British Journal of Radiology (2004) 77, 104-110
© 2004 British Institute of Radiology
doi: 10.1259/bjr/53300812
Ultrashort echo time (UTE) MRI of the spine in thalassaemia
M A Hall-Craggs, FRCR
1
J Porter, FRCP
2
P D Gatehouse, PhD
3 and
G M Bydder, FRCR
4
Departments of 1 Imaging and 2 Haematology, University College London Hospitals NHS Trust, University College Hospital, Grafton Way, London W1CE 6DB, 3 The Cardiac Magnetic Resonance Unit, Royal Brompton Hospital, Sydney Street, London SW3 6NP, UK and 4 Department of Radiology, University of California San Diego, 200 West Arbor Drive, San Diego, CA 92103-8756, USA
Correspondence: Professor G M Bydder
 |
Abstract
|
|---|
Back pain is common in adult patients with homozygous thalassaemia, and degenerative disc disease is increasingly recognised as a cause. Ultrashort echo time (UTE) pulse sequences, which are sensitive to the presence of short T2 relaxation components in tissue produced by iron deposition and other processes, were used to examine the lower thoracic and lumbar spine in symptomatic patients with 
-thalassaemia major or intermedia. Three patients were studied with fat suppressed as well as both fat suppressed and long T2 suppressed UTE (TE=0.08 ms) pulse sequences. Conventional 2D Fourier transformation T1 and T2 weighted scans were also performed for comparison. Normal controls showed narrow high signal areas in the region of the end-plate and annulus fibrosus. Patients showed hyperintense bands adjacent to the vertebral end plate in lower thoracic and lumbar spine discs using a UTE sequence with both long T2 component and fat suppression. The extent of the changes was most marked in the patient with the most severe degenerative change. In the patient with minimal disease, findings of this type were present in discs which did not show evidence of degeneration with conventional MR imaging. High signal changes of a type previously not described were observed in each patient. The effect may be due to organic iron entering the disc and decreasing its T1 and T2, but susceptibility effects from iron in the vertebral bodies, fibrosis and other causes also need to be considered.
|
Introduction
|
|---|
Back pain is a common symptom in adult patients with homozygous thalassaemia. The causes include osteoporotic compression fractures, extramedullary haemapoiesis with expansion of bone marrow and pressure on cortical bone, as well as spinal cord compression, scoliosis and desferrioamine (DFO) induced bone dysplasia [1–7]. Much less attention has been directed to disease of intervertebral discs, but Hartkamp et al [8] have observed a tendency to disc narrowing in thalassaemia on plain films, and Ho et al [9] have described increased severity of degenerative disc disease in thalassaemia compared with age matched controls. In their study the distribution of disc disease in thalassaemic patients was atypical, with equal involvement of the lower thoracic and lumbar intervertebral discs, as compared with a predilection for degenerative change in the lowest two lumbar discs in the control population. The cause of this increased severity and unusual distribution is not known.
We have implemented a pulse sequence with an ultrashort echo time (UTE) (0.08 ms) [10, 11]. With this sequence it is possible to detect short T2 relaxation components in tissues before they decay to a level where they are not observable with conventional spin echo pulse sequences. This sequence has been useful for demonstrating increased signal from short T2 components in conditions such as angiomas, haemorrhage, haemachromotosis of the liver and chronic fibrosis [11]. In this paper we report our experience with this technique in imaging the lower thoracic and lumbar spine in three patients with thalassaemia.
|
Subjects and methods
|
|---|
All studies were approved by the Institutional Review Board. The basic pulse sequence employed a half radiofrequency (rf) excitation followed by radial imaging of k-space from the centre out. This was followed by another half rf excitation with the gradient reversed and repeated radial mapping. The two sets of data were added to give a single line of k-space and the process was repeated through 360° in 512 steps. The data were mapped onto a 512 x 512 grid and reconstructed by 2D Fourier transformation (2DFT) to give a gradient echo like image. Four sets of images with echo times (TEs) of 0.08 ms, 5.95 ms, 11.08 ms and 17.70 ms were produced with a repetition time (TR) of 500 ms. In addition to the basic sequence, a version with frequency based fat suppression was employed. Long T2 suppression sequences were also used in which a rectangular 90° pulse was employed before the half rf excitation pulses to selectively excite tissues or fluids with a long T2 after which the signal was dephased by use of a gradient. This was used in combination with fat suppression. The abbreviations used to describe the sequences are listed in Table 1
. Fields of view (FOVs) of 30–40 cm were employed on a Sonata 1.5 T MR system (Siemens, Erlangen, Germany) although the system was designed for small FOV cardiac and brain imaging. Slice thickness was 4–6 mm. 8–12 multiple sagittal slices were obtained simultaneously with a flip angle (for long T2 components) of 80° and a slice gap of 10%. Conventional T1 weighted spin echo (TR/TE=500/8 ms) and T2 weighted fast spin echo (TR/TEeff=2500/91 ms) scans were also performed with the same FOV and slice thickness. Studies were conducted using a two element phased array spine coil.
