Left ventricular noncompaction in patients with β-thalassemia: Uncovering a previously unrecognized abnormality


  • Conflict of interest: Nothing to report


Left ventricular noncompaction (LVNC) is a rare cardiomyopathy with potentially serious outcomes. It results in multiple and excessive trabeculations, deep intertrabecular recesses, and a thickened ventricular myocardium with two distinct layers, compacted and noncompacted. The condition is most commonly congenital; however, acquired forms have also been described. A recent report of LVNC detected in a β-thalassemia twin suggested an association with cardiac siderosis. In a cross-sectional study of 135 transfusion-dependent patients with β-thalassemia (130 major and 5 intermedia, mean age 29.6 ± 7.7 years, 49.6% males) presenting for cardiac iron assessment by magnetic resonance imaging (MRI), we evaluated the prevalence and risk factors for LVNC. None of the patients had neuromuscular or congenital heart disease. Eighteen patients (13.3%; 95% confidence interval [CI] = 8.6–20.1) fulfilled the preassigned strict criteria for LVNC on cardiac MRI. There were no statistically significant differences between patients with and without LVNC with respect to demographics; hemoglobin levels; splenectomy status; systemic, hepatic, and cardiac iron overload indices; hepatic disease and infection studies; or iron chelator type. Patients with LVNC were more likely to have heart failure (adjusted odds ratio = 1.77; 95% CI = 0.29–10.89); although with high uncertainty. Patients with β-thalassemia have a higher prevalence of LVNC than normal individuals. As this finding could not be explained by conventional risk factors in this patient population, further investigation of the underlying mechanisms of LVNC is warranted. This remains crucial for an entity with adverse cardiac outcomes, especially in patients with β-thalassemia where cardiac disease remains a primary cause of mortality. Am. J. Hematol., 2012. © 2012 Wiley Periodicals, Inc.


Left ventricular noncompaction (LVNC), once also known as hypertrabeculation or spongy myocardium, is a rare congenital disorder that is commonly attributed to intrauterine arrest of normal compaction during the endomyocardial morphogenesis. It results in multiple and excessive trabeculations, deep intertrabecular recesses, and a thickened ventricular myocardium with two distinct layers, compacted and noncompacted [1]. Although initially described and recognized as a rare entity, more recent reports highlight higher prevalence that could be attributed to better understanding of the disease and improvement in imaging technology [2]. Several diagnostic criteria for LVNC have been placed relying on echocardiographic [3–6], pathoanatomic [7], or cardiac magnetic resonance imaging (MRI) [8, 9] findings. Although the genetic origins of this mostly familial disease are still under investigation, LVNC is predominantly associated with autosomal dominant patterns of inheritance [10]. X-linked inheritance and de novo mutations have been described [11, 12]. LVNC has been linked to a number of congenital cardiac and sometimes noncardiac disorders [13], particularly neuromuscular disease [14]. Most frequently, LVNC is associated with mitochondrial disorders [15], Barth syndrome [16], and hypertrophic cardiomyopathy due to mutations in myosin heavy-chain 7 (MYH7), α-cardiac actin (ACTC), troponin T Type 2 (TnnT2), myosin binding protein C (MYBPC3) [17], or the 1p36 deletion syndrome [18], and so forth [11]. Nonetheless, acquired cases have also been reported in adults with no obvious congenital disease, and the term isolated LVNC was used to describe such cases [19]. Its prevalence is <0.14% in adults referred for echocardiography [20]; however, figures may vary depending on the diagnostic criteria used and due to similarity of its features to other diseases of the myocardium and endocardium [21]. Some reports have also shown that it can occur as a transient phenomenon during myocarditis [22, 23]. Few reports described cases of isolated LVNC in hematological disorders: one case of spherocytosis [24] and another with essential thrombocythemia [25]. Recently, a first report of LVNC in identical twins with β-thalassemia major (TM) and cardiac iron overload has been published [26]. In this line, this study aims to explore the prevalence and risk factors for LVNC in patients with β-thalassemia presenting for cardiac iron assessment by MRI.


