Objective
Friedreich ataxia (FRDA) is an autosomal recessive condition due to a GAA triplet expansion in the FXN gene that causes increased left ventricular (LV) wall thickness and can progress to LV systolic dysfunction. However, the changes in myocardial function that occur before a reduction in LV ejection fraction are incompletely understood.
Methods
LV long-axis function was assessed by measurement of tissue Doppler imaging (TDI) peak systolic (S`), early diastolic (E`), and atrial velocities (A`) at the septal and lateral borders of the mitral annulus in 60 subjects homozygous for a GAA expansion in the FXN gene who had preserved LV ejection fraction. Comparison was made with 60 sex- and age-matched controls. TDI velocities at 5 years were compared with baseline values in 17 FRDA subjects with follow-up studies who still had preserved ejection fraction.
Results
S` and E` were reduced in FRDA subjects at both the septal and the lateral mitral annular borders. Lateral E` was independently and inversely related to age, blood pressure, septal wall thickness, and the number of GAA repeats in the smaller allele of the FXN gene, whereas septal E` was not correlated with GAA repeat number. At 5 years, there was a reduction in lateral S` and E` but no change in septal TDI velocities.
Conclusion
Subjects with FRDA have impairment of septal and lateral long-axis LV function, but there also seem to be regional differences in the effects of this condition that are at least partly related to the degree of genetic abnormality.
The clinical features of Friedreich ataxia (FRDA) not only include a progressive neurodegenerative disease and an increased incidence of diabetes but also a high incidence of cardiac abnormalities. The most common echocardiographic finding in FRDA is increased thickness of the left ventricular (LV) wall, but subjects with FRDA are also susceptible to the development of a dilated cardiomyopathy. The clinical presentation of heart disease in FRDA can be with arrhythmias, symptoms of heart failure, and even sudden death. Although cardiac disease is probably the most common cause of death in FRDA, the timing of development and rapidity of progression of the cardiac disease in FRDA are highly variable.
FRDA is inherited in an autosomal recessive manner, with ∼98% of mutant alleles in FRDA containing an unstable expansion of a GAA trinucleotide in intron 1 of the FXN gene, which encodes for the protein frataxin. The number of GAA repeats is associated with earlier symptom onset and increasing cardiac disease severity. Most studies have reported a more important role for the size of GAA repeats in the smaller allele (GAA1), although some studies have also suggested a role for GAA repeats in the larger allele (GAA2). A biological basis for the importance of GAA1 is provided by the finding that GAA1 is inversely correlated with the residual frataxin level in lymphoblasts from FRDA subjects. The deficiency of mitochondrial frataxin results in an abnormality of mitochondrial iron transport and increased sensitivity to oxidative damage.
Important questions still remain about the effects of FRDA and GAA repeats on LV structure and function, particularly in the early stages of cardiac involvement when therapies are likely to have the greatest potential to influence the course of the disease. Long-axis LV function is of particular interest in FRDA because abnormalities of long-axis function have been shown to be an early change in a number of conditions, including myotonic dystrophy and hypertrophic cardiomyopathy, even before the development of LV hypertrophy. Long-axis function is of additional importance because it can be simply and reliably measured noninvasively with the widely available echocardiographic technique of pulsed-wave tissue Doppler imaging (TDI). However, the effects of FRDA on long-axis LV function are unclear because findings from previous studies have been inconsistent.
The present study, performed in a cohort of FRDA subjects with preserved LV ejection fraction (EF) and free of cardiac symptoms and diabetes, had the following aims: (1) to determine if there are early abnormalities of long-axis LV function in FRDA, using peak systolic and diastolic velocities measured by pulsed-wave TDI at the septal and lateral borders of the mitral annulus; (2) to determine the relation of TDI mitral annular velocities in FRDA with GAA1, GAA2, and the extent of LV structural change; and (3) to investigate whether medium-term changes in TDI velocities occur in FRDA in the absence of a deterioration in LV EF.