Three normal male controls with no significant history of back pain (aged 29, 33 and 58 years) were studied. Scans from two patients with severe degenerative disc disease (aged 54 and 71 years, respectively, both female) who were examined with the same sequences were reviewed. Three patients with thalassaemia were studied. Their clinical details are included below.
|
Results
|
|---|
A fat and long T2 suppressed UTE (FLUTE) image from the 33-year-old normal control is shown in Figure 1. The disc signal is low except for faint higher signal lines in the region of the annulus fibrosus and end plate. High signal is seen posteriorly both superiorly and inferiorly related to the spine array coils. There is also some spatial distortion and other artefacts due to the limited region of uniform gradient performance on the MR system. A similar pattern of low disc signal with a faint increase in the region of the annulus fibrosus was seen in the other two normal controls.
View larger version (114K):
[in this window]
[in a new window]
|
Figure 1. Normal spine in a 33-year-old volunteer. Sagittal fat and long T2 suppressed ultrashort echo time (FLUTE) (repetition time/echo time=500/0.08 ms) image. The lumbar and lower thoracic discs have a faint high signal in the region of the annulus fibrosus and low signal in the nucleus pulposus. (High signal is seen posteriorly in the lower thoracic/upper lumbar region and in the sacral region adjacent to the spine coil. Artefacts associated with the small useful field of view are seen in the pelvic region and superiorly).
|
|
A FLUTE image from the 71-year-old female patient with severe degenerative disc disease is shown in Figure 2. There is loss of disc height and alignment within the spine. The central disc signal is low with only small areas of increased signal from the annulus fibrosus and other sites. Artefacts due to the spine coil were noted again. Changes of the same type were seen in the other patient with degenerative disease.
View larger version (108K):
[in this window]
[in a new window]
|
Figure 2. Advanced degenerative disc disease in a 71-year-old woman. Fat and long T2 suppressed ultrashort echo time (FLUTE) (repetition time/echo time=500/0.08 ms) image. There is considerable loss of alignment and disc height. Signal is seen in the annulus fibrosus at T12/L1 disc but elsewhere there are only small irregular areas of higher signal.
|
|
Case 1
A male patient of Cypriot origin aged 29 years with -thalassaemia major has received monthly blood transfusions since the age of 6 months. The current transfusion regimen maintains a mean haemoglobin (Hb) of 12 g dl-1 with a minimum pre-transfusion Hb of 9.5 g dl-1. He receives iron chelation therapy with DFO currently at a dose of 3 G five times per week over 12 h. DFO was commenced at the age of 4 years. Back pain, particularly in the lower lumbar area has been a problem for many years and appears worse before blood transfusion. His current serum ferritin is 1500 µg l-1 and his estimated liver iron (derived from MR imaging T2* measurement) is 1.3 mg g-1 dry wt. Dual-energy X-ray absorptiometry (DXA) shows no osteoporosis and he has not required hormone replacement treatment for hypogonadism.
His conventional T1 and T2 weighted MR scans (Figure 3a, b) showed reduced signal in the bone marrow and some expansion of the vertebral bodies with loss of disc height at all levels. The T1 weighted scan (Figure 3a) shows relatively high signal areas in some discs. The T2 weighted scan (Figure 3b) show some areas of high signal centrally and at the disc margins in an abnormal pattern. The fat suppressed UTE (FUTE) image (Figure 3c) showed intermediate signal from bone marrow with slightly higher signal from most discs except for a low signal (probably vacuum phenomenon) in the L4/L5 disc and to a lesser extent centrally and anteriorly in other discs (e.g. L2/L3, L1/L2 and T12/L1).