This was a cross-sectional study of all consecutive patients with β-thalassemia presenting to our center for cardiac iron assessment by MRI in the period between September 12, 2007 and September 24, 2008. According to our practice guidelines, patients eligible for cardiac iron assessment by MRI include all transfusion-dependent patients with TM as well as regularly transfused patients with β-thalassemia intermedia (TI) who have received at least four red blood cell transfusions during the year prior to cardiac MRI assessment. Cardiac MRI examinations were carried out only in patients with sinus rhythm. The study received approval from the institutional review board, and all patients signed a written informed consent. Patients' records were reviewed using Webthal®, a computerized clinical record for patients with thalassemia attending Italian centers, for data on demographics (age and sex), splenectomy status, iron chelator use, history of hepatitis C virus (HCV) infection, and history of neuromuscular disease. Patients' body mass index (BMI) and body surface area were also measured.

Cardiac assessment

For each patient, cardiac assessment included a full patient history inquiring for cardiovascular symptoms, history of cardiac thromboembolic events, history of congenital heart disease, and family history of cardiovascular disease. Patients with evidence of heart failure were also classified according to the New York Heart Association Functional Classification (NYHA) [27]. A full physical examination was also performed. All patients also underwent electrocardiography (ECG), and most of the patients had Holter monitoring to evaluate rhythm abnormalities. Patients with neuromuscular diseases or congenital heart diseases were excluded from the study.

All cardiac MRI studies were performed on Philips Achieva 1.5-Tesla MR systems (Philips Healthcare, Best, The Netherlands), equipped with a five-element SENSE cardiac coil using the same scanning protocol. All sequences were ECG gated. All images were acquired in end-expiratory breath old. After the acquisition of scout images, fast imaging with steady-state free precession loops were acquired in two-chamber, short-axis, and four-chamber views, and parallel slices in short-axis planes with zero interslice gap to cover the entire ventricle (retrospective gating; slice thickness = 8 mm; echo time [TE] = 1.67 msec; repetition time [TR] = 3.3 msec; flip angle = 60°). When LVNC was suspected, black-blood T2-weighted images were acquired in two-chamber, four-chamber, and in short-axis views. A diagnosis of LVNC was established according to the presence of all the following criteria:

  • Segmental left ventricular wall thickening with a thin compacted epicardial layer and a thicker noncompacted endocardial layer.

  • Measurement of end-diastolic thickness of noncompact and of compact layers with noncompact/compact ratio ≥2.5 [8, 9]. Measurements of thickness were never done at the apical segment (Cerqueira segment 17) [28].

  • End-diastolic thickness of compact layer in site of noncompact ≤ 4.5 mm.

  • Hyperintensity of signal in black-blood T2-weighted images in the site of hypertrabeculation. This finding is highly sensitive but has low specificity because it could also be present in other pathologies.

Left ventricular volumes and systolic function were determined evaluating all short-axis images with dedicated software (Philips Viewforum) for planimetry of endocardial borders at end-diastole and end-systole [29]. Left ventricular volume values were also indexed to body surface area [30].

Patients were also evaluated for cardiac siderosis with myocardial T2* by using single short-axis, mid-ventricular slices acquired at eight TEs (2, 3, 4, 6, 9, 12, 15, and 18 msec). A gradient-echo sequence was used (flip angle = 20°; pixel dimension = 2.1 mm; repetition time = 21 msec). Images at different TEs were coregistered to correct for variations in end-expiratory cardiac position. Signal decay curves were measured using a full-thickness region of interest in the interventricular septum. The trend line was fitted to a monoexponential decay model, with an equation of the following form:

equation image

where S = fitted signal; S0 = initial amplitude, and T2* = relaxation constant [31, 32].