Materials and Methods
Study Subjects
The study group comprised 60 adult subjects with FRDA due to homozygosity for GAA triplet repeat expansions in intron 1 of the FXN gene who were prospectively recruited from a multidisciplinary FRDA clinic and who did not meet the exclusion criteria. FRDA subjects were excluded if they had diabetes because of previous evidence that TDI mitral annular velocities are reduced in diabetes. FRDA subjects were also excluded if they had a history of heart failure or hypertension, a cardiac rhythm other than sinus, a blood pressure (BP) >140/90 mm Hg at the time of the echocardiographic examination, or an EF ≤ 50%. Sixty age- and sex-matched healthy individuals with no known cardiac disease and with normal BP provided a control group.
From the FRDA cohort, individuals were identified who had a follow-up echocardiographic study performed on the same echocardiographic platform >4 years and <6 years after their baseline study and who had not developed an impairment of EF (<50%) by the time of the follow-up study. Of the 60 subjects with FRDA, 3 died within 5 years (2 after developing a cardiomyopathy with reduced EF and 1 suddenly). One subject was found to have a deterioration in LV EF, whereas 26 were known to be alive at 5 years but either did not attend for any follow-up echocardiography ( n = 5) or did not have a follow-up echocardiogram available in the range of 4–6 years after their baseline study ( n = 21). Of the remaining 30 subjects, 13 remained asymptomatic and were not known to have deterioration in LV ejection but had not reached 4 years from the time of their baseline echocardiogram. This left 17 subjects who had echocardiograms performed 60 ± 4 months after the baseline echocardiogram. The study protocol was approved by the Southern Health Human Research and Ethics Committee and all subjects gave informed consent.
Echocardiography
Transthoracic echocardiography was performed using a Sonos 5500 ultrasound machine (Philips, Amsterdam, The Netherlands) and measurements were performed off-line using Xcelera V1.2 L4 SP2 (Philips). Nearly all studies were performed by the one echocardiographer (LD). M-mode images of the left ventricle were obtained in the parasternal long-axis view just distal to the mitral valve leaflet tips after alignment of the cursor perpendicular to the LV wall. Two-dimensional images were used to facilitate identification of the endocardium and standard M-mode measurements of LV septal wall thickness (SWT), posterior wall thickness (PWT) and LV end-diastolic diameter (LVEDD) were obtained. LV mass was calculated using the modified formula of Devereux et al. and divided by body surface area (BSA) to calculate LV mass index (LVMI). Relative wall thickness (RWT) was calculated as 2 times the PWT divided by the LVEDD. Cutoffs for normal values of SWT and LVMI were determined separately for male and female subjects and defined by the mean value in the control group ± 2 SD.
Four- and two-chamber two-dimensional loops of LV contraction were recorded and used for measurement of LV end-diastolic (LVEDV) and end-systolic volumes and the calculation of EF using the biplane method of discs. Left atrial volume was calculated using the single plane area length method from the apical four-chamber view and indexed to BSA (left atrial volume index).
LV inflow velocities were recorded using pulsed-wave Doppler in the apical four-chamber view with the sample volume at the level of the mitral leaflet tips. Peak velocity (E) and deceleration time (DT) of early diastolic filling and peak velocity (A) of the atrial phase of diastolic filling were measured. Isovolumic relaxation time was measured from a continuous-wave Doppler signal with the cursor directed between the aortic and mitral valves to show the aortic valve closure and LV inflow signals.
TDI
Pulsed-wave TDI was performed in the apical four-chamber view as previously described. TDI velocities of longitudinal mitral annular motion were recorded at end-expiration apnea at both the septal and lateral mitral annular borders after optimizing parallel alignment of the ultrasound beam. Spectral pulsed-wave Doppler was used with instrument settings adjusted to record the high-frequency/low-velocity tissue signals using an appropriate sample volume size. Peak systolic velocity (S`) and early (E`) and late (A`) diastolic velocities were measured off-line, and five to six measurements were averaged. Interobserver variability for measurements of TDI velocities in our laboratory have been reported and vary between 5.5% and 7.7%. Satisfactory tissue Doppler signals were acquired in all subjects.