The FLUTE image (Figure 3d
) shows obvious high signal bands parallel to the end-plates with some reduction of signal centrally in the regions corresponding to the high signal areas in Figure 3b. The relatively high signal in the discs on the T1 weighted image (Figure 3a) paralleled the higher signal seen on the T2 weighted image (Figure 3b). This is probably due to an increase in long T2 components and may have been due to inflammatory or reactive changes. Suppression of the long T2 components with the FLUTE sequence allows features specifically attributable to short T2 components to be identified in Figure 3d. The reduction in signal in the central disc region on the FLUTE sequence at some levels may be due to concurrent increase in T2 components (which are then suppressed) and reduction in short T2 components.
Case 2
A female patient aged 33 years with -thalassaemia major has received monthly blood transfusions since the age of 1 year. The current transfusion regimen maintains a mean Hb of 12 g dl-1 with a minimum pre-transfusion Hb of 9.5 g dl-1 with two units four times per week. She has chronic hepatitis C which as not been eradicated with either interferon alone or interferon plus rabvirin. DFO was commenced at the age of 5 years and compliance was poor until the age of 16 years. At this age she developed iron induced cardiomyopathy which was successfully treated with continuous intravenous DFO. She currently receives continuous iron chelation therapy with DFO at a dose of 36 mg kg-1 day-2. Back pain, particularly in the lower lumbar area has been a problem for the last 4 years. Pain in the left knee has also been an intermittent feature. Her current serum ferritin level is 1500 µg l-1 with an estimated liver iron (derived from MR imaging T2*) of 1.3 mg g-1 dry wt. DXA scan shows osteoporosis in both lumbar spine and femur and she receives hormonal replacement for hypogonadism. She has received calitriol for persistent hypocalcaemia for 4 years. She has required insulin since 1990 for diabetes mellitus. Conventional T2 weighted (TR/TEeff=2500/91 ms) (Figure 4a) and FLUTE (TR/TE=500/0.08 ms) (Figure 4b) images are shown. The bone marrow signal is moderately reduced in (Figure 4a) where there is loss of disc signal except for an increase centrally at L2/L3 with reactive changes in the adjacent vertebral body. The FLUTE image (Figure 4b) shows high signal in the T10/T11 disc with successively less marked changes in T11/T12 and T12/L1 discs. Low signal is seen centrally at L2/L3 disc and other disc levels.
View larger version (117K):
[in this window]
[in a new window]
|
Figure 4. Case 2: Female aged 33 years with -thalassaemia major. (a) T2 weighted (repetition time (TR)/echo time (TE)eff=2500/91 ms) and (b) fat and long T2 suppressed ultrashort echo time (FLUTE) (TR/TE=500/0.08 ms) images. The marrow signal is moderately reduced in (a) where there is loss of disc signal except at L2/L3 which shows high signal centrally with adjacent reactive changes. The FLUTE image (b) shows high signal in the T10/T11 disc with less marked change in T11/T12 and T12/L1. Low signal is seen centrally in the discs at the lower levels.
|
|
Case 3
A 43-year-old female with late presenting (at age 40 years) thalassaemia intermedia due to interaction of heterozygosity of IVS-110 (G-A) and a triplicated alpha genotype has never received a blood transfusion and has not had a splenectomy though she has marked splenic enlargement (12 cm). Her steady state Hb lies between 7.6 g dl-1 and 9.1 g dl-1. There is evidence of mild hepatic iron accumulation (estimated liver iron of 3.2 mg g-1 dry wt by T2* MR imaging) and a raised ferritin of 400 µg l-1. DXA scan shows only mild osteopenia. She presented with lumbar back pain and conventional MR imaging showed masses of extramedullary haemopoiesis at T9/10 and at S1/2 (a 3–4 cm presacral mass). Her conventional T2 weighted scan (Figure 5a) showed low signal in the bone marrow with normal disc height and disc signal. The FLUTE scan showed high signal bands adjacent to the end plates at T11/T12 and T12/L1 where no degenerative changes were apparent on Figure 5a.
