of LVGFI with MIO, traditional functional parameters of the left ventricle, myocardial fibrosis, and heart failure in a large cohort of welltreated β-TM patients.

Study population
We considered 1352 adult β-TM patients (708 females, 32.79 ± 7.16 years), consecutively enrolled in the Myocardial Iron Overload in Thalassemia (MIOT) project [23]. The MIOT project was a Network constituted by 70 thalassemia centers and 10 validated MRI centers, where MRI exams were performed using homogeneous, standardized, and validated procedures [24]. All centers were linked by a web-based database, collecting all patients' demographic, clinical, and instrumental data [23].
All patients were regularly transfused since early childhood to maintain a pre-transfusion hemoglobin concentration above 9-10 g/dl and were chelated. CMR scanning was performed within one week before a regular scheduled blood transfusion.
Moreover, we studied 112 healthy subjects (50 females, mean age 32.09 ± 6.08 years) who constituted the control population. Inclusion criteria were: normal electrocardiogram, no history of cardiac diseases or symptoms, no cardiovascular risk factors, no known systemic diseases, and no absolute contraindications to the CMR.
All subjects gave written informed consent. The study complied with the Declaration of Helsinki and was approved by the institutional ethics committee.

CMR
All patients underwent CMR using conventional clinical 1.5T scanners of three main vendors (GE Healthcare, Milwaukee, WI; Philips, Best, Netherlands; Siemens, Erlangen, Germany). Breath-holding in end-expiration and ECG-gating were used.
For MIO assessment, three parallel short-axis views (basal, medium, and apical) of the left ventricle (LV) were acquired at 10 echo times by a T2* gradient-echo multiecho sequence [15]. Image analysis was performed using a custom-written, previously validated software (HIPPO MIOT®) [25]. The software provided the T2* value on each of 16 LV segments, according to the standard AHA/ACC model [26]. Susceptibility artifacts were corrected using an appropriate correction map [25]. The global heart T2* value was obtained by averaging all segmental values. The intersite, inter-study, intra-observer, and inter-observer variability of the proposed methodology had been previously assessed [24,27].
For the evaluation of cardiac function, steady-state free procession (SSFP) cines were acquired during 8-second breath holds in the vertical and horizontal long axis planes, whit subsequent contiguous 8-mm short axis slices from the atrio-ventricular ring to the apex [28]. Images analysis was performed using MASS® software (Medis, Leiden, The Netherlands). The analysis was based on the manual definition of the endocardial and epicardial borders of the LV wall in end-diastolic and end-systolic phases for each slice. Moreover, the papillary muscles were delineated and were considered myocardial mass rather than part of the blood pool. For the calculation of LV end-diastolic and end-systolic LV volumes (LVEDV and LVESV, respectively), no geometric assumption of the ventricle shape was needed. The LV stroke volume (LVSV) was calculated as the difference between LVEDV and LVESV. The LVEF was given by the ratio between the LVSV and the LVEDV. The intercenter variability for the quantification of LV function had been previously reported [29]. The LVGFI was calculated as the ratio between the LVSV and the global volume, multiplied by 100 and consequently expressed as a percentage [20]. The global volume was the sum of the LV mean cavity volume [(LVEDV + LVESV)/2] and the myocardium volume. LV myocardial volume was calculated as LV myocardial mass divided through the specific myocardial density (1.05 g/mL).
Late Gadolinium Enhanced (LGE) short-axis images were acquired 10-18 min after Gadobutrol (Gadovist®; Bayer Schering Pharma; Berlin, Germany) intravenous administration at the standard dose of 0.2 mmol/kg using a fast gradient-echo inversion recovery sequence to detect myocardial fibrosis. Also, vertical, horizontal, and oblique long-axis views were acquired. LGE was considered present when visualised in two different views [30].
LGE images were not acquired in patients with a glomerular filtration rate < 30mL/min/1.73m 2 and in patients who refused the contrast medium administration.

Diagnostic criteria
A T2* measurement of 20ms was taken as a "conservative" normal value for the segmental and global T2* values [11,25,31].
Diabetes mellitus was defined as fasting plasma glucose ≥ 126 mg/dl or 2-h plasma glucose ≥ 200 mg/dl during an oral glucose tolerance test (OGTT) or a random plasma glucose ≥ 200 mg/dl with classic symptoms of hyperglycaemia or hyperglycaemic crisis [32].
HF was identified based on symptoms, signs, biomarkers, and instrumental parameters, according to the current guidelines [33].