Analysis of GAA Expansion
The number of GAA repeats in the FXN gene was measured as previously described.
Statistics
Data were analyzed using standard statistical software (SYSTAT version 12; Systat Software, Chicago, IL). Continuous variables are presented as mean ± SD. Group differences in continuous variables between FRDA subjects and controls were assessed with an unpaired Student t test. Changes in continuous echocardiographic variables over time were assessed with a paired Student t test. Multivariate linear regression analyses of TDI velocities were performed in the combined FRDA and control groups, with FRDA and controls included in the analysis as dummy variables to determine if the effect of FRDA was independent of clinical variables that differed between the two groups.
Linear regression analysis was performed in FRDA subjects to determine correlations of LV structure (SWT, LVEDD, LVMI, and LVEDV) with age, disease duration, BP, and BSA, and of TDI velocities with LV structure, variables found to be related to TDI velocities in previous studies (age, heart rate [HR], BSA, and BP), and disease-specific markers (age at disease onset, disease duration, GAA1, and GAA2). Adjustment for age was performed for those variables known or suspected to be affected by age (BSA, BP, E`, and A`). Because men have a higher SWT and LVMI than women, sex was included as a dummy variable to see if it altered models including SWT or LVMI, but this inclusion had no significant effect on any of the models. The independent nature of various relationships was determined by multivariate analyses that included variables with P values < .15 on univariate analyses. The partial correlation coefficient (β) value is provided in those multivariate analyses where an aim was to determine the degree of contribution of various independent variables to the dependent variable. The adjusted value for r 2 was used to estimate the degree of variability in a dependent variable explained by a multivariate model. Apart from decisions regarding inclusion of variables in multivariate models, a P value of < .05 was considered significant.
Results
Subject Characteristics
Characteristics of the FRDA subjects and controls are shown in Table 1 . The groups were well matched for age, sex, height, weight, BSA, BMI, and BP, but HR was higher in FRDA subjects.
| FRDA ( n = 60) | Controls ( n = 60) | P | |
|---|---|---|---|
| Age (y) | 31 ± 9 | 30 ± 7 | .67 |
| Male/female | 27/33 | 27/33 | – |
| Height (cm) | 171 ± 9 | 171 ± 8 | .57 |
| Weight (kg) | 69 ± 13 | 68 ± 11 | .90 |
| BSA (m 2 ) | 1.80 ± 0.19 | 1.80 ± 0.17 | .62 |
| BMI (kg/m 2 ) | 23.6 ± 4.4 | 23.2 ± 2.7 | .91 |
| HR (beats/min) | 75 ± 11 | 63 ± 11 | <.001 |
| Systolic BP (mm Hg) | 110 ± 12 | 110 ± 11 | .99 |
| Diastolic BP (mm Hg) | 69 ± 9 | 66 ± 8 | .25 |
| GAA1 | 670 ± 178 | ||
| GAA2 | 916 ± 152 | ||
| Age at disease onset (y) | 15 ± 6 | ||
| Disease duration (y) | 16 (1–39) |
LV and Left Atrial Dimensions
There were no FRDA subjects with significant asymmetry between LV SWT and PWT. In comparison with controls, both SWT and PWT were higher in FRDA subjects ( Table 2 ). Because LVEDD in FRDA was also lower than controls, there was a proportionally greater increase in RWT than LVMI in FRDA subjects. An example of increased RWT in a subject with FRDA is shown in Figure 1 A . FRDA subjects had a lower LVEDV but similar EF to controls. In FRDA subjects, male subjects had a higher LVEDD ( P < . 02), SWT ( P < . 05), LVEDV ( P < . 001), and LVMI ( P = .006) than female subjects, but there was no difference in RWT between male and female subjects. By using a cutoff for increased RWT of > 0.42 and sex-specific cutoffs of > 2 SD above control values for SWT and LVMI, 45% of FRDA subjects had increased SWT, 68% had increased RWT, and only 22% had increased LVMI. Left atrial volume index in FRDA was similar to controls.