Laboratory and hepatic measurements

All patients subjected to cardiac MRI evaluation also had laboratory measurements for total hemoglobin (Hb), serum ferritin (automated immunofluorescence assay; B.R.A.H.A.M.S. Ferritin KRYPTOR), and alanine aminotransferase levels. All studies were performed at the same laboratory. The liver iron concentration (LIC) was also measured noninvasively by a Superconducting Quantum Interference Device biomagnetic susceptometer (Model 5700 Tristan Technology, San Diego, CA) [33]. Liver stiffness was evaluated by transient elastography (FibroScan®; Echosens, Paris, France) [34].

Statistical analysis

Data are presented as means ± standard deviation or percentages. Bivariate comparisons between patients with evidence of LVNC and those without were made using the Student's t-test for continuous variables and the χ2 and Fisher's exact tests for categorical variables. Multivariate logistic regression was used to evaluate the independent effect of LVNC diagnosis on cardiac outcomes, while adjusting for clinically relevant confounders. All P-values are two sided with the level of significance set at 0.05.


A total of 135 patients (130 TM and 5 TI) were included in this analysis. The mean age of patients was 29.6 ± 7.7 years (range, 10.7–48.2 years) with an equal sex distribution (49.6% males). Patients' characteristics are summarized in Table I. None of the patients had congenital heart disease or neuromuscular disorders. Moreover, none of the patients had ever received fetal Hb induction therapy. A considerable proportion of patients had heart T2* values signifying higher risk of subsequent cardiomyopathy (<20 msec, n = 41 [30.4%] and <10 msec, n = 7 [5.2%]). None of the patients had a history of cardiac thromboembolism. A total of 12 (8.9%) patients had established heart failure and 35 (25.9%) patients had evidence of rhythm abnormalities on ECG (Table I).

Table I. Patients' Characteristics (n = 135)
  • PCR-RNA, polymerase chain reaction-ribonucleic acid; NYHA, New York Heart Association Functional Classification; ECG, electrocardiography; LVEF, left ventricular ejection fraction; LVEDV, left ventricular end-diastolic volume; BSA, body surface area; LIC, liver iron concentration; ALT, alanine aminotransferase; SD, standard deviation.

  • a

    Data were missing for four patients.

  • b

    The detected abnormalities included: left anterior fascicular block (n = 4); negative T waves (n = 8); Wolff-Parkinson-White syndrome or pattern (n = 4); early repolarization (n = 4); long QT syndrome or pattern (n = 3); pathological Q waves (n = 2); left ventricular hypertrophy pattern (n = 3); atrial premature contractions (n = 3); ventricular premature contractions (n = 1); left bundle branch block (n = 1); right bundle branch block (n = 5; 2 complete and 3 incomplete); S1, S2, and S3 syndrome or pattern (n = 2).

Mean age ± SD (range) (years)29.6 ± 7.7 (10.7–48.2)
Male:female, n (%)67 (49.6):68 (50.4)
Mean body mass index ± SD (range) (kg/m2)21.4 ± 3.0 (13.9–34.2)
Mean body surface area ± SD (range) (m2)1.6 ± 0.2 (0.5–2.0)
Splenectomized, n (%)95 (70.4)
Iron chelation therapy, n (%)132 (97.8)
 Deferoxamine35 (25.9)
 Deferiprone31 (23.0)
 Deferoxamine/deferiprone combination40 (29.6)
 Deferasirox26 (19.3)
Duration on current chelator ± SD (range) (years)7.0 ± 8.8 (<1–37.9)
Hepatitis C virus PCR-RNA positive, n (%)32/131a (24.4)
Neuromuscular disease, n (%)0 (0.0)
Cardiac assessment
 Family history of cardiovascular disease0 (0.0)
 Congenital heart disease0 (0.0)
 Heart failure, n (%)12 (8.9)
  NYHA I10 (7.4)
  NYHA II2 (1.5)
 History of cardiac thromboembolic events, n (%)0 (0.0)
 ECG abnormality, n (%)35b (25.9)
 Mean LVEF ± SD (range) (%)59.7 ± 6.5 (37.0–65.0)
 Mean LVEDV ± SD (range) (mL)134.7 ± 38.0 (72.9–341.5)
 Mean LVEDV indexed to BSA ± SD (range) (mL)86.9 ± 31.3 (56.4–358.9)
 Mean heart T2* ± SD (range) (msec)30.9 ± 14.4 (5.8–61.0)
Laboratory and hepatic measurements
 Mean total hemoglobin ± SD (range) (g/dL)10.9 ± 0.8 (8.0–12.5)
 Mean serum ferritin ± SD (range) (ng/mL)1640.7 ± 1775.7 (30.0–12396.0)
 Mean LIC ± SD (range) (μg Fe/g wet weight)1175.0 ± 967.5 (0.0–5136.0)
 Mean ALT ± SD (range) (IU/L)48.9 ± 63.3(6.0–494.0)
 Mean liver stiffness ± SD (range) (kPa)7.0 ± 6.5 (2.5–66.4)

