Materials and methods

Study population

We studied retrospectively 115 patients with TM (40 men) (mean (SD) age 27 (8) years) consecutively examined at our MRI Laboratory. All patients had been regularly transfused since early childhood and started undergoing chelation therapy from the mid-to-late 1970s onwards, while patients born after the seventies had received chelation therapy from early childhood. Over the past year the mean (SD) serum ferritin level was 1842 (1621) ng/ml and the mean (SD) pre-transfusion haemoglobin was 9.85 (0.72) g/dl. Among the 89 patients with TM examined for the presence of anti-HCV (hepatitis C virus) antibodies, 63 (71%) were positive. All patients were <50 years old. All patients had a low (<10%) pre-test likelihood of coronary artery disease (CAD) considering the conventional cardiac risk factors as smoking, family history of coronary artery disease, hypertension, diabetes mellitus dyslipidaemia, obesity, heavy alcohol consumption, oral contraceptive use and menopausal age.23 Fifteen patients (13%) had a history of heart failure requiring medication and 12 patients (10%) had a history of documented arrhythmias requiring medication. Seven patients (6%) showed pulmonary hypertension (trans-tricuspid velocity jet >3.0 m/s). Table 1 summarises the characteristics of all the patients.

CMR scanning was performed within 1 week before regular scheduled blood transfusion. None of the patients were in decompensated heart failure at the time of scanning. A 12-lead ECG was performed within 1 month from CMR in the absence of clinical cardiovascular events and was read blindly by the consensus of two cardiologists (with 27 years and 12 years of experience) who were unaware of the results of the CMR examination results. All patients gave written informed consent to the protocol. The project was approved by the institutional ethics committee.

Cardiovascular magnetic resonance

CMR was performed using a 1.5 T MR scanner (GE Signa/Excite, Milwaukee, Wisconsin, USA). An eight-element cardiac phased-array receiver surface coil with breath holding in expiration and ECG gating was used for signal reception.

Steady-state free procession cines were acquired during 8 s breath holds in sequential 8 mm short-axis slices from the atrioventricular ring to the apex to assess biventricular function parameters quantitatively in a standard way,19 using MASS software (Medis, Leiden, The Netherlands). Segmental wall motion was visually assessed as 0, normal; 1, moderate hypokinesis; 2, severe hypokinesis; 3, akinesis; 4, dyskinesis.

For the measurement of myocardial iron overload, we used a multislice multiecho T2* approach, as previously described.16 17 18 Three parallel short-axis views of the left ventricle were obtained by T2* gradient-echo multiecho sequence. For signal analysis on T2* images, we used dedicated software (HIPPO MIOT IFC-CNR) to provide the T2* value on each of the 16 segments of the left ventricle, according to the standard AHA/ACC model,24 as well as the global T2* value averaged over all segmental T2* values18 and the T2* value in the mid-ventricular segment averaged over the mid-anterior septum and the mid-inferior septum. A T2* measurement >20 ms was taken as a “conservative” normal value for all 16 segments and for the global T2* value.

Contrast DE images were acquired in the same view as used for cine CMR from 10 to 15 min after the gadobutrol (1.0 mol/l) (0.2 mmol/kg) intravenous administration, using a fast gradient-echo inversion recovery sequence. In short-axis views we started with a basal slice from below the aortic outflow tract to just before the apical slices. Vertical, horizontal and oblique long-axis views were also acquired. Inversion times were adjusted to null for the normal myocardium (from 210 ms to 300 ms) with voxel size of 1.6×1.25×8.0 mm. The extent of DE was first evaluated visually using a two-point scale (enhancement absent or present). Enhancement was considered present whenever it was visualised in two different views. In addition, the extent of DE areas was quantified using semiautomatic, previously validated software.25 In each image, the boundaries of myocardium and DE areas were semiautomatically traced and manually corrected. A 17-segment model, according to the standard AHA/ACC model24 was adopted where 16 segments were derived from the short-axis images (same myocardium segmentation as used to analyse the T2* sequences) and the 17th segment, corresponding to the apex, was obtained from the long-axis view. The transmural extent of DE was defined as the extent of DE >75% in each segment through the ventricular wall. DE areas were expressed as a percentage of the entire left ventricle myocardium. The contrast-to-noise ratio was calculated by objective pixel quantification as the contrast between the hyperenhanced tissue and the normal myocardium divided by the noise intensity evaluated in the image background.