| FRDA ( n = 60) | Controls ( n = 60) | P | |
|---|---|---|---|
| LVEDD (cm) | 4.56 ± 0.55 | 4.95 ± 0.42 | <.001 |
| SWT (cm) | 1.10 ± 0.24 | 0.83 ± 0.15 | <.001 |
| PWT (cm) | 1.10 ± 0.20 | 0.83 ± 0.14 | <.001 |
| RWT | 0.49 ± 0.13 | 0.34 ± 0.05 | <.001 |
| Fractional shortening (%) | 41 ± 8 | 37 ± 6 | <.001 |
| LVMI (g/m 2 ) | 99 ± 29 | 78 ± 20 | <.001 |
| LVEDV (mL) | 79 ± 23 | 95 ± 24 | <.001 |
| EF (%) | 65 ± 6 | 64 ± 5 | .70 |
| Left atrial volume index (mL/m 2 ) | 26 ± 8 | 28 ± 7 | .31 |
Given the symmetry of LV thickening in FRDA subjects, only SWT has been used as the measure of LV wall thickness in the univariate and multivariate analyses. On univariate analysis of the FRDA group, neither age nor symptom duration was related to LVEDD or SWT, but both were inversely correlated with LVMI ( r = −0.27, P < . 05 and r = −0.30, P < . 02, respectively). LVEDV was inversely correlated with symptom duration ( r = −0.31, P = .015) but not with age. There was no relation of BSA or BP with SWT and no relation of BP with LVEDD, SWT, LVMI, or LVEDV. As expected, LVEDD and LVEDV were both positively correlated with BSA ( r = 0.52, P < . 001 and r = 0.39, P = .002, respectively).
Standard Doppler Measurements
The comparison of standard Doppler data in FRDA subjects and controls is shown in Table 3 . The peak E velocity was lower, and the isovolumic relaxation time was higher in FRDA subjects than controls, but the peak A-wave velocity, the E/A ratio, and the DT were similar.
| FRDA ( n = 60) | Controls ( n = 60) | P | |
|---|---|---|---|
| Transmitral E velocity (cm/s) | 76 ± 14 | 89 ± 20 | .002 |
| Transmitral A velocity (cm/s) | 46 ± 13 | 48 ± 14 | .38 |
| E/A ratio | 1.8 ± 0.7 | 1.9 ± 0.5 | .26 |
| DT (ms) | 176 ± 31 | 176 ± 23 | .96 |
| IVRT (ms) | 76 ± 10 | 70 ± 8 | <.001 |
Tissue Doppler Velocities
Examples of TDI septal and lateral mitral annular velocities in a subject with FRDA are shown in Figure 1 B and C. Compared with controls, S`, E`, and A` were all lower in FRDA subjects at both the septal and lateral borders of the mitral annulus ( Figure 2 A and B). Of the TDI velocities, E` showed the most pronounced difference between FRDA and control subjects, with the lateral E` 34% lower and the septal E` 27% lower in FRDA, whereas the lateral S` and septal S` were 20% and 16% lower, respectively. Even after adjusting for HR in multivariate analyses, FRDA remained a significant predictor of lower S`, E`, and A` at both annular borders ( P < . 001 for all).
In the FRDA group, there was a weak positive correlation between EF and septal S` ( r = 0.26, P < . 05), but no relation of EF with lateral S` ( P = .33). Univariate correlations of septal and lateral S`, E`, and A` velocities with age, symptom duration, HR, BP, LVEDD, SWT, LVMI, and LVEDV in FRDA subjects are presented in Table 4 . Septal S` was not significantly related to any of the variables, but there was a borderline significant inverse correlation with SWT and LVMI. Lateral S` was positively correlated with LVEDD and LVEDV and inversely correlated with SWT and symptom duration. Both LVEDV (β = 0.35, P = .003) and SWT (β = −0.35, P = .003) were independent predictors of lateral S`, together explaining 26% of the variability.