Statistical analysis
All data were analyzed using SPSS version 27.0 and R version 4.2.1 statistical packages.
Continuous variables were described as mean ± standard deviation (SD) and categorical variables were expressed as frequencies and percentages.
The normality of distribution of the parameters was assessed by using the Kolmogorov-Smirnov test.
For continuous values with normal distribution, comparisons between groups were made by independent-samples t-test (for 2 groups) or one-way ANOVA (for more than 2 groups). Wilcoxon's signed rank test or Kruskal-Wallis test were applied for continuous values with non-normal distribution. χ2 testing was performed for categorical data. Bonferroni post hoc test was used for multiple comparisons between pairs of groups.
Correlation analysis was performed using Pearson's test or Spearman's test where appropriate. The cocor−package of R was used to compare the strength of two overlapping correlations (the same variable was part of both correlations) [34].
Univariate and stepwise multivariate regression analyses were performed to identify determinants of LVGFI. Multivariate regression was performed using only variables with a p−value <0.05 in univariate regression analyses.
The receiver operating characteristic (ROC) analysis was performed to examine the diagnostic ability of LVGFI and LVEF and the results were presented as areas under the curve (AUCs) with 95% confidence intervals (CIs). The optimal cutoff value was calculated using the Youden index method. The Delong's test was used to compare the statistical differences between AUCs.
In all tests, a 2-tailed probability value of 0.05 was considered statistically significant. healthy subjects. No difference in terms of sex and age was present but TM patients showed a significantly lower body mass index. Per inclusion criteria, healthy subjects had no CVRF and the frequency of diabetes was significantly increased in TM patients. LV end-diastolic and end-systolic volume indexes were significantly higher in TM patients than in healthy subjects while no significant difference was found in LV mass index, LVEF, and LVGFI.
Representative examples of LVGFI are shown in Fig. 1.
Significant MIO (global heart T2* <20ms) was found in 370 (27.4%) TM patients. Both LVGFI and LVEF were significantly lower in patients with significant MIO than in patients without significant MIO and in healthy subjects while no significant difference was detected between TM patients without significant MIO and healthy subjects (Fig. 2).

Demographic and clinical correlates of LVGFI in TM patients
Demographic, clinical, and CMR characteristics of TM patients are summarized in Table 1.
LVGFI was not correlated with age (R = 0.007; p = 0.810) but was significantly lower in males than in females (44.42 ± 7.50% vs. 47.62 ± 7.27%; p < 0.0001). No association was detected between gender and LVEF. Table 1 shows the comparison between TM patients and  The contrast medium was administrated in 1099 patients (81.3%) and replacement myocardial fibrosis was detected in 203 (20.7%) of them. Two patients showed an ischemic pattern while the remaining 201 patients had a non-ischemic pattern of myocardial fibrosis, involving the septum in the 85.5% of cases. Compared to LGE-negative patients, patients with replacement myocardial fibrosis had a significantly lower LVGFI (44.78 ± 8.72% vs. 46.72 ± 7.30%; p = 0.031) (Fig. 4), besides comparable global heart T2* values.

Predictors of the LVGFI
The independent predictors of the LVGFI were assessed by performing a stepwise regression analysis including all significant variables in univariate regression analysis with the LVGFI as the dependent variable (Table 2). Male sex, diabetes mellitus, significant MIO, and replacement myocardial fibrosis were the strongest predictors of the LVGFI (F = 44.27; p < 0.0001).