On cardiac MRI assessment, 18 patients (13.3%; 95% confidence interval [CI] = 8.6–20.1) fulfilled the criteria for LVNC (Table II). Five of the 18 patients with LVNC also had evidence of a right ventricular hypertrabeculation pattern. Figure 1 displays cardiac MRI characteristics of a patient with LVNC. None of the patients with LVNC were the outcome of a consanguineous marriage. Two patients with LVNC also had a sibling evaluated in the study without evidence of LVNC. For another patient, both parents were also scanned in this study. The mother was found to have a hypertrabeculated pattern of left ventricular myocardium; however, it did not fulfill all criteria for the diagnosis of LVNC.

Figure 1.

Cardiac MRI of a 31-year-old female patient with β-thalassemia major and evidence of LVNC. (A1–A3) Steady-state free precession (SSFP) diastolic images. In these images, there is an evident hypertrabeculation. SSFP images were used to measure C and NC and NC/C ratio. (B1–B3) Fast-spin-echo (FSE)-T2-weighted diastolic images. In T2-weighted images within NC layer, there is a hyperintense signal due to very slow motion of blood. (A1 and B1) Two-chamber view; (A2 and B2) four-chamber view; and (A3 and B3) short-axis view.

Table II. Characteristics of Patients with Left Ventricular Noncompaction (LVNC)
No.Age (years)SexLVEF (%)LVEDV (mL)LVEDV/BSA (mL)NC (mm)C (mm)NC/CNYHAECG alterations
  1. M, male; F, female; LVEF, left ventricular ejection fraction; LVEDV, left ventricular end-diastolic volume; LVEDV/BSA, left ventricular ejection fraction indexed to body surface area; NC, end-diastolic thickness of noncompact layer; C, end-diastolic thickness of compact layer; NYHA, New York Heart Association Functional Classification; ECG, electrocardiography; LAFB, left anterior fascicular block; T−, negative T wave (leads where the pattern is present); APC, atrial premature contractions; RBBB I, right bundle branch block incomplete; SD, standard deviation.

831.1F52101.078.213053543.68ILAFB + T− (II, III, aVF, V4–6)
1730.6M57167.885.613503643.70RBBB I
N (%)11 M (61.1)2 (11.1)3/18 (16.7)

Risk factors for LVNC

Patients with LVNC and left ventricular compaction (LVC) were comparable in age, sex distribution, and BMI. Moreover, they had similar mean total Hb levels and proportions of patients with history of splenectomy and HCV infection. Although patients with LVNC were less commonly iron chelated, both patient groups had similar mean levels for markers of systemic, cardiac, and hepatic siderosis (Table III).

Table III. Comparison of Risk Factors Between the Left Ventricular Noncompaction (LVNC) and the Left Ventricular Compaction (LVC) Groups
ParameterLVNC group (n = 18)LVC group (n = 117)P-value
  1. PCR-RNA, polymerase chain reaction-ribonucleic acid; LIC, liver iron concentration; ALT, alanine aminotransferase; SD, standard deviation.