| Septal S` | Lateral S` | Septal E` | Lateral E` | Septal A` | Lateral A` | |
|---|---|---|---|---|---|---|
| Age | 0.19 † | −0.20 † | −0.30 ∗ | −0.51 ‡ | 0.49 ‡ | 0.49 ‡ |
| Symptom duration | 0.03 | −0.25 ∗ | −0.29 ∗ | −0.50 ‡ | 0.31 ∗ | 0.28 ∗ |
| BSA | 0.12 | 0.02 | −0.12 | −0.15 | 0.23 † | 0.36 ∗ |
| HR | 0.11 | 0.08 | –0.03 | 0.12 | 0.20 † | 0.05 |
| Systolic BP | 0.03 | −0.13 | −0.38 ∗ | −0.28 ∗ | 0.22 † | 0.28 ∗ |
| Diastolic BP | 0.14 | −0.08 | −0.27 ∗ | −0.33 ∗ | 0.29 ∗ | 0.31 ∗ |
| LVEDD | 0.09 | 0.27 ∗ | 0.16 | 0.35 ∗ | 0.07 | 0.24 † |
| SWT | −0.25 † | −0.40 ‡ | −0.37 ∗ | −0.33 ∗ | −0.27 ∗ | −0.30 ∗ |
| LVMI | −0.24 † | −0.23 † | −0.25 † | −0.07 | −0.29 ∗ | −0.25 † |
| LVEDV | 0.11 | 0.39 ‡ | 0.18 | 0.41 ‡ | 0.02 | 0.20 |
Both septal and lateral E` were inversely correlated with age, symptom duration, systolic BP, diastolic BP, and SWT. Lateral E`, but not septal E`, was positively correlated with LVEDD and LVEDV. On multivariate analysis, a model of septal E` that included age (β = −0.27, P = .02), systolic BP (β = −0.28, P = .017), and SWT (β = −0.40, P < . 001) explained 30% of the variability in septal E`. A model of lateral E` that included age (β = −0.51, P < . 001), LVEDV (β = 0.27, P = .007), and SWT (β = −0.37, P < . 001) explained 48% of the variability in lateral E`.
Septal and lateral A` were both positively correlated with age, symptom duration, systolic BP, and diastolic BP, and inversely correlated with SWT. Septal A` was inversely correlated with LVMI, and the relation between lateral A` and LVMI was borderline significant. On multivariate analysis, septal A` was independently and positively correlated with age (β = 0.50, P < . 001), HR (β = 0.32, P = .004), and SWT (β = −0.25, P = .02), these variables explaining 34% of the variability in septal A`, whereas lateral A` was only independently related to age.
Relation of GAA Repeat Length with LV Structure and Function
The relationships of GAA1 and GAA2 with age at symptom onset, HR, LVEDD, SWT, LVMI, LVEDV, EF, septal S`, and lateral S` were assessed by linear regression analysis. GAA2 was not related to any of the variables at the P < . 15 level (results not shown). The univariate relations of GAA1 with those variables are shown in Table 5 . GAA1 was inversely correlated with the age at symptom onset but not related to any of the other variables. The relationships of GAA1 and GAA2 with variables also correlated with age in this cohort (BSA, BP, septal and lateral E` and A`) were assessed in multivariate analyses after adjusting for age. GAA2 was not correlated with any of the variables (results not shown). Age-adjusted standard correlation coefficients of relationships of GAA1 with variables known to be affected by age are shown in Table 6 . Unexpectedly, GAA1 was inversely correlated with systolic BP and diastolic BP, and the possible confounding effects of these associations were therefore considered by including BP in further multivariate analyses. There was no correlation evident between GAA1 and septal E` or A` after adjusting for age. Lateral A` was inversely correlated with GAA1 after adjustment for age ( P < . 04), but was no longer significant after adjustment for BSA, SWT, or LVEDD. There was a borderline significant inverse correlation between lateral E` and GAA1 after adjusting for age, which became significant ( P < . 02) after including either systolic or diastolic BP in the model. GAA1 remained a significant predictor of lateral E` even after the addition of either SWT or LVMI to this model ( P < . 05), but was no longer a significant independent predictor of lateral E` with the inclusion of both SWT and LVEDD.