LVGFI and history of heart failure
Eighty-six patients (6.4%) patients had a history of heart failure.
At ROC curve analysis, a LVGFI ≤ 44.9% predicted the presence of a positive history of HF with a sensitivity of 67.4% and a specificity of 60.4% (p < 0.0001). The AUC was 0.67 (95% CIs = 0.64-0.69). At ROC curve analysis, a LVEF ≤ 58.0% predicted the presence of a positive history of HF with a sensitivity of 51.2% and a specificity of 60.3% (p = 0.001). The AUC was 0.62 (95% CIs = 0.59-0.64). The Delong's test showed a significant difference among the AUCs (p = 0.039) (Fig. 5B).
No association was detected between LVGFI and age at start of regular transfusions or chelation, splenectomy, and hemoglobin levels. LVGFI showed a weak inverse correlation with mean serum ferritin over the previous year (R=-0.150; p = 0.001).
Data about the presence of diabetes mellitus were available for 1292 patients, already diagnosed with diabetes or tested for blood glucose in the 6 months preceding the CMR scan. The prevalence of diabetes was 10.4%. LVGFI was significantly lower in patients with diabetes than in patients without diabetes (43.39 ± 8.79% vs. 46.42 ± 7.29%; p < 0.0001) while no significant difference was detected in terms of LVEF.
Global heart T2* values were significantly associated with both LVGFI (R = 0.266; p < 0.0001) and LVEF (R = 0.183; p < 0.0001), but the correlation with LVGFI was significantly stronger (p = 0.0001). The number of segments with T2*<20ms was more strongly associated (p = 0.003) with the LVGFI (R=-0.256; p < 0.0001) than with the LVEF (R=-0.199; p < 0.0001).  In TM patients, as in healthy subjects and in other patient populations, the male sex was associated with lower values of LVGFI, which can be explained by the different morphology of male and female hearts [22,40,41].
Besides male sex, diabetes, significant MIO, and replacement myocardial fibrosis were associated with worse LVGFI.
It should be recognized that global heart T2* values showed only a weak correlation with LV function markers, likely because, although iron could be removed by chelation treatment [42,43], the induced heart damage could be progressive and not totally reversible. Moreover, heart damage in thalassemia does not result only from iron overload, but other factors like nutritional deficiencies, genetic factors, diabetes, and other endocrinopathies can play a role [16,[44][45][46]. Anyway, LVGFI seems to be more closely related to cardiac iron burden and distribution than LVEF. Our findings can be explained by the fact that although the LVGFI is strongly related to LVEF, it carries additional data. In fact, it includes information on physiological adaptation as well as pathological remodelling by measures of both cavity size and myocardial mass [20]. Increased thickness of the ventricular wall is one of the first, and still reversible, cardiac alterations due to iron deposition in the myocardium [47]. The increase in LV wall thickness may be explained by the fact that iron deposition in the myocytes causes them to hypertrophy. Later, with increasing iron overload, left dilated cardiomyopathy develops and ventricular function becomes impaired [47,48]. In fact, significant LVEF changes appear later and are preceded by significant compensatory modifications in LV mass and volumes to preserve systolic function. In presence of an increased LV mass and increased relative wall thickness, the LVGFI decreases while the LVEF remains unchanged since it does not account for LV mass. Only in presence of an increase in the size of the ventricular cavity both functional indices are significantly decreased. However, the LVGFI is still more decreased than LVEF.
In agreement with previous studies, we confirmed the lack of correlation between myocardial fibrosis and heart iron [30,49]. Conversely, our findings suggest that myocardial fibrosis can have a negative impact on ventricular remodelling.
Our study clearly demonstrated for the first time an association between LVGFI and HF in TM. Importantly, the LVGFI was shown to provide a superior discriminatory ability compared with the LVEF, suggesting that the LVGFI is a useful functional parameter of the LV also in the thalassaemic population. Apart from the benefit of the inclusion of structural aspects of cardiac remodelling, another contributory factor to the improved performance of LVGFI may be its association with diabetes mellitus. In a

Discussion
This CMR multicenter study is the first that evaluated in TM patients the distribution, clinical correlates, and diagnostic capability of the LVGFI, a relatively new marker of cardiac performance.
We did not find a significant reduction of the mean LVGFI or LVEF in TM patients when compared with control subjects of identical age and sex distribution, while TM was associated with significantly higher LV volumes. Despite regular transfusion therapy, TM represents a chronically anemic condition, characterized by an increase in blood volumes (increased preload) and a decrease in systemic vascular resistance (decreased afterload) [35]. The anatomical-functional expression of this hemodynamic state is the enlargement of cardiac cavities and the increase of LVEF [36][37][38]. Myocardial iron overload is initially expressed as diastolic LV dysfunction but in end stage disease it may increase ventricular dimensions and decrease systolic function [11,39]. Indeed, when we categorized our patients based on the presence of significant MIO, both LVEF and LVGFI were reduced in TM patients with MIO versus patients without MIO and healthy subjects, while there was no difference between patients without MIO and healthy subjects. large retrospective historical TM cohort, diabetes was associated with a significantly higher risk of myocardial fibrosis, HF, and hyperkinetic arrhythmias, independently from MIO [44].

Limitations
A major limitation of this study is that it is a cross-sectional analysis. Prospective studies on large cohorts of TM patients are recommended to evaluate if LVGFI is a strong, independent, antecedent predictor of HF and if it can provide incremental prognostic value in comparison with LVEF.
We did not measure myocardial deformation (strain), which offers a more accurate and direct measure of myocardial function than EF [50]. Although feature tracking(FT) CMR allows quantification of myocardial deformation on routine SSFP cine images, the dedicated post-processing FT software packages were not available in the MIOT centers.

Conclusion
In TM patients the LVGFI is negatively affected by male sex, diabetes mellitus, myocardial iron overload, and myocardial fibrosis and provides incremental diagnostic value for the detection of HF compared with the LVEF. This simple and reliable LV functional index, integrating structural components of adverse cardiac remodelling in the assessment of LV cardiac performance, may potentially improve the risk stratification of TM patients.