Mean age ± SD (years)29.4 ± 5.329.6 ± 8.10.904
Male, n (%)11 (61.1)56 (47.9)0.323
Mean body mass index ± SD (kg/m2)20.6 ± 2.221.5 ± 3.10.238
Splenectomized, n (%)12 (66.7)83 (70.9)0.783
Iron chelation therapy, n (%)16 (88.9)116 (99.1)0.047
 Deferoxamine6 (33.3)29 (24.8)0.563
 Deferiprone5 (27.8)35 (29.9)1.000
 Deferoxamine/deferiprone combination4 (22.2)27 (23.1)1.000
 Deferasirox1 (5.6)25 (21.4)0.196
Duration on current chelator ± SD (years)9.3 ± 9.66.7 ± 8.60.264
Hepatitis C virus PCR-RNA positive, n (%)5 (27.8)27 (23.9)0.770
Mean heart T2* ± SD (msec)32.0 ± 14.730.7 ± 14.50.734
Mean total hemoglobin ± SD (g/dL)11.0 ± 1.010.9 ± 0.80.785
Mean serum ferritin ± SD (ng/mL)1255.1 ± 1658.31700.6 ± 1792.60.324
Mean LIC ± SD (μg Fe/g wet weight)1068.2 ± 961.81191.5 ± 971.50.617
Mean ALT ± SD (IU/L)43.4 ± 36.949.7 ± 66.60.698
Mean liver stiffness ± SD (kPa)5.9 ± 1.57.2 ± 6.90.468

LVNC and cardiac outcomes

Two patients with LVNC had evidence of heart failure, and three had minor ECG abnormalities. When compared with patients with LVC, patients with LVNC were more likely to have heart failure (odds ratio [OR] = 1.34; 95% CI = 0.27–6.67); although with high uncertainty. After adjusting for age, sex, cardiac T2* MRI, LIC, and total Hb level, the observed effect was larger but with greater uncertainty (OR = 1.77; 95% CI = 0.29–10.89). Nonetheless, patients with LVNC were less likely to have rhythm abnormalities on ECG when compared with patients with LVC (OR = 0.53; 95% CI = 0.14–1.96). After similar adjustment, results remained essentially unchanged (OR = 0.57; 95% CI = 0.14–2.40).


This is the first report to document that isolated LVNC is a common finding in patients with β-thalassemia. The observed prevalence of 13.3% is considerably higher than most reports from the nonthalassemic literature (<1%) [19]. However, the possibility that our findings could be attributed to a higher genetic background of LVNC in the Italian population when compared with previous reports in the general literature cannot be excluded. LVNC was associated with higher occurrence of heart disease but not rhythm abnormalities, although with uncertainty that could be attributed to the sample size. In previous reports from the nonthalassemic population, isolated LVNC was associated with significant morbidity, including end-stage heart failure, cerebrovascular events due to cardiac emboli, and ventricular arrhythmias; however, the incidence of these outcomes varies widely between different studies [19]. The rates of mortality and heart transplantation have ranged from as high as 40% [5] to as low as 3% [35]. Additionally, the use of different diagnostic criteria to identify LVNC further illustrates the interpretation of outcomes [36].

It should be noted that our finding of a high prevalence of LVNC and the potential mechanisms underlying this abnormality are only relevant to a similar group of regularly transfused and iron-chelated patients with thalassemia. Among the evaluated thalassemia (and its management)-related parameters, we could not identify any risk factors for this finding; however, this should be confirmed in larger longitudinal studies. It may be postulated that ineffective erythropoiesis and chronic anemia/hemolysis may lead to chronic oxidative and inflammatory stress triggering myocardial remodeling and transformation from compact musculature to the spongy myocardium, thus leading to an acquired form of LVNC. In one reported case of a 23-year-old male with hereditary spherocytosis, LVNC was detected on presentation with muscular weakness due to skeletal myopathy and symptoms of heart failure according to NYHA functional class II. The authors failed to identify any apparent risk factors, and the patient had no evidence of virus persistence, myocarditis, or other specific cardiomyopathy on endomyocardial biopsy [24]. The only common feature between the described patient and our cohort is the chronic exposure to anemia. Nonetheless, in this study, we could not identify a difference in age or total Hb level in patients who had evidence of LVNC and those who did not. Nutritional disturbances in thalassemia such as carnitine, selenium, and thiamine deficiencies might also be potential risk factors for cardiovascular disease [37]; however, we were unable to evaluate this in our study.