View larger version (119K):
[in this window]
[in a new window]
|
Figure 5. Case 3: Female aged 43 years with thalassaemia intermedia. (a) T2 weighted (repetition time (TR)/echo time (TE)eff=2500/91 ms) and (b) fat and long T2 suppressed ultrashort TE (FLUTE) (TR/TE=500/0.08 ms images. The T2 weighted images show a decreased marrow signal and loss of height in the body of T12 but normal disc signal at all levels. The FLUTE scan showed increased signal intensity bands at T11/T12 and T12/L1.
|
|
|
Discussion
|
|---|
All three patients showed high signal intensity bands parallel to the vertebral end plates in discs in the lower thoracic and upper lumbar spine. This was most obvious in the patient with the most severe degenerative disease (case 1) but was also present in the patient without MR signs of degenerative disc disease (case 3). The distribution corresponded with that previously described for degenerative disc disease in thalassaemia [9]. The parallel bands of high signal are very unusual and to our knowledge have not previously been described in MR studies of disc disease. The nearest equivalent is the appearance produced by gadolinium chelates 4–8 h after intravenous administration. The transport of this agent from the vertebral end-plate into the disc results in an increase in signal parallel to the end-plates [12].
The cause of the high signal bands is not known but there are several possibilities. Increased iron deposition in the intervertebral discs could shorten the T1 and T2* of this tissue. With conventional sequences the decrease in T2* could mean that the signal was low or undetectable. However with UTE sequences the signal may be detected before it has significantly decayed and the concurrent shortening of T1 could result in high signal relative to surrounding tissues particularly when the signal from fat and long T2 components was reduced (as with the FLUTE sequence). The distribution within discs would be consistent with diffusion of iron into the disc through the vertebral end plates.
Another possibility is increased fibrous connective tissue. In advanced degenerative disease this may be apparent in the disc. However the changes seen in the patients with thalassaemia were more marked than in those seen in the patients with degenerative disease. Also the third case had high signal bands present without degenerative changes.
Another possibility is susceptibility effects extending into the disc due to iron deposition in the adjacent vertebral bodies. These effects are usually localized around interfaces and decrease rapidly with distance. They typically result in a loss of signal rather than the increased signal seen with the FLUTE sequences. In general UTE sequences are resistant to susceptibility effects because of their very short TE.
Calcification could have produced a susceptibility effect, but the changes were much greater than those usually seen with calcification. Another possibility is a vacuum phenomenon producing susceptibility changes. This type of change is usually seen centrally in the disc in advanced degeneration. In addition high signal changes are seen around the cleft [13]. The low signal areas seen in Figure 3c are consistent with a vacuum phenomenon but the high signal seen in Figure 3d was not concentrated around these. High signal was present in the other two cases without evidence of the vacuum phenomenon.
Excessive iron has a toxic effect in many tissues and it is possible that it may be instrumental in causing premature disc disease in patients with thalassaemia. The degenerative process may be accelerated and be seen concurrently with iron deposition. The long T2 components seen on conventional scans may represent inflammatory or reactive changes within discs associated with this. Concurrent disease processes of this type which increase T2 may mask the effects of others that produce a short T2.
It would be of considerable interest to examine disc tissue histologically and ascertain its iron content as well as any evidence of fibrosis or inflammatory change. There do not appear to be reports in the literature of this having been done. Future studies in thalassaemic patients might investigate how the observed changes in the disc parallel the clinical history, markers of iron overload and treatment. It is possible that in other diseases which produce increased iron deposition in tissues such as haemochromatosis may show changes in discs. It is also possible that similar effects may be detectable in other related connective tissues such as tendons, ligaments, menisci and articular cartilage.
UTE pulse sequences may prove to have wider application in musculoskeletal disease. With conventional 2DFT imaging there is a delay after the initial rf excitation when the slice selection gradient is used to rephase the signal. Time is also required for the phase encoding pulses, the initial dephasing lobe of the frequency encoding pulse and the first half of the data acquisition before the centre of k-space is reached.
It is possible to avoid the need for rephasing of the slice selective rf excitation pulse by first collecting the data with the slice selection gradient in one direction and adding this to data collected in the same way with the slice selection gradient reversed. At the end of this process the signal is effectively in-phase and data sampling can, in principle, begin as soon as the rf pulse and the slice selection gradient are ramped down to zero. The use of radial imaging of k-space with the acquisition starting in the centre of k-space (where no gradient is required for the initial encoding) means that there is no need for a phase-encoding gradient, a read-dephasing gradient or additional time to get back to the centre of k-space in the read direction. Data sampling can continue while the gradient is being ramped up (although sampling during ramping takes longer than when the gradient is fully ramped up by a factor of about two) as well as after the gradient has reached its plateau. The gradient also needs to be ramped down very quickly at end of the rf excitation since persistence of the gradient after the end of the rf pulse may result in dephasing of the signal.