The cine and contrast enhanced images were evaluated blindly by the consensus of two skilled observers (with 8 years and 4 years of experience) who were unaware of the results of the other modality.

The mean scanning time was 40 min.

Statistical analysis

All data were analysed using SPSS version 13.0 statistical packages. All continuous variables are expressed as mean (SD). Correlation analysis was performed using the Spearman test. Comparisons between groups were made by independent-samples t test. Wilcoxon rank sum test was applied for continuous values with non-normal distribution (ie, T2* data and number of cardiac risk factors). A χ² test was performed for non-continuous variables, where appropriate. A two-tailed probability value of 0.05 was considered statistically significant.

Results

Contrast DE CMR

DE areas were detected in 28 patients (24%). The resulting contrast-to-noise ratio was 5.9 (5.7). The extent of DE areas was 3.9 (2.4)% of the total left myocardial mass. Among the 28 patients with areas of DE, 17 (61%) had two or more foci of fibrosis (fig 1) and 10 (36%) had fibrosis in the inferoseptal junction (fig 1). Of the 61 areas of fibrosis, 21 (34%) involved the interventricular septum.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Patchy foci of enhancement in one patient with thalassaemia major. Delayed enhancement involved the mid-lateral segment (A) and the inferoseptal junction (B).

In 26/28 patients (93%) the location of DE was epi-mesocardial, and not transmural or otherwise similar to the enhancement seen in CAD (fig 1). Of 26 patients, three patients showed hypokinetic segments corresponding to DE areas.

In two patients (7%) DE followed the coronary distribution (from subendocardium to subepicardium). One patient showed transmural DE in the apical region of the left ventricle with corresponding wall thinning and akinesia (fig 2). The other patient showed transmural DE in the apex, in the anterior and lateral distal segments with corresponding wall thinning and akinesia. The overall extent of DE in these two latter patients was significantly higher than in the other 21 patients, although statistical analysis was not performed owing to the small number of patients (9.5 (1.4)% vs 3.5 (1.8)%).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Patients with thalassaemia major with no myocardial iron overload (all 16 segments T2* values >20 ms) (A) and transmural delayed enhancement (black arrows) following coronary distribution in the apical region (B, C).

DE areas were detected more often at mid-ventricular level (54%) than at basal (26%) and apical level (18%); particularly, the mid-level was significantly more involved than the apical level (p<0.001). The mean (SD) number of DE segments per patient was 2.17 (1.33).

Contrast delayed enhanced versus myocardial iron overload and biventricular function parameters by CMR

Myocardial fibrosis or necrosis did not significantly affect T2* values. The global T2* value in patients with enhanced tissue was comparable to that of patients without enhanced tissue (the DE group 26 (15) ms vs the no-DE group 24 (12), p = 0.65). In addition, T2* values were comparable in segments with enhancement and segments without enhancement (the enhanced segments 32 (13) ms vs the no-enhanced segments 25 (14) ms, p = 0.24).

There were no significant differences between the DE group and the non-DE group in left and right ventricular ejection fraction (EF), left and right end-diastolic and end-systolic volume indexes, and in left ventricular mass index and cardiac index.