In the only previous report describing LVNC in an identical twin with TM, the authors speculated about a potential effect of cardiac siderosis in the increased thickness of the myocardium [26]. In our patients, there was no significant association between LVNC and cardiac T2* MRI values. However, the absence of evidence does not necessarily mean evidence of absence. Accumulation of toxic iron species within myocytes (evidence on cardiac T2* MRI) may not be necessary to induce cardiac abnormalities, and initial exposure to toxic iron species like nontransferrin-bound iron may be enough to cause damage to cardiac tissue. This would imply that even without MRI evidence of cardiac siderosis in such patients, they may still be at risk of iron-related damage [38]. Thus, the effect of iron overload on the occurrence of LVNC cannot be fully excluded.

The question remains whether β-thalassemia inheritance may be associated with alterations in embryonic cardiac development, that is, the LVNC described in this report is a congenital rather than an acquired disorder. No association between thalassemia syndromes and cardiac anomalies has ever been described. However, one report revealed that overexpression of rtEa4 or hEb propeptides of the proinsulin-like growth factor-I in zebrafish embryos disrupts both heart development and red cell production by reducing the levels of GATA 5 and NKX 2.5 (heart development) as well as GATA 1 and GATA 2 (red cell production) mRNA [39]. Whether similar genetic variants exist in people with thalassemia and whether they could explain the observation of LVNC merit further study. However, the lack of male predominance supports an acquired rather than primary genetic etiology to our observation as male predominance is well documented in genetic LVNC [19]. Moreover, as developmentally β-thalassemia does not manifest until after embryogenesis, the occurrence of LVNC in β-thalassemia is likely to be a postnatal rather than a congenital manifestation.

A false diagnosis of normal left ventricular trabeculations as LVNC in our study is very unlikely, as we applied strict diagnostic criteria. However, we cannot exclude some overestimation of the prevalence. Recent data indicate that by using the standard criterion of noncompact/compact ratio (≥2.3), a prevalence of LVNC up to 6% may be found on cardiac MRI in normal subjects [40]. We used a threshold ratio of ≥2.5, which slightly protects from this potential bias, and in any case, the observed prevalence in this report is considerably larger than that reported in normal subjects. Nonetheless, the limitation of our study is the cross-sectional nature of data collection. LVNC may have occurred in these patients years before the diagnosis was made, and the exact chronological relationship between the occurrence of LVNC and the evaluated risk factors cannot be appropriately determined. Moreover, we have evaluated a sample of patients who presented for cardiac MRI assessment. Although cardiac iron measurement is now done routinely for all patients with TM [41], some patients may have been selected for measurement based on preexisting cardiac risk factors. If such factors are determined to be linked to LVNC occurrence, this could explain the high prevalence of this finding in our study. The results of a longitudinal follow-up of our series, which is in progress, will provide further information in this regard.

The practical interest of our findings stems from the fact that cardiac disease is still the main cause of death in patients with TM, even in those optimally treated [42]. The pathogenesis of heart disease in TM has been primarily attributed to iron overload and the secondary chronic oxidative damage, as well as chronic anemia and viral myocarditis [43]. However, not all patients acquiring these factors develop heart disease, and the opposite remains true, which leaves other undisclosed factors to be determined [38]. Our study serves as an eye-opener for a novel finding in patients with β-thalassemia. Longer follow-up studies are needed to clarify the long-term prognosis in this patient population and to better understand the mechanisms underlying this cardiac abnormality.