The pulse sequence detects the free induction decay (FID) directly. There is no echo since the signal is not refocused and each half excitation is not fully rephased. It is only after they are added that the k-space data is in phase.
While simple UTE sequences are effective for imaging tissues with a majority of short T2 components, some form of long T2 component reduction is necessary to selectively image short T2 components in tissues in which they are a minority. The initial approach to this problem was to use a long, e.g. 10 ms, rectangular 90° pulse followed by a dephasing gradient. A second method is to subtract a later echo image from the first (UTE) one and produce a difference image. Tissues or fluids with a long T2 have their signal attenuated by this procedure, while tissues which have a short T2 and decay rapidly between the two echoes are highlighted on the resulting difference image [14].
This approach is of particular value in imaging tissues such as tendons, ligaments and menisci which have a majority of short T2 relaxation components but it is also of value for detecting signal from short T2 components in tissues in which they are a minority.
Received for publication June 9, 2003.
Revision received September 15, 2003.
Accepted for publication November 6, 2003.
|
References
|
|---|
- Angastiniotis M, Pavlides N, Aristidou K, et al. Bone pain in thalassaemia: assessment of DEXA and MRI findings. J Pediatr Endocrinol Metab 1998;11:779–84.
- Papavasiliou C, Gouliamos A, et al. CT and MRI of symptomatic spinal involvement by extramedullary haemopoiesis. Clin Radiol 1990;42:91–2.[CrossRef][Medline]
- Ayding
z U, Oto A, Cila A. Spinal cord compression due to epidural extramedullary haematopoiesis in thalassaemia: MRI. Neuroradiology 1997;39:870–2.[CrossRef][Medline]
- Papanastasiou DA, Baikousis A, Sdougos G, et al. Correlative analysis of the sagittal profile of the spine in patients with -thalassaemia and in healthy persons. J Spinal Disord 2000;13:113–7.[CrossRef][Medline]
- Brill PW, Winchester P, Giardina PJ, et al. Deferoxamine-induced bone dysplasia in patients with thalassaemia major. AJR Am J Roentgenol 1991;156:561–5.[Abstract/Free Full Text]
- Naselli A, Vignolo M, Di Battista E, et al. Long term follow up of skeletal dysplasia in thalassaemia major. J Pediatr Endocrinol Metab 1998;11:817–25.
- Levin TL, Sheth S, Berdon WE, et al. Deferoxamine-induced platyspondyly in hypertransfused thalassemic patients. Pediatr Radiol 1995;25:122–4.
- Hartkamp MJ, Babyn PS, Oliveri F. Spinal deformities in deferoxamine-treated homozygous beta-thalassaemia major patients. Pediatr Radiol 1993;23:525–8.[CrossRef][Medline]
- Ho C, Hall-Craggs M, Porter J, Desai S, Renfrew I. Degenerative disc disease as a cause of back pain in the thalassaemic population: a case-control study using MRI and plain radiology. Eur J Radiol (in press).
- Bergin CJ, Pauly JM, Macovski A. Lung parenchyma: projection reconstruction imaging. Radiology 1991;179:777–81.[Abstract/Free Full Text]
- Gatehouse PD, Bydder GM. Magnetic resonance imaging of short T2 components in tissues. Clin Radiol 2003;58:1–19.[CrossRef][Medline]
- Bydder GM. New approaches to magnetic resonance imaging of intervertebral discs, tendons, ligament and menisci. Spine 2002;27:1264–8.[CrossRef][Medline]
- Kim YS, Suh K, Chang Y, Lee JJ, Lee S, Kang D. Abnormal high signal intensity in the intervertebral disc with vacuum phenomenon on T1 weighted MR imaging: is it caused by susceptibility artefact? Radiology 1998;209(P):314.
- Robson MD, Gatehouse PD, Bydder M, Bydder GM. Magnetic resonance: an introduction to ultrashort echo-time imaging. J Comput Assist Tomogr 2003;27:825–46.[CrossRef][Medline]
This article has been cited by other articles:
|
|
|
|
 
M. T. Modic and J. S. Ross
Lumbar Degenerative Disk Disease
Radiology,
October 1, 2007;
245(1):
43 - 61.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