Contrast delayed enhanced: clinical correlates

The DE group was statistically significantly older than the non-DE group (DE group 31 (7.7) years vs the no-DE group 26 (7.7) years, p = 0.004). There was no significant difference between the DE group and the no-DE group in gender and Hb pre-transfusion levels. We detected a statistically significant correlation between the presence of fibrosis or necrosis and the number of cardiac risk factors (p = 0.006) as well as the presence of DE and history of cardiac complications (heart failure or documented arrhythmias requiring medication or pulmonary hypertension) (χ² = 10.2; p = 0.001). Moreover, the patients with DE and a history of cardiac complications showed an extent of DE areas (expressed as a percentage of the entire left ventricle myocardium) statistically significantly higher than that of the patients with DE and history free of cardiac complications (4.9 (2.7)% vs 3.0 (1.5)%, p = 0.004). We did not detect a statistically significant correlation between pulmonary hypertension and the presence of DE areas (p = NS). We found a statistically significant correlation between the presence of DE and the presence of anti-HCV antibodies (χ² = 3.9; p = 0.04).

Contrast DE and ECG changes

A significant correlation was found between the presence of myocardial fibrosis or necrosis and changes in ECG (χ² = 14.9; p<0.001). The most common findings in patients with enhanced tissue were T-wave inversion, flat T wave and right bundle branch block. The sensitivity, specificity, negative predictive value and positive predictive value of ECG in detecting myocardial fibrosis were respectively 79%, 66%, 90% and 42%.

Table 2 summarises the clinical and instrumental correlates in the DE and non-DE groups.

Discussion

Late gadolinium enhancement occurs in areas of expanded extracellular space owing to higher regional gadolinium concentration and slower distribution kinetics than in normal myocardium.20 With the possible exception of cardiac amyloidosis,26 DE has been shown to correspond to myocardial scarring.22 However, the expanded extracellular space in the myocardium in patients with thalassaemia might be caused by protein interstitial infiltration like ferritin or haemosiderin. We found comparable values of myocardial iron overload in segments with enhancement confronted with segments without enhancement which seems to exclude the possibility that expanded extracellular space in the myocardium of patients with thalassaemia major could be caused by protein infiltration like ferritin or haemosiderin. Thus, such regions of enhancement in thalassaemia cardiomyopathy seem to represent increased myocardial scarring known to occur in this condition.3 4 5 9 10 11 On the other hand, although it was postulated that T2* measurements in the heart could be susceptible to fibrosis in addition to tissue iron concentration,27 our findings further validate “heart T2*” as the equivalent of “heart iron”, pointing out that myocardial fibrosis does not affect T2* values.

In this study we observed that DE areas are a common finding (24%) among patients with TM.

Most of our study population showed a non-specific patchy, epi-mesocardial, non-transmural pattern of fibrosis (fig 1). Two patients showed transmural necrosis with corresponding wall thinning, akinesia (fig 2) and an overall extent of DE significantly higher than that seen in patients with non-specific pattern. Both patients had non-significant myocardial iron overloading (fig 2). In both patients, subsequent to CMR, the myocardial perfusion stress imaging with radionuclides showed fixed defects corresponding to the DE areas and non-reversible defects; the coronary angiography showed unobstructed coronary arteries. Thus, CMR had a key role in the clinical management of these patients, differentiating heart failure resulting from CAD and heart failure resulting from myocardial iron overload.

Studies of idiopathic or secondary non-thalassaemic haemochromatosis9 10 11 have shown absent or mild myocardial fibrosis using myocardial biopsy “in vivo”. Although a myocardial iron burden is common to these study groups, our thalassaemia population differed in age sex, and physiopathological mechanisms. On the other hand, in comparison with previous autoptic studies which reported marked and extensive myocardial fibrosis in patients with thalassaemia,3 4 5 our patients are a unique population belonging to an unselected cohort in a completely different transfusional and chelation era and they are likely to be representative of the majority of patients with TM, today.

We did not detect significant differences between the fibrosis group and the non-fibrosis group in the biventricular function parameters; in our study group the small range of EFs (from normal to moderately reduced) and the low prevalence of patients with a significant reduction of EF probably explain these findings.

Several explanations might account for the lack of correlation between myocardial fibrosis and myocardial iron overload in our study. First, genetically determined variables could affect susceptibility to heart failure in the presence of iron burden.28 Second, in our study the thalassaemia population were compliant and had been well treated for a long period and had a mild degree of myocardial iron overload. Third, although iron could be removed by intensive chelation treatment, the induced heart damage could be irreversible and progressive. Histological studies in the pre-chelation era postulated that simple storage of iron in cardiac muscle does not, itself, cause replacement fibrosis.4 In this respect, Sanyal et al have shown the absence of fibrosis in the heart of one patient in the presence of extensive iron deposition.9 In addition, heart damage in thalassaemia can also result from myocarditis7 or pulmonary hypertension.8

We found no significant correlation between pulmonary hypertension and fibrosis. However, the low prevalence of patients with significant pulmonary hypertension in our study group prevents exclusion of its role in the pathogenesis of fibrosis in patients with TM.

In our study patients had no history of myocarditis. The significant correlation between the presence of DE and the presence of anti-HCV antibodies may simply suggest that the HCV virus is a causal agent in the pathogenesis of fibrosis in patients with TM; HCV infection has recently been noted in patients with cardiomyopathies and myocarditis29 and our patients with TM showed a high prevalence of anti-HCV antibodies. Only immunohistology and PCR analysis of myocardial specimens could really answer this question. The use of antiviral therapy against hepatitis C virus could be reinforced in patients with TM, if specific studies confirmed that the HCV infection contributes to the development of myocardial fibrosis.

The DE group was significantly older, confirming the heart damage in thalassaemia and particularly the presence of scarring as a time-dependent process. In TM traditional cardiovascular risk factors seem to be implicated in the pathogenesis of myocardial scarring, considering the correlation found in our cohort of patients with TM between myocardial scarring and cardiovascular risk factors, particularly given the lack of obstructive disease. Thus to prevent myocardial fibrosis, tight control of cardiovascular risk factors is recommended for patients with TM where treatment strategies usually focus on chelation therapy.

Although the aetiopathogenesis of myocardial fibrosis remains unclear in patients with TM, this finding seems to have a clinical impact considering the correlation with a history of cardiac complications (heart failure, pulmonary hypertension or arrhythmias) and the increased DE areas in patients with cardiac complications.

Correlation analysis disclosed positive significant values between the presence of DE and ECG changes (p<0.001). ECG was abnormal in a high proportion of the patients with TM with myocardial fibrosis/necrosis, demonstrating a high sensitivity. The high negative predictive value of ECG seems to exclude with good accuracy the presence of fibrosis/necrosis in a patient with a normal ECG. Thus, ECG changes could be a good marker to predict myocardial fibrosis. ECG, owing to its low cost, could be used as a guide for performing a DE CMR examination, particularly in countries where there is an high prevalence of patients with thalassaemia and the availability of CMR is poor.

Limitations

To date, there has been no histological correlation of DE in patients with thalassaemia. We have planned prospective histological evaluations in our study group where it will be possible to correlate DE with histological specimens. Current CMR techniques are unlikely to detect diffuse microscopic fibrosis and hence a consistent number of patients with thalassaemia will show absent enhancement.30 Clinical and MRI follow-up are being carried out on this cohort of patients in order to obtain a better understanding of the implications of the presence of myocardial scarring in the clinical management of patients with TM. Unfortunately, immunohistology and PCR analysis of myocardial specimens were not available to link HCV infection as a causal agent to the pathogenesis of myocardial scarring in patients with TM.

Conclusion

This represents the first study that has documented myocardial scarring non-invasively “in vivo” in patients with thalassaemia by CMR. By this technique it is possible to discriminate in the same examination heart failure resulting from iron overload and heart failure resulting from other causes (eg, CAD). DE areas were detectable in a significant percentage of patients. The presence of myocardial fibrosis/necrosis seems to be a time-dependent process and correlates with cardiovascular risk factors, a history of cardiac complications and anti-HCV antibodies. Conversely, no significant association was found between the presence of DE areas and myocardial iron overloading. ECG changes showed a significant accuracy in predicting myocardial DE. Further studies are needed to determine the prognostic and therapeutic implications of DE in thalassaemic patients.